korg/linux
0
0
Fork 0
mirror of https://mirrors.bfsu.edu.cn/git/linux.git synced 2024-11-26 05:34:13 +08:00
Code Issues Packages Projects Releases Wiki Activity
a0df926d33
linux/fs/btrfs/scrub.c

4407 lines
114 KiB
C
Raw Normal View History

btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
* Copyright (C) 2011, 2012 STRATO. All rights reserved.
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public
* License v2 as published by the Free Software Foundation.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*/
#include <linux/blkdev.h>
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
#include <linux/ratelimit.h>
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
#include "ctree.h"
#include "volumes.h"
#include "disk-io.h"
#include "ordered-data.h"
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
#include "transaction.h"
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
#include "backref.h"
btrfs: integrating raid-repair and scrub-fixup-nodatasum This ties nodatasum fixup in scrub together with raid repair patches. While both series are working fine alone, scrub will report uncorrectable errors if they occur in a nodatasum extent *and* the page is in the page cache. Previously, we would have triggered readpage to find good data and do the repair. However, readpage wouldn't read anything in the case where the page is up to date in the cache. So, we simply take that good data we have and call repair_io_failure directly (unless the page in the cache is dirty). Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-08-05 00:11:04 +08:00
#include "extent_io.h"
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
#include "dev-replace.h"
Btrfs: integrate integrity check module into btrfs This is the last part of the patch series. It modifies the btrfs code to use the integrity check module if configured to do so with the define BTRFS_FS_CHECK_INTEGRITY. If this define is not set, the only effective change is that code is added that handles the mount option to activate the integrity check. If the mount option is set and the define BTRFS_FS_CHECK_INTEGRITY is not set, that code complains in the log and the mount fails with EINVAL. Add the mount option to activate the usage of the integrity check code. Add invocation of btrfs integrity check code init and cleanup function on mount and umount, respectively. Add hook to call btrfs integrity check code version of submit_bh/submit_bio. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2011-11-09 20:44:05 +08:00
#include "check-integrity.h"
Btrfs: use rcu to protect device->name Al pointed out that we can just toss out the old name on a device and add a new one arbitrarily, so anybody who uses device->name in printk could possibly use free'd memory. Instead of adding locking around all of this he suggested doing it with RCU, so I've introduced a struct rcu_string that does just that and have gone through and protected all accesses to device->name that aren't under the uuid_mutex with rcu_read_lock(). This protects us and I will use it for dealing with removing the device that we used to mount the file system in a later patch. Thanks, Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <josef@redhat.com>
2012-06-05 02:03:51 +08:00
#include "rcu-string.h"
Btrfs: RAID5 and RAID6 This builds on David Woodhouse's original Btrfs raid5/6 implementation. The code has changed quite a bit, blame Chris Mason for any bugs. Read/modify/write is done after the higher levels of the filesystem have prepared a given bio. This means the higher layers are not responsible for building full stripes, and they don't need to query for the topology of the extents that may get allocated during delayed allocation runs. It also means different files can easily share the same stripe. But, it does expose us to incorrect parity if we crash or lose power while doing a read/modify/write cycle. This will be addressed in a later commit. Scrub is unable to repair crc errors on raid5/6 chunks. Discard does not work on raid5/6 (yet) The stripe size is fixed at 64KiB per disk. This will be tunable in a later commit. Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-01-30 07:40:14 +08:00
#include "raid56.h"
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
* This is only the first step towards a full-features scrub. It reads all
* extent and super block and verifies the checksums. In case a bad checksum
* is found or the extent cannot be read, good data will be written back if
* any can be found.
*
* Future enhancements:
* - In case an unrepairable extent is encountered, track which files are
* affected and report them
* - track and record media errors, throw out bad devices
* - add a mode to also read unallocated space
*/
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
struct scrub_block;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
/*
* the following three values only influence the performance.
* The last one configures the number of parallel and outstanding I/O
* operations. The first two values configure an upper limit for the number
* of (dynamically allocated) pages that are added to a bio.
*/
#define SCRUB_PAGES_PER_RD_BIO 32 /* 128k per bio */
#define SCRUB_PAGES_PER_WR_BIO 32 /* 128k per bio */
#define SCRUB_BIOS_PER_SCTX 64 /* 8MB per device in flight */
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
/*
* the following value times PAGE_SIZE needs to be large enough to match the
* largest node/leaf/sector size that shall be supported.
* Values larger than BTRFS_STRIPE_LEN are not supported.
*/
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
#define SCRUB_MAX_PAGES_PER_BLOCK 16 /* 64k per node/leaf/sector */
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
struct scrub_recover {
atomic_t refs;
struct btrfs_bio *bbio;
u64 map_length;
};
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct scrub_page {
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
struct scrub_block *sblock;
struct page *page;
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
struct btrfs_device *dev;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
struct list_head list;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u64 flags; /* extent flags */
u64 generation;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u64 logical;
u64 physical;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
u64 physical_for_dev_replace;
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_t refs;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
struct {
unsigned int mirror_num:8;
unsigned int have_csum:1;
unsigned int io_error:1;
};
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u8 csum[BTRFS_CSUM_SIZE];
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
struct scrub_recover *recover;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
};
struct scrub_bio {
int index;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_device *dev;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct bio *bio;
int err;
u64 logical;
u64 physical;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
#if SCRUB_PAGES_PER_WR_BIO >= SCRUB_PAGES_PER_RD_BIO
struct scrub_page *pagev[SCRUB_PAGES_PER_WR_BIO];
#else
struct scrub_page *pagev[SCRUB_PAGES_PER_RD_BIO];
#endif
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
int page_count;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
int next_free;
struct btrfs_work work;
};
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
struct scrub_block {
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
struct scrub_page *pagev[SCRUB_MAX_PAGES_PER_BLOCK];
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
int page_count;
atomic_t outstanding_pages;
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_t refs; /* free mem on transition to zero */
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
struct scrub_parity *sparity;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
struct {
unsigned int header_error:1;
unsigned int checksum_error:1;
unsigned int no_io_error_seen:1;
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
unsigned int generation_error:1; /* also sets header_error */
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
/* The following is for the data used to check parity */
/* It is for the data with checksum */
unsigned int data_corrected:1;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
};
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
struct btrfs_work work;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
};
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
/* Used for the chunks with parity stripe such RAID5/6 */
struct scrub_parity {
struct scrub_ctx *sctx;
struct btrfs_device *scrub_dev;
u64 logic_start;
u64 logic_end;
int nsectors;
int stripe_len;
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_t refs;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
struct list_head spages;
/* Work of parity check and repair */
struct btrfs_work work;
/* Mark the parity blocks which have data */
unsigned long *dbitmap;
/*
* Mark the parity blocks which have data, but errors happen when
* read data or check data
*/
unsigned long *ebitmap;
unsigned long bitmap[0];
};
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct scrub_wr_ctx {
struct scrub_bio *wr_curr_bio;
struct btrfs_device *tgtdev;
int pages_per_wr_bio; /* <= SCRUB_PAGES_PER_WR_BIO */
atomic_t flush_all_writes;
struct mutex wr_lock;
};
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx {
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct scrub_bio *bios[SCRUB_BIOS_PER_SCTX];
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_root *dev_root;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
int first_free;
int curr;
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
atomic_t bios_in_flight;
atomic_t workers_pending;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
spinlock_t list_lock;
wait_queue_head_t list_wait;
u16 csum_size;
struct list_head csum_list;
atomic_t cancel_req;
btrfs: add readonly flag setting the readonly flag prevents writes in case an error is detected Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-23 23:34:19 +08:00
int readonly;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
int pages_per_rd_bio;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u32 sectorsize;
u32 nodesize;
Btrfs: disallow some operations on the device replace target device This patch adds some code to disallow operations on the device that is used as the target for the device replace operation. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 01:29:28 +08:00
int is_dev_replace;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct scrub_wr_ctx wr_ctx;
Btrfs: disallow some operations on the device replace target device This patch adds some code to disallow operations on the device that is used as the target for the device replace operation. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 01:29:28 +08:00
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
* statistics
*/
struct btrfs_scrub_progress stat;
spinlock_t stat_lock;
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
/*
* Use a ref counter to avoid use-after-free issues. Scrub workers
* decrement bios_in_flight and workers_pending and then do a wakeup
* on the list_wait wait queue. We must ensure the main scrub task
* doesn't free the scrub context before or while the workers are
* doing the wakeup() call.
*/
atomic_t refs;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
};
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
struct scrub_fixup_nodatasum {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_device *dev;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
u64 logical;
struct btrfs_root *root;
struct btrfs_work work;
int mirror_num;
};
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
struct scrub_nocow_inode {
u64 inum;
u64 offset;
u64 root;
struct list_head list;
};
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct scrub_copy_nocow_ctx {
struct scrub_ctx *sctx;
u64 logical;
u64 len;
int mirror_num;
u64 physical_for_dev_replace;
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
struct list_head inodes;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct btrfs_work work;
};
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
struct scrub_warning {
struct btrfs_path *path;
u64 extent_item_size;
const char *errstr;
sector_t sector;
u64 logical;
struct btrfs_device *dev;
};
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
static void scrub_pending_bio_inc(struct scrub_ctx *sctx);
static void scrub_pending_bio_dec(struct scrub_ctx *sctx);
static void scrub_pending_trans_workers_inc(struct scrub_ctx *sctx);
static void scrub_pending_trans_workers_dec(struct scrub_ctx *sctx);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_handle_errored_block(struct scrub_block *sblock_to_check);
Btrfs: Simplify scrub_setup_recheck_block()'s argument scrub_setup_recheck_block() have many arguments but most of them can be get from one of them, we can remove them to make code clean. Some other cleanup for that function also included in this patch. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:42 +08:00
static int scrub_setup_recheck_block(struct scrub_block *original_sblock,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct scrub_block *sblocks_for_recheck);
Btrfs: in scrub repair code, simplify alloc error handling In the scrub repair code, the code is changed to handle memory allocation errors a little bit smarter. The change is to handle it just like a read error. This simplifies the code and removes a couple of lines of code, since the code to handle read errors is there anyway. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:16:26 +08:00
static void scrub_recheck_block(struct btrfs_fs_info *fs_info,
btrfs: Reduce unnecessary arguments in scrub_recheck_block We don't need pass so many arguments for recheck sblock now, this patch cleans them. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:32:06 +08:00
struct scrub_block *sblock,
int retry_failed_mirror);
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
static void scrub_recheck_block_checksum(struct scrub_block *sblock);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
Btrfs: Remove noneed force_write in scrub_write_block_to_dev_replace It is always 1 in this place, because !1 case was already jumped out in previous code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:36 +08:00
struct scrub_block *sblock_good);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
struct scrub_block *sblock_good,
int page_num, int force_write);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static void scrub_write_block_to_dev_replace(struct scrub_block *sblock);
static int scrub_write_page_to_dev_replace(struct scrub_block *sblock,
int page_num);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_checksum_data(struct scrub_block *sblock);
static int scrub_checksum_tree_block(struct scrub_block *sblock);
static int scrub_checksum_super(struct scrub_block *sblock);
static void scrub_block_get(struct scrub_block *sblock);
static void scrub_block_put(struct scrub_block *sblock);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
static void scrub_page_get(struct scrub_page *spage);
static void scrub_page_put(struct scrub_page *spage);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
static void scrub_parity_get(struct scrub_parity *sparity);
static void scrub_parity_put(struct scrub_parity *sparity);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static int scrub_add_page_to_rd_bio(struct scrub_ctx *sctx,
struct scrub_page *spage);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
static int scrub_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
u64 physical, struct btrfs_device *dev, u64 flags,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
u64 gen, int mirror_num, u8 *csum, int force,
u64 physical_for_dev_replace);
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
static void scrub_bio_end_io(struct bio *bio);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static void scrub_bio_end_io_worker(struct btrfs_work *work);
static void scrub_block_complete(struct scrub_block *sblock);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static void scrub_remap_extent(struct btrfs_fs_info *fs_info,
u64 extent_logical, u64 extent_len,
u64 *extent_physical,
struct btrfs_device **extent_dev,
int *extent_mirror_num);
static int scrub_setup_wr_ctx(struct scrub_ctx *sctx,
struct scrub_wr_ctx *wr_ctx,
struct btrfs_fs_info *fs_info,
struct btrfs_device *dev,
int is_dev_replace);
static void scrub_free_wr_ctx(struct scrub_wr_ctx *wr_ctx);
static int scrub_add_page_to_wr_bio(struct scrub_ctx *sctx,
struct scrub_page *spage);
static void scrub_wr_submit(struct scrub_ctx *sctx);
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
static void scrub_wr_bio_end_io(struct bio *bio);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static void scrub_wr_bio_end_io_worker(struct btrfs_work *work);
static int write_page_nocow(struct scrub_ctx *sctx,
u64 physical_for_dev_replace, struct page *page);
static int copy_nocow_pages_for_inode(u64 inum, u64 offset, u64 root,
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
struct scrub_copy_nocow_ctx *ctx);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static int copy_nocow_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
int mirror_num, u64 physical_for_dev_replace);
static void copy_nocow_pages_worker(struct btrfs_work *work);
Btrfs: wrap repeated code into scrub_blocked_if_needed() Just wrap same code into one function scrub_blocked_if_needed(). This make a change that we will move waiting (@workers_pending = 0) before we can wake up commiting transaction(atomic_inc(@scrub_paused)), we must take carefully to not deadlock here. Thread 1 Thread 2 |->btrfs_commit_transaction() |->set trans type(COMMIT_DOING) |->btrfs_scrub_paused()(blocked) |->join_transaction(blocked) Move btrfs_scrub_paused() before setting trans type which means we can still join a transaction when commiting_transaction is blocked. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Suggested-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:16:53 +08:00
static void __scrub_blocked_if_needed(struct btrfs_fs_info *fs_info);
Btrfs: fix wrong super generation mismatch when scrubbing supers We came a race condition when scrubbing superblocks, the story is: In commiting transaction, we will update @last_trans_commited after writting superblocks, if scrubber start after writting superblocks and before updating @last_trans_commited, generation mismatch happens! We fix this by checking @scrub_pause_req, and we won't start a srubber until commiting transaction is finished.(after btrfs_scrub_continue() finished.) Reported-by: Sebastian Ochmann <ochmann@informatik.uni-bonn.de> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Reviewed-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:15:19 +08:00
static void scrub_blocked_if_needed(struct btrfs_fs_info *fs_info);
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
static void scrub_put_ctx(struct scrub_ctx *sctx);
Btrfs: minor cleanup in scrub Just a minor cleanup commit in preparation for the big block changes. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:26 +08:00
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
static void scrub_pending_bio_inc(struct scrub_ctx *sctx)
{
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
atomic_inc(&sctx->refs);
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
atomic_inc(&sctx->bios_in_flight);
}
static void scrub_pending_bio_dec(struct scrub_ctx *sctx)
{
atomic_dec(&sctx->bios_in_flight);
wake_up(&sctx->list_wait);
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
scrub_put_ctx(sctx);
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
}
Btrfs: wrap repeated code into scrub_blocked_if_needed() Just wrap same code into one function scrub_blocked_if_needed(). This make a change that we will move waiting (@workers_pending = 0) before we can wake up commiting transaction(atomic_inc(@scrub_paused)), we must take carefully to not deadlock here. Thread 1 Thread 2 |->btrfs_commit_transaction() |->set trans type(COMMIT_DOING) |->btrfs_scrub_paused()(blocked) |->join_transaction(blocked) Move btrfs_scrub_paused() before setting trans type which means we can still join a transaction when commiting_transaction is blocked. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Suggested-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:16:53 +08:00
static void __scrub_blocked_if_needed(struct btrfs_fs_info *fs_info)
Btrfs: fix wrong super generation mismatch when scrubbing supers We came a race condition when scrubbing superblocks, the story is: In commiting transaction, we will update @last_trans_commited after writting superblocks, if scrubber start after writting superblocks and before updating @last_trans_commited, generation mismatch happens! We fix this by checking @scrub_pause_req, and we won't start a srubber until commiting transaction is finished.(after btrfs_scrub_continue() finished.) Reported-by: Sebastian Ochmann <ochmann@informatik.uni-bonn.de> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Reviewed-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:15:19 +08:00
{
while (atomic_read(&fs_info->scrub_pause_req)) {
mutex_unlock(&fs_info->scrub_lock);
wait_event(fs_info->scrub_pause_wait,
atomic_read(&fs_info->scrub_pause_req) == 0);
mutex_lock(&fs_info->scrub_lock);
}
}
btrfs: Separate scrub_blocked_if_needed() to scrub_pause_on/off() It can reduce current duplicated code which is similar to scrub_blocked_if_needed() but can not call it because little different. It also used by my next patch which is in same case. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:28 +08:00
static void scrub_pause_on(struct btrfs_fs_info *fs_info)
Btrfs: wrap repeated code into scrub_blocked_if_needed() Just wrap same code into one function scrub_blocked_if_needed(). This make a change that we will move waiting (@workers_pending = 0) before we can wake up commiting transaction(atomic_inc(@scrub_paused)), we must take carefully to not deadlock here. Thread 1 Thread 2 |->btrfs_commit_transaction() |->set trans type(COMMIT_DOING) |->btrfs_scrub_paused()(blocked) |->join_transaction(blocked) Move btrfs_scrub_paused() before setting trans type which means we can still join a transaction when commiting_transaction is blocked. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Suggested-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:16:53 +08:00
{
atomic_inc(&fs_info->scrubs_paused);
wake_up(&fs_info->scrub_pause_wait);
btrfs: Separate scrub_blocked_if_needed() to scrub_pause_on/off() It can reduce current duplicated code which is similar to scrub_blocked_if_needed() but can not call it because little different. It also used by my next patch which is in same case. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:28 +08:00
}
Btrfs: wrap repeated code into scrub_blocked_if_needed() Just wrap same code into one function scrub_blocked_if_needed(). This make a change that we will move waiting (@workers_pending = 0) before we can wake up commiting transaction(atomic_inc(@scrub_paused)), we must take carefully to not deadlock here. Thread 1 Thread 2 |->btrfs_commit_transaction() |->set trans type(COMMIT_DOING) |->btrfs_scrub_paused()(blocked) |->join_transaction(blocked) Move btrfs_scrub_paused() before setting trans type which means we can still join a transaction when commiting_transaction is blocked. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Suggested-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:16:53 +08:00
btrfs: Separate scrub_blocked_if_needed() to scrub_pause_on/off() It can reduce current duplicated code which is similar to scrub_blocked_if_needed() but can not call it because little different. It also used by my next patch which is in same case. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:28 +08:00
static void scrub_pause_off(struct btrfs_fs_info *fs_info)
{
Btrfs: wrap repeated code into scrub_blocked_if_needed() Just wrap same code into one function scrub_blocked_if_needed(). This make a change that we will move waiting (@workers_pending = 0) before we can wake up commiting transaction(atomic_inc(@scrub_paused)), we must take carefully to not deadlock here. Thread 1 Thread 2 |->btrfs_commit_transaction() |->set trans type(COMMIT_DOING) |->btrfs_scrub_paused()(blocked) |->join_transaction(blocked) Move btrfs_scrub_paused() before setting trans type which means we can still join a transaction when commiting_transaction is blocked. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Suggested-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:16:53 +08:00
mutex_lock(&fs_info->scrub_lock);
__scrub_blocked_if_needed(fs_info);
atomic_dec(&fs_info->scrubs_paused);
mutex_unlock(&fs_info->scrub_lock);
wake_up(&fs_info->scrub_pause_wait);
}
btrfs: Separate scrub_blocked_if_needed() to scrub_pause_on/off() It can reduce current duplicated code which is similar to scrub_blocked_if_needed() but can not call it because little different. It also used by my next patch which is in same case. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:28 +08:00
static void scrub_blocked_if_needed(struct btrfs_fs_info *fs_info)
{
scrub_pause_on(fs_info);
scrub_pause_off(fs_info);
}
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
/*
* used for workers that require transaction commits (i.e., for the
* NOCOW case)
*/
static void scrub_pending_trans_workers_inc(struct scrub_ctx *sctx)
{
struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
atomic_inc(&sctx->refs);
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
/*
* increment scrubs_running to prevent cancel requests from
* completing as long as a worker is running. we must also
* increment scrubs_paused to prevent deadlocking on pause
* requests used for transactions commits (as the worker uses a
* transaction context). it is safe to regard the worker
* as paused for all matters practical. effectively, we only
* avoid cancellation requests from completing.
*/
mutex_lock(&fs_info->scrub_lock);
atomic_inc(&fs_info->scrubs_running);
atomic_inc(&fs_info->scrubs_paused);
mutex_unlock(&fs_info->scrub_lock);
Btrfs: wake up @scrub_pause_wait as much as we can check if @scrubs_running=@scrubs_paused condition inside wait_event() is not an atomic operation which means we may inc/dec @scrub_running/ paused at any time. Let's wake up @scrub_pause_wait as much as we can to let commit transaction blocked less. An example below: Thread1 Thread2 |->scrub_blocked_if_needed() |->scrub_pending_trans_workers_inc |->increase @scrub_paused |->increase @scrub_running |->wake up scrub_pause_wait list |->scrub blocked |->increase @scrub_paused Thread3 is commiting transaction which is blocked at btrfs_scrub_pause(). So after Thread2 increase @scrub_paused, we meet the condition @scrub_paused=@scrub_running, but transaction will be still blocked until another calling to wake up @scrub_pause_wait. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-19 19:24:19 +08:00
/*
* check if @scrubs_running=@scrubs_paused condition
* inside wait_event() is not an atomic operation.
* which means we may inc/dec @scrub_running/paused
* at any time. Let's wake up @scrub_pause_wait as
* much as we can to let commit transaction blocked less.
*/
wake_up(&fs_info->scrub_pause_wait);
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
atomic_inc(&sctx->workers_pending);
}
/* used for workers that require transaction commits */
static void scrub_pending_trans_workers_dec(struct scrub_ctx *sctx)
{
struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
/*
* see scrub_pending_trans_workers_inc() why we're pretending
* to be paused in the scrub counters
*/
mutex_lock(&fs_info->scrub_lock);
atomic_dec(&fs_info->scrubs_running);
atomic_dec(&fs_info->scrubs_paused);
mutex_unlock(&fs_info->scrub_lock);
atomic_dec(&sctx->workers_pending);
wake_up(&fs_info->scrub_pause_wait);
wake_up(&sctx->list_wait);
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
scrub_put_ctx(sctx);
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
static void scrub_free_csums(struct scrub_ctx *sctx)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
while (!list_empty(&sctx->csum_list)) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct btrfs_ordered_sum *sum;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sum = list_first_entry(&sctx->csum_list,
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct btrfs_ordered_sum, list);
list_del(&sum->list);
kfree(sum);
}
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
static noinline_for_stack void scrub_free_ctx(struct scrub_ctx *sctx)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
int i;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
if (!sctx)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
scrub_free_wr_ctx(&sctx->wr_ctx);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/* this can happen when scrub is cancelled */
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
if (sctx->curr != -1) {
struct scrub_bio *sbio = sctx->bios[sctx->curr];
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
for (i = 0; i < sbio->page_count; i++) {
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
WARN_ON(!sbio->pagev[i]->page);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
scrub_block_put(sbio->pagev[i]->sblock);
}
bio_put(sbio->bio);
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
for (i = 0; i < SCRUB_BIOS_PER_SCTX; ++i) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_bio *sbio = sctx->bios[i];
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (!sbio)
break;
kfree(sbio);
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
scrub_free_csums(sctx);
kfree(sctx);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
static void scrub_put_ctx(struct scrub_ctx *sctx)
{
if (atomic_dec_and_test(&sctx->refs))
scrub_free_ctx(sctx);
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
static noinline_for_stack
Btrfs: disallow some operations on the device replace target device This patch adds some code to disallow operations on the device that is used as the target for the device replace operation. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 01:29:28 +08:00
struct scrub_ctx *scrub_setup_ctx(struct btrfs_device *dev, int is_dev_replace)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
int i;
struct btrfs_fs_info *fs_info = dev->dev_root->fs_info;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
int ret;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
sctx = kzalloc(sizeof(*sctx), GFP_KERNEL);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
if (!sctx)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
goto nomem;
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
atomic_set(&sctx->refs, 1);
Btrfs: disallow some operations on the device replace target device This patch adds some code to disallow operations on the device that is used as the target for the device replace operation. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 01:29:28 +08:00
sctx->is_dev_replace = is_dev_replace;
block: remove bio_get_nr_vecs() We can always fill up the bio now, no need to estimate the possible size based on queue parameters. Acked-by: Steven Whitehouse <swhiteho@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> [hch: rebased and wrote a changelog] Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ming Lin <ming.l@ssi.samsung.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-19 20:31:01 +08:00
sctx->pages_per_rd_bio = SCRUB_PAGES_PER_RD_BIO;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sctx->curr = -1;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
sctx->dev_root = dev->dev_root;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
for (i = 0; i < SCRUB_BIOS_PER_SCTX; ++i) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct scrub_bio *sbio;
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
sbio = kzalloc(sizeof(*sbio), GFP_KERNEL);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (!sbio)
goto nomem;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sctx->bios[i] = sbio;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
sbio->index = i;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sbio->sctx = sctx;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
sbio->page_count = 0;
Btrfs: fix task hang under heavy compressed write This has been reported and discussed for a long time, and this hang occurs in both 3.15 and 3.16. Btrfs now migrates to use kernel workqueue, but it introduces this hang problem. Btrfs has a kind of work queued as an ordered way, which means that its ordered_func() must be processed in the way of FIFO, so it usually looks like -- normal_work_helper(arg) work = container_of(arg, struct btrfs_work, normal_work); work->func() <---- (we name it work X) for ordered_work in wq->ordered_list ordered_work->ordered_func() ordered_work->ordered_free() The hang is a rare case, first when we find free space, we get an uncached block group, then we go to read its free space cache inode for free space information, so it will file a readahead request btrfs_readpages() for page that is not in page cache __do_readpage() submit_extent_page() btrfs_submit_bio_hook() btrfs_bio_wq_end_io() submit_bio() end_workqueue_bio() <--(ret by the 1st endio) queue a work(named work Y) for the 2nd also the real endio() So the hang occurs when work Y's work_struct and work X's work_struct happens to share the same address. A bit more explanation, A,B,C -- struct btrfs_work arg -- struct work_struct kthread: worker_thread() pick up a work_struct from @worklist process_one_work(arg) worker->current_work = arg; <-- arg is A->normal_work worker->current_func(arg) normal_work_helper(arg) A = container_of(arg, struct btrfs_work, normal_work); A->func() A->ordered_func() A->ordered_free() <-- A gets freed B->ordered_func() submit_compressed_extents() find_free_extent() load_free_space_inode() ... <-- (the above readhead stack) end_workqueue_bio() btrfs_queue_work(work C) B->ordered_free() As if work A has a high priority in wq->ordered_list and there are more ordered works queued after it, such as B->ordered_func(), its memory could have been freed before normal_work_helper() returns, which means that kernel workqueue code worker_thread() still has worker->current_work pointer to be work A->normal_work's, ie. arg's address. Meanwhile, work C is allocated after work A is freed, work C->normal_work and work A->normal_work are likely to share the same address(I confirmed this with ftrace output, so I'm not just guessing, it's rare though). When another kthread picks up work C->normal_work to process, and finds our kthread is processing it(see find_worker_executing_work()), it'll think work C as a collision and skip then, which ends up nobody processing work C. So the situation is that our kthread is waiting forever on work C. Besides, there're other cases that can lead to deadlock, but the real problem is that all btrfs workqueue shares one work->func, -- normal_work_helper, so this makes each workqueue to have its own helper function, but only a wraper pf normal_work_helper. With this patch, I no long hit the above hang. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-08-15 23:36:53 +08:00
btrfs_init_work(&sbio->work, btrfs_scrub_helper,
scrub_bio_end_io_worker, NULL, NULL);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (i != SCRUB_BIOS_PER_SCTX - 1)
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sctx->bios[i]->next_free = i + 1;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
else
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sctx->bios[i]->next_free = -1;
}
sctx->first_free = 0;
sctx->nodesize = dev->dev_root->nodesize;
sctx->sectorsize = dev->dev_root->sectorsize;
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
atomic_set(&sctx->bios_in_flight, 0);
atomic_set(&sctx->workers_pending, 0);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
atomic_set(&sctx->cancel_req, 0);
sctx->csum_size = btrfs_super_csum_size(fs_info->super_copy);
INIT_LIST_HEAD(&sctx->csum_list);
spin_lock_init(&sctx->list_lock);
spin_lock_init(&sctx->stat_lock);
init_waitqueue_head(&sctx->list_wait);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
ret = scrub_setup_wr_ctx(sctx, &sctx->wr_ctx, fs_info,
fs_info->dev_replace.tgtdev, is_dev_replace);
if (ret) {
scrub_free_ctx(sctx);
return ERR_PTR(ret);
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
return sctx;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
nomem:
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
scrub_free_ctx(sctx);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return ERR_PTR(-ENOMEM);
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root,
void *warn_ctx)
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
{
u64 isize;
u32 nlink;
int ret;
int i;
struct extent_buffer *eb;
struct btrfs_inode_item *inode_item;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct scrub_warning *swarn = warn_ctx;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
struct btrfs_fs_info *fs_info = swarn->dev->dev_root->fs_info;
struct inode_fs_paths *ipath = NULL;
struct btrfs_root *local_root;
struct btrfs_key root_key;
btrfs: expand btrfs_find_item if found_key is NULL If the found_key is NULL, then btrfs_find_item becomes a verbose wrapper for simple btrfs_search_slot. After we've removed all such callers, passing a NULL key is not valid anymore. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-01-03 02:36:14 +08:00
struct btrfs_key key;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
root_key.objectid = root;
root_key.type = BTRFS_ROOT_ITEM_KEY;
root_key.offset = (u64)-1;
local_root = btrfs_read_fs_root_no_name(fs_info, &root_key);
if (IS_ERR(local_root)) {
ret = PTR_ERR(local_root);
goto err;
}
btrfs: cleanup, remove inode_item_info helper It's only a simple wrapper around btrfs_find_item, the locally defined key is not used. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-01-03 01:55:46 +08:00
/*
* this makes the path point to (inum INODE_ITEM ioff)
*/
btrfs: expand btrfs_find_item if found_key is NULL If the found_key is NULL, then btrfs_find_item becomes a verbose wrapper for simple btrfs_search_slot. After we've removed all such callers, passing a NULL key is not valid anymore. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-01-03 02:36:14 +08:00
key.objectid = inum;
key.type = BTRFS_INODE_ITEM_KEY;
key.offset = 0;
ret = btrfs_search_slot(NULL, local_root, &key, swarn->path, 0, 0);
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
if (ret) {
btrfs_release_path(swarn->path);
goto err;
}
eb = swarn->path->nodes[0];
inode_item = btrfs_item_ptr(eb, swarn->path->slots[0],
struct btrfs_inode_item);
isize = btrfs_inode_size(eb, inode_item);
nlink = btrfs_inode_nlink(eb, inode_item);
btrfs_release_path(swarn->path);
ipath = init_ipath(4096, local_root, swarn->path);
btrfs scrub: handle -ENOMEM from init_ipath() init_ipath() can return an ERR_PTR(-ENOMEM). Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
2011-11-16 16:28:01 +08:00
if (IS_ERR(ipath)) {
ret = PTR_ERR(ipath);
ipath = NULL;
goto err;
}
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
ret = paths_from_inode(inum, ipath);
if (ret < 0)
goto err;
/*
* we deliberately ignore the bit ipath might have been too small to
* hold all of the paths here
*/
for (i = 0; i < ipath->fspath->elem_cnt; ++i)
btrfs: switch message printers to _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 15:01:03 +08:00
btrfs_warn_in_rcu(fs_info, "%s at logical %llu on dev "
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
"%s, sector %llu, root %llu, inode %llu, offset %llu, "
btrfs: switch message printers to _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 15:01:03 +08:00
"length %llu, links %u (path: %s)", swarn->errstr,
Btrfs: use rcu to protect device->name Al pointed out that we can just toss out the old name on a device and add a new one arbitrarily, so anybody who uses device->name in printk could possibly use free'd memory. Instead of adding locking around all of this he suggested doing it with RCU, so I've introduced a struct rcu_string that does just that and have gone through and protected all accesses to device->name that aren't under the uuid_mutex with rcu_read_lock(). This protects us and I will use it for dealing with removing the device that we used to mount the file system in a later patch. Thanks, Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <josef@redhat.com>
2012-06-05 02:03:51 +08:00
swarn->logical, rcu_str_deref(swarn->dev->name),
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
(unsigned long long)swarn->sector, root, inum, offset,
min(isize - offset, (u64)PAGE_SIZE), nlink,
btrfs: Fix up 32/64-bit compatibility for new ioctls This patch casts to unsigned long before casting to a pointer and fixes the following warnings: fs/btrfs/extent_io.c:2289:20: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast] fs/btrfs/ioctl.c:2933:37: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast] fs/btrfs/ioctl.c:2937:21: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast] fs/btrfs/ioctl.c:3020:21: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast] fs/btrfs/scrub.c:275:4: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast] fs/btrfs/backref.c:686:27: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast] Signed-off-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-11-20 20:31:57 +08:00
(char *)(unsigned long)ipath->fspath->val[i]);
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
free_ipath(ipath);
return 0;
err:
btrfs: switch message printers to _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 15:01:03 +08:00
btrfs_warn_in_rcu(fs_info, "%s at logical %llu on dev "
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
"%s, sector %llu, root %llu, inode %llu, offset %llu: path "
btrfs: switch message printers to _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 15:01:03 +08:00
"resolving failed with ret=%d", swarn->errstr,
Btrfs: use rcu to protect device->name Al pointed out that we can just toss out the old name on a device and add a new one arbitrarily, so anybody who uses device->name in printk could possibly use free'd memory. Instead of adding locking around all of this he suggested doing it with RCU, so I've introduced a struct rcu_string that does just that and have gone through and protected all accesses to device->name that aren't under the uuid_mutex with rcu_read_lock(). This protects us and I will use it for dealing with removing the device that we used to mount the file system in a later patch. Thanks, Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <josef@redhat.com>
2012-06-05 02:03:51 +08:00
swarn->logical, rcu_str_deref(swarn->dev->name),
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
(unsigned long long)swarn->sector, root, inum, offset, ret);
free_ipath(ipath);
return 0;
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static void scrub_print_warning(const char *errstr, struct scrub_block *sblock)
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
{
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_device *dev;
struct btrfs_fs_info *fs_info;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
struct btrfs_path *path;
struct btrfs_key found_key;
struct extent_buffer *eb;
struct btrfs_extent_item *ei;
struct scrub_warning swarn;
Btrfs: fix a bug in parsing return value in logical resolve In logical resolve, we parse extent_from_logical()'s 'ret' as a kind of flag. It is possible to lose our errors because (-EXXXX & BTRFS_EXTENT_FLAG_TREE_BLOCK) is true. I'm not sure if it is on purpose, it just looks too hacky if it is. I'd rather use a real flag and a 'ret' to catch errors. Acked-by: Jan Schmidt <list.btrfs@jan-o-sch.net> Signed-off-by: Liu Bo <liub.liubo@gmail.com>
2012-09-08 10:01:28 +08:00
unsigned long ptr = 0;
u64 extent_item_pos;
u64 flags = 0;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
u64 ref_root;
Btrfs: fix a bug in parsing return value in logical resolve In logical resolve, we parse extent_from_logical()'s 'ret' as a kind of flag. It is possible to lose our errors because (-EXXXX & BTRFS_EXTENT_FLAG_TREE_BLOCK) is true. I'm not sure if it is on purpose, it just looks too hacky if it is. I'd rather use a real flag and a 'ret' to catch errors. Acked-by: Jan Schmidt <list.btrfs@jan-o-sch.net> Signed-off-by: Liu Bo <liub.liubo@gmail.com>
2012-09-08 10:01:28 +08:00
u32 item_size;
btrfs: scrub: silence an uninitialized variable warning It's basically harmless if "ref_level" isn't initialized since it's only used for an error message, but it causes a static checker warning. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: David Sterba <dsterba@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2016-03-11 16:08:56 +08:00
u8 ref_level = 0;
Btrfs: fix a bug in parsing return value in logical resolve In logical resolve, we parse extent_from_logical()'s 'ret' as a kind of flag. It is possible to lose our errors because (-EXXXX & BTRFS_EXTENT_FLAG_TREE_BLOCK) is true. I'm not sure if it is on purpose, it just looks too hacky if it is. I'd rather use a real flag and a 'ret' to catch errors. Acked-by: Jan Schmidt <list.btrfs@jan-o-sch.net> Signed-off-by: Liu Bo <liub.liubo@gmail.com>
2012-09-08 10:01:28 +08:00
int ret;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
WARN_ON(sblock->page_count < 1);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
dev = sblock->pagev[0]->dev;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
fs_info = sblock->sctx->dev_root->fs_info;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
path = btrfs_alloc_path();
btrfs: remove unused members from struct scrub_warning Signed-off-by: David Sterba <dsterba@suse.cz>
2014-07-30 07:25:30 +08:00
if (!path)
return;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
swarn.sector = (sblock->pagev[0]->physical) >> 9;
swarn.logical = sblock->pagev[0]->logical;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
swarn.errstr = errstr;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
swarn.dev = NULL;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
Btrfs: fix a bug in parsing return value in logical resolve In logical resolve, we parse extent_from_logical()'s 'ret' as a kind of flag. It is possible to lose our errors because (-EXXXX & BTRFS_EXTENT_FLAG_TREE_BLOCK) is true. I'm not sure if it is on purpose, it just looks too hacky if it is. I'd rather use a real flag and a 'ret' to catch errors. Acked-by: Jan Schmidt <list.btrfs@jan-o-sch.net> Signed-off-by: Liu Bo <liub.liubo@gmail.com>
2012-09-08 10:01:28 +08:00
ret = extent_from_logical(fs_info, swarn.logical, path, &found_key,
&flags);
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
if (ret < 0)
goto out;
Btrfs: new backref walking code The old backref iteration code could only safely be used on commit roots. Besides this limitation, it had bugs in finding the roots for these references. This commit replaces large parts of it by btrfs_find_all_roots() which a) really finds all roots and the correct roots, b) works correctly under heavy file system load, c) considers delayed refs. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-12-02 21:56:41 +08:00
extent_item_pos = swarn.logical - found_key.objectid;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
swarn.extent_item_size = found_key.offset;
eb = path->nodes[0];
ei = btrfs_item_ptr(eb, path->slots[0], struct btrfs_extent_item);
item_size = btrfs_item_size_nr(eb, path->slots[0]);
Btrfs: fix a bug in parsing return value in logical resolve In logical resolve, we parse extent_from_logical()'s 'ret' as a kind of flag. It is possible to lose our errors because (-EXXXX & BTRFS_EXTENT_FLAG_TREE_BLOCK) is true. I'm not sure if it is on purpose, it just looks too hacky if it is. I'd rather use a real flag and a 'ret' to catch errors. Acked-by: Jan Schmidt <list.btrfs@jan-o-sch.net> Signed-off-by: Liu Bo <liub.liubo@gmail.com>
2012-09-08 10:01:28 +08:00
if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
do {
Btrfs: fix scrub_print_warning to handle skinny metadata extents The skinny extents are intepreted incorrectly in scrub_print_warning(), and end up hitting the BUG() in btrfs_extent_inline_ref_size. Reported-by: Konstantinos Skarlatos <k.skarlatos@gmail.com> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-06-09 10:54:07 +08:00
ret = tree_backref_for_extent(&ptr, eb, &found_key, ei,
item_size, &ref_root,
&ref_level);
btrfs: switch message printers to _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 15:01:03 +08:00
btrfs_warn_in_rcu(fs_info,
"%s at logical %llu on dev %s, "
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
"sector %llu: metadata %s (level %d) in tree "
btrfs: switch message printers to _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 15:01:03 +08:00
"%llu", errstr, swarn.logical,
Btrfs: use rcu to protect device->name Al pointed out that we can just toss out the old name on a device and add a new one arbitrarily, so anybody who uses device->name in printk could possibly use free'd memory. Instead of adding locking around all of this he suggested doing it with RCU, so I've introduced a struct rcu_string that does just that and have gone through and protected all accesses to device->name that aren't under the uuid_mutex with rcu_read_lock(). This protects us and I will use it for dealing with removing the device that we used to mount the file system in a later patch. Thanks, Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <josef@redhat.com>
2012-06-05 02:03:51 +08:00
rcu_str_deref(dev->name),
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
(unsigned long long)swarn.sector,
ref_level ? "node" : "leaf",
ret < 0 ? -1 : ref_level,
ret < 0 ? -1 : ref_root);
} while (ret != 1);
Btrfs: don't drop path when printing out tree errors in scrub A user reported a panic where we were panicing somewhere in tree_backref_for_extent from scrub_print_warning. He only captured the trace but looking at scrub_print_warning we drop the path right before we mess with the extent buffer to print out a bunch of stuff, which isn't right. So fix this by dropping the path after we use the eb if we need to. Thanks, Cc: stable@vger.kernel.org Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-03-29 22:09:34 +08:00
btrfs_release_path(path);
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
} else {
Btrfs: don't drop path when printing out tree errors in scrub A user reported a panic where we were panicing somewhere in tree_backref_for_extent from scrub_print_warning. He only captured the trace but looking at scrub_print_warning we drop the path right before we mess with the extent buffer to print out a bunch of stuff, which isn't right. So fix this by dropping the path after we use the eb if we need to. Thanks, Cc: stable@vger.kernel.org Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-03-29 22:09:34 +08:00
btrfs_release_path(path);
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
swarn.path = path;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
swarn.dev = dev;
Btrfs: fix regression in scrub path resolving In commit 4692cf58 we introduced new backref walking code for btrfs. This assumes we're searching live roots, which requires a transaction context. While scrubbing, however, we must not join a transaction because this could deadlock with the commit path. Additionally, what scrub really wants to do is resolving a logical address in the commit root it's currently checking. This patch adds support for logical to path resolving on commit roots and makes scrub use that. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2012-03-24 00:32:28 +08:00
iterate_extent_inodes(fs_info, found_key.objectid,
extent_item_pos, 1,
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
scrub_print_warning_inode, &swarn);
}
out:
btrfs_free_path(path);
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static int scrub_fixup_readpage(u64 inum, u64 offset, u64 root, void *fixup_ctx)
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
{
btrfs: integrating raid-repair and scrub-fixup-nodatasum This ties nodatasum fixup in scrub together with raid repair patches. While both series are working fine alone, scrub will report uncorrectable errors if they occur in a nodatasum extent *and* the page is in the page cache. Previously, we would have triggered readpage to find good data and do the repair. However, readpage wouldn't read anything in the case where the page is up to date in the cache. So, we simply take that good data we have and call repair_io_failure directly (unless the page in the cache is dirty). Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-08-05 00:11:04 +08:00
struct page *page = NULL;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
unsigned long index;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct scrub_fixup_nodatasum *fixup = fixup_ctx;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
int ret;
btrfs: integrating raid-repair and scrub-fixup-nodatasum This ties nodatasum fixup in scrub together with raid repair patches. While both series are working fine alone, scrub will report uncorrectable errors if they occur in a nodatasum extent *and* the page is in the page cache. Previously, we would have triggered readpage to find good data and do the repair. However, readpage wouldn't read anything in the case where the page is up to date in the cache. So, we simply take that good data we have and call repair_io_failure directly (unless the page in the cache is dirty). Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-08-05 00:11:04 +08:00
int corrected = 0;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
struct btrfs_key key;
btrfs: integrating raid-repair and scrub-fixup-nodatasum This ties nodatasum fixup in scrub together with raid repair patches. While both series are working fine alone, scrub will report uncorrectable errors if they occur in a nodatasum extent *and* the page is in the page cache. Previously, we would have triggered readpage to find good data and do the repair. However, readpage wouldn't read anything in the case where the page is up to date in the cache. So, we simply take that good data we have and call repair_io_failure directly (unless the page in the cache is dirty). Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-08-05 00:11:04 +08:00
struct inode *inode = NULL;
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
struct btrfs_fs_info *fs_info;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
u64 end = offset + PAGE_SIZE - 1;
struct btrfs_root *local_root;
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
int srcu_index;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
key.objectid = root;
key.type = BTRFS_ROOT_ITEM_KEY;
key.offset = (u64)-1;
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
fs_info = fixup->root->fs_info;
srcu_index = srcu_read_lock(&fs_info->subvol_srcu);
local_root = btrfs_read_fs_root_no_name(fs_info, &key);
if (IS_ERR(local_root)) {
srcu_read_unlock(&fs_info->subvol_srcu, srcu_index);
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
return PTR_ERR(local_root);
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
}
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
key.type = BTRFS_INODE_ITEM_KEY;
key.objectid = inum;
key.offset = 0;
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
inode = btrfs_iget(fs_info->sb, &key, local_root, NULL);
srcu_read_unlock(&fs_info->subvol_srcu, srcu_index);
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
if (IS_ERR(inode))
return PTR_ERR(inode);
index = offset >> PAGE_CACHE_SHIFT;
page = find_or_create_page(inode->i_mapping, index, GFP_NOFS);
btrfs: integrating raid-repair and scrub-fixup-nodatasum This ties nodatasum fixup in scrub together with raid repair patches. While both series are working fine alone, scrub will report uncorrectable errors if they occur in a nodatasum extent *and* the page is in the page cache. Previously, we would have triggered readpage to find good data and do the repair. However, readpage wouldn't read anything in the case where the page is up to date in the cache. So, we simply take that good data we have and call repair_io_failure directly (unless the page in the cache is dirty). Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-08-05 00:11:04 +08:00
if (!page) {
ret = -ENOMEM;
goto out;
}
if (PageUptodate(page)) {
if (PageDirty(page)) {
/*
* we need to write the data to the defect sector. the
* data that was in that sector is not in memory,
* because the page was modified. we must not write the
* modified page to that sector.
*
* TODO: what could be done here: wait for the delalloc
* runner to write out that page (might involve
* COW) and see whether the sector is still
* referenced afterwards.
*
* For the meantime, we'll treat this error
* incorrectable, although there is a chance that a
* later scrub will find the bad sector again and that
* there's no dirty page in memory, then.
*/
ret = -EIO;
goto out;
}
Btrfs: modify clean_io_failure and make it suit direct io We could not use clean_io_failure in the direct IO path because it got the filesystem information from the page structure, but the page in the direct IO bio didn't have the filesystem information in its structure. So we need modify it and pass all the information it need by parameters. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-12 18:44:01 +08:00
ret = repair_io_failure(inode, offset, PAGE_SIZE,
btrfs: integrating raid-repair and scrub-fixup-nodatasum This ties nodatasum fixup in scrub together with raid repair patches. While both series are working fine alone, scrub will report uncorrectable errors if they occur in a nodatasum extent *and* the page is in the page cache. Previously, we would have triggered readpage to find good data and do the repair. However, readpage wouldn't read anything in the case where the page is up to date in the cache. So, we simply take that good data we have and call repair_io_failure directly (unless the page in the cache is dirty). Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-08-05 00:11:04 +08:00
fixup->logical, page,
Btrfs: modify repair_io_failure and make it suit direct io The original code of repair_io_failure was just used for buffered read, because it got some filesystem data from page structure, it is safe for the page in the page cache. But when we do a direct read, the pages in bio are not in the page cache, that is there is no filesystem data in the page structure. In order to implement direct read data repair, we need modify repair_io_failure and pass all filesystem data it need by function parameters. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-12 18:44:00 +08:00
offset - page_offset(page),
btrfs: integrating raid-repair and scrub-fixup-nodatasum This ties nodatasum fixup in scrub together with raid repair patches. While both series are working fine alone, scrub will report uncorrectable errors if they occur in a nodatasum extent *and* the page is in the page cache. Previously, we would have triggered readpage to find good data and do the repair. However, readpage wouldn't read anything in the case where the page is up to date in the cache. So, we simply take that good data we have and call repair_io_failure directly (unless the page in the cache is dirty). Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-08-05 00:11:04 +08:00
fixup->mirror_num);
unlock_page(page);
corrected = !ret;
} else {
/*
* we need to get good data first. the general readpage path
* will call repair_io_failure for us, we just have to make
* sure we read the bad mirror.
*/
ret = set_extent_bits(&BTRFS_I(inode)->io_tree, offset, end,
EXTENT_DAMAGED, GFP_NOFS);
if (ret) {
/* set_extent_bits should give proper error */
WARN_ON(ret > 0);
if (ret > 0)
ret = -EFAULT;
goto out;
}
ret = extent_read_full_page(&BTRFS_I(inode)->io_tree, page,
btrfs_get_extent,
fixup->mirror_num);
wait_on_page_locked(page);
corrected = !test_range_bit(&BTRFS_I(inode)->io_tree, offset,
end, EXTENT_DAMAGED, 0, NULL);
if (!corrected)
clear_extent_bits(&BTRFS_I(inode)->io_tree, offset, end,
EXTENT_DAMAGED, GFP_NOFS);
}
out:
if (page)
put_page(page);
btrfs: Remove unnecessary check for NULL iput() already checks for the inode being NULL, thus it's unnecessary to check before calling. Signed-off-by: Tobias Klauser <tklauser@distanz.ch> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-25 20:58:05 +08:00
iput(inode);
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
if (ret < 0)
return ret;
if (ret == 0 && corrected) {
/*
* we only need to call readpage for one of the inodes belonging
* to this extent. so make iterate_extent_inodes stop
*/
return 1;
}
return -EIO;
}
static void scrub_fixup_nodatasum(struct btrfs_work *work)
{
int ret;
struct scrub_fixup_nodatasum *fixup;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
struct btrfs_trans_handle *trans = NULL;
struct btrfs_path *path;
int uncorrectable = 0;
fixup = container_of(work, struct scrub_fixup_nodatasum, work);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sctx = fixup->sctx;
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
path = btrfs_alloc_path();
if (!path) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
++sctx->stat.malloc_errors;
spin_unlock(&sctx->stat_lock);
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
uncorrectable = 1;
goto out;
}
trans = btrfs_join_transaction(fixup->root);
if (IS_ERR(trans)) {
uncorrectable = 1;
goto out;
}
/*
* the idea is to trigger a regular read through the standard path. we
* read a page from the (failed) logical address by specifying the
* corresponding copynum of the failed sector. thus, that readpage is
* expected to fail.
* that is the point where on-the-fly error correction will kick in
* (once it's finished) and rewrite the failed sector if a good copy
* can be found.
*/
ret = iterate_inodes_from_logical(fixup->logical, fixup->root->fs_info,
path, scrub_fixup_readpage,
fixup);
if (ret < 0) {
uncorrectable = 1;
goto out;
}
WARN_ON(ret != 1);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
++sctx->stat.corrected_errors;
spin_unlock(&sctx->stat_lock);
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
out:
if (trans && !IS_ERR(trans))
btrfs_end_transaction(trans, fixup->root);
if (uncorrectable) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
++sctx->stat.uncorrectable_errors;
spin_unlock(&sctx->stat_lock);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
btrfs_dev_replace_stats_inc(
&sctx->dev_root->fs_info->dev_replace.
num_uncorrectable_read_errors);
btrfs: switch message printers to ratelimited _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 16:43:10 +08:00
btrfs_err_rl_in_rcu(sctx->dev_root->fs_info,
"unable to fixup (nodatasum) error at logical %llu on dev %s",
Btrfs: Remove superfluous casts from u64 to unsigned long long u64 is "unsigned long long" on all architectures now, so there's no need to cast it when formatting it using the "ll" length modifier. Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-08-20 19:20:07 +08:00
fixup->logical, rcu_str_deref(fixup->dev->name));
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
}
btrfs_free_path(path);
kfree(fixup);
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
scrub_pending_trans_workers_dec(sctx);
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
}
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
static inline void scrub_get_recover(struct scrub_recover *recover)
{
atomic_inc(&recover->refs);
}
static inline void scrub_put_recover(struct scrub_recover *recover)
{
if (atomic_dec_and_test(&recover->refs)) {
Btrfs: add ref_count and free function for btrfs_bio 1: ref_count is simple than current RBIO_HOLD_BBIO_MAP_BIT flag to keep btrfs_bio's memory in raid56 recovery implement. 2: free function for bbio will make code clean and flexible, plus forced data type checking in compile. Changelog v1->v2: Rename following by David Sterba's suggestion: put_btrfs_bio() -> btrfs_put_bio() get_btrfs_bio() -> btrfs_get_bio() bbio->ref_count -> bbio->refs Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:34 +08:00
btrfs_put_bbio(recover->bbio);
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
kfree(recover);
}
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
* scrub_handle_errored_block gets called when either verification of the
* pages failed or the bio failed to read, e.g. with EIO. In the latter
* case, this function handles all pages in the bio, even though only one
* may be bad.
* The goal of this function is to repair the errored block by using the
* contents of one of the mirrors.
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
*/
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx = sblock_to_check->sctx;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_device *dev;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
struct btrfs_fs_info *fs_info;
u64 length;
u64 logical;
unsigned int failed_mirror_index;
unsigned int is_metadata;
unsigned int have_csum;
struct scrub_block *sblocks_for_recheck; /* holds one for each mirror */
struct scrub_block *sblock_bad;
int ret;
int mirror_index;
int page_num;
int success;
btrfs scrub: print paths of corrupted files While scrubbing, we may encounter various errors. Previously, a logical address was printed to the log only. Now, all paths belonging to that address are resolved and printed separately. That should work for hardlinks as well as reflinks. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:59:12 +08:00
static DEFINE_RATELIMIT_STATE(_rs, DEFAULT_RATELIMIT_INTERVAL,
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
DEFAULT_RATELIMIT_BURST);
BUG_ON(sblock_to_check->page_count < 1);
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
fs_info = sctx->dev_root->fs_info;
Btrfs: fix BUG() in scrub when first superblock reading gives EIO This fixes a very special case that can be reproduced by just disconnecting a disk at runtime, and without unmounting the filesystem first, start scrub on the filesystem with the disconnected disk. All read and write EIOs are handled correctly, only the first superblock is an exception and gives a BUG() in a subfunction. The BUG() is correct, it would crash later otherwise. The subfunction must not be called for superblocks and this is what the fix changes. Reported-by: Joeri Vanthienen <mail@joerivanthienen.be> Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-15 02:57:29 +08:00
if (sblock_to_check->pagev[0]->flags & BTRFS_EXTENT_FLAG_SUPER) {
/*
* if we find an error in a super block, we just report it.
* They will get written with the next transaction commit
* anyway
*/
spin_lock(&sctx->stat_lock);
++sctx->stat.super_errors;
spin_unlock(&sctx->stat_lock);
return 0;
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
length = sblock_to_check->page_count * PAGE_SIZE;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
logical = sblock_to_check->pagev[0]->logical;
BUG_ON(sblock_to_check->pagev[0]->mirror_num < 1);
failed_mirror_index = sblock_to_check->pagev[0]->mirror_num - 1;
is_metadata = !(sblock_to_check->pagev[0]->flags &
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
BTRFS_EXTENT_FLAG_DATA);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
have_csum = sblock_to_check->pagev[0]->have_csum;
dev = sblock_to_check->pagev[0]->dev;
btrfs scrub: added unverified_errors In normal operation, scrub is reading data sequentially in large portions. In case of an i/o error, we try to find the corrupted area(s) by issuing page sized read requests. With this commit we increment the unverified_errors counter if all of the small size requests succeed. Userland patches carrying such conspicous events to the administrator should already be around. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:56:13 +08:00
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (sctx->is_dev_replace && !is_metadata && !have_csum) {
sblocks_for_recheck = NULL;
goto nodatasum_case;
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/*
* read all mirrors one after the other. This includes to
* re-read the extent or metadata block that failed (that was
* the cause that this fixup code is called) another time,
* page by page this time in order to know which pages
* caused I/O errors and which ones are good (for all mirrors).
* It is the goal to handle the situation when more than one
* mirror contains I/O errors, but the errors do not
* overlap, i.e. the data can be repaired by selecting the
* pages from those mirrors without I/O error on the
* particular pages. One example (with blocks >= 2 * PAGE_SIZE)
* would be that mirror #1 has an I/O error on the first page,
* the second page is good, and mirror #2 has an I/O error on
* the second page, but the first page is good.
* Then the first page of the first mirror can be repaired by
* taking the first page of the second mirror, and the
* second page of the second mirror can be repaired by
* copying the contents of the 2nd page of the 1st mirror.
* One more note: if the pages of one mirror contain I/O
* errors, the checksum cannot be verified. In order to get
* the best data for repairing, the first attempt is to find
* a mirror without I/O errors and with a validated checksum.
* Only if this is not possible, the pages are picked from
* mirrors with I/O errors without considering the checksum.
* If the latter is the case, at the end, the checksum of the
* repaired area is verified in order to correctly maintain
* the statistics.
*/
btrfs: cleanup, use kmalloc_array/kcalloc array helpers Convert kmalloc(nr * size, ..) to kmalloc_array that does additional overflow checks, the zeroing variant is kcalloc. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-02-21 01:00:26 +08:00
sblocks_for_recheck = kcalloc(BTRFS_MAX_MIRRORS,
sizeof(*sblocks_for_recheck), GFP_NOFS);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (!sblocks_for_recheck) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
sctx->stat.read_errors++;
sctx->stat.uncorrectable_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_READ_ERRS);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
goto out;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/* setup the context, map the logical blocks and alloc the pages */
Btrfs: Simplify scrub_setup_recheck_block()'s argument scrub_setup_recheck_block() have many arguments but most of them can be get from one of them, we can remove them to make code clean. Some other cleanup for that function also included in this patch. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:42 +08:00
ret = scrub_setup_recheck_block(sblock_to_check, sblocks_for_recheck);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (ret) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.read_errors++;
sctx->stat.uncorrectable_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_READ_ERRS);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
goto out;
}
BUG_ON(failed_mirror_index >= BTRFS_MAX_MIRRORS);
sblock_bad = sblocks_for_recheck + failed_mirror_index;
btrfs scrub: added unverified_errors In normal operation, scrub is reading data sequentially in large portions. In case of an i/o error, we try to find the corrupted area(s) by issuing page sized read requests. With this commit we increment the unverified_errors counter if all of the small size requests succeed. Userland patches carrying such conspicous events to the administrator should already be around. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:56:13 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/* build and submit the bios for the failed mirror, check checksums */
btrfs: Reduce unnecessary arguments in scrub_recheck_block We don't need pass so many arguments for recheck sblock now, this patch cleans them. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:32:06 +08:00
scrub_recheck_block(fs_info, sblock_bad, 1);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (!sblock_bad->header_error && !sblock_bad->checksum_error &&
sblock_bad->no_io_error_seen) {
/*
* the error disappeared after reading page by page, or
* the area was part of a huge bio and other parts of the
* bio caused I/O errors, or the block layer merged several
* read requests into one and the error is caused by a
* different bio (usually one of the two latter cases is
* the cause)
*/
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.unverified_errors++;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
sblock_to_check->data_corrected = 1;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_unlock(&sctx->stat_lock);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (sctx->is_dev_replace)
scrub_write_block_to_dev_replace(sblock_bad);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
goto out;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (!sblock_bad->no_io_error_seen) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.read_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (__ratelimit(&_rs))
scrub_print_warning("i/o error", sblock_to_check);
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_READ_ERRS);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
} else if (sblock_bad->checksum_error) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.csum_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (__ratelimit(&_rs))
scrub_print_warning("checksum error", sblock_to_check);
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
btrfs_dev_stat_inc_and_print(dev,
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
BTRFS_DEV_STAT_CORRUPTION_ERRS);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
} else if (sblock_bad->header_error) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.verify_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (__ratelimit(&_rs))
scrub_print_warning("checksum/header error",
sblock_to_check);
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
if (sblock_bad->generation_error)
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
btrfs_dev_stat_inc_and_print(dev,
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
BTRFS_DEV_STAT_GENERATION_ERRS);
else
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
btrfs_dev_stat_inc_and_print(dev,
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
BTRFS_DEV_STAT_CORRUPTION_ERRS);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: do not inc uncorrectable_errors counter on ro scrubs Currently if we discover an error when scrubbing in ro mode we a) blindly increment the uncorrectable_errors counter, and b) spam the dmesg with the 'unable to fixup (regular) error at ...' message, even though a) we haven't tried to determine if the error is correctable or not, and b) we haven't tried to fixup anything. Fix this. Cc: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-11-04 01:06:38 +08:00
if (sctx->readonly) {
ASSERT(!sctx->is_dev_replace);
goto out;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (!is_metadata && !have_csum) {
struct scrub_fixup_nodatasum *fixup_nodatasum;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
WARN_ON(sctx->is_dev_replace);
Btrfs: Fix a jump typo of nodatasum_case to avoid wrong WARN_ON() if (sctx->is_dev_replace && !is_metadata && !have_csum) { ... goto nodatasum_case; } ... nodatasum_case: WARN_ON(sctx->is_dev_replace); In above code, nodatasum_case marker should be moved after WARN_ON(). Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:35 +08:00
nodatasum_case:
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/*
* !is_metadata and !have_csum, this means that the data
* might not be COW'ed, that it might be modified
* concurrently. The general strategy to work on the
* commit root does not help in the case when COW is not
* used.
*/
fixup_nodatasum = kzalloc(sizeof(*fixup_nodatasum), GFP_NOFS);
if (!fixup_nodatasum)
goto did_not_correct_error;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
fixup_nodatasum->sctx = sctx;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
fixup_nodatasum->dev = dev;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
fixup_nodatasum->logical = logical;
fixup_nodatasum->root = fs_info->extent_root;
fixup_nodatasum->mirror_num = failed_mirror_index + 1;
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
scrub_pending_trans_workers_inc(sctx);
Btrfs: fix task hang under heavy compressed write This has been reported and discussed for a long time, and this hang occurs in both 3.15 and 3.16. Btrfs now migrates to use kernel workqueue, but it introduces this hang problem. Btrfs has a kind of work queued as an ordered way, which means that its ordered_func() must be processed in the way of FIFO, so it usually looks like -- normal_work_helper(arg) work = container_of(arg, struct btrfs_work, normal_work); work->func() <---- (we name it work X) for ordered_work in wq->ordered_list ordered_work->ordered_func() ordered_work->ordered_free() The hang is a rare case, first when we find free space, we get an uncached block group, then we go to read its free space cache inode for free space information, so it will file a readahead request btrfs_readpages() for page that is not in page cache __do_readpage() submit_extent_page() btrfs_submit_bio_hook() btrfs_bio_wq_end_io() submit_bio() end_workqueue_bio() <--(ret by the 1st endio) queue a work(named work Y) for the 2nd also the real endio() So the hang occurs when work Y's work_struct and work X's work_struct happens to share the same address. A bit more explanation, A,B,C -- struct btrfs_work arg -- struct work_struct kthread: worker_thread() pick up a work_struct from @worklist process_one_work(arg) worker->current_work = arg; <-- arg is A->normal_work worker->current_func(arg) normal_work_helper(arg) A = container_of(arg, struct btrfs_work, normal_work); A->func() A->ordered_func() A->ordered_free() <-- A gets freed B->ordered_func() submit_compressed_extents() find_free_extent() load_free_space_inode() ... <-- (the above readhead stack) end_workqueue_bio() btrfs_queue_work(work C) B->ordered_free() As if work A has a high priority in wq->ordered_list and there are more ordered works queued after it, such as B->ordered_func(), its memory could have been freed before normal_work_helper() returns, which means that kernel workqueue code worker_thread() still has worker->current_work pointer to be work A->normal_work's, ie. arg's address. Meanwhile, work C is allocated after work A is freed, work C->normal_work and work A->normal_work are likely to share the same address(I confirmed this with ftrace output, so I'm not just guessing, it's rare though). When another kthread picks up work C->normal_work to process, and finds our kthread is processing it(see find_worker_executing_work()), it'll think work C as a collision and skip then, which ends up nobody processing work C. So the situation is that our kthread is waiting forever on work C. Besides, there're other cases that can lead to deadlock, but the real problem is that all btrfs workqueue shares one work->func, -- normal_work_helper, so this makes each workqueue to have its own helper function, but only a wraper pf normal_work_helper. With this patch, I no long hit the above hang. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-08-15 23:36:53 +08:00
btrfs_init_work(&fixup_nodatasum->work, btrfs_scrub_helper,
scrub_fixup_nodatasum, NULL, NULL);
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
btrfs_queue_work(fs_info->scrub_workers,
&fixup_nodatasum->work);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
goto out;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/*
* now build and submit the bios for the other mirrors, check
Btrfs: in scrub repair code, optimize the reading of mirrors In case that disk blocks need to be repaired (rewritten), the current code at first (for simplicity reasons) reads all alternate mirrors in the first step, afterwards selects the best one in a second step. This is now changed to read one alternate mirror after the other and to leave the loop early when a perfect mirror is found. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:14:21 +08:00
* checksums.
* First try to pick the mirror which is completely without I/O
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
* errors and also does not have a checksum error.
* If one is found, and if a checksum is present, the full block
* that is known to contain an error is rewritten. Afterwards
* the block is known to be corrected.
* If a mirror is found which is completely correct, and no
* checksum is present, only those pages are rewritten that had
* an I/O error in the block to be repaired, since it cannot be
* determined, which copy of the other pages is better (and it
* could happen otherwise that a correct page would be
* overwritten by a bad one).
*/
for (mirror_index = 0;
mirror_index < BTRFS_MAX_MIRRORS &&
sblocks_for_recheck[mirror_index].page_count > 0;
mirror_index++) {
Btrfs: in scrub repair code, optimize the reading of mirrors In case that disk blocks need to be repaired (rewritten), the current code at first (for simplicity reasons) reads all alternate mirrors in the first step, afterwards selects the best one in a second step. This is now changed to read one alternate mirror after the other and to leave the loop early when a perfect mirror is found. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:14:21 +08:00
struct scrub_block *sblock_other;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
Btrfs: in scrub repair code, optimize the reading of mirrors In case that disk blocks need to be repaired (rewritten), the current code at first (for simplicity reasons) reads all alternate mirrors in the first step, afterwards selects the best one in a second step. This is now changed to read one alternate mirror after the other and to leave the loop early when a perfect mirror is found. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:14:21 +08:00
if (mirror_index == failed_mirror_index)
continue;
sblock_other = sblocks_for_recheck + mirror_index;
/* build and submit the bios, check checksums */
btrfs: Reduce unnecessary arguments in scrub_recheck_block We don't need pass so many arguments for recheck sblock now, this patch cleans them. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:32:06 +08:00
scrub_recheck_block(fs_info, sblock_other, 0);
Btrfs: in scrub repair code, simplify alloc error handling In the scrub repair code, the code is changed to handle memory allocation errors a little bit smarter. The change is to handle it just like a read error. This simplifies the code and removes a couple of lines of code, since the code to handle read errors is there anyway. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:16:26 +08:00
if (!sblock_other->header_error &&
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
!sblock_other->checksum_error &&
sblock_other->no_io_error_seen) {
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (sctx->is_dev_replace) {
scrub_write_block_to_dev_replace(sblock_other);
Btrfs: Remove noneed force_write in scrub_write_block_to_dev_replace It is always 1 in this place, because !1 case was already jumped out in previous code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:36 +08:00
goto corrected_error;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
} else {
ret = scrub_repair_block_from_good_copy(
Btrfs: Remove noneed force_write in scrub_write_block_to_dev_replace It is always 1 in this place, because !1 case was already jumped out in previous code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:36 +08:00
sblock_bad, sblock_other);
if (!ret)
goto corrected_error;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: Combine per-page recover in dev-replace and scrub The code are similar, combine them to make code clean and easy to maintenance. Some lost condition are also completed with benefit of this combination. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:41 +08:00
if (sblock_bad->no_io_error_seen && !sctx->is_dev_replace)
goto did_not_correct_error;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
/*
* In case of I/O errors in the area that is supposed to be
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
* repaired, continue by picking good copies of those pages.
* Select the good pages from mirrors to rewrite bad pages from
* the area to fix. Afterwards verify the checksum of the block
* that is supposed to be repaired. This verification step is
* only done for the purpose of statistic counting and for the
* final scrub report, whether errors remain.
* A perfect algorithm could make use of the checksum and try
* all possible combinations of pages from the different mirrors
* until the checksum verification succeeds. For example, when
* the 2nd page of mirror #1 faces I/O errors, and the 2nd page
* of mirror #2 is readable but the final checksum test fails,
* then the 2nd page of mirror #3 could be tried, whether now
* the final checksum succeedes. But this would be a rare
* exception and is therefore not implemented. At least it is
* avoided that the good copy is overwritten.
* A more useful improvement would be to pick the sectors
* without I/O error based on sector sizes (512 bytes on legacy
* disks) instead of on PAGE_SIZE. Then maybe 512 byte of one
* mirror could be repaired by taking 512 byte of a different
* mirror, even if other 512 byte sectors in the same PAGE_SIZE
* area are unreadable.
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
*/
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
success = 1;
Btrfs: Combine per-page recover in dev-replace and scrub The code are similar, combine them to make code clean and easy to maintenance. Some lost condition are also completed with benefit of this combination. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:41 +08:00
for (page_num = 0; page_num < sblock_bad->page_count;
page_num++) {
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
struct scrub_page *page_bad = sblock_bad->pagev[page_num];
Btrfs: Combine per-page recover in dev-replace and scrub The code are similar, combine them to make code clean and easy to maintenance. Some lost condition are also completed with benefit of this combination. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:41 +08:00
struct scrub_block *sblock_other = NULL;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
Btrfs: Combine per-page recover in dev-replace and scrub The code are similar, combine them to make code clean and easy to maintenance. Some lost condition are also completed with benefit of this combination. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:41 +08:00
/* skip no-io-error page in scrub */
if (!page_bad->io_error && !sctx->is_dev_replace)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
continue;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
Btrfs: Combine per-page recover in dev-replace and scrub The code are similar, combine them to make code clean and easy to maintenance. Some lost condition are also completed with benefit of this combination. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:41 +08:00
/* try to find no-io-error page in mirrors */
if (page_bad->io_error) {
for (mirror_index = 0;
mirror_index < BTRFS_MAX_MIRRORS &&
sblocks_for_recheck[mirror_index].page_count > 0;
mirror_index++) {
if (!sblocks_for_recheck[mirror_index].
pagev[page_num]->io_error) {
sblock_other = sblocks_for_recheck +
mirror_index;
break;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
}
Btrfs: Combine per-page recover in dev-replace and scrub The code are similar, combine them to make code clean and easy to maintenance. Some lost condition are also completed with benefit of this combination. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:41 +08:00
if (!sblock_other)
success = 0;
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: Combine per-page recover in dev-replace and scrub The code are similar, combine them to make code clean and easy to maintenance. Some lost condition are also completed with benefit of this combination. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:41 +08:00
if (sctx->is_dev_replace) {
/*
* did not find a mirror to fetch the page
* from. scrub_write_page_to_dev_replace()
* handles this case (page->io_error), by
* filling the block with zeros before
* submitting the write request
*/
if (!sblock_other)
sblock_other = sblock_bad;
if (scrub_write_page_to_dev_replace(sblock_other,
page_num) != 0) {
btrfs_dev_replace_stats_inc(
&sctx->dev_root->
fs_info->dev_replace.
num_write_errors);
success = 0;
}
} else if (sblock_other) {
ret = scrub_repair_page_from_good_copy(sblock_bad,
sblock_other,
page_num, 0);
if (0 == ret)
page_bad->io_error = 0;
else
success = 0;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: Combine per-page recover in dev-replace and scrub The code are similar, combine them to make code clean and easy to maintenance. Some lost condition are also completed with benefit of this combination. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:41 +08:00
if (success && !sctx->is_dev_replace) {
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (is_metadata || have_csum) {
/*
* need to verify the checksum now that all
* sectors on disk are repaired (the write
* request for data to be repaired is on its way).
* Just be lazy and use scrub_recheck_block()
* which re-reads the data before the checksum
* is verified, but most likely the data comes out
* of the page cache.
*/
btrfs: Reduce unnecessary arguments in scrub_recheck_block We don't need pass so many arguments for recheck sblock now, this patch cleans them. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:32:06 +08:00
scrub_recheck_block(fs_info, sblock_bad, 1);
Btrfs: in scrub repair code, simplify alloc error handling In the scrub repair code, the code is changed to handle memory allocation errors a little bit smarter. The change is to handle it just like a read error. This simplifies the code and removes a couple of lines of code, since the code to handle read errors is there anyway. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:16:26 +08:00
if (!sblock_bad->header_error &&
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
!sblock_bad->checksum_error &&
sblock_bad->no_io_error_seen)
goto corrected_error;
else
goto did_not_correct_error;
} else {
corrected_error:
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.corrected_errors++;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
sblock_to_check->data_corrected = 1;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_unlock(&sctx->stat_lock);
btrfs: switch message printers to ratelimited _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 16:43:10 +08:00
btrfs_err_rl_in_rcu(fs_info,
"fixed up error at logical %llu on dev %s",
Btrfs: Remove superfluous casts from u64 to unsigned long long u64 is "unsigned long long" on all architectures now, so there's no need to cast it when formatting it using the "ll" length modifier. Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-08-20 19:20:07 +08:00
logical, rcu_str_deref(dev->name));
btrfs: add readonly flag setting the readonly flag prevents writes in case an error is detected Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-23 23:34:19 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
} else {
did_not_correct_error:
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.uncorrectable_errors++;
spin_unlock(&sctx->stat_lock);
btrfs: switch message printers to ratelimited _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 16:43:10 +08:00
btrfs_err_rl_in_rcu(fs_info,
"unable to fixup (regular) error at logical %llu on dev %s",
Btrfs: Remove superfluous casts from u64 to unsigned long long u64 is "unsigned long long" on all architectures now, so there's no need to cast it when formatting it using the "ll" length modifier. Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-08-20 19:20:07 +08:00
logical, rcu_str_deref(dev->name));
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
out:
if (sblocks_for_recheck) {
for (mirror_index = 0; mirror_index < BTRFS_MAX_MIRRORS;
mirror_index++) {
struct scrub_block *sblock = sblocks_for_recheck +
mirror_index;
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
struct scrub_recover *recover;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
int page_index;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
for (page_index = 0; page_index < sblock->page_count;
page_index++) {
sblock->pagev[page_index]->sblock = NULL;
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
recover = sblock->pagev[page_index]->recover;
if (recover) {
scrub_put_recover(recover);
sblock->pagev[page_index]->recover =
NULL;
}
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
scrub_page_put(sblock->pagev[page_index]);
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
kfree(sblocks_for_recheck);
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return 0;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: Make raid_map array be inlined in btrfs_bio structure It can make code more simple and clear, we need not care about free bbio and raid_map together. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:33 +08:00
static inline int scrub_nr_raid_mirrors(struct btrfs_bio *bbio)
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
{
Btrfs: Include map_type in raid_bio Corrent code use many kinds of "clever" way to determine operation target's raid type, as: raid_map != NULL or raid_map[MAX_NR] == RAID[56]_Q_STRIPE To make code easy to maintenance, this patch put raid type into bbio, and we can always get raid type from bbio with a "stupid" way. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:43 +08:00
if (bbio->map_type & BTRFS_BLOCK_GROUP_RAID5)
return 2;
else if (bbio->map_type & BTRFS_BLOCK_GROUP_RAID6)
return 3;
else
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
return (int)bbio->num_stripes;
}
Btrfs: Include map_type in raid_bio Corrent code use many kinds of "clever" way to determine operation target's raid type, as: raid_map != NULL or raid_map[MAX_NR] == RAID[56]_Q_STRIPE To make code easy to maintenance, this patch put raid type into bbio, and we can always get raid type from bbio with a "stupid" way. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:43 +08:00
static inline void scrub_stripe_index_and_offset(u64 logical, u64 map_type,
u64 *raid_map,
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
u64 mapped_length,
int nstripes, int mirror,
int *stripe_index,
u64 *stripe_offset)
{
int i;
Btrfs: Introduce BTRFS_BLOCK_GROUP_RAID56_MASK to check raid56 simply So we can check raid56 with: (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) instead of long: (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:44 +08:00
if (map_type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
/* RAID5/6 */
for (i = 0; i < nstripes; i++) {
if (raid_map[i] == RAID6_Q_STRIPE ||
raid_map[i] == RAID5_P_STRIPE)
continue;
if (logical >= raid_map[i] &&
logical < raid_map[i] + mapped_length)
break;
}
*stripe_index = i;
*stripe_offset = logical - raid_map[i];
} else {
/* The other RAID type */
*stripe_index = mirror;
*stripe_offset = 0;
}
}
Btrfs: Simplify scrub_setup_recheck_block()'s argument scrub_setup_recheck_block() have many arguments but most of them can be get from one of them, we can remove them to make code clean. Some other cleanup for that function also included in this patch. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:42 +08:00
static int scrub_setup_recheck_block(struct scrub_block *original_sblock,
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
struct scrub_block *sblocks_for_recheck)
{
Btrfs: Simplify scrub_setup_recheck_block()'s argument scrub_setup_recheck_block() have many arguments but most of them can be get from one of them, we can remove them to make code clean. Some other cleanup for that function also included in this patch. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:42 +08:00
struct scrub_ctx *sctx = original_sblock->sctx;
struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
u64 length = original_sblock->page_count * PAGE_SIZE;
u64 logical = original_sblock->pagev[0]->logical;
btrfs: scrub: setup all fields for sblock_to_check scrub_setup_recheck_block() isn't setup all necessary fields for sblock_to_check because history reason. So current code need more arguments in severial functions, and more local variables, just to passing these lacked values to necessary place. This patch setup above fields to sblock_to_check in scrub_setup_recheck_block(), for: 1: more cleanup for function arg, local variable 2: to make sblock_to_check complete, then we can use sblock_to_check without concern about some uninitialized member. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-19 22:39:18 +08:00
u64 generation = original_sblock->pagev[0]->generation;
u64 flags = original_sblock->pagev[0]->flags;
u64 have_csum = original_sblock->pagev[0]->have_csum;
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
struct scrub_recover *recover;
struct btrfs_bio *bbio;
u64 sublen;
u64 mapped_length;
u64 stripe_offset;
int stripe_index;
Btrfs: Simplify scrub_setup_recheck_block()'s argument scrub_setup_recheck_block() have many arguments but most of them can be get from one of them, we can remove them to make code clean. Some other cleanup for that function also included in this patch. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:42 +08:00
int page_index = 0;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
int mirror_index;
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
int nmirrors;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
int ret;
/*
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
* note: the two members refs and outstanding_pages
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
* are not used (and not set) in the blocks that are used for
* the recheck procedure
*/
while (length > 0) {
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
sublen = min_t(u64, length, PAGE_SIZE);
mapped_length = sublen;
bbio = NULL;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/*
* with a length of PAGE_SIZE, each returned stripe
* represents one mirror
*/
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
ret = btrfs_map_sblock(fs_info, REQ_GET_READ_MIRRORS, logical,
Btrfs: Make raid_map array be inlined in btrfs_bio structure It can make code more simple and clear, we need not care about free bbio and raid_map together. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:33 +08:00
&mapped_length, &bbio, 0, 1);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (ret || !bbio || mapped_length < sublen) {
Btrfs: add ref_count and free function for btrfs_bio 1: ref_count is simple than current RBIO_HOLD_BBIO_MAP_BIT flag to keep btrfs_bio's memory in raid56 recovery implement. 2: free function for bbio will make code clean and flexible, plus forced data type checking in compile. Changelog v1->v2: Rename following by David Sterba's suggestion: put_btrfs_bio() -> btrfs_put_bio() get_btrfs_bio() -> btrfs_get_bio() bbio->ref_count -> bbio->refs Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:34 +08:00
btrfs_put_bbio(bbio);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return -EIO;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
recover = kzalloc(sizeof(struct scrub_recover), GFP_NOFS);
if (!recover) {
Btrfs: add ref_count and free function for btrfs_bio 1: ref_count is simple than current RBIO_HOLD_BBIO_MAP_BIT flag to keep btrfs_bio's memory in raid56 recovery implement. 2: free function for bbio will make code clean and flexible, plus forced data type checking in compile. Changelog v1->v2: Rename following by David Sterba's suggestion: put_btrfs_bio() -> btrfs_put_bio() get_btrfs_bio() -> btrfs_get_bio() bbio->ref_count -> bbio->refs Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:34 +08:00
btrfs_put_bbio(bbio);
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
return -ENOMEM;
}
atomic_set(&recover->refs, 1);
recover->bbio = bbio;
recover->map_length = mapped_length;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
BUG_ON(page_index >= SCRUB_PAGES_PER_RD_BIO);
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
Btrfs: Simplify scrub_setup_recheck_block()'s argument scrub_setup_recheck_block() have many arguments but most of them can be get from one of them, we can remove them to make code clean. Some other cleanup for that function also included in this patch. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:42 +08:00
nmirrors = min(scrub_nr_raid_mirrors(bbio), BTRFS_MAX_MIRRORS);
Btrfs: Include map_type in raid_bio Corrent code use many kinds of "clever" way to determine operation target's raid type, as: raid_map != NULL or raid_map[MAX_NR] == RAID[56]_Q_STRIPE To make code easy to maintenance, this patch put raid type into bbio, and we can always get raid type from bbio with a "stupid" way. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:43 +08:00
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
for (mirror_index = 0; mirror_index < nmirrors;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
mirror_index++) {
struct scrub_block *sblock;
struct scrub_page *page;
sblock = sblocks_for_recheck + mirror_index;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
sblock->sctx = sctx;
btrfs: scrub: setup all fields for sblock_to_check scrub_setup_recheck_block() isn't setup all necessary fields for sblock_to_check because history reason. So current code need more arguments in severial functions, and more local variables, just to passing these lacked values to necessary place. This patch setup above fields to sblock_to_check in scrub_setup_recheck_block(), for: 1: more cleanup for function arg, local variable 2: to make sblock_to_check complete, then we can use sblock_to_check without concern about some uninitialized member. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-19 22:39:18 +08:00
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
page = kzalloc(sizeof(*page), GFP_NOFS);
if (!page) {
leave_nomem:
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
scrub_put_recover(recover);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return -ENOMEM;
}
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
scrub_page_get(page);
sblock->pagev[page_index] = page;
btrfs: scrub: setup all fields for sblock_to_check scrub_setup_recheck_block() isn't setup all necessary fields for sblock_to_check because history reason. So current code need more arguments in severial functions, and more local variables, just to passing these lacked values to necessary place. This patch setup above fields to sblock_to_check in scrub_setup_recheck_block(), for: 1: more cleanup for function arg, local variable 2: to make sblock_to_check complete, then we can use sblock_to_check without concern about some uninitialized member. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-19 22:39:18 +08:00
page->sblock = sblock;
page->flags = flags;
page->generation = generation;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
page->logical = logical;
btrfs: scrub: setup all fields for sblock_to_check scrub_setup_recheck_block() isn't setup all necessary fields for sblock_to_check because history reason. So current code need more arguments in severial functions, and more local variables, just to passing these lacked values to necessary place. This patch setup above fields to sblock_to_check in scrub_setup_recheck_block(), for: 1: more cleanup for function arg, local variable 2: to make sblock_to_check complete, then we can use sblock_to_check without concern about some uninitialized member. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-19 22:39:18 +08:00
page->have_csum = have_csum;
if (have_csum)
memcpy(page->csum,
original_sblock->pagev[0]->csum,
sctx->csum_size);
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
Btrfs: Include map_type in raid_bio Corrent code use many kinds of "clever" way to determine operation target's raid type, as: raid_map != NULL or raid_map[MAX_NR] == RAID[56]_Q_STRIPE To make code easy to maintenance, this patch put raid type into bbio, and we can always get raid type from bbio with a "stupid" way. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:43 +08:00
scrub_stripe_index_and_offset(logical,
bbio->map_type,
bbio->raid_map,
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
mapped_length,
Btrfs: fix a out-of-bound access of raid_map We add the number of stripes on target devices into bbio->num_stripes if we are under device replacement, and we just sort the raid_map of those stripes that not on the target devices, so if when we need real raid_map, we need skip the stripes on the target devices. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:31 +08:00
bbio->num_stripes -
bbio->num_tgtdevs,
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
mirror_index,
&stripe_index,
&stripe_offset);
page->physical = bbio->stripes[stripe_index].physical +
stripe_offset;
page->dev = bbio->stripes[stripe_index].dev;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
BUG_ON(page_index >= original_sblock->page_count);
page->physical_for_dev_replace =
original_sblock->pagev[page_index]->
physical_for_dev_replace;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
/* for missing devices, dev->bdev is NULL */
page->mirror_num = mirror_index + 1;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
sblock->page_count++;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
page->page = alloc_page(GFP_NOFS);
if (!page->page)
goto leave_nomem;
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
scrub_get_recover(recover);
page->recover = recover;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
scrub_put_recover(recover);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
length -= sublen;
logical += sublen;
page_index++;
}
return 0;
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
}
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
struct scrub_bio_ret {
struct completion event;
int error;
};
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
static void scrub_bio_wait_endio(struct bio *bio)
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
{
struct scrub_bio_ret *ret = bio->bi_private;
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
ret->error = bio->bi_error;
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
complete(&ret->event);
}
static inline int scrub_is_page_on_raid56(struct scrub_page *page)
{
Btrfs: Include map_type in raid_bio Corrent code use many kinds of "clever" way to determine operation target's raid type, as: raid_map != NULL or raid_map[MAX_NR] == RAID[56]_Q_STRIPE To make code easy to maintenance, this patch put raid type into bbio, and we can always get raid type from bbio with a "stupid" way. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:43 +08:00
return page->recover &&
Btrfs: Introduce BTRFS_BLOCK_GROUP_RAID56_MASK to check raid56 simply So we can check raid56 with: (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) instead of long: (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:44 +08:00
(page->recover->bbio->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK);
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
}
static int scrub_submit_raid56_bio_wait(struct btrfs_fs_info *fs_info,
struct bio *bio,
struct scrub_page *page)
{
struct scrub_bio_ret done;
int ret;
init_completion(&done.event);
done.error = 0;
bio->bi_iter.bi_sector = page->logical >> 9;
bio->bi_private = &done;
bio->bi_end_io = scrub_bio_wait_endio;
ret = raid56_parity_recover(fs_info->fs_root, bio, page->recover->bbio,
page->recover->map_length,
Btrfs, raid56: fix use-after-free problem in the final device replace procedure on raid56 The commit c404e0dc (Btrfs: fix use-after-free in the finishing procedure of the device replace) fixed a use-after-free problem which happened when removing the source device at the end of device replace, but at that time, btrfs didn't support device replace on raid56, so we didn't fix the problem on the raid56 profile. Currently, we implemented device replace for raid56, so we need kick that problem out before we enable that function for raid56. The fix method is very simple, we just increase the bio per-cpu counter before we submit a raid56 io, and decrease the counter when the raid56 io ends. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-25 16:39:28 +08:00
page->mirror_num, 0);
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
if (ret)
return ret;
wait_for_completion(&done.event);
if (done.error)
return -EIO;
return 0;
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/*
* this function will check the on disk data for checksum errors, header
* errors and read I/O errors. If any I/O errors happen, the exact pages
* which are errored are marked as being bad. The goal is to enable scrub
* to take those pages that are not errored from all the mirrors so that
* the pages that are errored in the just handled mirror can be repaired.
*/
Btrfs: in scrub repair code, simplify alloc error handling In the scrub repair code, the code is changed to handle memory allocation errors a little bit smarter. The change is to handle it just like a read error. This simplifies the code and removes a couple of lines of code, since the code to handle read errors is there anyway. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:16:26 +08:00
static void scrub_recheck_block(struct btrfs_fs_info *fs_info,
btrfs: Reduce unnecessary arguments in scrub_recheck_block We don't need pass so many arguments for recheck sblock now, this patch cleans them. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:32:06 +08:00
struct scrub_block *sblock,
int retry_failed_mirror)
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
{
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
int page_num;
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
sblock->no_io_error_seen = 1;
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
for (page_num = 0; page_num < sblock->page_count; page_num++) {
struct bio *bio;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
struct scrub_page *page = sblock->pagev[page_num];
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
if (page->dev->bdev == NULL) {
Btrfs: fix crash in scrub repair code when device is missing Fix that when scrub tries to repair an I/O or checksum error and one of the devices containing the mirror is missing, it crashes in bio_add_page because the bdev is a NULL pointer for missing devices. Reported-by: Marco L. Crociani <marco.crociani@gmail.com> Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-05-05 03:16:07 +08:00
page->io_error = 1;
sblock->no_io_error_seen = 0;
continue;
}
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
WARN_ON(!page->page);
Btrfs: use a btrfs bioset instead of abusing bio internals Btrfs has been pointer tagging bi_private and using bi_bdev to store the stripe index and mirror number of failed IOs. As bios bubble back up through the call chain, we use these to decide if and how to retry our IOs. They are also used to count IO failures on a per device basis. Recently a bio tracepoint was added lead to crashes because we were abusing bi_bdev. This commit adds a btrfs bioset, and creates explicit fields for the mirror number and stripe index. The plan is to extend this structure for all of the fields currently in struct btrfs_bio, which will mean one less kmalloc in our IO path. Signed-off-by: Chris Mason <chris.mason@fusionio.com> Reported-by: Tejun Heo <tj@kernel.org>
2013-05-18 06:30:14 +08:00
bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
Btrfs: in scrub repair code, simplify alloc error handling In the scrub repair code, the code is changed to handle memory allocation errors a little bit smarter. The change is to handle it just like a read error. This simplifies the code and removes a couple of lines of code, since the code to handle read errors is there anyway. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:16:26 +08:00
if (!bio) {
page->io_error = 1;
sblock->no_io_error_seen = 0;
continue;
}
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
bio->bi_bdev = page->dev->bdev;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
Btrfs: in scrub repair code, simplify alloc error handling In the scrub repair code, the code is changed to handle memory allocation errors a little bit smarter. The change is to handle it just like a read error. This simplifies the code and removes a couple of lines of code, since the code to handle read errors is there anyway. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:16:26 +08:00
bio_add_page(bio, page->page, PAGE_SIZE, 0);
Btrfs, scrub: repair the common data on RAID5/6 if it is corrupted This patch implement the RAID5/6 common data repair function, the implementation is similar to the scrub on the other RAID such as RAID1, the differentia is that we don't read the data from the mirror, we use the data repair function of RAID5/6. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-10-23 14:42:50 +08:00
if (!retry_failed_mirror && scrub_is_page_on_raid56(page)) {
if (scrub_submit_raid56_bio_wait(fs_info, bio, page))
sblock->no_io_error_seen = 0;
} else {
bio->bi_iter.bi_sector = page->physical >> 9;
if (btrfsic_submit_bio_wait(READ, bio))
sblock->no_io_error_seen = 0;
}
block: submit_bio_wait() conversions It was being open coded in a few places. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Neil Brown <neilb@suse.de> Cc: Chris Mason <chris.mason@fusionio.com> Acked-by: NeilBrown <neilb@suse.de>
2013-11-24 14:33:32 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
bio_put(bio);
}
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (sblock->no_io_error_seen)
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
scrub_recheck_block_checksum(sblock);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: fix wrong fsid check of scrub All the metadata in the seed devices has the same fsid as the fsid of the seed filesystem which is on the seed device, so we should check them by the current filesystem. Fix it. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-24 11:37:08 +08:00
static inline int scrub_check_fsid(u8 fsid[],
struct scrub_page *spage)
{
struct btrfs_fs_devices *fs_devices = spage->dev->fs_devices;
int ret;
ret = memcmp(fsid, fs_devices->fsid, BTRFS_UUID_SIZE);
return !ret;
}
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
static void scrub_recheck_block_checksum(struct scrub_block *sblock)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
sblock->header_error = 0;
sblock->checksum_error = 0;
sblock->generation_error = 0;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
if (sblock->pagev[0]->flags & BTRFS_EXTENT_FLAG_DATA)
scrub_checksum_data(sblock);
else
scrub_checksum_tree_block(sblock);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_repair_block_from_good_copy(struct scrub_block *sblock_bad,
Btrfs: Remove noneed force_write in scrub_write_block_to_dev_replace It is always 1 in this place, because !1 case was already jumped out in previous code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:36 +08:00
struct scrub_block *sblock_good)
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
{
int page_num;
int ret = 0;
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
for (page_num = 0; page_num < sblock_bad->page_count; page_num++) {
int ret_sub;
btrfs scrub: make fixups sync btrfs scrub - make fixups sync, don't reuse fixup bios Fixups are already sync for csum failures, this patch makes them sync for EIO case as well. Fixups are now sharing pages with the parent sbio - instead of allocating a separate page to do a fixup we grab the page from the sbio buffer. Fixup bios are no longer reused. struct fixup is no longer needed, instead pass [sbio pointer, index]. Originally this was added to look at the possibility of sharing the code between drive swap and scrub, but it actually fixes a serious bug in scrub code where errors that could be corrected were ignored and reported as uncorrectable. btrfs scrub - restore bios properly after media errors The current code reallocates a bio after a media error. This is a temporary measure introduced in v3 after a serious problem related to bio reuse was found in v2 of scrub patchset. Basically we did not reset bv_offset and bv_len fields of the bio_vec structure. They are changed in case I/O error happens, for example, at offset 512 or 1024 into the page. Also bi_flags field wasn't properly setup before reusing the bio. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-04-09 19:27:01 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
ret_sub = scrub_repair_page_from_good_copy(sblock_bad,
sblock_good,
Btrfs: Remove noneed force_write in scrub_write_block_to_dev_replace It is always 1 in this place, because !1 case was already jumped out in previous code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:36 +08:00
page_num, 1);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (ret_sub)
ret = ret_sub;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return ret;
}
static int scrub_repair_page_from_good_copy(struct scrub_block *sblock_bad,
struct scrub_block *sblock_good,
int page_num, int force_write)
{
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
struct scrub_page *page_bad = sblock_bad->pagev[page_num];
struct scrub_page *page_good = sblock_good->pagev[page_num];
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
BUG_ON(page_bad->page == NULL);
BUG_ON(page_good->page == NULL);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (force_write || sblock_bad->header_error ||
sblock_bad->checksum_error || page_bad->io_error) {
struct bio *bio;
int ret;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (!page_bad->dev->bdev) {
btrfs: switch message printers to ratelimited variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 17:01:36 +08:00
btrfs_warn_rl(sblock_bad->sctx->dev_root->fs_info,
Btrfs: convert printk to btrfs_ and fix BTRFS prefix Convert all applicable cases of printk and pr_* to the btrfs_* macros. Fix all uses of the BTRFS prefix. Signed-off-by: Frank Holton <fholton@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-21 00:37:06 +08:00
"scrub_repair_page_from_good_copy(bdev == NULL) "
btrfs: switch message printers to ratelimited variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 17:01:36 +08:00
"is unexpected");
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
return -EIO;
}
Btrfs: use a btrfs bioset instead of abusing bio internals Btrfs has been pointer tagging bi_private and using bi_bdev to store the stripe index and mirror number of failed IOs. As bios bubble back up through the call chain, we use these to decide if and how to retry our IOs. They are also used to count IO failures on a per device basis. Recently a bio tracepoint was added lead to crashes because we were abusing bi_bdev. This commit adds a btrfs bioset, and creates explicit fields for the mirror number and stripe index. The plan is to extend this structure for all of the fields currently in struct btrfs_bio, which will mean one less kmalloc in our IO path. Signed-off-by: Chris Mason <chris.mason@fusionio.com> Reported-by: Tejun Heo <tj@kernel.org>
2013-05-18 06:30:14 +08:00
bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
Btrfs: check return value of bio_alloc() properly bio_alloc() has the possibility of returning NULL. So, it is necessary to check the return value. Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-04-13 04:03:56 +08:00
if (!bio)
return -EIO;
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
bio->bi_bdev = page_bad->dev->bdev;
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 06:44:27 +08:00
bio->bi_iter.bi_sector = page_bad->physical >> 9;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
ret = bio_add_page(bio, page_good->page, PAGE_SIZE, 0);
if (PAGE_SIZE != ret) {
bio_put(bio);
return -EIO;
btrfs scrub: added unverified_errors In normal operation, scrub is reading data sequentially in large portions. In case of an i/o error, we try to find the corrupted area(s) by issuing page sized read requests. With this commit we increment the unverified_errors counter if all of the small size requests succeed. Userland patches carrying such conspicous events to the administrator should already be around. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:56:13 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
block: submit_bio_wait() conversions It was being open coded in a few places. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Neil Brown <neilb@suse.de> Cc: Chris Mason <chris.mason@fusionio.com> Acked-by: NeilBrown <neilb@suse.de>
2013-11-24 14:33:32 +08:00
if (btrfsic_submit_bio_wait(WRITE, bio)) {
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
btrfs_dev_stat_inc_and_print(page_bad->dev,
BTRFS_DEV_STAT_WRITE_ERRS);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
btrfs_dev_replace_stats_inc(
&sblock_bad->sctx->dev_root->fs_info->
dev_replace.num_write_errors);
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
bio_put(bio);
return -EIO;
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
bio_put(bio);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return 0;
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static void scrub_write_block_to_dev_replace(struct scrub_block *sblock)
{
int page_num;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
/*
* This block is used for the check of the parity on the source device,
* so the data needn't be written into the destination device.
*/
if (sblock->sparity)
return;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
for (page_num = 0; page_num < sblock->page_count; page_num++) {
int ret;
ret = scrub_write_page_to_dev_replace(sblock, page_num);
if (ret)
btrfs_dev_replace_stats_inc(
&sblock->sctx->dev_root->fs_info->dev_replace.
num_write_errors);
}
}
static int scrub_write_page_to_dev_replace(struct scrub_block *sblock,
int page_num)
{
struct scrub_page *spage = sblock->pagev[page_num];
BUG_ON(spage->page == NULL);
if (spage->io_error) {
void *mapped_buffer = kmap_atomic(spage->page);
memset(mapped_buffer, 0, PAGE_CACHE_SIZE);
flush_dcache_page(spage->page);
kunmap_atomic(mapped_buffer);
}
return scrub_add_page_to_wr_bio(sblock->sctx, spage);
}
static int scrub_add_page_to_wr_bio(struct scrub_ctx *sctx,
struct scrub_page *spage)
{
struct scrub_wr_ctx *wr_ctx = &sctx->wr_ctx;
struct scrub_bio *sbio;
int ret;
mutex_lock(&wr_ctx->wr_lock);
again:
if (!wr_ctx->wr_curr_bio) {
wr_ctx->wr_curr_bio = kzalloc(sizeof(*wr_ctx->wr_curr_bio),
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
GFP_KERNEL);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (!wr_ctx->wr_curr_bio) {
mutex_unlock(&wr_ctx->wr_lock);
return -ENOMEM;
}
wr_ctx->wr_curr_bio->sctx = sctx;
wr_ctx->wr_curr_bio->page_count = 0;
}
sbio = wr_ctx->wr_curr_bio;
if (sbio->page_count == 0) {
struct bio *bio;
sbio->physical = spage->physical_for_dev_replace;
sbio->logical = spage->logical;
sbio->dev = wr_ctx->tgtdev;
bio = sbio->bio;
if (!bio) {
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
bio = btrfs_io_bio_alloc(GFP_KERNEL,
wr_ctx->pages_per_wr_bio);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (!bio) {
mutex_unlock(&wr_ctx->wr_lock);
return -ENOMEM;
}
sbio->bio = bio;
}
bio->bi_private = sbio;
bio->bi_end_io = scrub_wr_bio_end_io;
bio->bi_bdev = sbio->dev->bdev;
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 06:44:27 +08:00
bio->bi_iter.bi_sector = sbio->physical >> 9;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
sbio->err = 0;
} else if (sbio->physical + sbio->page_count * PAGE_SIZE !=
spage->physical_for_dev_replace ||
sbio->logical + sbio->page_count * PAGE_SIZE !=
spage->logical) {
scrub_wr_submit(sctx);
goto again;
}
ret = bio_add_page(sbio->bio, spage->page, PAGE_SIZE, 0);
if (ret != PAGE_SIZE) {
if (sbio->page_count < 1) {
bio_put(sbio->bio);
sbio->bio = NULL;
mutex_unlock(&wr_ctx->wr_lock);
return -EIO;
}
scrub_wr_submit(sctx);
goto again;
}
sbio->pagev[sbio->page_count] = spage;
scrub_page_get(spage);
sbio->page_count++;
if (sbio->page_count == wr_ctx->pages_per_wr_bio)
scrub_wr_submit(sctx);
mutex_unlock(&wr_ctx->wr_lock);
return 0;
}
static void scrub_wr_submit(struct scrub_ctx *sctx)
{
struct scrub_wr_ctx *wr_ctx = &sctx->wr_ctx;
struct scrub_bio *sbio;
if (!wr_ctx->wr_curr_bio)
return;
sbio = wr_ctx->wr_curr_bio;
wr_ctx->wr_curr_bio = NULL;
WARN_ON(!sbio->bio->bi_bdev);
scrub_pending_bio_inc(sctx);
/* process all writes in a single worker thread. Then the block layer
* orders the requests before sending them to the driver which
* doubled the write performance on spinning disks when measured
* with Linux 3.5 */
btrfsic_submit_bio(WRITE, sbio->bio);
}
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
static void scrub_wr_bio_end_io(struct bio *bio)
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
{
struct scrub_bio *sbio = bio->bi_private;
struct btrfs_fs_info *fs_info = sbio->dev->dev_root->fs_info;
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
sbio->err = bio->bi_error;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
sbio->bio = bio;
Btrfs: fix task hang under heavy compressed write This has been reported and discussed for a long time, and this hang occurs in both 3.15 and 3.16. Btrfs now migrates to use kernel workqueue, but it introduces this hang problem. Btrfs has a kind of work queued as an ordered way, which means that its ordered_func() must be processed in the way of FIFO, so it usually looks like -- normal_work_helper(arg) work = container_of(arg, struct btrfs_work, normal_work); work->func() <---- (we name it work X) for ordered_work in wq->ordered_list ordered_work->ordered_func() ordered_work->ordered_free() The hang is a rare case, first when we find free space, we get an uncached block group, then we go to read its free space cache inode for free space information, so it will file a readahead request btrfs_readpages() for page that is not in page cache __do_readpage() submit_extent_page() btrfs_submit_bio_hook() btrfs_bio_wq_end_io() submit_bio() end_workqueue_bio() <--(ret by the 1st endio) queue a work(named work Y) for the 2nd also the real endio() So the hang occurs when work Y's work_struct and work X's work_struct happens to share the same address. A bit more explanation, A,B,C -- struct btrfs_work arg -- struct work_struct kthread: worker_thread() pick up a work_struct from @worklist process_one_work(arg) worker->current_work = arg; <-- arg is A->normal_work worker->current_func(arg) normal_work_helper(arg) A = container_of(arg, struct btrfs_work, normal_work); A->func() A->ordered_func() A->ordered_free() <-- A gets freed B->ordered_func() submit_compressed_extents() find_free_extent() load_free_space_inode() ... <-- (the above readhead stack) end_workqueue_bio() btrfs_queue_work(work C) B->ordered_free() As if work A has a high priority in wq->ordered_list and there are more ordered works queued after it, such as B->ordered_func(), its memory could have been freed before normal_work_helper() returns, which means that kernel workqueue code worker_thread() still has worker->current_work pointer to be work A->normal_work's, ie. arg's address. Meanwhile, work C is allocated after work A is freed, work C->normal_work and work A->normal_work are likely to share the same address(I confirmed this with ftrace output, so I'm not just guessing, it's rare though). When another kthread picks up work C->normal_work to process, and finds our kthread is processing it(see find_worker_executing_work()), it'll think work C as a collision and skip then, which ends up nobody processing work C. So the situation is that our kthread is waiting forever on work C. Besides, there're other cases that can lead to deadlock, but the real problem is that all btrfs workqueue shares one work->func, -- normal_work_helper, so this makes each workqueue to have its own helper function, but only a wraper pf normal_work_helper. With this patch, I no long hit the above hang. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-08-15 23:36:53 +08:00
btrfs_init_work(&sbio->work, btrfs_scrubwrc_helper,
scrub_wr_bio_end_io_worker, NULL, NULL);
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
btrfs_queue_work(fs_info->scrub_wr_completion_workers, &sbio->work);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
}
static void scrub_wr_bio_end_io_worker(struct btrfs_work *work)
{
struct scrub_bio *sbio = container_of(work, struct scrub_bio, work);
struct scrub_ctx *sctx = sbio->sctx;
int i;
WARN_ON(sbio->page_count > SCRUB_PAGES_PER_WR_BIO);
if (sbio->err) {
struct btrfs_dev_replace *dev_replace =
&sbio->sctx->dev_root->fs_info->dev_replace;
for (i = 0; i < sbio->page_count; i++) {
struct scrub_page *spage = sbio->pagev[i];
spage->io_error = 1;
btrfs_dev_replace_stats_inc(&dev_replace->
num_write_errors);
}
}
for (i = 0; i < sbio->page_count; i++)
scrub_page_put(sbio->pagev[i]);
bio_put(sbio->bio);
kfree(sbio);
scrub_pending_bio_dec(sctx);
}
static int scrub_checksum(struct scrub_block *sblock)
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
{
u64 flags;
int ret;
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
/*
* No need to initialize these stats currently,
* because this function only use return value
* instead of these stats value.
*
* Todo:
* always use stats
*/
sblock->header_error = 0;
sblock->generation_error = 0;
sblock->checksum_error = 0;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
WARN_ON(sblock->page_count < 1);
flags = sblock->pagev[0]->flags;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
ret = 0;
if (flags & BTRFS_EXTENT_FLAG_DATA)
ret = scrub_checksum_data(sblock);
else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK)
ret = scrub_checksum_tree_block(sblock);
else if (flags & BTRFS_EXTENT_FLAG_SUPER)
(void)scrub_checksum_super(sblock);
else
WARN_ON(1);
if (ret)
scrub_handle_errored_block(sblock);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
return ret;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_checksum_data(struct scrub_block *sblock)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx = sblock->sctx;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u8 csum[BTRFS_CSUM_SIZE];
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u8 *on_disk_csum;
struct page *page;
void *buffer;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u32 crc = ~(u32)0;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u64 len;
int index;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
BUG_ON(sblock->page_count < 1);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
if (!sblock->pagev[0]->have_csum)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return 0;
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
on_disk_csum = sblock->pagev[0]->csum;
page = sblock->pagev[0]->page;
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
buffer = kmap_atomic(page);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
len = sctx->sectorsize;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
index = 0;
for (;;) {
u64 l = min_t(u64, len, PAGE_SIZE);
Btrfs: cleanup unused arguments of btrfs_csum_data Argument 'root' is no more used in btrfs_csum_data(). Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-03-14 22:57:45 +08:00
crc = btrfs_csum_data(buffer, crc, l);
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
kunmap_atomic(buffer);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
len -= l;
if (len == 0)
break;
index++;
BUG_ON(index >= sblock->page_count);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
BUG_ON(!sblock->pagev[index]->page);
page = sblock->pagev[index]->page;
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
buffer = kmap_atomic(page);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs_csum_final(crc, csum);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
if (memcmp(csum, on_disk_csum, sctx->csum_size))
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
sblock->checksum_error = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
return sblock->checksum_error;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_checksum_tree_block(struct scrub_block *sblock)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx = sblock->sctx;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct btrfs_header *h;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_root *root = sctx->dev_root;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct btrfs_fs_info *fs_info = root->fs_info;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u8 calculated_csum[BTRFS_CSUM_SIZE];
u8 on_disk_csum[BTRFS_CSUM_SIZE];
struct page *page;
void *mapped_buffer;
u64 mapped_size;
void *p;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u32 crc = ~(u32)0;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u64 len;
int index;
BUG_ON(sblock->page_count < 1);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
page = sblock->pagev[0]->page;
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
mapped_buffer = kmap_atomic(page);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
h = (struct btrfs_header *)mapped_buffer;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
memcpy(on_disk_csum, h->csum, sctx->csum_size);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
* we don't use the getter functions here, as we
* a) don't have an extent buffer and
* b) the page is already kmapped
*/
btrfs: Cleanup for using BTRFS_SETGET_STACK instead of raw convert Some codes still use the cpu_to_lexx instead of the BTRFS_SETGET_STACK_FUNCS declared in ctree.h. Also added some BTRFS_SETGET_STACK_FUNCS for btrfs_header btrfs_timespec and other structures. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Reviewed-by: Miao Xie <miaoxie@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-07-16 11:19:18 +08:00
if (sblock->pagev[0]->logical != btrfs_stack_header_bytenr(h))
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
sblock->header_error = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
if (sblock->pagev[0]->generation != btrfs_stack_header_generation(h)) {
sblock->header_error = 1;
sblock->generation_error = 1;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: fix wrong fsid check of scrub All the metadata in the seed devices has the same fsid as the fsid of the seed filesystem which is on the seed device, so we should check them by the current filesystem. Fix it. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-24 11:37:08 +08:00
if (!scrub_check_fsid(h->fsid, sblock->pagev[0]))
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
sblock->header_error = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (memcmp(h->chunk_tree_uuid, fs_info->chunk_tree_uuid,
BTRFS_UUID_SIZE))
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
sblock->header_error = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
len = sctx->nodesize - BTRFS_CSUM_SIZE;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE;
p = ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE;
index = 0;
for (;;) {
u64 l = min_t(u64, len, mapped_size);
Btrfs: cleanup unused arguments of btrfs_csum_data Argument 'root' is no more used in btrfs_csum_data(). Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-03-14 22:57:45 +08:00
crc = btrfs_csum_data(p, crc, l);
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
kunmap_atomic(mapped_buffer);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
len -= l;
if (len == 0)
break;
index++;
BUG_ON(index >= sblock->page_count);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
BUG_ON(!sblock->pagev[index]->page);
page = sblock->pagev[index]->page;
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
mapped_buffer = kmap_atomic(page);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
mapped_size = PAGE_SIZE;
p = mapped_buffer;
}
btrfs_csum_final(crc, calculated_csum);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
if (memcmp(calculated_csum, on_disk_csum, sctx->csum_size))
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
sblock->checksum_error = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
return sblock->header_error || sblock->checksum_error;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static int scrub_checksum_super(struct scrub_block *sblock)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
struct btrfs_super_block *s;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx = sblock->sctx;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u8 calculated_csum[BTRFS_CSUM_SIZE];
u8 on_disk_csum[BTRFS_CSUM_SIZE];
struct page *page;
void *mapped_buffer;
u64 mapped_size;
void *p;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u32 crc = ~(u32)0;
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
int fail_gen = 0;
int fail_cor = 0;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u64 len;
int index;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
BUG_ON(sblock->page_count < 1);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
page = sblock->pagev[0]->page;
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
mapped_buffer = kmap_atomic(page);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
s = (struct btrfs_super_block *)mapped_buffer;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
memcpy(on_disk_csum, s->csum, sctx->csum_size);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: Cleanup for using BTRFS_SETGET_STACK instead of raw convert Some codes still use the cpu_to_lexx instead of the BTRFS_SETGET_STACK_FUNCS declared in ctree.h. Also added some BTRFS_SETGET_STACK_FUNCS for btrfs_header btrfs_timespec and other structures. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Reviewed-by: Miao Xie <miaoxie@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-07-16 11:19:18 +08:00
if (sblock->pagev[0]->logical != btrfs_super_bytenr(s))
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
++fail_cor;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: Cleanup for using BTRFS_SETGET_STACK instead of raw convert Some codes still use the cpu_to_lexx instead of the BTRFS_SETGET_STACK_FUNCS declared in ctree.h. Also added some BTRFS_SETGET_STACK_FUNCS for btrfs_header btrfs_timespec and other structures. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Reviewed-by: Miao Xie <miaoxie@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-07-16 11:19:18 +08:00
if (sblock->pagev[0]->generation != btrfs_super_generation(s))
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
++fail_gen;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: fix wrong fsid check of scrub All the metadata in the seed devices has the same fsid as the fsid of the seed filesystem which is on the seed device, so we should check them by the current filesystem. Fix it. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-24 11:37:08 +08:00
if (!scrub_check_fsid(s->fsid, sblock->pagev[0]))
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
++fail_cor;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
len = BTRFS_SUPER_INFO_SIZE - BTRFS_CSUM_SIZE;
mapped_size = PAGE_SIZE - BTRFS_CSUM_SIZE;
p = ((u8 *)mapped_buffer) + BTRFS_CSUM_SIZE;
index = 0;
for (;;) {
u64 l = min_t(u64, len, mapped_size);
Btrfs: cleanup unused arguments of btrfs_csum_data Argument 'root' is no more used in btrfs_csum_data(). Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-03-14 22:57:45 +08:00
crc = btrfs_csum_data(p, crc, l);
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
kunmap_atomic(mapped_buffer);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
len -= l;
if (len == 0)
break;
index++;
BUG_ON(index >= sblock->page_count);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
BUG_ON(!sblock->pagev[index]->page);
page = sblock->pagev[index]->page;
Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs fixes and features from Chris Mason: "We've merged in the error handling patches from SuSE. These are already shipping in the sles kernel, and they give btrfs the ability to abort transactions and go readonly on errors. It involves a lot of churn as they clarify BUG_ONs, and remove the ones we now properly deal with. Josef reworked the way our metadata interacts with the page cache. page->private now points to the btrfs extent_buffer object, which makes everything faster. He changed it so we write an whole extent buffer at a time instead of allowing individual pages to go down,, which will be important for the raid5/6 code (for the 3.5 merge window ;) Josef also made us more aggressive about dropping pages for metadata blocks that were freed due to COW. Overall, our metadata caching is much faster now. We've integrated my patch for metadata bigger than the page size. This allows metadata blocks up to 64KB in size. In practice 16K and 32K seem to work best. For workloads with lots of metadata, this cuts down the size of the extent allocation tree dramatically and fragments much less. Scrub was updated to support the larger block sizes, which ended up being a fairly large change (thanks Stefan Behrens). We also have an assortment of fixes and updates, especially to the balancing code (Ilya Dryomov), the back ref walker (Jan Schmidt) and the defragging code (Liu Bo)." Fixed up trivial conflicts in fs/btrfs/scrub.c that were just due to removal of the second argument to k[un]map_atomic() in commit 7ac687d9e047. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (75 commits) Btrfs: update the checks for mixed block groups with big metadata blocks Btrfs: update to the right index of defragment Btrfs: do not bother to defrag an extent if it is a big real extent Btrfs: add a check to decide if we should defrag the range Btrfs: fix recursive defragment with autodefrag option Btrfs: fix the mismatch of page->mapping Btrfs: fix race between direct io and autodefrag Btrfs: fix deadlock during allocating chunks Btrfs: show useful info in space reservation tracepoint Btrfs: don't use crc items bigger than 4KB Btrfs: flush out and clean up any block device pages during mount btrfs: disallow unequal data/metadata blocksize for mixed block groups Btrfs: enhance superblock sanity checks Btrfs: change scrub to support big blocks Btrfs: minor cleanup in scrub Btrfs: introduce common define for max number of mirrors Btrfs: fix infinite loop in btrfs_shrink_device() Btrfs: fix memory leak in resolver code Btrfs: allow dup for data chunks in mixed mode Btrfs: validate target profiles only if we are going to use them ...
2012-03-31 03:44:29 +08:00
mapped_buffer = kmap_atomic(page);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
mapped_size = PAGE_SIZE;
p = mapped_buffer;
}
btrfs_csum_final(crc, calculated_csum);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
if (memcmp(calculated_csum, on_disk_csum, sctx->csum_size))
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
++fail_cor;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
if (fail_cor + fail_gen) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
* if we find an error in a super block, we just report it.
* They will get written with the next transaction commit
* anyway
*/
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
++sctx->stat.super_errors;
spin_unlock(&sctx->stat_lock);
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
if (fail_cor)
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
btrfs_dev_stat_inc_and_print(sblock->pagev[0]->dev,
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
BTRFS_DEV_STAT_CORRUPTION_ERRS);
else
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
btrfs_dev_stat_inc_and_print(sblock->pagev[0]->dev,
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
BTRFS_DEV_STAT_GENERATION_ERRS);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: add device counters for detected IO and checksum errors The goal is to detect when drives start to get an increased error rate, when drives should be replaced soon. Therefore statistic counters are added that count IO errors (read, write and flush). Additionally, the software detected errors like checksum errors and corrupted blocks are counted. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de>
2012-05-25 22:06:08 +08:00
return fail_cor + fail_gen;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static void scrub_block_get(struct scrub_block *sblock)
{
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_inc(&sblock->refs);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
static void scrub_block_put(struct scrub_block *sblock)
{
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
if (atomic_dec_and_test(&sblock->refs)) {
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
int i;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (sblock->sparity)
scrub_parity_put(sblock->sparity);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
for (i = 0; i < sblock->page_count; i++)
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
scrub_page_put(sblock->pagev[i]);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
kfree(sblock);
}
}
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
static void scrub_page_get(struct scrub_page *spage)
{
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_inc(&spage->refs);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
}
static void scrub_page_put(struct scrub_page *spage)
{
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
if (atomic_dec_and_test(&spage->refs)) {
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
if (spage->page)
__free_page(spage->page);
kfree(spage);
}
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
static void scrub_submit(struct scrub_ctx *sctx)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
struct scrub_bio *sbio;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
if (sctx->curr == -1)
Btrfs: minor cleanup in scrub Just a minor cleanup commit in preparation for the big block changes. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:26 +08:00
return;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sbio = sctx->bios[sctx->curr];
sctx->curr = -1;
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
scrub_pending_bio_inc(sctx);
Btrfs: remove misleading handling of missing device scrub scrub_submit() claims that it can handle a bio with a NULL block device, but this is misleading, as calling bio_add_page() on a bio with a NULL ->bi_bdev would've already crashed. Delete this, as we're about to properly handle a missing block device. Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:48 +08:00
btrfsic_submit_bio(READ, sbio->bio);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static int scrub_add_page_to_rd_bio(struct scrub_ctx *sctx,
struct scrub_page *spage)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
struct scrub_block *sblock = spage->sblock;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct scrub_bio *sbio;
Btrfs: handle bio_add_page failure gracefully in scrub Currently scrub fails with ENOMEM when bio_add_page fails. Unfortunately dm based targets accept only one page per bio, thus making scrub always fails. This patch just submits the current bio when an error is encountered and starts a new one. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-11-11 21:17:10 +08:00
int ret;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
again:
/*
* grab a fresh bio or wait for one to become available
*/
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
while (sctx->curr == -1) {
spin_lock(&sctx->list_lock);
sctx->curr = sctx->first_free;
if (sctx->curr != -1) {
sctx->first_free = sctx->bios[sctx->curr]->next_free;
sctx->bios[sctx->curr]->next_free = -1;
sctx->bios[sctx->curr]->page_count = 0;
spin_unlock(&sctx->list_lock);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
} else {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_unlock(&sctx->list_lock);
wait_event(sctx->list_wait, sctx->first_free != -1);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sbio = sctx->bios[sctx->curr];
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (sbio->page_count == 0) {
Btrfs: handle bio_add_page failure gracefully in scrub Currently scrub fails with ENOMEM when bio_add_page fails. Unfortunately dm based targets accept only one page per bio, thus making scrub always fails. This patch just submits the current bio when an error is encountered and starts a new one. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-11-11 21:17:10 +08:00
struct bio *bio;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
sbio->physical = spage->physical;
sbio->logical = spage->logical;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
sbio->dev = spage->dev;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
bio = sbio->bio;
if (!bio) {
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
bio = btrfs_io_bio_alloc(GFP_KERNEL,
sctx->pages_per_rd_bio);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (!bio)
return -ENOMEM;
sbio->bio = bio;
}
Btrfs: handle bio_add_page failure gracefully in scrub Currently scrub fails with ENOMEM when bio_add_page fails. Unfortunately dm based targets accept only one page per bio, thus making scrub always fails. This patch just submits the current bio when an error is encountered and starts a new one. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-11-11 21:17:10 +08:00
bio->bi_private = sbio;
bio->bi_end_io = scrub_bio_end_io;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
bio->bi_bdev = sbio->dev->bdev;
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 06:44:27 +08:00
bio->bi_iter.bi_sector = sbio->physical >> 9;
Btrfs: handle bio_add_page failure gracefully in scrub Currently scrub fails with ENOMEM when bio_add_page fails. Unfortunately dm based targets accept only one page per bio, thus making scrub always fails. This patch just submits the current bio when an error is encountered and starts a new one. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-11-11 21:17:10 +08:00
sbio->err = 0;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
} else if (sbio->physical + sbio->page_count * PAGE_SIZE !=
spage->physical ||
sbio->logical + sbio->page_count * PAGE_SIZE !=
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
spage->logical ||
sbio->dev != spage->dev) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
scrub_submit(sctx);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
goto again;
}
Btrfs: handle bio_add_page failure gracefully in scrub Currently scrub fails with ENOMEM when bio_add_page fails. Unfortunately dm based targets accept only one page per bio, thus making scrub always fails. This patch just submits the current bio when an error is encountered and starts a new one. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-11-11 21:17:10 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
sbio->pagev[sbio->page_count] = spage;
ret = bio_add_page(sbio->bio, spage->page, PAGE_SIZE, 0);
if (ret != PAGE_SIZE) {
if (sbio->page_count < 1) {
bio_put(sbio->bio);
sbio->bio = NULL;
return -EIO;
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
scrub_submit(sctx);
Btrfs: handle bio_add_page failure gracefully in scrub Currently scrub fails with ENOMEM when bio_add_page fails. Unfortunately dm based targets accept only one page per bio, thus making scrub always fails. This patch just submits the current bio when an error is encountered and starts a new one. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-11-11 21:17:10 +08:00
goto again;
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
scrub_block_get(sblock); /* one for the page added to the bio */
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
atomic_inc(&sblock->outstanding_pages);
sbio->page_count++;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (sbio->page_count == sctx->pages_per_rd_bio)
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
scrub_submit(sctx);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return 0;
}
Merge branch 'for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs updates from Chris Mason: "This has Jeff Mahoney's long standing trim patch that fixes corners where trims were missing. Omar has some raid5/6 fixes, especially for using scrub and device replace when devices are missing. Zhao Lie continues cleaning and fixing things, this series fixes some really hard to hit corners in xfstests. I had to pull it last merge window due to some deadlocks, but those are now resolved. I added support for Tejun's new blkio controllers. It seems to work well for single devices, we'll expand to multi-device as well" * 'for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (47 commits) btrfs: fix compile when block cgroups are not enabled Btrfs: fix file read corruption after extent cloning and fsync Btrfs: check if previous transaction aborted to avoid fs corruption btrfs: use __GFP_NOFAIL in alloc_btrfs_bio btrfs: Prevent from early transaction abort btrfs: Remove unused arguments in tree-log.c btrfs: Remove useless condition in start_log_trans() Btrfs: add support for blkio controllers Btrfs: remove unused mutex from struct 'btrfs_fs_info' Btrfs: fix parity scrub of RAID 5/6 with missing device Btrfs: fix device replace of a missing RAID 5/6 device Btrfs: add RAID 5/6 BTRFS_RBIO_REBUILD_MISSING operation Btrfs: count devices correctly in readahead during RAID 5/6 replace Btrfs: remove misleading handling of missing device scrub btrfs: fix clone / extent-same deadlocks Btrfs: fix defrag to merge tail file extent Btrfs: fix warning in backref walking btrfs: Add WARN_ON() for double lock in btrfs_tree_lock() btrfs: Remove root argument in extent_data_ref_count() btrfs: Fix wrong comment of btrfs_alloc_tree_block() ...
2015-09-06 06:14:43 +08:00
static void scrub_missing_raid56_end_io(struct bio *bio)
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
{
struct scrub_block *sblock = bio->bi_private;
struct btrfs_fs_info *fs_info = sblock->sctx->dev_root->fs_info;
Merge branch 'for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs Pull btrfs updates from Chris Mason: "This has Jeff Mahoney's long standing trim patch that fixes corners where trims were missing. Omar has some raid5/6 fixes, especially for using scrub and device replace when devices are missing. Zhao Lie continues cleaning and fixing things, this series fixes some really hard to hit corners in xfstests. I had to pull it last merge window due to some deadlocks, but those are now resolved. I added support for Tejun's new blkio controllers. It seems to work well for single devices, we'll expand to multi-device as well" * 'for-linus-4.3' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (47 commits) btrfs: fix compile when block cgroups are not enabled Btrfs: fix file read corruption after extent cloning and fsync Btrfs: check if previous transaction aborted to avoid fs corruption btrfs: use __GFP_NOFAIL in alloc_btrfs_bio btrfs: Prevent from early transaction abort btrfs: Remove unused arguments in tree-log.c btrfs: Remove useless condition in start_log_trans() Btrfs: add support for blkio controllers Btrfs: remove unused mutex from struct 'btrfs_fs_info' Btrfs: fix parity scrub of RAID 5/6 with missing device Btrfs: fix device replace of a missing RAID 5/6 device Btrfs: add RAID 5/6 BTRFS_RBIO_REBUILD_MISSING operation Btrfs: count devices correctly in readahead during RAID 5/6 replace Btrfs: remove misleading handling of missing device scrub btrfs: fix clone / extent-same deadlocks Btrfs: fix defrag to merge tail file extent Btrfs: fix warning in backref walking btrfs: Add WARN_ON() for double lock in btrfs_tree_lock() btrfs: Remove root argument in extent_data_ref_count() btrfs: Fix wrong comment of btrfs_alloc_tree_block() ...
2015-09-06 06:14:43 +08:00
if (bio->bi_error)
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
sblock->no_io_error_seen = 0;
btrfs_queue_work(fs_info->scrub_workers, &sblock->work);
}
static void scrub_missing_raid56_worker(struct btrfs_work *work)
{
struct scrub_block *sblock = container_of(work, struct scrub_block, work);
struct scrub_ctx *sctx = sblock->sctx;
u64 logical;
struct btrfs_device *dev;
logical = sblock->pagev[0]->logical;
dev = sblock->pagev[0]->dev;
btrfs: Reduce unnecessary arguments in scrub_recheck_block We don't need pass so many arguments for recheck sblock now, this patch cleans them. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:32:06 +08:00
if (sblock->no_io_error_seen)
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
scrub_recheck_block_checksum(sblock);
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
if (!sblock->no_io_error_seen) {
spin_lock(&sctx->stat_lock);
sctx->stat.read_errors++;
spin_unlock(&sctx->stat_lock);
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
btrfs_err_rl_in_rcu(sctx->dev_root->fs_info,
btrfs: switch message printers to ratelimited _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 16:43:10 +08:00
"IO error rebuilding logical %llu for dev %s",
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
logical, rcu_str_deref(dev->name));
} else if (sblock->header_error || sblock->checksum_error) {
spin_lock(&sctx->stat_lock);
sctx->stat.uncorrectable_errors++;
spin_unlock(&sctx->stat_lock);
btrfs: Use scrub_checksum_data and scrub_checksum_tree_block for scrub_recheck_block_checksum We can use existing scrub_checksum_data() and scrub_checksum_tree_block() for scrub_recheck_block_checksum(), instead of write duplicated code. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 21:18:02 +08:00
btrfs_err_rl_in_rcu(sctx->dev_root->fs_info,
btrfs: switch message printers to ratelimited _in_rcu variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 16:43:10 +08:00
"failed to rebuild valid logical %llu for dev %s",
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
logical, rcu_str_deref(dev->name));
} else {
scrub_write_block_to_dev_replace(sblock);
}
scrub_block_put(sblock);
if (sctx->is_dev_replace &&
atomic_read(&sctx->wr_ctx.flush_all_writes)) {
mutex_lock(&sctx->wr_ctx.wr_lock);
scrub_wr_submit(sctx);
mutex_unlock(&sctx->wr_ctx.wr_lock);
}
scrub_pending_bio_dec(sctx);
}
static void scrub_missing_raid56_pages(struct scrub_block *sblock)
{
struct scrub_ctx *sctx = sblock->sctx;
struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
u64 length = sblock->page_count * PAGE_SIZE;
u64 logical = sblock->pagev[0]->logical;
struct btrfs_bio *bbio;
struct bio *bio;
struct btrfs_raid_bio *rbio;
int ret;
int i;
ret = btrfs_map_sblock(fs_info, REQ_GET_READ_MIRRORS, logical, &length,
&bbio, 0, 1);
if (ret || !bbio || !bbio->raid_map)
goto bbio_out;
if (WARN_ON(!sctx->is_dev_replace ||
!(bbio->map_type & BTRFS_BLOCK_GROUP_RAID56_MASK))) {
/*
* We shouldn't be scrubbing a missing device. Even for dev
* replace, we should only get here for RAID 5/6. We either
* managed to mount something with no mirrors remaining or
* there's a bug in scrub_remap_extent()/btrfs_map_block().
*/
goto bbio_out;
}
bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
if (!bio)
goto bbio_out;
bio->bi_iter.bi_sector = logical >> 9;
bio->bi_private = sblock;
bio->bi_end_io = scrub_missing_raid56_end_io;
rbio = raid56_alloc_missing_rbio(sctx->dev_root, bio, bbio, length);
if (!rbio)
goto rbio_out;
for (i = 0; i < sblock->page_count; i++) {
struct scrub_page *spage = sblock->pagev[i];
raid56_add_scrub_pages(rbio, spage->page, spage->logical);
}
btrfs_init_work(&sblock->work, btrfs_scrub_helper,
scrub_missing_raid56_worker, NULL, NULL);
scrub_block_get(sblock);
scrub_pending_bio_inc(sctx);
raid56_submit_missing_rbio(rbio);
return;
rbio_out:
bio_put(bio);
bbio_out:
btrfs_put_bbio(bbio);
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
static int scrub_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
u64 physical, struct btrfs_device *dev, u64 flags,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
u64 gen, int mirror_num, u8 *csum, int force,
u64 physical_for_dev_replace)
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
{
struct scrub_block *sblock;
int index;
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
sblock = kzalloc(sizeof(*sblock), GFP_KERNEL);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (!sblock) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return -ENOMEM;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
/* one ref inside this function, plus one for each page added to
* a bio later on */
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_set(&sblock->refs, 1);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sblock->sctx = sctx;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
sblock->no_io_error_seen = 1;
for (index = 0; len > 0; index++) {
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
struct scrub_page *spage;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u64 l = min_t(u64, len, PAGE_SIZE);
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
spage = kzalloc(sizeof(*spage), GFP_KERNEL);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
if (!spage) {
leave_nomem:
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
scrub_block_put(sblock);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return -ENOMEM;
}
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
BUG_ON(index >= SCRUB_MAX_PAGES_PER_BLOCK);
scrub_page_get(spage);
sblock->pagev[index] = spage;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
spage->sblock = sblock;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
spage->dev = dev;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
spage->flags = flags;
spage->generation = gen;
spage->logical = logical;
spage->physical = physical;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
spage->physical_for_dev_replace = physical_for_dev_replace;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
spage->mirror_num = mirror_num;
if (csum) {
spage->have_csum = 1;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
memcpy(spage->csum, csum, sctx->csum_size);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
} else {
spage->have_csum = 0;
}
sblock->page_count++;
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
spage->page = alloc_page(GFP_KERNEL);
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
if (!spage->page)
goto leave_nomem;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
len -= l;
logical += l;
physical += l;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
physical_for_dev_replace += l;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
WARN_ON(sblock->page_count == 0);
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
if (dev->missing) {
/*
* This case should only be hit for RAID 5/6 device replace. See
* the comment in scrub_missing_raid56_pages() for details.
*/
scrub_missing_raid56_pages(sblock);
} else {
for (index = 0; index < sblock->page_count; index++) {
struct scrub_page *spage = sblock->pagev[index];
int ret;
btrfs: scrub: don't reuse bios and pages The current scrub implementation reuses bios and pages as often as possible, allocating them only on start and releasing them when finished. This leads to more problems with the block layer than it's worth. The elevator gets confused when there are more pages added to the bio than bi_size suggests. This patch completely rips out the reuse of bios and pages and allocates them freshly for each submit. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Maosn <chris.mason@oracle.com>
2011-05-29 03:57:55 +08:00
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
ret = scrub_add_page_to_rd_bio(sctx, spage);
if (ret) {
scrub_block_put(sblock);
return ret;
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: fix device replace of a missing RAID 5/6 device The original implementation of device replace on RAID 5/6 seems to have missed support for replacing a missing device. When this is attempted, we end up calling bio_add_page() on a bio with a NULL ->bi_bdev, which crashes when we try to dereference it. This happens because btrfs_map_block() has no choice but to return us the missing device because RAID 5/6 don't have any alternate mirrors to read from, and a missing device has a NULL bdev. The idea implemented here is to handle the missing device case separately, which better only happen when we're replacing a missing RAID 5/6 device. We use the new BTRFS_RBIO_REBUILD_MISSING operation to reconstruct the data from parity, check it with scrub_recheck_block_checksum(), and write it out with scrub_write_block_to_dev_replace(). Reported-by: Philip <bugzilla@philip-seeger.de> Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=96141 Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:51 +08:00
if (force)
scrub_submit(sctx);
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/* last one frees, either here or in bio completion for last page */
scrub_block_put(sblock);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return 0;
}
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
static void scrub_bio_end_io(struct bio *bio)
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
{
struct scrub_bio *sbio = bio->bi_private;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_fs_info *fs_info = sbio->dev->dev_root->fs_info;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
sbio->err = bio->bi_error;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
sbio->bio = bio;
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
btrfs_queue_work(fs_info->scrub_workers, &sbio->work);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
static void scrub_bio_end_io_worker(struct btrfs_work *work)
{
struct scrub_bio *sbio = container_of(work, struct scrub_bio, work);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx = sbio->sctx;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
int i;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
BUG_ON(sbio->page_count > SCRUB_PAGES_PER_RD_BIO);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
if (sbio->err) {
for (i = 0; i < sbio->page_count; i++) {
struct scrub_page *spage = sbio->pagev[i];
spage->io_error = 1;
spage->sblock->no_io_error_seen = 0;
}
}
/* now complete the scrub_block items that have all pages completed */
for (i = 0; i < sbio->page_count; i++) {
struct scrub_page *spage = sbio->pagev[i];
struct scrub_block *sblock = spage->sblock;
if (atomic_dec_and_test(&sblock->outstanding_pages))
scrub_block_complete(sblock);
scrub_block_put(sblock);
}
bio_put(sbio->bio);
sbio->bio = NULL;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->list_lock);
sbio->next_free = sctx->first_free;
sctx->first_free = sbio->index;
spin_unlock(&sctx->list_lock);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (sctx->is_dev_replace &&
atomic_read(&sctx->wr_ctx.flush_all_writes)) {
mutex_lock(&sctx->wr_ctx.wr_lock);
scrub_wr_submit(sctx);
mutex_unlock(&sctx->wr_ctx.wr_lock);
}
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
scrub_pending_bio_dec(sctx);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
static inline void __scrub_mark_bitmap(struct scrub_parity *sparity,
unsigned long *bitmap,
u64 start, u64 len)
{
btrfs: cleanup, use correct type in div_u64_rem div_u64_rem expects u32 for divisior and reminder. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-02-21 01:42:11 +08:00
u32 offset;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
int nsectors;
int sectorsize = sparity->sctx->dev_root->sectorsize;
if (len >= sparity->stripe_len) {
bitmap_set(bitmap, 0, sparity->nsectors);
return;
}
start -= sparity->logic_start;
btrfs: replace remaining do_div calls with div_u64 variants Switch to div_u64_rem that does type checking and has more obvious semantics than do_div. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-02-21 01:43:47 +08:00
start = div_u64_rem(start, sparity->stripe_len, &offset);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
offset /= sectorsize;
nsectors = (int)len / sectorsize;
if (offset + nsectors <= sparity->nsectors) {
bitmap_set(bitmap, offset, nsectors);
return;
}
bitmap_set(bitmap, offset, sparity->nsectors - offset);
bitmap_set(bitmap, 0, nsectors - (sparity->nsectors - offset));
}
static inline void scrub_parity_mark_sectors_error(struct scrub_parity *sparity,
u64 start, u64 len)
{
__scrub_mark_bitmap(sparity, sparity->ebitmap, start, len);
}
static inline void scrub_parity_mark_sectors_data(struct scrub_parity *sparity,
u64 start, u64 len)
{
__scrub_mark_bitmap(sparity, sparity->dbitmap, start, len);
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
static void scrub_block_complete(struct scrub_block *sblock)
{
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
int corrupted = 0;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (!sblock->no_io_error_seen) {
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
corrupted = 1;
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
scrub_handle_errored_block(sblock);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
} else {
/*
* if has checksum error, write via repair mechanism in
* dev replace case, otherwise write here in dev replace
* case.
*/
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
corrupted = scrub_checksum(sblock);
if (!corrupted && sblock->sctx->is_dev_replace)
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
scrub_write_block_to_dev_replace(sblock);
}
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (sblock->sparity && corrupted && !sblock->data_corrected) {
u64 start = sblock->pagev[0]->logical;
u64 end = sblock->pagev[sblock->page_count - 1]->logical +
PAGE_SIZE;
scrub_parity_mark_sectors_error(sblock->sparity,
start, end - start);
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
btrfs: Remove len argument from scrub_find_csum It is useless. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 22:03:02 +08:00
static int scrub_find_csum(struct scrub_ctx *sctx, u64 logical, u8 *csum)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
struct btrfs_ordered_sum *sum = NULL;
Btrfs: remove btrfs_sector_sum structure Using the structure btrfs_sector_sum to keep the checksum value is unnecessary, because the extents that btrfs_sector_sum points to are continuous, we can find out the expected checksums by btrfs_ordered_sum's bytenr and the offset, so we can remove btrfs_sector_sum's bytenr. After removing bytenr, there is only one member in the structure, so it makes no sense to keep the structure, just remove it, and use a u32 array to store the checksum value. By this change, we don't use the while loop to get the checksums one by one. Now, we can get several checksum value at one time, it improved the performance by ~74% on my SSD (31MB/s -> 54MB/s). test command: # dd if=/dev/zero of=/mnt/btrfs/file0 bs=1M count=1024 oflag=sync Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-19 10:36:09 +08:00
unsigned long index;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
unsigned long num_sectors;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
while (!list_empty(&sctx->csum_list)) {
sum = list_first_entry(&sctx->csum_list,
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct btrfs_ordered_sum, list);
if (sum->bytenr > logical)
return 0;
if (sum->bytenr + sum->len > logical)
break;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
++sctx->stat.csum_discards;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
list_del(&sum->list);
kfree(sum);
sum = NULL;
}
if (!sum)
return 0;
Btrfs: remove btrfs_sector_sum structure Using the structure btrfs_sector_sum to keep the checksum value is unnecessary, because the extents that btrfs_sector_sum points to are continuous, we can find out the expected checksums by btrfs_ordered_sum's bytenr and the offset, so we can remove btrfs_sector_sum's bytenr. After removing bytenr, there is only one member in the structure, so it makes no sense to keep the structure, just remove it, and use a u32 array to store the checksum value. By this change, we don't use the while loop to get the checksums one by one. Now, we can get several checksum value at one time, it improved the performance by ~74% on my SSD (31MB/s -> 54MB/s). test command: # dd if=/dev/zero of=/mnt/btrfs/file0 bs=1M count=1024 oflag=sync Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-19 10:36:09 +08:00
index = ((u32)(logical - sum->bytenr)) / sctx->sectorsize;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
num_sectors = sum->len / sctx->sectorsize;
Btrfs: remove btrfs_sector_sum structure Using the structure btrfs_sector_sum to keep the checksum value is unnecessary, because the extents that btrfs_sector_sum points to are continuous, we can find out the expected checksums by btrfs_ordered_sum's bytenr and the offset, so we can remove btrfs_sector_sum's bytenr. After removing bytenr, there is only one member in the structure, so it makes no sense to keep the structure, just remove it, and use a u32 array to store the checksum value. By this change, we don't use the while loop to get the checksums one by one. Now, we can get several checksum value at one time, it improved the performance by ~74% on my SSD (31MB/s -> 54MB/s). test command: # dd if=/dev/zero of=/mnt/btrfs/file0 bs=1M count=1024 oflag=sync Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-19 10:36:09 +08:00
memcpy(csum, sum->sums + index, sctx->csum_size);
if (index == num_sectors - 1) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
list_del(&sum->list);
kfree(sum);
}
Btrfs: remove btrfs_sector_sum structure Using the structure btrfs_sector_sum to keep the checksum value is unnecessary, because the extents that btrfs_sector_sum points to are continuous, we can find out the expected checksums by btrfs_ordered_sum's bytenr and the offset, so we can remove btrfs_sector_sum's bytenr. After removing bytenr, there is only one member in the structure, so it makes no sense to keep the structure, just remove it, and use a u32 array to store the checksum value. By this change, we don't use the while loop to get the checksums one by one. Now, we can get several checksum value at one time, it improved the performance by ~74% on my SSD (31MB/s -> 54MB/s). test command: # dd if=/dev/zero of=/mnt/btrfs/file0 bs=1M count=1024 oflag=sync Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-19 10:36:09 +08:00
return 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
/* scrub extent tries to collect up to 64 kB for each bio */
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
static int scrub_extent(struct scrub_ctx *sctx, u64 logical, u64 len,
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
u64 physical, struct btrfs_device *dev, u64 flags,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
u64 gen, int mirror_num, u64 physical_for_dev_replace)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
int ret;
u8 csum[BTRFS_CSUM_SIZE];
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u32 blocksize;
if (flags & BTRFS_EXTENT_FLAG_DATA) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
blocksize = sctx->sectorsize;
spin_lock(&sctx->stat_lock);
sctx->stat.data_extents_scrubbed++;
sctx->stat.data_bytes_scrubbed += len;
spin_unlock(&sctx->stat_lock);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
} else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
blocksize = sctx->nodesize;
spin_lock(&sctx->stat_lock);
sctx->stat.tree_extents_scrubbed++;
sctx->stat.tree_bytes_scrubbed += len;
spin_unlock(&sctx->stat_lock);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
} else {
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
blocksize = sctx->sectorsize;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
WARN_ON(1);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
while (len) {
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
u64 l = min_t(u64, len, blocksize);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
int have_csum = 0;
if (flags & BTRFS_EXTENT_FLAG_DATA) {
/* push csums to sbio */
btrfs: Remove len argument from scrub_find_csum It is useless. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 22:03:02 +08:00
have_csum = scrub_find_csum(sctx, logical, csum);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (have_csum == 0)
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
++sctx->stat.no_csum;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (sctx->is_dev_replace && !have_csum) {
ret = copy_nocow_pages(sctx, logical, l,
mirror_num,
physical_for_dev_replace);
goto behind_scrub_pages;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
ret = scrub_pages(sctx, logical, l, physical, dev, flags, gen,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
mirror_num, have_csum ? csum : NULL, 0,
physical_for_dev_replace);
behind_scrub_pages:
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (ret)
return ret;
len -= l;
logical += l;
physical += l;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
physical_for_dev_replace += l;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
return 0;
}
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
static int scrub_pages_for_parity(struct scrub_parity *sparity,
u64 logical, u64 len,
u64 physical, struct btrfs_device *dev,
u64 flags, u64 gen, int mirror_num, u8 *csum)
{
struct scrub_ctx *sctx = sparity->sctx;
struct scrub_block *sblock;
int index;
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
sblock = kzalloc(sizeof(*sblock), GFP_KERNEL);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (!sblock) {
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
return -ENOMEM;
}
/* one ref inside this function, plus one for each page added to
* a bio later on */
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_set(&sblock->refs, 1);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
sblock->sctx = sctx;
sblock->no_io_error_seen = 1;
sblock->sparity = sparity;
scrub_parity_get(sparity);
for (index = 0; len > 0; index++) {
struct scrub_page *spage;
u64 l = min_t(u64, len, PAGE_SIZE);
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
spage = kzalloc(sizeof(*spage), GFP_KERNEL);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (!spage) {
leave_nomem:
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
scrub_block_put(sblock);
return -ENOMEM;
}
BUG_ON(index >= SCRUB_MAX_PAGES_PER_BLOCK);
/* For scrub block */
scrub_page_get(spage);
sblock->pagev[index] = spage;
/* For scrub parity */
scrub_page_get(spage);
list_add_tail(&spage->list, &sparity->spages);
spage->sblock = sblock;
spage->dev = dev;
spage->flags = flags;
spage->generation = gen;
spage->logical = logical;
spage->physical = physical;
spage->mirror_num = mirror_num;
if (csum) {
spage->have_csum = 1;
memcpy(spage->csum, csum, sctx->csum_size);
} else {
spage->have_csum = 0;
}
sblock->page_count++;
btrfs: scrub: use GFP_KERNEL on the submission path Scrub is not on the critical writeback path we don't need to use GFP_NOFS for all allocations. The failures are handled and stats passed back to userspace. Let's use GFP_KERNEL on the paths where everything is ok, ie. setup the global structures and the IO submission paths. Functions that do the repair and fixups still use GFP_NOFS as we might want to skip any other filesystem activity if we encounter an error. This could turn out to be unnecessary, but requires more review compared to the easy cases in this patch. Signed-off-by: David Sterba <dsterba@suse.com>
2016-02-11 17:49:42 +08:00
spage->page = alloc_page(GFP_KERNEL);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (!spage->page)
goto leave_nomem;
len -= l;
logical += l;
physical += l;
}
WARN_ON(sblock->page_count == 0);
for (index = 0; index < sblock->page_count; index++) {
struct scrub_page *spage = sblock->pagev[index];
int ret;
ret = scrub_add_page_to_rd_bio(sctx, spage);
if (ret) {
scrub_block_put(sblock);
return ret;
}
}
/* last one frees, either here or in bio completion for last page */
scrub_block_put(sblock);
return 0;
}
static int scrub_extent_for_parity(struct scrub_parity *sparity,
u64 logical, u64 len,
u64 physical, struct btrfs_device *dev,
u64 flags, u64 gen, int mirror_num)
{
struct scrub_ctx *sctx = sparity->sctx;
int ret;
u8 csum[BTRFS_CSUM_SIZE];
u32 blocksize;
Btrfs: fix parity scrub of RAID 5/6 with missing device When testing the previous patch, Zhao Lei reported a similar bug when attempting to scrub a degraded RAID 5/6 filesystem with a missing device, leading to NULL pointer dereferences from the RAID 5/6 parity scrubbing code. The first cause was the same as in the previous patch: attempting to call bio_add_page() on a missing block device. To fix this, scrub_extent_for_parity() can just mark the sectors on the missing device as errors instead of attempting to read from it. Additionally, the code uses scrub_remap_extent() to map the extent of the corresponding data stripe, but the extent wasn't already mapped. If scrub_remap_extent() finds a missing block device, it doesn't initialize extent_dev, so we're left with a NULL struct btrfs_device. The solution is to use btrfs_map_block() directly. Reported-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:52 +08:00
if (dev->missing) {
scrub_parity_mark_sectors_error(sparity, logical, len);
return 0;
}
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (flags & BTRFS_EXTENT_FLAG_DATA) {
blocksize = sctx->sectorsize;
} else if (flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) {
blocksize = sctx->nodesize;
} else {
blocksize = sctx->sectorsize;
WARN_ON(1);
}
while (len) {
u64 l = min_t(u64, len, blocksize);
int have_csum = 0;
if (flags & BTRFS_EXTENT_FLAG_DATA) {
/* push csums to sbio */
btrfs: Remove len argument from scrub_find_csum It is useless. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-24 22:03:02 +08:00
have_csum = scrub_find_csum(sctx, logical, csum);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (have_csum == 0)
goto skip;
}
ret = scrub_pages_for_parity(sparity, logical, l, physical, dev,
flags, gen, mirror_num,
have_csum ? csum : NULL);
if (ret)
return ret;
Btrfs, scrub: uninitialized variable in scrub_extent_for_parity() The only way that "ret" is set is when we call scrub_pages_for_parity() so the skip to "if (ret) " test doesn't make sense and causes a static checker warning. Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-12-13 03:30:00 +08:00
skip:
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
len -= l;
logical += l;
physical += l;
}
return 0;
}
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
/*
* Given a physical address, this will calculate it's
* logical offset. if this is a parity stripe, it will return
* the most left data stripe's logical offset.
*
* return 0 if it is a data stripe, 1 means parity stripe.
*/
static int get_raid56_logic_offset(u64 physical, int num,
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
struct map_lookup *map, u64 *offset,
u64 *stripe_start)
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
{
int i;
int j = 0;
u64 stripe_nr;
u64 last_offset;
btrfs: cleanup, use correct type in div_u64_rem div_u64_rem expects u32 for divisior and reminder. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-02-21 01:42:11 +08:00
u32 stripe_index;
u32 rot;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
last_offset = (physical - map->stripes[num].physical) *
nr_data_stripes(map);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (stripe_start)
*stripe_start = last_offset;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
*offset = last_offset;
for (i = 0; i < nr_data_stripes(map); i++) {
*offset = last_offset + i * map->stripe_len;
btrfs: cleanup 64bit/32bit divs, provably bounded values The divisor is derived from nodesize or PAGE_SIZE, fits into 32bit type. Get rid of a few more do_div instances. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-01-17 00:26:13 +08:00
stripe_nr = div_u64(*offset, map->stripe_len);
stripe_nr = div_u64(stripe_nr, nr_data_stripes(map));
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
/* Work out the disk rotation on this stripe-set */
btrfs: replace remaining do_div calls with div_u64 variants Switch to div_u64_rem that does type checking and has more obvious semantics than do_div. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-02-21 01:43:47 +08:00
stripe_nr = div_u64_rem(stripe_nr, map->num_stripes, &rot);
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
/* calculate which stripe this data locates */
rot += i;
Btrfs: fix compile warnings on on avr32 platform fs/btrfs/scrub.c: In function 'get_raid56_logic_offset': fs/btrfs/scrub.c:2269: warning: comparison of distinct pointer types lacks a cast fs/btrfs/scrub.c:2269: warning: right shift count >= width of type fs/btrfs/scrub.c:2269: warning: passing argument 1 of '__div64_32' from incompatible pointer type Since @rot is an int type, we should not use do_div(), fix it. Reported-by: kbuild test robot <fengguang.wu@intel.com> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-11 18:32:25 +08:00
stripe_index = rot % map->num_stripes;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
if (stripe_index == num)
return 0;
if (stripe_index < num)
j++;
}
*offset = last_offset + j * map->stripe_len;
return 1;
}
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
static void scrub_free_parity(struct scrub_parity *sparity)
{
struct scrub_ctx *sctx = sparity->sctx;
struct scrub_page *curr, *next;
int nbits;
nbits = bitmap_weight(sparity->ebitmap, sparity->nsectors);
if (nbits) {
spin_lock(&sctx->stat_lock);
sctx->stat.read_errors += nbits;
sctx->stat.uncorrectable_errors += nbits;
spin_unlock(&sctx->stat_lock);
}
list_for_each_entry_safe(curr, next, &sparity->spages, list) {
list_del_init(&curr->list);
scrub_page_put(curr);
}
kfree(sparity);
}
btrfs: Fix lockdep warning of wr_ctx->wr_lock in scrub_free_wr_ctx() lockdep report following warning in test: [25176.843958] ================================= [25176.844519] [ INFO: inconsistent lock state ] [25176.845047] 4.1.0-rc3 #22 Tainted: G W [25176.845591] --------------------------------- [25176.846153] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage. [25176.846713] fsstress/26661 [HC0[0]:SC1[1]:HE1:SE0] takes: [25176.847246] (&wr_ctx->wr_lock){+.?...}, at: [<ffffffffa04cdc6d>] scrub_free_ctx+0x2d/0xf0 [btrfs] [25176.847838] {SOFTIRQ-ON-W} state was registered at: [25176.848396] [<ffffffff810bf460>] __lock_acquire+0x6a0/0xe10 [25176.848955] [<ffffffff810bfd1e>] lock_acquire+0xce/0x2c0 [25176.849491] [<ffffffff816489af>] mutex_lock_nested+0x7f/0x410 [25176.850029] [<ffffffffa04d04ff>] scrub_stripe+0x4df/0x1080 [btrfs] [25176.850575] [<ffffffffa04d11b1>] scrub_chunk.isra.19+0x111/0x130 [btrfs] [25176.851110] [<ffffffffa04d144c>] scrub_enumerate_chunks+0x27c/0x510 [btrfs] [25176.851660] [<ffffffffa04d3b87>] btrfs_scrub_dev+0x1c7/0x6c0 [btrfs] [25176.852189] [<ffffffffa04e918e>] btrfs_dev_replace_start+0x36e/0x450 [btrfs] [25176.852771] [<ffffffffa04a98e0>] btrfs_ioctl+0x1e10/0x2d20 [btrfs] [25176.853315] [<ffffffff8121c5b8>] do_vfs_ioctl+0x318/0x570 [25176.853868] [<ffffffff8121c851>] SyS_ioctl+0x41/0x80 [25176.854406] [<ffffffff8164da17>] system_call_fastpath+0x12/0x6f [25176.854935] irq event stamp: 51506 [25176.855511] hardirqs last enabled at (51506): [<ffffffff810d4ce5>] vprintk_emit+0x225/0x5e0 [25176.856059] hardirqs last disabled at (51505): [<ffffffff810d4b77>] vprintk_emit+0xb7/0x5e0 [25176.856642] softirqs last enabled at (50886): [<ffffffff81067a23>] __do_softirq+0x363/0x640 [25176.857184] softirqs last disabled at (50949): [<ffffffff8106804d>] irq_exit+0x10d/0x120 [25176.857746] other info that might help us debug this: [25176.858845] Possible unsafe locking scenario: [25176.859981] CPU0 [25176.860537] ---- [25176.861059] lock(&wr_ctx->wr_lock); [25176.861705] <Interrupt> [25176.862272] lock(&wr_ctx->wr_lock); [25176.862881] *** DEADLOCK *** Reason: Above warning is caused by: Interrupt -> bio_endio() -> ... -> scrub_put_ctx() -> scrub_free_ctx() *1 -> ... -> mutex_lock(&wr_ctx->wr_lock); scrub_put_ctx() is allowed to be called in end_bio interrupt, but in code design, it will never call scrub_free_ctx(sctx) in interrupe context(above *1), because btrfs_scrub_dev() get one additional reference of sctx->refs, which makes scrub_free_ctx() only called withine btrfs_scrub_dev(). Now the code runs out of our wish, because free sequence in scrub_pending_bio_dec() have a gap. Current code: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| -> scrub_free_ctx() | -----------------------------------+----------------------------------- We expected: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) | -> scrub_free_ctx() -----------------------------------+----------------------------------- Fix: Move scrub_pending_bio_dec() to a workqueue, to avoid this function run in interrupt context. Tested by check tracelog in debug. Changelog v1->v2: Use workqueue instead of adjust function call sequence in v1, because v1 will introduce a bug pointed out by: Filipe David Manana <fdmanana@gmail.com> Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Reviewed-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-04 20:09:15 +08:00
static void scrub_parity_bio_endio_worker(struct btrfs_work *work)
{
struct scrub_parity *sparity = container_of(work, struct scrub_parity,
work);
struct scrub_ctx *sctx = sparity->sctx;
scrub_free_parity(sparity);
scrub_pending_bio_dec(sctx);
}
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
static void scrub_parity_bio_endio(struct bio *bio)
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
{
struct scrub_parity *sparity = (struct scrub_parity *)bio->bi_private;
block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hannes Reinecke <hare@suse.de> Reviewed-by: NeilBrown <neilb@suse.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-07-20 21:29:37 +08:00
if (bio->bi_error)
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
bitmap_or(sparity->ebitmap, sparity->ebitmap, sparity->dbitmap,
sparity->nsectors);
bio_put(bio);
btrfs: Fix lockdep warning of wr_ctx->wr_lock in scrub_free_wr_ctx() lockdep report following warning in test: [25176.843958] ================================= [25176.844519] [ INFO: inconsistent lock state ] [25176.845047] 4.1.0-rc3 #22 Tainted: G W [25176.845591] --------------------------------- [25176.846153] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage. [25176.846713] fsstress/26661 [HC0[0]:SC1[1]:HE1:SE0] takes: [25176.847246] (&wr_ctx->wr_lock){+.?...}, at: [<ffffffffa04cdc6d>] scrub_free_ctx+0x2d/0xf0 [btrfs] [25176.847838] {SOFTIRQ-ON-W} state was registered at: [25176.848396] [<ffffffff810bf460>] __lock_acquire+0x6a0/0xe10 [25176.848955] [<ffffffff810bfd1e>] lock_acquire+0xce/0x2c0 [25176.849491] [<ffffffff816489af>] mutex_lock_nested+0x7f/0x410 [25176.850029] [<ffffffffa04d04ff>] scrub_stripe+0x4df/0x1080 [btrfs] [25176.850575] [<ffffffffa04d11b1>] scrub_chunk.isra.19+0x111/0x130 [btrfs] [25176.851110] [<ffffffffa04d144c>] scrub_enumerate_chunks+0x27c/0x510 [btrfs] [25176.851660] [<ffffffffa04d3b87>] btrfs_scrub_dev+0x1c7/0x6c0 [btrfs] [25176.852189] [<ffffffffa04e918e>] btrfs_dev_replace_start+0x36e/0x450 [btrfs] [25176.852771] [<ffffffffa04a98e0>] btrfs_ioctl+0x1e10/0x2d20 [btrfs] [25176.853315] [<ffffffff8121c5b8>] do_vfs_ioctl+0x318/0x570 [25176.853868] [<ffffffff8121c851>] SyS_ioctl+0x41/0x80 [25176.854406] [<ffffffff8164da17>] system_call_fastpath+0x12/0x6f [25176.854935] irq event stamp: 51506 [25176.855511] hardirqs last enabled at (51506): [<ffffffff810d4ce5>] vprintk_emit+0x225/0x5e0 [25176.856059] hardirqs last disabled at (51505): [<ffffffff810d4b77>] vprintk_emit+0xb7/0x5e0 [25176.856642] softirqs last enabled at (50886): [<ffffffff81067a23>] __do_softirq+0x363/0x640 [25176.857184] softirqs last disabled at (50949): [<ffffffff8106804d>] irq_exit+0x10d/0x120 [25176.857746] other info that might help us debug this: [25176.858845] Possible unsafe locking scenario: [25176.859981] CPU0 [25176.860537] ---- [25176.861059] lock(&wr_ctx->wr_lock); [25176.861705] <Interrupt> [25176.862272] lock(&wr_ctx->wr_lock); [25176.862881] *** DEADLOCK *** Reason: Above warning is caused by: Interrupt -> bio_endio() -> ... -> scrub_put_ctx() -> scrub_free_ctx() *1 -> ... -> mutex_lock(&wr_ctx->wr_lock); scrub_put_ctx() is allowed to be called in end_bio interrupt, but in code design, it will never call scrub_free_ctx(sctx) in interrupe context(above *1), because btrfs_scrub_dev() get one additional reference of sctx->refs, which makes scrub_free_ctx() only called withine btrfs_scrub_dev(). Now the code runs out of our wish, because free sequence in scrub_pending_bio_dec() have a gap. Current code: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| -> scrub_free_ctx() | -----------------------------------+----------------------------------- We expected: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) | -> scrub_free_ctx() -----------------------------------+----------------------------------- Fix: Move scrub_pending_bio_dec() to a workqueue, to avoid this function run in interrupt context. Tested by check tracelog in debug. Changelog v1->v2: Use workqueue instead of adjust function call sequence in v1, because v1 will introduce a bug pointed out by: Filipe David Manana <fdmanana@gmail.com> Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Reviewed-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-04 20:09:15 +08:00
btrfs_init_work(&sparity->work, btrfs_scrubparity_helper,
scrub_parity_bio_endio_worker, NULL, NULL);
btrfs_queue_work(sparity->sctx->dev_root->fs_info->scrub_parity_workers,
&sparity->work);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
}
static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
{
struct scrub_ctx *sctx = sparity->sctx;
struct bio *bio;
struct btrfs_raid_bio *rbio;
struct scrub_page *spage;
struct btrfs_bio *bbio = NULL;
u64 length;
int ret;
if (!bitmap_andnot(sparity->dbitmap, sparity->dbitmap, sparity->ebitmap,
sparity->nsectors))
goto out;
btrfs: Fix calculate typo caused by ambiguous meaning of logic_end For example, in scrub_raid56_parity(), following lines are used to judge is all data processed: place1: if (key.objectid > logic_end) ... place2: if (logic_start >= logic_end) ... ... (place2 is typo, is should be ">", it is copied from other place, where logic_end's meaning is different, long story...) We can fix above typo directly, but the root reason is ambiguous meaning of logic_end in scrub raid56 parity. In other place, XXX_end is pointed to data which is not included, and we need to process segment of [XXX_start, XXX_end). But for scrub raid56 parity, logic_end is pointed to lattest data need to process, and introduced many "+ 1" and "- 1" in code as below: length = sparity->logic_end - sparity->logic_start + 1 logic_end - logic_start + 1 stripe_logical + increment - 1 This patch changed logic_end's meaning to make it in normal understanding in raid56 parity functions and data struct alone with above bugfix. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 15:42:26 +08:00
length = sparity->logic_end - sparity->logic_start;
Btrfs, replace: write raid56 parity into the replace target device This function reused the code of parity scrub, and we just write the right parity or corrected parity into the target device before the parity scrub end. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-14 17:45:42 +08:00
ret = btrfs_map_sblock(sctx->dev_root->fs_info, WRITE,
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
sparity->logic_start,
Btrfs: Make raid_map array be inlined in btrfs_bio structure It can make code more simple and clear, we need not care about free bbio and raid_map together. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:33 +08:00
&length, &bbio, 0, 1);
if (ret || !bbio || !bbio->raid_map)
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
goto bbio_out;
bio = btrfs_io_bio_alloc(GFP_NOFS, 0);
if (!bio)
goto bbio_out;
bio->bi_iter.bi_sector = sparity->logic_start >> 9;
bio->bi_private = sparity;
bio->bi_end_io = scrub_parity_bio_endio;
rbio = raid56_parity_alloc_scrub_rbio(sctx->dev_root, bio, bbio,
Btrfs: Make raid_map array be inlined in btrfs_bio structure It can make code more simple and clear, we need not care about free bbio and raid_map together. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:33 +08:00
length, sparity->scrub_dev,
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
sparity->dbitmap,
sparity->nsectors);
if (!rbio)
goto rbio_out;
list_for_each_entry(spage, &sparity->spages, list)
Btrfs: add RAID 5/6 BTRFS_RBIO_REBUILD_MISSING operation The current RAID 5/6 recovery code isn't quite prepared to handle missing devices. In particular, it expects a bio that we previously attempted to use in the read path, meaning that it has valid pages allocated. However, missing devices have a NULL blkdev, and we can't call bio_add_page() on a bio with a NULL blkdev. We could do manual manipulation of bio->bi_io_vec, but that's pretty gross. So instead, add a separate path that allows us to manually add pages to the rbio. Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:50 +08:00
raid56_add_scrub_pages(rbio, spage->page, spage->logical);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
scrub_pending_bio_inc(sctx);
raid56_parity_submit_scrub_rbio(rbio);
return;
rbio_out:
bio_put(bio);
bbio_out:
Btrfs: add ref_count and free function for btrfs_bio 1: ref_count is simple than current RBIO_HOLD_BBIO_MAP_BIT flag to keep btrfs_bio's memory in raid56 recovery implement. 2: free function for bbio will make code clean and flexible, plus forced data type checking in compile. Changelog v1->v2: Rename following by David Sterba's suggestion: put_btrfs_bio() -> btrfs_put_bio() get_btrfs_bio() -> btrfs_get_bio() bbio->ref_count -> bbio->refs Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:34 +08:00
btrfs_put_bbio(bbio);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
bitmap_or(sparity->ebitmap, sparity->ebitmap, sparity->dbitmap,
sparity->nsectors);
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
out:
scrub_free_parity(sparity);
}
static inline int scrub_calc_parity_bitmap_len(int nsectors)
{
btrfs: Fix calculation of rbio->dbitmap's size calculation Current code is trying to calculate rbio->dbitmap's size to make it align to sizeof(long), but implement haven't achived this object, it is align to sizeof(char) instead. This patch fixed above calculation, and use sizeof(long) instead of fixed "8" to increate compatibility. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-12-08 19:55:57 +08:00
return DIV_ROUND_UP(nsectors, BITS_PER_LONG) * sizeof(long);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
}
static void scrub_parity_get(struct scrub_parity *sparity)
{
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_inc(&sparity->refs);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
}
static void scrub_parity_put(struct scrub_parity *sparity)
{
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
if (!atomic_dec_and_test(&sparity->refs))
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
return;
scrub_parity_check_and_repair(sparity);
}
static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
struct map_lookup *map,
struct btrfs_device *sdev,
struct btrfs_path *path,
u64 logic_start,
u64 logic_end)
{
struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
struct btrfs_root *root = fs_info->extent_root;
struct btrfs_root *csum_root = fs_info->csum_root;
struct btrfs_extent_item *extent;
Btrfs: fix parity scrub of RAID 5/6 with missing device When testing the previous patch, Zhao Lei reported a similar bug when attempting to scrub a degraded RAID 5/6 filesystem with a missing device, leading to NULL pointer dereferences from the RAID 5/6 parity scrubbing code. The first cause was the same as in the previous patch: attempting to call bio_add_page() on a missing block device. To fix this, scrub_extent_for_parity() can just mark the sectors on the missing device as errors instead of attempting to read from it. Additionally, the code uses scrub_remap_extent() to map the extent of the corresponding data stripe, but the extent wasn't already mapped. If scrub_remap_extent() finds a missing block device, it doesn't initialize extent_dev, so we're left with a NULL struct btrfs_device. The solution is to use btrfs_map_block() directly. Reported-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:52 +08:00
struct btrfs_bio *bbio = NULL;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
u64 flags;
int ret;
int slot;
struct extent_buffer *l;
struct btrfs_key key;
u64 generation;
u64 extent_logical;
u64 extent_physical;
u64 extent_len;
Btrfs: fix parity scrub of RAID 5/6 with missing device When testing the previous patch, Zhao Lei reported a similar bug when attempting to scrub a degraded RAID 5/6 filesystem with a missing device, leading to NULL pointer dereferences from the RAID 5/6 parity scrubbing code. The first cause was the same as in the previous patch: attempting to call bio_add_page() on a missing block device. To fix this, scrub_extent_for_parity() can just mark the sectors on the missing device as errors instead of attempting to read from it. Additionally, the code uses scrub_remap_extent() to map the extent of the corresponding data stripe, but the extent wasn't already mapped. If scrub_remap_extent() finds a missing block device, it doesn't initialize extent_dev, so we're left with a NULL struct btrfs_device. The solution is to use btrfs_map_block() directly. Reported-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:52 +08:00
u64 mapped_length;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
struct btrfs_device *extent_dev;
struct scrub_parity *sparity;
int nsectors;
int bitmap_len;
int extent_mirror_num;
int stop_loop = 0;
nsectors = map->stripe_len / root->sectorsize;
bitmap_len = scrub_calc_parity_bitmap_len(nsectors);
sparity = kzalloc(sizeof(struct scrub_parity) + 2 * bitmap_len,
GFP_NOFS);
if (!sparity) {
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
return -ENOMEM;
}
sparity->stripe_len = map->stripe_len;
sparity->nsectors = nsectors;
sparity->sctx = sctx;
sparity->scrub_dev = sdev;
sparity->logic_start = logic_start;
sparity->logic_end = logic_end;
Rename all ref_count to refs in struct refs is better than ref_count to record a struct's ref count. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Suggested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:45 +08:00
atomic_set(&sparity->refs, 1);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
INIT_LIST_HEAD(&sparity->spages);
sparity->dbitmap = sparity->bitmap;
sparity->ebitmap = (void *)sparity->bitmap + bitmap_len;
ret = 0;
while (logic_start < logic_end) {
if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
key.type = BTRFS_METADATA_ITEM_KEY;
else
key.type = BTRFS_EXTENT_ITEM_KEY;
key.objectid = logic_start;
key.offset = (u64)-1;
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
if (ret < 0)
goto out;
if (ret > 0) {
ret = btrfs_previous_extent_item(root, path, 0);
if (ret < 0)
goto out;
if (ret > 0) {
btrfs_release_path(path);
ret = btrfs_search_slot(NULL, root, &key,
path, 0, 0);
if (ret < 0)
goto out;
}
}
stop_loop = 0;
while (1) {
u64 bytes;
l = path->nodes[0];
slot = path->slots[0];
if (slot >= btrfs_header_nritems(l)) {
ret = btrfs_next_leaf(root, path);
if (ret == 0)
continue;
if (ret < 0)
goto out;
stop_loop = 1;
break;
}
btrfs_item_key_to_cpu(l, &key, slot);
btrfs: Bypass unrelated items before accessing its contents in scrub When we access extent_root in scrub_stripe() and scrub_raid56_parity(), we need bypass unrelated tree item firstly before using its contents to do other condition. It is not a bug fix, only making code sequence in logic. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-22 13:14:48 +08:00
if (key.type != BTRFS_EXTENT_ITEM_KEY &&
key.type != BTRFS_METADATA_ITEM_KEY)
goto next;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (key.type == BTRFS_METADATA_ITEM_KEY)
bytes = root->nodesize;
else
bytes = key.offset;
if (key.objectid + bytes <= logic_start)
goto next;
btrfs: Fix calculate typo caused by ambiguous meaning of logic_end For example, in scrub_raid56_parity(), following lines are used to judge is all data processed: place1: if (key.objectid > logic_end) ... place2: if (logic_start >= logic_end) ... ... (place2 is typo, is should be ">", it is copied from other place, where logic_end's meaning is different, long story...) We can fix above typo directly, but the root reason is ambiguous meaning of logic_end in scrub raid56 parity. In other place, XXX_end is pointed to data which is not included, and we need to process segment of [XXX_start, XXX_end). But for scrub raid56 parity, logic_end is pointed to lattest data need to process, and introduced many "+ 1" and "- 1" in code as below: length = sparity->logic_end - sparity->logic_start + 1 logic_end - logic_start + 1 stripe_logical + increment - 1 This patch changed logic_end's meaning to make it in normal understanding in raid56 parity functions and data struct alone with above bugfix. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 15:42:26 +08:00
if (key.objectid >= logic_end) {
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
stop_loop = 1;
break;
}
while (key.objectid >= logic_start + map->stripe_len)
logic_start += map->stripe_len;
extent = btrfs_item_ptr(l, slot,
struct btrfs_extent_item);
flags = btrfs_extent_flags(l, extent);
generation = btrfs_extent_generation(l, extent);
btrfs: Fix scrub panic when leaf crosses stripes Scrub panic in following operation: mkfs.ext4 /dev/vdh btrfs-convert /dev/vdh mount /dev/vdh /mnt/tmp1 btrfs scrub start -B /dev/vdh (panic) Reason: 1: In some case, leaf created by btrfs-convert was splited into 2 strips. 2: Scrub bypassed part of above wrong leaf data, but remain data caused panic in scrub_checksum_tree_block(). For reason 1: we can get following information after some simple operation. a. mkfs.ext4 /dev/vdh btrfs-convert /dev/vdh b. btrfs-debug-tree /dev/vdh we can see following item in extent tree: item 25 key (27054080 METADATA_ITEM 0) itemoff 15083 itemsize 33 Its logical address is [27054080, 27070464) and acrossed 2 strips: [27000832, 27066368) [27066368, 27131904) Will be fixed in btrfs-progs(btrfs-convert, btrfsck, ...) For reason 2: Scrub is trying to do a "bypass" in this case, but the result is "panic", because current code lacks of some condition in bypass, and let some wrong leaf data escaped. This patch fixed above scrub code. Before patch: # btrfs scrub start -B /dev/vdh (panic) After patch: # btrfs scrub start -B /dev/vdh scrub done for 353cec8f-da31-4a94-aa35-be72d997b06e ... # dmesg ... [ 59.088697] BTRFS error (device vdh): scrub: tree block 27054080 spanning stripes, ignored. logical=27000832 [ 59.089929] BTRFS error (device vdh): scrub: tree block 27054080 spanning stripes, ignored. logical=27066368 # Reported-by: Chris Murphy <lists@colorremedies.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-23 12:29:49 +08:00
if ((flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) &&
(key.objectid < logic_start ||
key.objectid + bytes >
logic_start + map->stripe_len)) {
btrfs_err(fs_info, "scrub: tree block %llu spanning stripes, ignored. logical=%llu",
key.objectid, logic_start);
btrfs: scrub: set error stats when tree block spanning stripes It is better to show error stats to user when we found tree block spanning stripes. On a btrfs created by old version of btrfs-convert: Before patch: # btrfs scrub start -B /dev/vdh scrub done for 8b342d35-2904-41ab-b3cb-2f929709cf47 scrub started at Tue Aug 25 21:19:09 2015 and finished after 00:00:00 total bytes scrubbed: 53.54MiB with 0 errors # dmesg ... [ 128.711434] BTRFS error (device vdh): scrub: tree block 27054080 spanning stripes, ignored. logical=27000832 [ 128.712744] BTRFS error (device vdh): scrub: tree block 27054080 spanning stripes, ignored. logical=27066368 ... After patch: # btrfs scrub start -B /dev/vdh scrub done for ff7f844b-7a4e-4b1a-88a9-8252ab25be1b scrub started at Tue Aug 25 21:42:29 2015 and finished after 00:00:00 total bytes scrubbed: 53.60MiB with 2 errors error details: corrected errors: 0, uncorrectable errors: 2, unverified errors: 0 ERROR: There are uncorrectable errors. # dmesg ...omit... # Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-25 21:31:40 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.uncorrectable_errors++;
spin_unlock(&sctx->stat_lock);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
goto next;
}
again:
extent_logical = key.objectid;
extent_len = bytes;
if (extent_logical < logic_start) {
extent_len -= logic_start - extent_logical;
extent_logical = logic_start;
}
if (extent_logical + extent_len >
logic_start + map->stripe_len)
extent_len = logic_start + map->stripe_len -
extent_logical;
scrub_parity_mark_sectors_data(sparity, extent_logical,
extent_len);
Btrfs: fix parity scrub of RAID 5/6 with missing device When testing the previous patch, Zhao Lei reported a similar bug when attempting to scrub a degraded RAID 5/6 filesystem with a missing device, leading to NULL pointer dereferences from the RAID 5/6 parity scrubbing code. The first cause was the same as in the previous patch: attempting to call bio_add_page() on a missing block device. To fix this, scrub_extent_for_parity() can just mark the sectors on the missing device as errors instead of attempting to read from it. Additionally, the code uses scrub_remap_extent() to map the extent of the corresponding data stripe, but the extent wasn't already mapped. If scrub_remap_extent() finds a missing block device, it doesn't initialize extent_dev, so we're left with a NULL struct btrfs_device. The solution is to use btrfs_map_block() directly. Reported-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Omar Sandoval <osandov@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-20 02:52:52 +08:00
mapped_length = extent_len;
ret = btrfs_map_block(fs_info, READ, extent_logical,
&mapped_length, &bbio, 0);
if (!ret) {
if (!bbio || mapped_length < extent_len)
ret = -EIO;
}
if (ret) {
btrfs_put_bbio(bbio);
goto out;
}
extent_physical = bbio->stripes[0].physical;
extent_mirror_num = bbio->mirror_num;
extent_dev = bbio->stripes[0].dev;
btrfs_put_bbio(bbio);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
ret = btrfs_lookup_csums_range(csum_root,
extent_logical,
extent_logical + extent_len - 1,
&sctx->csum_list, 1);
if (ret)
goto out;
ret = scrub_extent_for_parity(sparity, extent_logical,
extent_len,
extent_physical,
extent_dev, flags,
generation,
extent_mirror_num);
btrfs: Free checksum list on scrub_extent() fail When scrub_extent() failed, we need to free previois created checksum list. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 12:22:30 +08:00
scrub_free_csums(sctx);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
if (ret)
goto out;
if (extent_logical + extent_len <
key.objectid + bytes) {
logic_start += map->stripe_len;
if (logic_start >= logic_end) {
stop_loop = 1;
break;
}
if (logic_start < key.objectid + bytes) {
cond_resched();
goto again;
}
}
next:
path->slots[0]++;
}
btrfs_release_path(path);
if (stop_loop)
break;
logic_start += map->stripe_len;
}
out:
if (ret < 0)
scrub_parity_mark_sectors_error(sparity, logic_start,
btrfs: Fix calculate typo caused by ambiguous meaning of logic_end For example, in scrub_raid56_parity(), following lines are used to judge is all data processed: place1: if (key.objectid > logic_end) ... place2: if (logic_start >= logic_end) ... ... (place2 is typo, is should be ">", it is copied from other place, where logic_end's meaning is different, long story...) We can fix above typo directly, but the root reason is ambiguous meaning of logic_end in scrub raid56 parity. In other place, XXX_end is pointed to data which is not included, and we need to process segment of [XXX_start, XXX_end). But for scrub raid56 parity, logic_end is pointed to lattest data need to process, and introduced many "+ 1" and "- 1" in code as below: length = sparity->logic_end - sparity->logic_start + 1 logic_end - logic_start + 1 stripe_logical + increment - 1 This patch changed logic_end's meaning to make it in normal understanding in raid56 parity functions and data struct alone with above bugfix. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 15:42:26 +08:00
logic_end - logic_start);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
scrub_parity_put(sparity);
scrub_submit(sctx);
mutex_lock(&sctx->wr_ctx.wr_lock);
scrub_wr_submit(sctx);
mutex_unlock(&sctx->wr_ctx.wr_lock);
btrfs_release_path(path);
return ret < 0 ? ret : 0;
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct map_lookup *map,
struct btrfs_device *scrub_dev,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
int num, u64 base, u64 length,
int is_dev_replace)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
struct btrfs_path *path, *ppath;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct btrfs_root *root = fs_info->extent_root;
struct btrfs_root *csum_root = fs_info->csum_root;
struct btrfs_extent_item *extent;
btrfs: scrub: add explicit plugging With the removal of the implicit plugging scrub ends up doing more and smaller I/O than necessary. This patch adds explicit plugging per chunk. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-05-29 04:58:38 +08:00
struct blk_plug plug;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u64 flags;
int ret;
int slot;
u64 nstripes;
struct extent_buffer *l;
struct btrfs_key key;
u64 physical;
u64 logical;
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
u64 logic_end;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
u64 physical_end;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u64 generation;
btrfs scrub: use int for mirror_num, not u64 the rest of the code uses int mirror_num, and so should scrub Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-17 21:55:21 +08:00
int mirror_num;
btrfs: use readahead API for scrub Scrub uses a simple tree-enumeration to bring the relevant portions of the extent- and csum-tree into the page cache before starting the scrub-I/O. This is now replaced by using the new readahead-API. During readahead the scrub is being accounted as paused, so it won't hold off transaction commits. This change raises the average disk bandwith utilisation on my test volume from 70% to 90%. On another volume, the time for a test run went down from 89s to 43s. Changes v5: - reada1/2 are now of type struct reada_control * Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-10 18:39:23 +08:00
struct reada_control *reada1;
struct reada_control *reada2;
struct btrfs_key key_start;
struct btrfs_key key_end;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
u64 increment = map->stripe_len;
u64 offset;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
u64 extent_logical;
u64 extent_physical;
u64 extent_len;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
u64 stripe_logical;
u64 stripe_end;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct btrfs_device *extent_dev;
int extent_mirror_num;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
int stop_loop = 0;
Btrfs: RAID5 and RAID6 This builds on David Woodhouse's original Btrfs raid5/6 implementation. The code has changed quite a bit, blame Chris Mason for any bugs. Read/modify/write is done after the higher levels of the filesystem have prepared a given bio. This means the higher layers are not responsible for building full stripes, and they don't need to query for the topology of the extents that may get allocated during delayed allocation runs. It also means different files can easily share the same stripe. But, it does expose us to incorrect parity if we crash or lose power while doing a read/modify/write cycle. This will be addressed in a later commit. Scrub is unable to repair crc errors on raid5/6 chunks. Discard does not work on raid5/6 (yet) The stripe size is fixed at 64KiB per disk. This will be tunable in a later commit. Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-01-30 07:40:14 +08:00
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
physical = map->stripes[num].physical;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
offset = 0;
btrfs: cleanup 64bit/32bit divs, provably bounded values The divisor is derived from nodesize or PAGE_SIZE, fits into 32bit type. Get rid of a few more do_div instances. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-01-17 00:26:13 +08:00
nstripes = div_u64(length, map->stripe_len);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
offset = map->stripe_len * num;
increment = map->stripe_len * map->num_stripes;
btrfs scrub: bugfix: mirror_num off by one Fix the mirror_num determination in scrub_stripe. The rest of the scrub code did not use mirror_num for anything important and that error went unnoticed. The nodatasum fixup patch of this set depends on a correct mirror_num. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:56:54 +08:00
mirror_num = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
} else if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
int factor = map->num_stripes / map->sub_stripes;
offset = map->stripe_len * (num / map->sub_stripes);
increment = map->stripe_len * factor;
btrfs scrub: bugfix: mirror_num off by one Fix the mirror_num determination in scrub_stripe. The rest of the scrub code did not use mirror_num for anything important and that error went unnoticed. The nodatasum fixup patch of this set depends on a correct mirror_num. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:56:54 +08:00
mirror_num = num % map->sub_stripes + 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
} else if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
increment = map->stripe_len;
btrfs scrub: bugfix: mirror_num off by one Fix the mirror_num determination in scrub_stripe. The rest of the scrub code did not use mirror_num for anything important and that error went unnoticed. The nodatasum fixup patch of this set depends on a correct mirror_num. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:56:54 +08:00
mirror_num = num % map->num_stripes + 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
increment = map->stripe_len;
btrfs scrub: bugfix: mirror_num off by one Fix the mirror_num determination in scrub_stripe. The rest of the scrub code did not use mirror_num for anything important and that error went unnoticed. The nodatasum fixup patch of this set depends on a correct mirror_num. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:56:54 +08:00
mirror_num = num % map->num_stripes + 1;
Btrfs: Introduce BTRFS_BLOCK_GROUP_RAID56_MASK to check raid56 simply So we can check raid56 with: (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) instead of long: (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:44 +08:00
} else if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
get_raid56_logic_offset(physical, num, map, &offset, NULL);
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
increment = map->stripe_len * nr_data_stripes(map);
mirror_num = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
} else {
increment = map->stripe_len;
btrfs scrub: bugfix: mirror_num off by one Fix the mirror_num determination in scrub_stripe. The rest of the scrub code did not use mirror_num for anything important and that error went unnoticed. The nodatasum fixup patch of this set depends on a correct mirror_num. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 01:56:54 +08:00
mirror_num = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
ppath = btrfs_alloc_path();
if (!ppath) {
Btrfs: fix incorrect freeing in scrub_stripe The address that should be freed is not 'ppath' but 'path'. Signed-off-by: Tsutomu Itoh <t-itoh@jp.fujitsu.com> Reviewed-by: Miao Xie <miaoxie@huawei.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-09 16:37:52 +08:00
btrfs_free_path(path);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
return -ENOMEM;
}
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/*
* work on commit root. The related disk blocks are static as
* long as COW is applied. This means, it is save to rewrite
* them to repair disk errors without any race conditions
*/
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
path->search_commit_root = 1;
path->skip_locking = 1;
btrfs: fix raid56 scrub failed in xfstests btrfs/072 The xfstests btrfs/072 reports uncorrectable read errors in dmesg, because scrub forgets to use commit_root for parity scrub routine and scrub attempts to scrub those extents items whose contents are not fully on disk. To fix it, we just add the @search_commit_root flag back. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Reviewed-by: Miao Xie <miaoxie@huawei.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-09 09:39:40 +08:00
ppath->search_commit_root = 1;
ppath->skip_locking = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
btrfs: use readahead API for scrub Scrub uses a simple tree-enumeration to bring the relevant portions of the extent- and csum-tree into the page cache before starting the scrub-I/O. This is now replaced by using the new readahead-API. During readahead the scrub is being accounted as paused, so it won't hold off transaction commits. This change raises the average disk bandwith utilisation on my test volume from 70% to 90%. On another volume, the time for a test run went down from 89s to 43s. Changes v5: - reada1/2 are now of type struct reada_control * Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-10 18:39:23 +08:00
* trigger the readahead for extent tree csum tree and wait for
* completion. During readahead, the scrub is officially paused
* to not hold off transaction commits
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
*/
logical = base + offset;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
physical_end = physical + nstripes * map->stripe_len;
Btrfs: Introduce BTRFS_BLOCK_GROUP_RAID56_MASK to check raid56 simply So we can check raid56 with: (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) instead of long: (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:44 +08:00
if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
get_raid56_logic_offset(physical_end, num,
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
map, &logic_end, NULL);
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
logic_end += base;
} else {
logic_end = logical + increment * nstripes;
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
wait_event(sctx->list_wait,
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
atomic_read(&sctx->bios_in_flight) == 0);
Btrfs: wrap repeated code into scrub_blocked_if_needed() Just wrap same code into one function scrub_blocked_if_needed(). This make a change that we will move waiting (@workers_pending = 0) before we can wake up commiting transaction(atomic_inc(@scrub_paused)), we must take carefully to not deadlock here. Thread 1 Thread 2 |->btrfs_commit_transaction() |->set trans type(COMMIT_DOING) |->btrfs_scrub_paused()(blocked) |->join_transaction(blocked) Move btrfs_scrub_paused() before setting trans type which means we can still join a transaction when commiting_transaction is blocked. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Suggested-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:16:53 +08:00
scrub_blocked_if_needed(fs_info);
btrfs: use readahead API for scrub Scrub uses a simple tree-enumeration to bring the relevant portions of the extent- and csum-tree into the page cache before starting the scrub-I/O. This is now replaced by using the new readahead-API. During readahead the scrub is being accounted as paused, so it won't hold off transaction commits. This change raises the average disk bandwith utilisation on my test volume from 70% to 90%. On another volume, the time for a test run went down from 89s to 43s. Changes v5: - reada1/2 are now of type struct reada_control * Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-10 18:39:23 +08:00
/* FIXME it might be better to start readahead at commit root */
key_start.objectid = logical;
key_start.type = BTRFS_EXTENT_ITEM_KEY;
key_start.offset = (u64)0;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
key_end.objectid = logic_end;
Btrfs: add a incompatible format change for smaller metadata extent refs We currently store the first key of the tree block inside the reference for the tree block in the extent tree. This takes up quite a bit of space. Make a new key type for metadata which holds the level as the offset and completely removes storing the btrfs_tree_block_info inside the extent ref. This reduces the size from 51 bytes to 33 bytes per extent reference for each tree block. In practice this results in a 30-35% decrease in the size of our extent tree, which means we COW less and can keep more of the extent tree in memory which makes our heavy metadata operations go much faster. This is not an automatic format change, you must enable it at mkfs time or with btrfstune. This patch deals with having metadata stored as either the old format or the new format so it is easy to convert. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-03-08 03:22:04 +08:00
key_end.type = BTRFS_METADATA_ITEM_KEY;
key_end.offset = (u64)-1;
btrfs: use readahead API for scrub Scrub uses a simple tree-enumeration to bring the relevant portions of the extent- and csum-tree into the page cache before starting the scrub-I/O. This is now replaced by using the new readahead-API. During readahead the scrub is being accounted as paused, so it won't hold off transaction commits. This change raises the average disk bandwith utilisation on my test volume from 70% to 90%. On another volume, the time for a test run went down from 89s to 43s. Changes v5: - reada1/2 are now of type struct reada_control * Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-10 18:39:23 +08:00
reada1 = btrfs_reada_add(root, &key_start, &key_end);
key_start.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
key_start.type = BTRFS_EXTENT_CSUM_KEY;
key_start.offset = logical;
key_end.objectid = BTRFS_EXTENT_CSUM_OBJECTID;
key_end.type = BTRFS_EXTENT_CSUM_KEY;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
key_end.offset = logic_end;
btrfs: use readahead API for scrub Scrub uses a simple tree-enumeration to bring the relevant portions of the extent- and csum-tree into the page cache before starting the scrub-I/O. This is now replaced by using the new readahead-API. During readahead the scrub is being accounted as paused, so it won't hold off transaction commits. This change raises the average disk bandwith utilisation on my test volume from 70% to 90%. On another volume, the time for a test run went down from 89s to 43s. Changes v5: - reada1/2 are now of type struct reada_control * Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-10 18:39:23 +08:00
reada2 = btrfs_reada_add(csum_root, &key_start, &key_end);
if (!IS_ERR(reada1))
btrfs_reada_wait(reada1);
if (!IS_ERR(reada2))
btrfs_reada_wait(reada2);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
* collect all data csums for the stripe to avoid seeking during
* the scrub. This might currently (crc32) end up to be about 1MB
*/
btrfs: scrub: add explicit plugging With the removal of the implicit plugging scrub ends up doing more and smaller I/O than necessary. This patch adds explicit plugging per chunk. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-05-29 04:58:38 +08:00
blk_start_plug(&plug);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
* now find all extents for each stripe and scrub them
*/
ret = 0;
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
while (physical < physical_end) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
* canceled?
*/
if (atomic_read(&fs_info->scrub_cancel_req) ||
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
atomic_read(&sctx->cancel_req)) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
ret = -ECANCELED;
goto out;
}
/*
* check to see if we have to pause
*/
if (atomic_read(&fs_info->scrub_pause_req)) {
/* push queued extents */
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
atomic_set(&sctx->wr_ctx.flush_all_writes, 1);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
scrub_submit(sctx);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
mutex_lock(&sctx->wr_ctx.wr_lock);
scrub_wr_submit(sctx);
mutex_unlock(&sctx->wr_ctx.wr_lock);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
wait_event(sctx->list_wait,
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
atomic_read(&sctx->bios_in_flight) == 0);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
atomic_set(&sctx->wr_ctx.flush_all_writes, 0);
Btrfs: fix wrong super generation mismatch when scrubbing supers We came a race condition when scrubbing superblocks, the story is: In commiting transaction, we will update @last_trans_commited after writting superblocks, if scrubber start after writting superblocks and before updating @last_trans_commited, generation mismatch happens! We fix this by checking @scrub_pause_req, and we won't start a srubber until commiting transaction is finished.(after btrfs_scrub_continue() finished.) Reported-by: Sebastian Ochmann <ochmann@informatik.uni-bonn.de> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Reviewed-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:15:19 +08:00
scrub_blocked_if_needed(fs_info);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
btrfs: Check cancel and pause in interval of scrub operation Old code checking cancel and pause request inside scrub stripe operation, like: loop() { if (parity) { scrub_parity_stripe(); continue; } check_cancel_and_pause() scrub_normal_stripe(); } Reason is when introduce raid56 stripe scrub, new code is inserted simplely to front of loop. Better to: loop() { check_cancel_and_pause() if (parity) scrub_parity_stripe(); else scrub_normal_stripe(); } This patch adjusted code place to realize above sequence. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 12:22:29 +08:00
if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
ret = get_raid56_logic_offset(physical, num, map,
&logical,
&stripe_logical);
logical += base;
if (ret) {
btrfs: Update out-of-date "skip parity stripe" comment Because btrfs support scrub raid56 parity stripe now. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-18 17:54:30 +08:00
/* it is parity strip */
btrfs: Check cancel and pause in interval of scrub operation Old code checking cancel and pause request inside scrub stripe operation, like: loop() { if (parity) { scrub_parity_stripe(); continue; } check_cancel_and_pause() scrub_normal_stripe(); } Reason is when introduce raid56 stripe scrub, new code is inserted simplely to front of loop. Better to: loop() { check_cancel_and_pause() if (parity) scrub_parity_stripe(); else scrub_normal_stripe(); } This patch adjusted code place to realize above sequence. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 12:22:29 +08:00
stripe_logical += base;
btrfs: Fix calculate typo caused by ambiguous meaning of logic_end For example, in scrub_raid56_parity(), following lines are used to judge is all data processed: place1: if (key.objectid > logic_end) ... place2: if (logic_start >= logic_end) ... ... (place2 is typo, is should be ">", it is copied from other place, where logic_end's meaning is different, long story...) We can fix above typo directly, but the root reason is ambiguous meaning of logic_end in scrub raid56 parity. In other place, XXX_end is pointed to data which is not included, and we need to process segment of [XXX_start, XXX_end). But for scrub raid56 parity, logic_end is pointed to lattest data need to process, and introduced many "+ 1" and "- 1" in code as below: length = sparity->logic_end - sparity->logic_start + 1 logic_end - logic_start + 1 stripe_logical + increment - 1 This patch changed logic_end's meaning to make it in normal understanding in raid56 parity functions and data struct alone with above bugfix. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 15:42:26 +08:00
stripe_end = stripe_logical + increment;
btrfs: Check cancel and pause in interval of scrub operation Old code checking cancel and pause request inside scrub stripe operation, like: loop() { if (parity) { scrub_parity_stripe(); continue; } check_cancel_and_pause() scrub_normal_stripe(); } Reason is when introduce raid56 stripe scrub, new code is inserted simplely to front of loop. Better to: loop() { check_cancel_and_pause() if (parity) scrub_parity_stripe(); else scrub_normal_stripe(); } This patch adjusted code place to realize above sequence. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 12:22:29 +08:00
ret = scrub_raid56_parity(sctx, map, scrub_dev,
ppath, stripe_logical,
stripe_end);
if (ret)
goto out;
goto skip;
}
}
Btrfs: fix missing skinny metadata check in scrub_stripe() Check if we support skinny metadata firstly and fix to use right type to search. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-12 21:38:32 +08:00
if (btrfs_fs_incompat(fs_info, SKINNY_METADATA))
key.type = BTRFS_METADATA_ITEM_KEY;
else
key.type = BTRFS_EXTENT_ITEM_KEY;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
key.objectid = logical;
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
key.offset = (u64)-1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
if (ret < 0)
goto out;
Btrfs: add a incompatible format change for smaller metadata extent refs We currently store the first key of the tree block inside the reference for the tree block in the extent tree. This takes up quite a bit of space. Make a new key type for metadata which holds the level as the offset and completely removes storing the btrfs_tree_block_info inside the extent ref. This reduces the size from 51 bytes to 33 bytes per extent reference for each tree block. In practice this results in a 30-35% decrease in the size of our extent tree, which means we COW less and can keep more of the extent tree in memory which makes our heavy metadata operations go much faster. This is not an automatic format change, you must enable it at mkfs time or with btrfstune. This patch deals with having metadata stored as either the old format or the new format so it is easy to convert. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-03-08 03:22:04 +08:00
btrfs: scrub: errors in tree enumeration due to the semantics of btrfs_search_slot the path can point to an invalid slot when ret > 0. This condition went unnoticed, which in turn could have led to an incomplete scrubbing. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-03 16:09:26 +08:00
if (ret > 0) {
Btrfs: fix to search previous metadata extent item since skinny metadata There is a bug that using btrfs_previous_item() to search metadata extent item. This is because in btrfs_previous_item(), we need type match, however, since skinny metada was introduced by josef, we may mix this two types. So just use btrfs_previous_item() is not working right. To keep btrfs_previous_item() like normal tree search, i introduce another function btrfs_previous_extent_item(). Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-01-12 21:38:33 +08:00
ret = btrfs_previous_extent_item(root, path, 0);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (ret < 0)
goto out;
btrfs: scrub: errors in tree enumeration due to the semantics of btrfs_search_slot the path can point to an invalid slot when ret > 0. This condition went unnoticed, which in turn could have led to an incomplete scrubbing. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-03 16:09:26 +08:00
if (ret > 0) {
/* there's no smaller item, so stick with the
* larger one */
btrfs_release_path(path);
ret = btrfs_search_slot(NULL, root, &key,
path, 0, 0);
if (ret < 0)
goto out;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
stop_loop = 0;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
while (1) {
Btrfs: add a incompatible format change for smaller metadata extent refs We currently store the first key of the tree block inside the reference for the tree block in the extent tree. This takes up quite a bit of space. Make a new key type for metadata which holds the level as the offset and completely removes storing the btrfs_tree_block_info inside the extent ref. This reduces the size from 51 bytes to 33 bytes per extent reference for each tree block. In practice this results in a 30-35% decrease in the size of our extent tree, which means we COW less and can keep more of the extent tree in memory which makes our heavy metadata operations go much faster. This is not an automatic format change, you must enable it at mkfs time or with btrfstune. This patch deals with having metadata stored as either the old format or the new format so it is easy to convert. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-03-08 03:22:04 +08:00
u64 bytes;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
l = path->nodes[0];
slot = path->slots[0];
if (slot >= btrfs_header_nritems(l)) {
ret = btrfs_next_leaf(root, path);
if (ret == 0)
continue;
if (ret < 0)
goto out;
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
stop_loop = 1;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
break;
}
btrfs_item_key_to_cpu(l, &key, slot);
btrfs: Bypass unrelated items before accessing its contents in scrub When we access extent_root in scrub_stripe() and scrub_raid56_parity(), we need bypass unrelated tree item firstly before using its contents to do other condition. It is not a bug fix, only making code sequence in logic. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-22 13:14:48 +08:00
if (key.type != BTRFS_EXTENT_ITEM_KEY &&
key.type != BTRFS_METADATA_ITEM_KEY)
goto next;
Btrfs: add a incompatible format change for smaller metadata extent refs We currently store the first key of the tree block inside the reference for the tree block in the extent tree. This takes up quite a bit of space. Make a new key type for metadata which holds the level as the offset and completely removes storing the btrfs_tree_block_info inside the extent ref. This reduces the size from 51 bytes to 33 bytes per extent reference for each tree block. In practice this results in a 30-35% decrease in the size of our extent tree, which means we COW less and can keep more of the extent tree in memory which makes our heavy metadata operations go much faster. This is not an automatic format change, you must enable it at mkfs time or with btrfstune. This patch deals with having metadata stored as either the old format or the new format so it is easy to convert. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-03-08 03:22:04 +08:00
if (key.type == BTRFS_METADATA_ITEM_KEY)
btrfs: use nodesize everywhere, kill leafsize The nodesize and leafsize were never of different values. Unify the usage and make nodesize the one. Cleanup the redundant checks and helpers. Shaves a few bytes from .text: text data bss dec hex filename 852418 24560 23112 900090 dbbfa btrfs.ko.before 851074 24584 23112 898770 db6d2 btrfs.ko.after Signed-off-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-06-05 01:22:26 +08:00
bytes = root->nodesize;
Btrfs: add a incompatible format change for smaller metadata extent refs We currently store the first key of the tree block inside the reference for the tree block in the extent tree. This takes up quite a bit of space. Make a new key type for metadata which holds the level as the offset and completely removes storing the btrfs_tree_block_info inside the extent ref. This reduces the size from 51 bytes to 33 bytes per extent reference for each tree block. In practice this results in a 30-35% decrease in the size of our extent tree, which means we COW less and can keep more of the extent tree in memory which makes our heavy metadata operations go much faster. This is not an automatic format change, you must enable it at mkfs time or with btrfstune. This patch deals with having metadata stored as either the old format or the new format so it is easy to convert. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-03-08 03:22:04 +08:00
else
bytes = key.offset;
if (key.objectid + bytes <= logical)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
goto next;
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
if (key.objectid >= logical + map->stripe_len) {
/* out of this device extent */
if (key.objectid >= logic_end)
stop_loop = 1;
break;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
extent = btrfs_item_ptr(l, slot,
struct btrfs_extent_item);
flags = btrfs_extent_flags(l, extent);
generation = btrfs_extent_generation(l, extent);
btrfs: Fix scrub panic when leaf crosses stripes Scrub panic in following operation: mkfs.ext4 /dev/vdh btrfs-convert /dev/vdh mount /dev/vdh /mnt/tmp1 btrfs scrub start -B /dev/vdh (panic) Reason: 1: In some case, leaf created by btrfs-convert was splited into 2 strips. 2: Scrub bypassed part of above wrong leaf data, but remain data caused panic in scrub_checksum_tree_block(). For reason 1: we can get following information after some simple operation. a. mkfs.ext4 /dev/vdh btrfs-convert /dev/vdh b. btrfs-debug-tree /dev/vdh we can see following item in extent tree: item 25 key (27054080 METADATA_ITEM 0) itemoff 15083 itemsize 33 Its logical address is [27054080, 27070464) and acrossed 2 strips: [27000832, 27066368) [27066368, 27131904) Will be fixed in btrfs-progs(btrfs-convert, btrfsck, ...) For reason 2: Scrub is trying to do a "bypass" in this case, but the result is "panic", because current code lacks of some condition in bypass, and let some wrong leaf data escaped. This patch fixed above scrub code. Before patch: # btrfs scrub start -B /dev/vdh (panic) After patch: # btrfs scrub start -B /dev/vdh scrub done for 353cec8f-da31-4a94-aa35-be72d997b06e ... # dmesg ... [ 59.088697] BTRFS error (device vdh): scrub: tree block 27054080 spanning stripes, ignored. logical=27000832 [ 59.089929] BTRFS error (device vdh): scrub: tree block 27054080 spanning stripes, ignored. logical=27066368 # Reported-by: Chris Murphy <lists@colorremedies.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-23 12:29:49 +08:00
if ((flags & BTRFS_EXTENT_FLAG_TREE_BLOCK) &&
(key.objectid < logical ||
key.objectid + bytes >
logical + map->stripe_len)) {
Btrfs: convert printk to btrfs_ and fix BTRFS prefix Convert all applicable cases of printk and pr_* to the btrfs_* macros. Fix all uses of the BTRFS prefix. Signed-off-by: Frank Holton <fholton@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-21 00:37:06 +08:00
btrfs_err(fs_info,
"scrub: tree block %llu spanning "
"stripes, ignored. logical=%llu",
Btrfs: Remove superfluous casts from u64 to unsigned long long u64 is "unsigned long long" on all architectures now, so there's no need to cast it when formatting it using the "ll" length modifier. Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-08-20 19:20:07 +08:00
key.objectid, logical);
btrfs: scrub: set error stats when tree block spanning stripes It is better to show error stats to user when we found tree block spanning stripes. On a btrfs created by old version of btrfs-convert: Before patch: # btrfs scrub start -B /dev/vdh scrub done for 8b342d35-2904-41ab-b3cb-2f929709cf47 scrub started at Tue Aug 25 21:19:09 2015 and finished after 00:00:00 total bytes scrubbed: 53.54MiB with 0 errors # dmesg ... [ 128.711434] BTRFS error (device vdh): scrub: tree block 27054080 spanning stripes, ignored. logical=27000832 [ 128.712744] BTRFS error (device vdh): scrub: tree block 27054080 spanning stripes, ignored. logical=27066368 ... After patch: # btrfs scrub start -B /dev/vdh scrub done for ff7f844b-7a4e-4b1a-88a9-8252ab25be1b scrub started at Tue Aug 25 21:42:29 2015 and finished after 00:00:00 total bytes scrubbed: 53.60MiB with 2 errors error details: corrected errors: 0, uncorrectable errors: 2, unverified errors: 0 ERROR: There are uncorrectable errors. # dmesg ...omit... # Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-25 21:31:40 +08:00
spin_lock(&sctx->stat_lock);
sctx->stat.uncorrectable_errors++;
spin_unlock(&sctx->stat_lock);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
goto next;
}
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
again:
extent_logical = key.objectid;
extent_len = bytes;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/*
* trim extent to this stripe
*/
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
if (extent_logical < logical) {
extent_len -= logical - extent_logical;
extent_logical = logical;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
if (extent_logical + extent_len >
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
logical + map->stripe_len) {
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
extent_len = logical + map->stripe_len -
extent_logical;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
extent_physical = extent_logical - logical + physical;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
extent_dev = scrub_dev;
extent_mirror_num = mirror_num;
if (is_dev_replace)
scrub_remap_extent(fs_info, extent_logical,
extent_len, &extent_physical,
&extent_dev,
&extent_mirror_num);
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
btrfs: Load only necessary csums into list in scrub We need not load csum of whole strip in scrub because strip is trimed before use, it is to say, what we really need to calculate csum is data between [extent_logical, extent_len). This patch changed to use above segment for btrfs_lookup_csums_range() in scrub_stripe() Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-22 13:14:47 +08:00
ret = btrfs_lookup_csums_range(csum_root,
extent_logical,
extent_logical +
extent_len - 1,
&sctx->csum_list, 1);
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
if (ret)
goto out;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
ret = scrub_extent(sctx, extent_logical, extent_len,
extent_physical, extent_dev, flags,
generation, extent_mirror_num,
Btrfs: fix wrong write offset when replacing a device Miao Xie reported the following issue: The filesystem was corrupted after we did a device replace. Steps to reproduce: # mkfs.btrfs -f -m single -d raid10 <device0>..<device3> # mount <device0> <mnt> # btrfs replace start -rfB 1 <device4> <mnt> # umount <mnt> # btrfsck <device4> The reason for the issue is that we changed the write offset by mistake, introduced by commit 625f1c8dc. We read the data from the source device at first, and then write the data into the corresponding place of the new device. In order to implement the "-r" option, the source location is remapped using btrfs_map_block(). The read takes place on the mapped location, and the write needs to take place on the unmapped location. Currently the write is using the mapped location, and this commit changes it back by undoing the change to the write address that the aforementioned commit added by mistake. Reported-by: Miao Xie <miaox@cn.fujitsu.com> Cc: <stable@vger.kernel.org> # 3.10+ Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-07-04 22:14:23 +08:00
extent_logical - logical + physical);
btrfs: Free checksum list on scrub_extent() fail When scrub_extent() failed, we need to free previois created checksum list. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 12:22:30 +08:00
scrub_free_csums(sctx);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (ret)
goto out;
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
if (extent_logical + extent_len <
key.objectid + bytes) {
Btrfs: Introduce BTRFS_BLOCK_GROUP_RAID56_MASK to check raid56 simply So we can check raid56 with: (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) instead of long: (map->type & (BTRFS_BLOCK_GROUP_RAID5 | BTRFS_BLOCK_GROUP_RAID6)) Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:44 +08:00
if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
/*
* loop until we find next data stripe
* or we have finished all stripes.
*/
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
loop:
physical += map->stripe_len;
ret = get_raid56_logic_offset(physical,
num, map, &logical,
&stripe_logical);
logical += base;
if (ret && physical < physical_end) {
stripe_logical += base;
stripe_end = stripe_logical +
btrfs: Fix calculate typo caused by ambiguous meaning of logic_end For example, in scrub_raid56_parity(), following lines are used to judge is all data processed: place1: if (key.objectid > logic_end) ... place2: if (logic_start >= logic_end) ... ... (place2 is typo, is should be ">", it is copied from other place, where logic_end's meaning is different, long story...) We can fix above typo directly, but the root reason is ambiguous meaning of logic_end in scrub raid56 parity. In other place, XXX_end is pointed to data which is not included, and we need to process segment of [XXX_start, XXX_end). But for scrub raid56 parity, logic_end is pointed to lattest data need to process, and introduced many "+ 1" and "- 1" in code as below: length = sparity->logic_end - sparity->logic_start + 1 logic_end - logic_start + 1 stripe_logical + increment - 1 This patch changed logic_end's meaning to make it in normal understanding in raid56 parity functions and data struct alone with above bugfix. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-07-21 15:42:26 +08:00
increment;
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
ret = scrub_raid56_parity(sctx,
map, scrub_dev, ppath,
stripe_logical,
stripe_end);
if (ret)
goto out;
goto loop;
}
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
} else {
physical += map->stripe_len;
logical += increment;
}
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
if (logical < key.objectid + bytes) {
cond_resched();
goto again;
}
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
if (physical >= physical_end) {
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
stop_loop = 1;
break;
}
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
next:
path->slots[0]++;
}
Merge branch 'for-chris' of git://git.kernel.org/pub/scm/linux/kernel/git/arne/btrfs-unstable-arne into inode_numbers Conflicts: fs/btrfs/Makefile fs/btrfs/ctree.h fs/btrfs/volumes.h Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-05-23 18:30:52 +08:00
btrfs_release_path(path);
Btrfs: scrub raid56 stripes in the right way Steps to reproduce: # mkfs.btrfs -f /dev/sda[8-11] -m raid5 -d raid5 # mount /dev/sda8 /mnt # btrfs scrub start -BR /mnt # echo $? <--unverified errors make return value be 3 This is because we don't setup right mapping between physical and logical address for raid56, which makes checksum mismatch. But we will find everthing is fine later when rechecking using btrfs_map_block(). This patch fixed the problem by settuping right mappings and we only verify data stripes' checksums. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-04-01 18:01:43 +08:00
skip:
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
logical += increment;
physical += map->stripe_len;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_lock(&sctx->stat_lock);
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
if (stop_loop)
sctx->stat.last_physical = map->stripes[num].physical +
length;
else
sctx->stat.last_physical = physical;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
spin_unlock(&sctx->stat_lock);
Btrfs: improve the loop of scrub_stripe 1) Right now scrub_stripe() is looping in some unnecessary cases: * when the found extent item's objectid has been out of the dev extent's range but we haven't finish scanning all the range within the dev extent * when all the items has been processed but we haven't finish scanning all the range within the dev extent In both cases, we can just finish the loop to save costs. 2) Besides, when the found extent item's length is larger than the stripe len(64k), we don't have to release the path and search again as it'll get at the same key used in the last loop, we can instead increase the logical cursor in place till all space of the extent is scanned. 3) And we use 0 as the key's offset to search btree, then get to previous item to find a smaller item, and again have to move to the next one to get the right item. Setting offset=-1 and previous_item() is the correct way. 4) As we won't find any checksum at offset unless this 'offset' is in a data extent, we can just find checksum when we're really going to scrub an extent. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-04-27 10:56:57 +08:00
if (stop_loop)
break;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
out:
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
/* push queued extents */
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
scrub_submit(sctx);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
mutex_lock(&sctx->wr_ctx.wr_lock);
scrub_wr_submit(sctx);
mutex_unlock(&sctx->wr_ctx.wr_lock);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: scrub: add explicit plugging With the removal of the implicit plugging scrub ends up doing more and smaller I/O than necessary. This patch adds explicit plugging per chunk. Signed-off-by: Arne Jansen <sensille@gmx.net> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-05-29 04:58:38 +08:00
blk_finish_plug(&plug);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs_free_path(path);
Btrfs, raid56: support parity scrub on raid56 The implementation is: - Read and check all the data with checksum in the same stripe. All the data which has checksum is COW data, and we are sure that it is not changed though we don't lock the stripe. because the space of that data just can be reclaimed after the current transction is committed, and then the fs can use it to store the other data, but when doing scrub, we hold the current transaction, that is that data can not be recovered, it is safe that read and check it out of the stripe lock. - Lock the stripe - Read out all the data without checksum and parity The data without checksum and the parity may be changed if we don't lock the stripe, so we need read it in the stripe lock context. - Check the parity - Re-calculate the new parity and write back it if the old parity is not right - Unlock the stripe If we can not read out the data or the data we read is corrupted, we will try to repair it. If the repair fails. we will mark the horizontal sub-stripe(pages on the same horizontal) as corrupted sub-stripe, and we will skip the parity check and repair of that horizontal sub-stripe. And in order to skip the horizontal sub-stripe that has no data, we introduce a bitmap. If there is some data on the horizontal sub-stripe, we will the relative bit to 1, and when we check and repair the parity, we will skip those horizontal sub-stripes that the relative bits is 0. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
2014-11-06 17:20:58 +08:00
btrfs_free_path(ppath);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return ret < 0 ? ret : 0;
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
static noinline_for_stack int scrub_chunk(struct scrub_ctx *sctx,
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_device *scrub_dev,
u64 chunk_offset, u64 length,
Btrfs: fix race between scrub and block group deletion Scrub can race with the cleaner kthread deleting block groups that are unused (and with relocation too) leading to a failure with error -EINVAL that gets returned to user space. The following diagram illustrates how it happens: CPU 1 CPU 2 cleaner kthread btrfs_delete_unused_bgs() gets block group X from fs_info->unused_bgs sets block group to RO btrfs_remove_chunk(bg X) deletes device extents scrub_enumerate_chunks() searches device tree using its commit root finds device extent for block group X gets block group X from the tree fs_info->block_group_cache_tree (via btrfs_lookup_block_group()) sets bg X to RO (again) btrfs_remove_block_group(bg X) deletes block group from fs_info->block_group_cache_tree removes extent map from fs_info->mapping_tree scrub_chunk(offset X) searches fs_info->mapping_tree for extent map starting at offset X --> doesn't find any such extent map --> returns -EINVAL and scrub errors out to userspace with -EINVAL Fix this by dealing with an extent map lookup failure as an indicator of block group deletion. Issue reproduced with fstest btrfs/071. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-11-19 18:57:20 +08:00
u64 dev_offset,
struct btrfs_block_group_cache *cache,
int is_dev_replace)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
struct btrfs_mapping_tree *map_tree =
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
&sctx->dev_root->fs_info->mapping_tree;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct map_lookup *map;
struct extent_map *em;
int i;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
int ret = 0;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
read_lock(&map_tree->map_tree.lock);
em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
read_unlock(&map_tree->map_tree.lock);
Btrfs: fix race between scrub and block group deletion Scrub can race with the cleaner kthread deleting block groups that are unused (and with relocation too) leading to a failure with error -EINVAL that gets returned to user space. The following diagram illustrates how it happens: CPU 1 CPU 2 cleaner kthread btrfs_delete_unused_bgs() gets block group X from fs_info->unused_bgs sets block group to RO btrfs_remove_chunk(bg X) deletes device extents scrub_enumerate_chunks() searches device tree using its commit root finds device extent for block group X gets block group X from the tree fs_info->block_group_cache_tree (via btrfs_lookup_block_group()) sets bg X to RO (again) btrfs_remove_block_group(bg X) deletes block group from fs_info->block_group_cache_tree removes extent map from fs_info->mapping_tree scrub_chunk(offset X) searches fs_info->mapping_tree for extent map starting at offset X --> doesn't find any such extent map --> returns -EINVAL and scrub errors out to userspace with -EINVAL Fix this by dealing with an extent map lookup failure as an indicator of block group deletion. Issue reproduced with fstest btrfs/071. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-11-19 18:57:20 +08:00
if (!em) {
/*
* Might have been an unused block group deleted by the cleaner
* kthread or relocation.
*/
spin_lock(&cache->lock);
if (!cache->removed)
ret = -EINVAL;
spin_unlock(&cache->lock);
return ret;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: cleanup, stop casting for extent_map->lookup everywhere Overloading extent_map->bdev to struct map_lookup * might have started out as a means to an end, but it's a pattern that's used all over the place now. Let's get rid of the casting and just add a union instead. Signed-off-by: Jeff Mahoney <jeffm@suse.com> Signed-off-by: David Sterba <dsterba@suse.com>
2015-06-03 22:55:48 +08:00
map = em->map_lookup;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (em->start != chunk_offset)
goto out;
if (em->len < length)
goto out;
for (i = 0; i < map->num_stripes; ++i) {
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
if (map->stripes[i].dev->bdev == scrub_dev->bdev &&
btrfs: don't check DUP chunks twice Because scrub enumerates the dev extent tree to find the chunks to scrub, it currently finds each DUP chunk twice and also scrubs it twice. This patch makes sure that scrub_chunk only checks that part of the chunk the dev extent has been found for. This only changes the behaviour for DUP chunks. Reported-and-tested-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Arne Jansen <sensille@gmx.net>
2012-02-09 22:09:02 +08:00
map->stripes[i].physical == dev_offset) {
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
ret = scrub_stripe(sctx, map, scrub_dev, i,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
chunk_offset, length,
is_dev_replace);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (ret)
goto out;
}
}
out:
free_extent_map(em);
return ret;
}
static noinline_for_stack
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
int scrub_enumerate_chunks(struct scrub_ctx *sctx,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct btrfs_device *scrub_dev, u64 start, u64 end,
int is_dev_replace)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
struct btrfs_dev_extent *dev_extent = NULL;
struct btrfs_path *path;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_root *root = sctx->dev_root;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
struct btrfs_fs_info *fs_info = root->fs_info;
u64 length;
u64 chunk_offset;
btrfs: Fix data checksum error cause by replace with io-load. xfstests btrfs/070 sometimes failed. In my test machine, its fail rate is about 30%. In another vm(vmware), its fail rate is about 50%. Reason: btrfs/070 do replace and defrag with fsstress simultaneously, after above operation, checksum error is found by scrub. Actually, it have no relationship with defrag operation, only replace with fsstress can trigger this bug. New data writen to target device have possibility rewrited by old data from source device by replace code in debug, to avoid above problem, we can set target block group to readonly in replace period, so new data requested by other operation will not write to same place with replace code. Before patch(4.1-rc3): 30% failed in 100 xfstests. After patch: 0% failed in 300 xfstests. It also happened in btrfs/071 as it's another scrub with IO load tests. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:30 +08:00
int ret = 0;
btrfs: Continue replace when set_block_ro failed xfstests/011 failed in node with small_size filesystem. Can be reproduced by following script: DEV_LIST="/dev/vdd /dev/vde" DEV_REPLACE="/dev/vdf" do_test() { local mkfs_opt="$1" local size="$2" dmesg -c >/dev/null umount $SCRATCH_MNT &>/dev/null echo mkfs.btrfs -f $mkfs_opt "${DEV_LIST[*]}" mkfs.btrfs -f $mkfs_opt "${DEV_LIST[@]}" || return 1 mount "${DEV_LIST[0]}" $SCRATCH_MNT echo -n "Writing big files" dd if=/dev/urandom of=$SCRATCH_MNT/t0 bs=1M count=1 >/dev/null 2>&1 for ((i = 1; i <= size; i++)); do echo -n . /bin/cp $SCRATCH_MNT/t0 $SCRATCH_MNT/t$i || return 1 done echo echo Start replace btrfs replace start -Bf "${DEV_LIST[0]}" "$DEV_REPLACE" $SCRATCH_MNT || { dmesg return 1 } return 0 } # Set size to value near fs size # for example, 1897 can trigger this bug in 2.6G device. # ./do_test "-d raid1 -m raid1" 1897 System will report replace fail with following warning in dmesg: [ 134.710853] BTRFS: dev_replace from /dev/vdd (devid 1) to /dev/vdf started [ 135.542390] BTRFS: btrfs_scrub_dev(/dev/vdd, 1, /dev/vdf) failed -28 [ 135.543505] ------------[ cut here ]------------ [ 135.544127] WARNING: CPU: 0 PID: 4080 at fs/btrfs/dev-replace.c:428 btrfs_dev_replace_start+0x398/0x440() [ 135.545276] Modules linked in: [ 135.545681] CPU: 0 PID: 4080 Comm: btrfs Not tainted 4.3.0 #256 [ 135.546439] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.8.2-0-g33fbe13 by qemu-project.org 04/01/2014 [ 135.547798] ffffffff81c5bfcf ffff88003cbb3d28 ffffffff817fe7b5 0000000000000000 [ 135.548774] ffff88003cbb3d60 ffffffff810a88f1 ffff88002b030000 00000000ffffffe4 [ 135.549774] ffff88003c080000 ffff88003c082588 ffff88003c28ab60 ffff88003cbb3d70 [ 135.550758] Call Trace: [ 135.551086] [<ffffffff817fe7b5>] dump_stack+0x44/0x55 [ 135.551737] [<ffffffff810a88f1>] warn_slowpath_common+0x81/0xc0 [ 135.552487] [<ffffffff810a89e5>] warn_slowpath_null+0x15/0x20 [ 135.553211] [<ffffffff81448c88>] btrfs_dev_replace_start+0x398/0x440 [ 135.554051] [<ffffffff81412c3e>] btrfs_ioctl+0x1d2e/0x25c0 [ 135.554722] [<ffffffff8114c7ba>] ? __audit_syscall_entry+0xaa/0xf0 [ 135.555506] [<ffffffff8111ab36>] ? current_kernel_time64+0x56/0xa0 [ 135.556304] [<ffffffff81201e3d>] do_vfs_ioctl+0x30d/0x580 [ 135.557009] [<ffffffff8114c7ba>] ? __audit_syscall_entry+0xaa/0xf0 [ 135.557855] [<ffffffff810011d1>] ? do_audit_syscall_entry+0x61/0x70 [ 135.558669] [<ffffffff8120d1c1>] ? __fget_light+0x61/0x90 [ 135.559374] [<ffffffff81202124>] SyS_ioctl+0x74/0x80 [ 135.559987] [<ffffffff81809857>] entry_SYSCALL_64_fastpath+0x12/0x6f [ 135.560842] ---[ end trace 2a5c1fc3205abbdd ]--- Reason: When big data writen to fs, the whole free space will be allocated for data chunk. And operation as scrub need to set_block_ro(), and when there is only one metadata chunk in system(or other metadata chunks are all full), the function will try to allocate a new chunk, and failed because no space in device. Fix: When set_block_ro failed for metadata chunk, it is not a problem because scrub_lock paused commit_trancaction in same time, and metadata are always cowed, so the on-the-fly writepages will not write data into same place with scrub/replace. Let replace continue in this case is no problem. Tested by above script, and xfstests/011, plus 100 times xfstests/070. Changelog v1->v2: 1: Add detail comments in source and commit-message. 2: Add dmesg detail into commit-message. 3: Limit return value of -ENOSPC to be passed. All suggested by: Filipe Manana <fdmanana@gmail.com> Suggested-by: Filipe Manana <fdmanana@gmail.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-11-17 18:46:17 +08:00
int ro_set;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
int slot;
struct extent_buffer *l;
struct btrfs_key key;
struct btrfs_key found_key;
struct btrfs_block_group_cache *cache;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct btrfs_dev_replace *dev_replace = &fs_info->dev_replace;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
btrfs: cleanup, use enum values for btrfs_path reada Replace the integers by enums for better readability. The value 2 does not have any meaning since a717531942f488209dded30f6bc648167bcefa72 "Btrfs: do less aggressive btree readahead" (2009-01-22). Signed-off-by: David Sterba <dsterba@suse.com>
2015-11-27 23:31:35 +08:00
path->reada = READA_FORWARD;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
path->search_commit_root = 1;
path->skip_locking = 1;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
key.objectid = scrub_dev->devid;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
key.offset = 0ull;
key.type = BTRFS_DEV_EXTENT_KEY;
while (1) {
ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
if (ret < 0)
btrfs: scrub: errors in tree enumeration due to the semantics of btrfs_search_slot the path can point to an invalid slot when ret > 0. This condition went unnoticed, which in turn could have led to an incomplete scrubbing. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-03 16:09:26 +08:00
break;
if (ret > 0) {
if (path->slots[0] >=
btrfs_header_nritems(path->nodes[0])) {
ret = btrfs_next_leaf(root, path);
btrfs: Fix data checksum error cause by replace with io-load. xfstests btrfs/070 sometimes failed. In my test machine, its fail rate is about 30%. In another vm(vmware), its fail rate is about 50%. Reason: btrfs/070 do replace and defrag with fsstress simultaneously, after above operation, checksum error is found by scrub. Actually, it have no relationship with defrag operation, only replace with fsstress can trigger this bug. New data writen to target device have possibility rewrited by old data from source device by replace code in debug, to avoid above problem, we can set target block group to readonly in replace period, so new data requested by other operation will not write to same place with replace code. Before patch(4.1-rc3): 30% failed in 100 xfstests. After patch: 0% failed in 300 xfstests. It also happened in btrfs/071 as it's another scrub with IO load tests. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:30 +08:00
if (ret < 0)
break;
if (ret > 0) {
ret = 0;
btrfs: scrub: errors in tree enumeration due to the semantics of btrfs_search_slot the path can point to an invalid slot when ret > 0. This condition went unnoticed, which in turn could have led to an incomplete scrubbing. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-03 16:09:26 +08:00
break;
btrfs: Fix data checksum error cause by replace with io-load. xfstests btrfs/070 sometimes failed. In my test machine, its fail rate is about 30%. In another vm(vmware), its fail rate is about 50%. Reason: btrfs/070 do replace and defrag with fsstress simultaneously, after above operation, checksum error is found by scrub. Actually, it have no relationship with defrag operation, only replace with fsstress can trigger this bug. New data writen to target device have possibility rewrited by old data from source device by replace code in debug, to avoid above problem, we can set target block group to readonly in replace period, so new data requested by other operation will not write to same place with replace code. Before patch(4.1-rc3): 30% failed in 100 xfstests. After patch: 0% failed in 300 xfstests. It also happened in btrfs/071 as it's another scrub with IO load tests. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:30 +08:00
}
} else {
ret = 0;
btrfs: scrub: errors in tree enumeration due to the semantics of btrfs_search_slot the path can point to an invalid slot when ret > 0. This condition went unnoticed, which in turn could have led to an incomplete scrubbing. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-03 16:09:26 +08:00
}
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
l = path->nodes[0];
slot = path->slots[0];
btrfs_item_key_to_cpu(l, &found_key, slot);
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
if (found_key.objectid != scrub_dev->devid)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
break;
btrfs: kill the key type accessor helpers btrfs_set_key_type and btrfs_key_type are used inconsistently along with open coded variants. Other members of btrfs_key are accessed directly without any helpers anyway. Signed-off-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-06-05 00:41:45 +08:00
if (found_key.type != BTRFS_DEV_EXTENT_KEY)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
break;
if (found_key.offset >= end)
break;
if (found_key.offset < key.offset)
break;
dev_extent = btrfs_item_ptr(l, slot, struct btrfs_dev_extent);
length = btrfs_dev_extent_length(l, dev_extent);
btrfs: Skip scrubbing removed chunks to avoid -ENOENT. When run scrub with balance, sometimes -ENOENT will be returned, since in scrub_enumerate_chunks() will search dev_extent in *COMMIT_ROOT*, but btrfs_lookup_block_group() will search block group in *MEMORY*, so if a chunk is removed but not committed, -ENOENT will be returned. However, there is no need to stop scrubbing since other chunks may be scrubbed without problem. So this patch changes the behavior to skip removed chunks and continue to scrub the rest. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-06-19 10:42:51 +08:00
if (found_key.offset + length <= start)
goto skip;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
chunk_offset = btrfs_dev_extent_chunk_offset(l, dev_extent);
/*
* get a reference on the corresponding block group to prevent
* the chunk from going away while we scrub it
*/
cache = btrfs_lookup_block_group(fs_info, chunk_offset);
btrfs: Skip scrubbing removed chunks to avoid -ENOENT. When run scrub with balance, sometimes -ENOENT will be returned, since in scrub_enumerate_chunks() will search dev_extent in *COMMIT_ROOT*, but btrfs_lookup_block_group() will search block group in *MEMORY*, so if a chunk is removed but not committed, -ENOENT will be returned. However, there is no need to stop scrubbing since other chunks may be scrubbed without problem. So this patch changes the behavior to skip removed chunks and continue to scrub the rest. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-06-19 10:42:51 +08:00
/* some chunks are removed but not committed to disk yet,
* continue scrubbing */
if (!cache)
goto skip;
btrfs: Fix data checksum error cause by replace with io-load. xfstests btrfs/070 sometimes failed. In my test machine, its fail rate is about 30%. In another vm(vmware), its fail rate is about 50%. Reason: btrfs/070 do replace and defrag with fsstress simultaneously, after above operation, checksum error is found by scrub. Actually, it have no relationship with defrag operation, only replace with fsstress can trigger this bug. New data writen to target device have possibility rewrited by old data from source device by replace code in debug, to avoid above problem, we can set target block group to readonly in replace period, so new data requested by other operation will not write to same place with replace code. Before patch(4.1-rc3): 30% failed in 100 xfstests. After patch: 0% failed in 300 xfstests. It also happened in btrfs/071 as it's another scrub with IO load tests. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:30 +08:00
/*
* we need call btrfs_inc_block_group_ro() with scrubs_paused,
* to avoid deadlock caused by:
* btrfs_inc_block_group_ro()
* -> btrfs_wait_for_commit()
* -> btrfs_commit_transaction()
* -> btrfs_scrub_pause()
*/
scrub_pause_on(fs_info);
ret = btrfs_inc_block_group_ro(root, cache);
scrub_pause_off(fs_info);
btrfs: Continue replace when set_block_ro failed xfstests/011 failed in node with small_size filesystem. Can be reproduced by following script: DEV_LIST="/dev/vdd /dev/vde" DEV_REPLACE="/dev/vdf" do_test() { local mkfs_opt="$1" local size="$2" dmesg -c >/dev/null umount $SCRATCH_MNT &>/dev/null echo mkfs.btrfs -f $mkfs_opt "${DEV_LIST[*]}" mkfs.btrfs -f $mkfs_opt "${DEV_LIST[@]}" || return 1 mount "${DEV_LIST[0]}" $SCRATCH_MNT echo -n "Writing big files" dd if=/dev/urandom of=$SCRATCH_MNT/t0 bs=1M count=1 >/dev/null 2>&1 for ((i = 1; i <= size; i++)); do echo -n . /bin/cp $SCRATCH_MNT/t0 $SCRATCH_MNT/t$i || return 1 done echo echo Start replace btrfs replace start -Bf "${DEV_LIST[0]}" "$DEV_REPLACE" $SCRATCH_MNT || { dmesg return 1 } return 0 } # Set size to value near fs size # for example, 1897 can trigger this bug in 2.6G device. # ./do_test "-d raid1 -m raid1" 1897 System will report replace fail with following warning in dmesg: [ 134.710853] BTRFS: dev_replace from /dev/vdd (devid 1) to /dev/vdf started [ 135.542390] BTRFS: btrfs_scrub_dev(/dev/vdd, 1, /dev/vdf) failed -28 [ 135.543505] ------------[ cut here ]------------ [ 135.544127] WARNING: CPU: 0 PID: 4080 at fs/btrfs/dev-replace.c:428 btrfs_dev_replace_start+0x398/0x440() [ 135.545276] Modules linked in: [ 135.545681] CPU: 0 PID: 4080 Comm: btrfs Not tainted 4.3.0 #256 [ 135.546439] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.8.2-0-g33fbe13 by qemu-project.org 04/01/2014 [ 135.547798] ffffffff81c5bfcf ffff88003cbb3d28 ffffffff817fe7b5 0000000000000000 [ 135.548774] ffff88003cbb3d60 ffffffff810a88f1 ffff88002b030000 00000000ffffffe4 [ 135.549774] ffff88003c080000 ffff88003c082588 ffff88003c28ab60 ffff88003cbb3d70 [ 135.550758] Call Trace: [ 135.551086] [<ffffffff817fe7b5>] dump_stack+0x44/0x55 [ 135.551737] [<ffffffff810a88f1>] warn_slowpath_common+0x81/0xc0 [ 135.552487] [<ffffffff810a89e5>] warn_slowpath_null+0x15/0x20 [ 135.553211] [<ffffffff81448c88>] btrfs_dev_replace_start+0x398/0x440 [ 135.554051] [<ffffffff81412c3e>] btrfs_ioctl+0x1d2e/0x25c0 [ 135.554722] [<ffffffff8114c7ba>] ? __audit_syscall_entry+0xaa/0xf0 [ 135.555506] [<ffffffff8111ab36>] ? current_kernel_time64+0x56/0xa0 [ 135.556304] [<ffffffff81201e3d>] do_vfs_ioctl+0x30d/0x580 [ 135.557009] [<ffffffff8114c7ba>] ? __audit_syscall_entry+0xaa/0xf0 [ 135.557855] [<ffffffff810011d1>] ? do_audit_syscall_entry+0x61/0x70 [ 135.558669] [<ffffffff8120d1c1>] ? __fget_light+0x61/0x90 [ 135.559374] [<ffffffff81202124>] SyS_ioctl+0x74/0x80 [ 135.559987] [<ffffffff81809857>] entry_SYSCALL_64_fastpath+0x12/0x6f [ 135.560842] ---[ end trace 2a5c1fc3205abbdd ]--- Reason: When big data writen to fs, the whole free space will be allocated for data chunk. And operation as scrub need to set_block_ro(), and when there is only one metadata chunk in system(or other metadata chunks are all full), the function will try to allocate a new chunk, and failed because no space in device. Fix: When set_block_ro failed for metadata chunk, it is not a problem because scrub_lock paused commit_trancaction in same time, and metadata are always cowed, so the on-the-fly writepages will not write data into same place with scrub/replace. Let replace continue in this case is no problem. Tested by above script, and xfstests/011, plus 100 times xfstests/070. Changelog v1->v2: 1: Add detail comments in source and commit-message. 2: Add dmesg detail into commit-message. 3: Limit return value of -ENOSPC to be passed. All suggested by: Filipe Manana <fdmanana@gmail.com> Suggested-by: Filipe Manana <fdmanana@gmail.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-11-17 18:46:17 +08:00
if (ret == 0) {
ro_set = 1;
} else if (ret == -ENOSPC) {
/*
* btrfs_inc_block_group_ro return -ENOSPC when it
* failed in creating new chunk for metadata.
* It is not a problem for scrub/replace, because
* metadata are always cowed, and our scrub paused
* commit_transactions.
*/
ro_set = 0;
} else {
btrfs_warn(fs_info, "failed setting block group ro, ret=%d\n",
ret);
btrfs: Fix data checksum error cause by replace with io-load. xfstests btrfs/070 sometimes failed. In my test machine, its fail rate is about 30%. In another vm(vmware), its fail rate is about 50%. Reason: btrfs/070 do replace and defrag with fsstress simultaneously, after above operation, checksum error is found by scrub. Actually, it have no relationship with defrag operation, only replace with fsstress can trigger this bug. New data writen to target device have possibility rewrited by old data from source device by replace code in debug, to avoid above problem, we can set target block group to readonly in replace period, so new data requested by other operation will not write to same place with replace code. Before patch(4.1-rc3): 30% failed in 100 xfstests. After patch: 0% failed in 300 xfstests. It also happened in btrfs/071 as it's another scrub with IO load tests. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:30 +08:00
btrfs_put_block_group(cache);
break;
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
dev_replace->cursor_right = found_key.offset + length;
dev_replace->cursor_left = found_key.offset;
dev_replace->item_needs_writeback = 1;
btrfs: Remove noused chunk_tree and chunk_objectid from scrub_enumerate_chunks and scrub_chunk These variables are not used from introduced version, remove them. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-19 15:02:40 +08:00
ret = scrub_chunk(sctx, scrub_dev, chunk_offset, length,
Btrfs: fix race between scrub and block group deletion Scrub can race with the cleaner kthread deleting block groups that are unused (and with relocation too) leading to a failure with error -EINVAL that gets returned to user space. The following diagram illustrates how it happens: CPU 1 CPU 2 cleaner kthread btrfs_delete_unused_bgs() gets block group X from fs_info->unused_bgs sets block group to RO btrfs_remove_chunk(bg X) deletes device extents scrub_enumerate_chunks() searches device tree using its commit root finds device extent for block group X gets block group X from the tree fs_info->block_group_cache_tree (via btrfs_lookup_block_group()) sets bg X to RO (again) btrfs_remove_block_group(bg X) deletes block group from fs_info->block_group_cache_tree removes extent map from fs_info->mapping_tree scrub_chunk(offset X) searches fs_info->mapping_tree for extent map starting at offset X --> doesn't find any such extent map --> returns -EINVAL and scrub errors out to userspace with -EINVAL Fix this by dealing with an extent map lookup failure as an indicator of block group deletion. Issue reproduced with fstest btrfs/071. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-11-19 18:57:20 +08:00
found_key.offset, cache, is_dev_replace);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
/*
* flush, submit all pending read and write bios, afterwards
* wait for them.
* Note that in the dev replace case, a read request causes
* write requests that are submitted in the read completion
* worker. Therefore in the current situation, it is required
* that all write requests are flushed, so that all read and
* write requests are really completed when bios_in_flight
* changes to 0.
*/
atomic_set(&sctx->wr_ctx.flush_all_writes, 1);
scrub_submit(sctx);
mutex_lock(&sctx->wr_ctx.wr_lock);
scrub_wr_submit(sctx);
mutex_unlock(&sctx->wr_ctx.wr_lock);
wait_event(sctx->list_wait,
atomic_read(&sctx->bios_in_flight) == 0);
btrfs: use scrub_pause_on/off() to reduce code in scrub_enumerate_chunks() Use new intruduced scrub_pause_on/off() can make this code block clean and more readable. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:29 +08:00
scrub_pause_on(fs_info);
Btrfs: device_replace: fix deadlock for nocow case commit cb7ab02156e4 cause a following deadlock found by xfstests,btrfs/011: Thread1 is commiting transaction which is blocked at btrfs_scrub_pause(). Thread2 is calling btrfs_file_aio_write() which has held inode's @i_mutex and commit transaction(blocked because Thread1 is committing transaction). Thread3 is copy_nocow_page worker which will also try to hold inode @i_mutex, so thread3 will wait Thread1 finished. Thread4 is waiting pending workers finished which will wait Thread3 finished. So the problem is like this: Thread1--->Thread4--->Thread3--->Thread2---->Thread1 Deadlock happens! we fix it by letting Thread1 go firstly, which means we won't block transaction commit while we are waiting pending workers finished. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-19 19:24:17 +08:00
/*
* must be called before we decrease @scrub_paused.
* make sure we don't block transaction commit while
* we are waiting pending workers finished.
*/
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
wait_event(sctx->list_wait,
atomic_read(&sctx->workers_pending) == 0);
Btrfs: device_replace: fix deadlock for nocow case commit cb7ab02156e4 cause a following deadlock found by xfstests,btrfs/011: Thread1 is commiting transaction which is blocked at btrfs_scrub_pause(). Thread2 is calling btrfs_file_aio_write() which has held inode's @i_mutex and commit transaction(blocked because Thread1 is committing transaction). Thread3 is copy_nocow_page worker which will also try to hold inode @i_mutex, so thread3 will wait Thread1 finished. Thread4 is waiting pending workers finished which will wait Thread3 finished. So the problem is like this: Thread1--->Thread4--->Thread3--->Thread2---->Thread1 Deadlock happens! we fix it by letting Thread1 go firstly, which means we won't block transaction commit while we are waiting pending workers finished. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-19 19:24:17 +08:00
atomic_set(&sctx->wr_ctx.flush_all_writes, 0);
btrfs: use scrub_pause_on/off() to reduce code in scrub_enumerate_chunks() Use new intruduced scrub_pause_on/off() can make this code block clean and more readable. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:29 +08:00
scrub_pause_off(fs_info);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
btrfs: Continue replace when set_block_ro failed xfstests/011 failed in node with small_size filesystem. Can be reproduced by following script: DEV_LIST="/dev/vdd /dev/vde" DEV_REPLACE="/dev/vdf" do_test() { local mkfs_opt="$1" local size="$2" dmesg -c >/dev/null umount $SCRATCH_MNT &>/dev/null echo mkfs.btrfs -f $mkfs_opt "${DEV_LIST[*]}" mkfs.btrfs -f $mkfs_opt "${DEV_LIST[@]}" || return 1 mount "${DEV_LIST[0]}" $SCRATCH_MNT echo -n "Writing big files" dd if=/dev/urandom of=$SCRATCH_MNT/t0 bs=1M count=1 >/dev/null 2>&1 for ((i = 1; i <= size; i++)); do echo -n . /bin/cp $SCRATCH_MNT/t0 $SCRATCH_MNT/t$i || return 1 done echo echo Start replace btrfs replace start -Bf "${DEV_LIST[0]}" "$DEV_REPLACE" $SCRATCH_MNT || { dmesg return 1 } return 0 } # Set size to value near fs size # for example, 1897 can trigger this bug in 2.6G device. # ./do_test "-d raid1 -m raid1" 1897 System will report replace fail with following warning in dmesg: [ 134.710853] BTRFS: dev_replace from /dev/vdd (devid 1) to /dev/vdf started [ 135.542390] BTRFS: btrfs_scrub_dev(/dev/vdd, 1, /dev/vdf) failed -28 [ 135.543505] ------------[ cut here ]------------ [ 135.544127] WARNING: CPU: 0 PID: 4080 at fs/btrfs/dev-replace.c:428 btrfs_dev_replace_start+0x398/0x440() [ 135.545276] Modules linked in: [ 135.545681] CPU: 0 PID: 4080 Comm: btrfs Not tainted 4.3.0 #256 [ 135.546439] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.8.2-0-g33fbe13 by qemu-project.org 04/01/2014 [ 135.547798] ffffffff81c5bfcf ffff88003cbb3d28 ffffffff817fe7b5 0000000000000000 [ 135.548774] ffff88003cbb3d60 ffffffff810a88f1 ffff88002b030000 00000000ffffffe4 [ 135.549774] ffff88003c080000 ffff88003c082588 ffff88003c28ab60 ffff88003cbb3d70 [ 135.550758] Call Trace: [ 135.551086] [<ffffffff817fe7b5>] dump_stack+0x44/0x55 [ 135.551737] [<ffffffff810a88f1>] warn_slowpath_common+0x81/0xc0 [ 135.552487] [<ffffffff810a89e5>] warn_slowpath_null+0x15/0x20 [ 135.553211] [<ffffffff81448c88>] btrfs_dev_replace_start+0x398/0x440 [ 135.554051] [<ffffffff81412c3e>] btrfs_ioctl+0x1d2e/0x25c0 [ 135.554722] [<ffffffff8114c7ba>] ? __audit_syscall_entry+0xaa/0xf0 [ 135.555506] [<ffffffff8111ab36>] ? current_kernel_time64+0x56/0xa0 [ 135.556304] [<ffffffff81201e3d>] do_vfs_ioctl+0x30d/0x580 [ 135.557009] [<ffffffff8114c7ba>] ? __audit_syscall_entry+0xaa/0xf0 [ 135.557855] [<ffffffff810011d1>] ? do_audit_syscall_entry+0x61/0x70 [ 135.558669] [<ffffffff8120d1c1>] ? __fget_light+0x61/0x90 [ 135.559374] [<ffffffff81202124>] SyS_ioctl+0x74/0x80 [ 135.559987] [<ffffffff81809857>] entry_SYSCALL_64_fastpath+0x12/0x6f [ 135.560842] ---[ end trace 2a5c1fc3205abbdd ]--- Reason: When big data writen to fs, the whole free space will be allocated for data chunk. And operation as scrub need to set_block_ro(), and when there is only one metadata chunk in system(or other metadata chunks are all full), the function will try to allocate a new chunk, and failed because no space in device. Fix: When set_block_ro failed for metadata chunk, it is not a problem because scrub_lock paused commit_trancaction in same time, and metadata are always cowed, so the on-the-fly writepages will not write data into same place with scrub/replace. Let replace continue in this case is no problem. Tested by above script, and xfstests/011, plus 100 times xfstests/070. Changelog v1->v2: 1: Add detail comments in source and commit-message. 2: Add dmesg detail into commit-message. 3: Limit return value of -ENOSPC to be passed. All suggested by: Filipe Manana <fdmanana@gmail.com> Suggested-by: Filipe Manana <fdmanana@gmail.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-11-17 18:46:17 +08:00
if (ro_set)
btrfs_dec_block_group_ro(root, cache);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
Btrfs: fix scrub preventing unused block groups from being deleted Currently scrub can race with the cleaner kthread when the later attempts to delete an unused block group, and the result is preventing the cleaner kthread from ever deleting later the block group - unless the block group becomes used and unused again. The following diagram illustrates that race: CPU 1 CPU 2 cleaner kthread btrfs_delete_unused_bgs() gets block group X from fs_info->unused_bgs and removes it from that list scrub_enumerate_chunks() searches device tree using its commit root finds device extent for block group X gets block group X from the tree fs_info->block_group_cache_tree (via btrfs_lookup_block_group()) sets bg X to RO sees the block group is already RO and therefore doesn't delete it nor adds it back to unused list So fix this by making scrub add the block group again to the list of unused block groups if the block group is still unused when it finished scrubbing it and it hasn't been removed already. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-11-19 19:45:48 +08:00
/*
* We might have prevented the cleaner kthread from deleting
* this block group if it was already unused because we raced
* and set it to RO mode first. So add it back to the unused
* list, otherwise it might not ever be deleted unless a manual
* balance is triggered or it becomes used and unused again.
*/
spin_lock(&cache->lock);
if (!cache->removed && !cache->ro && cache->reserved == 0 &&
btrfs_block_group_used(&cache->item) == 0) {
spin_unlock(&cache->lock);
spin_lock(&fs_info->unused_bgs_lock);
if (list_empty(&cache->bg_list)) {
btrfs_get_block_group(cache);
list_add_tail(&cache->bg_list,
&fs_info->unused_bgs);
}
spin_unlock(&fs_info->unused_bgs_lock);
} else {
spin_unlock(&cache->lock);
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs_put_block_group(cache);
if (ret)
break;
Btrfs: fix a scrub regression in case of write errors This regression was introduced by the device-replace patches. Scrub immediately stops checking those disks that have write errors. This is nothing that happens in the real world, but it is wrong since scrub is the tool to detect and repair defects. Fix it. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-28 01:39:51 +08:00
if (is_dev_replace &&
atomic64_read(&dev_replace->num_write_errors) > 0) {
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
ret = -EIO;
break;
}
if (sctx->stat.malloc_errors > 0) {
ret = -ENOMEM;
break;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: fix the dev-replace suspend sequence Replace progresses strictly from lower to higher offsets, and the progress is tracked in chunks, by storing the physical offset of the dev_extent which is being copied in the cursor_left field of btrfs_dev_replace_item. When we are done copying the chunk, left_cursor is updated to point one byte past the dev_extent, so that on resume we can skip the dev_extents that have already been copied. There is a major bug (which goes all the way back to the inception of dev-replace in 3.8) in the way left_cursor is bumped: the bump is done unconditionally, without any regard to the scrub_chunk return value. On suspend (and also on any kind of error) scrub_chunk returns early, i.e. without completing the copy. This leads to us skipping the chunk that hasn't been fully copied yet when resuming. Fix this by doing the cursor_left update only if scrub_chunk ret is 0. (On suspend scrub_chunk returns with -ECANCELED, so this fix covers both suspend and error cases.) Cc: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Ilya Dryomov <idryomov@gmail.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-10-07 18:42:57 +08:00
dev_replace->cursor_left = dev_replace->cursor_right;
dev_replace->item_needs_writeback = 1;
btrfs: Skip scrubbing removed chunks to avoid -ENOENT. When run scrub with balance, sometimes -ENOENT will be returned, since in scrub_enumerate_chunks() will search dev_extent in *COMMIT_ROOT*, but btrfs_lookup_block_group() will search block group in *MEMORY*, so if a chunk is removed but not committed, -ENOENT will be returned. However, there is no need to stop scrubbing since other chunks may be scrubbed without problem. So this patch changes the behavior to skip removed chunks and continue to scrub the rest. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-06-19 10:42:51 +08:00
skip:
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
key.offset = found_key.offset + length;
Merge branch 'for-chris' of git://git.kernel.org/pub/scm/linux/kernel/git/arne/btrfs-unstable-arne into inode_numbers Conflicts: fs/btrfs/Makefile fs/btrfs/ctree.h fs/btrfs/volumes.h Signed-off-by: Chris Mason <chris.mason@oracle.com>
2011-05-23 18:30:52 +08:00
btrfs_release_path(path);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
btrfs_free_path(path);
btrfs: scrub: errors in tree enumeration due to the semantics of btrfs_search_slot the path can point to an invalid slot when ret > 0. This condition went unnoticed, which in turn could have led to an incomplete scrubbing. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-03 16:09:26 +08:00
btrfs: Fix data checksum error cause by replace with io-load. xfstests btrfs/070 sometimes failed. In my test machine, its fail rate is about 30%. In another vm(vmware), its fail rate is about 50%. Reason: btrfs/070 do replace and defrag with fsstress simultaneously, after above operation, checksum error is found by scrub. Actually, it have no relationship with defrag operation, only replace with fsstress can trigger this bug. New data writen to target device have possibility rewrited by old data from source device by replace code in debug, to avoid above problem, we can set target block group to readonly in replace period, so new data requested by other operation will not write to same place with replace code. Before patch(4.1-rc3): 30% failed in 100 xfstests. After patch: 0% failed in 300 xfstests. It also happened in btrfs/071 as it's another scrub with IO load tests. Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-08-05 16:43:30 +08:00
return ret;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx,
struct btrfs_device *scrub_dev)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
int i;
u64 bytenr;
u64 gen;
int ret;
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
struct btrfs_root *root = sctx->dev_root;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: use bit operation for ->fs_state There is no lock to protect fs_info->fs_state, it will introduce some problems, such as the value may be covered by the other task when several tasks modify it. For example: Task0 - CPU0 Task1 - CPU1 mov %fs_state rax or $0x1 rax mov %fs_state rax or $0x2 rax mov rax %fs_state mov rax %fs_state The expected value is 3, but in fact, it is 2. Though this problem doesn't happen now (because there is only one flag currently), the code is error prone, if we add other flags, the above problem will happen to a certainty. Now we use bit operation for it to fix the above problem. In this way, we can make the code more robust and be easy to add new flags. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 18:14:48 +08:00
if (test_bit(BTRFS_FS_STATE_ERROR, &root->fs_info->fs_state))
btrfs: replace many BUG_ONs with proper error handling btrfs currently handles most errors with BUG_ON. This patch is a work-in- progress but aims to handle most errors other than internal logic errors and ENOMEM more gracefully. This iteration prevents most crashes but can run into lockups with the page lock on occasion when the timing "works out." Signed-off-by: Jeff Mahoney <jeffm@suse.com>
2012-03-12 23:03:00 +08:00
return -EIO;
Btrfs: fix wrong generation check of super block on a seed device The super block generation of the seed devices is not the same as the filesystem which sprouted from them because we don't update the super block on the seed devices when we change that new filesystem. So we should not use the generation of that new filesystem to check the super block generation on the seed devices, Fix it. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-24 11:37:09 +08:00
/* Seed devices of a new filesystem has their own generation. */
if (scrub_dev->fs_devices != root->fs_info->fs_devices)
gen = scrub_dev->generation;
else
gen = root->fs_info->last_trans_committed;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
bytenr = btrfs_sb_offset(i);
Btrfs: fix wrong disk size when writing super blocks total_size will be changed when resizing a device, and disk_total_size will be changed if resizing is successful. Meanwhile, the on-disk super blocks of the previous transaction might not be updated. Considering the consistency of the metadata in the previous transaction, We should use the size in the previous transaction to check if the super block is beyond the boundary of the device. Fix it. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-09-03 21:35:33 +08:00
if (bytenr + BTRFS_SUPER_INFO_SIZE >
scrub_dev->commit_total_bytes)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
break;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
ret = scrub_pages(sctx, bytenr, BTRFS_SUPER_INFO_SIZE, bytenr,
Btrfs: remove the block device pointer from the scrub context struct The block device is removed from the scrub context state structure. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in the scrub context struct and moved into the lower level scope of scrub_bio, fixup and page structures where the block device context is known. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 20:26:57 +08:00
scrub_dev, BTRFS_EXTENT_FLAG_SUPER, gen, i,
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
NULL, 1, bytenr);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (ret)
return ret;
}
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
wait_event(sctx->list_wait, atomic_read(&sctx->bios_in_flight) == 0);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return 0;
}
/*
* get a reference count on fs_info->scrub_workers. start worker if necessary
*/
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static noinline_for_stack int scrub_workers_get(struct btrfs_fs_info *fs_info,
int is_dev_replace)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
btrfs: use correct type for workqueue flags Through all the local wrappers to alloc_workqueue, __alloc_workqueue_key takes an unsigned int. Signed-off-by: David Sterba <dsterba@suse.cz>
2015-02-17 01:34:01 +08:00
unsigned int flags = WQ_FREEZABLE | WQ_UNBOUND;
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
int max_active = fs_info->thread_pool_size;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
btrfs: reinitialize scrub workers Scrub starts the workers each time a scrub starts and stops them after it finished. This patch adds an initialization for the workers before each start, otherwise the workers behave strangely. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-10 18:07:07 +08:00
if (fs_info->scrub_workers_refcnt == 0) {
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (is_dev_replace)
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
fs_info->scrub_workers =
btrfs: drop duplicate prefix from scrub workqueues The helper btrfs_alloc_workqueue will add the "btrfs-" prefix. Signed-off-by: David Sterba <dsterba@suse.com>
2015-12-02 01:09:12 +08:00
btrfs_alloc_workqueue("scrub", flags,
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
1, 4);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
else
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
fs_info->scrub_workers =
btrfs: drop duplicate prefix from scrub workqueues The helper btrfs_alloc_workqueue will add the "btrfs-" prefix. Signed-off-by: David Sterba <dsterba@suse.com>
2015-12-02 01:09:12 +08:00
btrfs_alloc_workqueue("scrub", flags,
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
max_active, 4);
btrfs: add error handling for scrub_workers_get() Although it is a rare case, we'd better free previous allocated memory on error. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-12 20:36:58 +08:00
if (!fs_info->scrub_workers)
goto fail_scrub_workers;
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
fs_info->scrub_wr_completion_workers =
btrfs: drop duplicate prefix from scrub workqueues The helper btrfs_alloc_workqueue will add the "btrfs-" prefix. Signed-off-by: David Sterba <dsterba@suse.com>
2015-12-02 01:09:12 +08:00
btrfs_alloc_workqueue("scrubwrc", flags,
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
max_active, 2);
btrfs: add error handling for scrub_workers_get() Although it is a rare case, we'd better free previous allocated memory on error. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-12 20:36:58 +08:00
if (!fs_info->scrub_wr_completion_workers)
goto fail_scrub_wr_completion_workers;
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
fs_info->scrub_nocow_workers =
btrfs: drop duplicate prefix from scrub workqueues The helper btrfs_alloc_workqueue will add the "btrfs-" prefix. Signed-off-by: David Sterba <dsterba@suse.com>
2015-12-02 01:09:12 +08:00
btrfs_alloc_workqueue("scrubnc", flags, 1, 0);
btrfs: add error handling for scrub_workers_get() Although it is a rare case, we'd better free previous allocated memory on error. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-12 20:36:58 +08:00
if (!fs_info->scrub_nocow_workers)
goto fail_scrub_nocow_workers;
btrfs: Fix lockdep warning of wr_ctx->wr_lock in scrub_free_wr_ctx() lockdep report following warning in test: [25176.843958] ================================= [25176.844519] [ INFO: inconsistent lock state ] [25176.845047] 4.1.0-rc3 #22 Tainted: G W [25176.845591] --------------------------------- [25176.846153] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage. [25176.846713] fsstress/26661 [HC0[0]:SC1[1]:HE1:SE0] takes: [25176.847246] (&wr_ctx->wr_lock){+.?...}, at: [<ffffffffa04cdc6d>] scrub_free_ctx+0x2d/0xf0 [btrfs] [25176.847838] {SOFTIRQ-ON-W} state was registered at: [25176.848396] [<ffffffff810bf460>] __lock_acquire+0x6a0/0xe10 [25176.848955] [<ffffffff810bfd1e>] lock_acquire+0xce/0x2c0 [25176.849491] [<ffffffff816489af>] mutex_lock_nested+0x7f/0x410 [25176.850029] [<ffffffffa04d04ff>] scrub_stripe+0x4df/0x1080 [btrfs] [25176.850575] [<ffffffffa04d11b1>] scrub_chunk.isra.19+0x111/0x130 [btrfs] [25176.851110] [<ffffffffa04d144c>] scrub_enumerate_chunks+0x27c/0x510 [btrfs] [25176.851660] [<ffffffffa04d3b87>] btrfs_scrub_dev+0x1c7/0x6c0 [btrfs] [25176.852189] [<ffffffffa04e918e>] btrfs_dev_replace_start+0x36e/0x450 [btrfs] [25176.852771] [<ffffffffa04a98e0>] btrfs_ioctl+0x1e10/0x2d20 [btrfs] [25176.853315] [<ffffffff8121c5b8>] do_vfs_ioctl+0x318/0x570 [25176.853868] [<ffffffff8121c851>] SyS_ioctl+0x41/0x80 [25176.854406] [<ffffffff8164da17>] system_call_fastpath+0x12/0x6f [25176.854935] irq event stamp: 51506 [25176.855511] hardirqs last enabled at (51506): [<ffffffff810d4ce5>] vprintk_emit+0x225/0x5e0 [25176.856059] hardirqs last disabled at (51505): [<ffffffff810d4b77>] vprintk_emit+0xb7/0x5e0 [25176.856642] softirqs last enabled at (50886): [<ffffffff81067a23>] __do_softirq+0x363/0x640 [25176.857184] softirqs last disabled at (50949): [<ffffffff8106804d>] irq_exit+0x10d/0x120 [25176.857746] other info that might help us debug this: [25176.858845] Possible unsafe locking scenario: [25176.859981] CPU0 [25176.860537] ---- [25176.861059] lock(&wr_ctx->wr_lock); [25176.861705] <Interrupt> [25176.862272] lock(&wr_ctx->wr_lock); [25176.862881] *** DEADLOCK *** Reason: Above warning is caused by: Interrupt -> bio_endio() -> ... -> scrub_put_ctx() -> scrub_free_ctx() *1 -> ... -> mutex_lock(&wr_ctx->wr_lock); scrub_put_ctx() is allowed to be called in end_bio interrupt, but in code design, it will never call scrub_free_ctx(sctx) in interrupe context(above *1), because btrfs_scrub_dev() get one additional reference of sctx->refs, which makes scrub_free_ctx() only called withine btrfs_scrub_dev(). Now the code runs out of our wish, because free sequence in scrub_pending_bio_dec() have a gap. Current code: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| -> scrub_free_ctx() | -----------------------------------+----------------------------------- We expected: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) | -> scrub_free_ctx() -----------------------------------+----------------------------------- Fix: Move scrub_pending_bio_dec() to a workqueue, to avoid this function run in interrupt context. Tested by check tracelog in debug. Changelog v1->v2: Use workqueue instead of adjust function call sequence in v1, because v1 will introduce a bug pointed out by: Filipe David Manana <fdmanana@gmail.com> Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Reviewed-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-04 20:09:15 +08:00
fs_info->scrub_parity_workers =
btrfs: drop duplicate prefix from scrub workqueues The helper btrfs_alloc_workqueue will add the "btrfs-" prefix. Signed-off-by: David Sterba <dsterba@suse.com>
2015-12-02 01:09:12 +08:00
btrfs_alloc_workqueue("scrubparity", flags,
btrfs: Fix lockdep warning of wr_ctx->wr_lock in scrub_free_wr_ctx() lockdep report following warning in test: [25176.843958] ================================= [25176.844519] [ INFO: inconsistent lock state ] [25176.845047] 4.1.0-rc3 #22 Tainted: G W [25176.845591] --------------------------------- [25176.846153] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage. [25176.846713] fsstress/26661 [HC0[0]:SC1[1]:HE1:SE0] takes: [25176.847246] (&wr_ctx->wr_lock){+.?...}, at: [<ffffffffa04cdc6d>] scrub_free_ctx+0x2d/0xf0 [btrfs] [25176.847838] {SOFTIRQ-ON-W} state was registered at: [25176.848396] [<ffffffff810bf460>] __lock_acquire+0x6a0/0xe10 [25176.848955] [<ffffffff810bfd1e>] lock_acquire+0xce/0x2c0 [25176.849491] [<ffffffff816489af>] mutex_lock_nested+0x7f/0x410 [25176.850029] [<ffffffffa04d04ff>] scrub_stripe+0x4df/0x1080 [btrfs] [25176.850575] [<ffffffffa04d11b1>] scrub_chunk.isra.19+0x111/0x130 [btrfs] [25176.851110] [<ffffffffa04d144c>] scrub_enumerate_chunks+0x27c/0x510 [btrfs] [25176.851660] [<ffffffffa04d3b87>] btrfs_scrub_dev+0x1c7/0x6c0 [btrfs] [25176.852189] [<ffffffffa04e918e>] btrfs_dev_replace_start+0x36e/0x450 [btrfs] [25176.852771] [<ffffffffa04a98e0>] btrfs_ioctl+0x1e10/0x2d20 [btrfs] [25176.853315] [<ffffffff8121c5b8>] do_vfs_ioctl+0x318/0x570 [25176.853868] [<ffffffff8121c851>] SyS_ioctl+0x41/0x80 [25176.854406] [<ffffffff8164da17>] system_call_fastpath+0x12/0x6f [25176.854935] irq event stamp: 51506 [25176.855511] hardirqs last enabled at (51506): [<ffffffff810d4ce5>] vprintk_emit+0x225/0x5e0 [25176.856059] hardirqs last disabled at (51505): [<ffffffff810d4b77>] vprintk_emit+0xb7/0x5e0 [25176.856642] softirqs last enabled at (50886): [<ffffffff81067a23>] __do_softirq+0x363/0x640 [25176.857184] softirqs last disabled at (50949): [<ffffffff8106804d>] irq_exit+0x10d/0x120 [25176.857746] other info that might help us debug this: [25176.858845] Possible unsafe locking scenario: [25176.859981] CPU0 [25176.860537] ---- [25176.861059] lock(&wr_ctx->wr_lock); [25176.861705] <Interrupt> [25176.862272] lock(&wr_ctx->wr_lock); [25176.862881] *** DEADLOCK *** Reason: Above warning is caused by: Interrupt -> bio_endio() -> ... -> scrub_put_ctx() -> scrub_free_ctx() *1 -> ... -> mutex_lock(&wr_ctx->wr_lock); scrub_put_ctx() is allowed to be called in end_bio interrupt, but in code design, it will never call scrub_free_ctx(sctx) in interrupe context(above *1), because btrfs_scrub_dev() get one additional reference of sctx->refs, which makes scrub_free_ctx() only called withine btrfs_scrub_dev(). Now the code runs out of our wish, because free sequence in scrub_pending_bio_dec() have a gap. Current code: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| -> scrub_free_ctx() | -----------------------------------+----------------------------------- We expected: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) | -> scrub_free_ctx() -----------------------------------+----------------------------------- Fix: Move scrub_pending_bio_dec() to a workqueue, to avoid this function run in interrupt context. Tested by check tracelog in debug. Changelog v1->v2: Use workqueue instead of adjust function call sequence in v1, because v1 will introduce a bug pointed out by: Filipe David Manana <fdmanana@gmail.com> Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Reviewed-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-04 20:09:15 +08:00
max_active, 2);
btrfs: add error handling for scrub_workers_get() Although it is a rare case, we'd better free previous allocated memory on error. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-12 20:36:58 +08:00
if (!fs_info->scrub_parity_workers)
goto fail_scrub_parity_workers;
btrfs: reinitialize scrub workers Scrub starts the workers each time a scrub starts and stops them after it finished. This patch adds an initialization for the workers before each start, otherwise the workers behave strangely. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-06-10 18:07:07 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
++fs_info->scrub_workers_refcnt;
btrfs: add error handling for scrub_workers_get() Although it is a rare case, we'd better free previous allocated memory on error. Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-12 20:36:58 +08:00
return 0;
fail_scrub_parity_workers:
btrfs_destroy_workqueue(fs_info->scrub_nocow_workers);
fail_scrub_nocow_workers:
btrfs_destroy_workqueue(fs_info->scrub_wr_completion_workers);
fail_scrub_wr_completion_workers:
btrfs_destroy_workqueue(fs_info->scrub_workers);
fail_scrub_workers:
return -ENOMEM;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
static noinline_for_stack void scrub_workers_put(struct btrfs_fs_info *fs_info)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (--fs_info->scrub_workers_refcnt == 0) {
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
btrfs_destroy_workqueue(fs_info->scrub_workers);
btrfs_destroy_workqueue(fs_info->scrub_wr_completion_workers);
btrfs_destroy_workqueue(fs_info->scrub_nocow_workers);
btrfs: Fix lockdep warning of wr_ctx->wr_lock in scrub_free_wr_ctx() lockdep report following warning in test: [25176.843958] ================================= [25176.844519] [ INFO: inconsistent lock state ] [25176.845047] 4.1.0-rc3 #22 Tainted: G W [25176.845591] --------------------------------- [25176.846153] inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage. [25176.846713] fsstress/26661 [HC0[0]:SC1[1]:HE1:SE0] takes: [25176.847246] (&wr_ctx->wr_lock){+.?...}, at: [<ffffffffa04cdc6d>] scrub_free_ctx+0x2d/0xf0 [btrfs] [25176.847838] {SOFTIRQ-ON-W} state was registered at: [25176.848396] [<ffffffff810bf460>] __lock_acquire+0x6a0/0xe10 [25176.848955] [<ffffffff810bfd1e>] lock_acquire+0xce/0x2c0 [25176.849491] [<ffffffff816489af>] mutex_lock_nested+0x7f/0x410 [25176.850029] [<ffffffffa04d04ff>] scrub_stripe+0x4df/0x1080 [btrfs] [25176.850575] [<ffffffffa04d11b1>] scrub_chunk.isra.19+0x111/0x130 [btrfs] [25176.851110] [<ffffffffa04d144c>] scrub_enumerate_chunks+0x27c/0x510 [btrfs] [25176.851660] [<ffffffffa04d3b87>] btrfs_scrub_dev+0x1c7/0x6c0 [btrfs] [25176.852189] [<ffffffffa04e918e>] btrfs_dev_replace_start+0x36e/0x450 [btrfs] [25176.852771] [<ffffffffa04a98e0>] btrfs_ioctl+0x1e10/0x2d20 [btrfs] [25176.853315] [<ffffffff8121c5b8>] do_vfs_ioctl+0x318/0x570 [25176.853868] [<ffffffff8121c851>] SyS_ioctl+0x41/0x80 [25176.854406] [<ffffffff8164da17>] system_call_fastpath+0x12/0x6f [25176.854935] irq event stamp: 51506 [25176.855511] hardirqs last enabled at (51506): [<ffffffff810d4ce5>] vprintk_emit+0x225/0x5e0 [25176.856059] hardirqs last disabled at (51505): [<ffffffff810d4b77>] vprintk_emit+0xb7/0x5e0 [25176.856642] softirqs last enabled at (50886): [<ffffffff81067a23>] __do_softirq+0x363/0x640 [25176.857184] softirqs last disabled at (50949): [<ffffffff8106804d>] irq_exit+0x10d/0x120 [25176.857746] other info that might help us debug this: [25176.858845] Possible unsafe locking scenario: [25176.859981] CPU0 [25176.860537] ---- [25176.861059] lock(&wr_ctx->wr_lock); [25176.861705] <Interrupt> [25176.862272] lock(&wr_ctx->wr_lock); [25176.862881] *** DEADLOCK *** Reason: Above warning is caused by: Interrupt -> bio_endio() -> ... -> scrub_put_ctx() -> scrub_free_ctx() *1 -> ... -> mutex_lock(&wr_ctx->wr_lock); scrub_put_ctx() is allowed to be called in end_bio interrupt, but in code design, it will never call scrub_free_ctx(sctx) in interrupe context(above *1), because btrfs_scrub_dev() get one additional reference of sctx->refs, which makes scrub_free_ctx() only called withine btrfs_scrub_dev(). Now the code runs out of our wish, because free sequence in scrub_pending_bio_dec() have a gap. Current code: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| -> scrub_free_ctx() | -----------------------------------+----------------------------------- We expected: -----------------------------------+----------------------------------- scrub_pending_bio_dec() | btrfs_scrub_dev -----------------------------------+----------------------------------- atomic_dec(&sctx->bios_in_flight); | wake_up(&sctx->list_wait); | scrub_put_ctx(sctx); | -> atomic_dec_and_test(&sctx->refs)| | scrub_put_ctx() | -> atomic_dec_and_test(&sctx->refs) | -> scrub_free_ctx() -----------------------------------+----------------------------------- Fix: Move scrub_pending_bio_dec() to a workqueue, to avoid this function run in interrupt context. Tested by check tracelog in debug. Changelog v1->v2: Use workqueue instead of adjust function call sequence in v1, because v1 will introduce a bug pointed out by: Filipe David Manana <fdmanana@gmail.com> Reported-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Reviewed-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-06-04 20:09:15 +08:00
btrfs_destroy_workqueue(fs_info->scrub_parity_workers);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
WARN_ON(fs_info->scrub_workers_refcnt < 0);
}
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
int btrfs_scrub_dev(struct btrfs_fs_info *fs_info, u64 devid, u64 start,
u64 end, struct btrfs_scrub_progress *progress,
Btrfs: disallow some operations on the device replace target device This patch adds some code to disallow operations on the device that is used as the target for the device replace operation. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 01:29:28 +08:00
int readonly, int is_dev_replace)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
int ret;
struct btrfs_device *dev;
Btrfs: don't write any data into a readonly device when scrub We should not write data into a readonly device especially seed device when doing scrub, skip those devices. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-24 11:37:07 +08:00
struct rcu_string *name;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
if (btrfs_fs_closing(fs_info))
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return -EINVAL;
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
if (fs_info->chunk_root->nodesize > BTRFS_STRIPE_LEN) {
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/*
* in this case scrub is unable to calculate the checksum
* the way scrub is implemented. Do not handle this
* situation at all because it won't ever happen.
*/
Btrfs: convert printk to btrfs_ and fix BTRFS prefix Convert all applicable cases of printk and pr_* to the btrfs_* macros. Fix all uses of the BTRFS prefix. Signed-off-by: Frank Holton <fholton@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-21 00:37:06 +08:00
btrfs_err(fs_info,
"scrub: size assumption nodesize <= BTRFS_STRIPE_LEN (%d <= %d) fails",
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
fs_info->chunk_root->nodesize, BTRFS_STRIPE_LEN);
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
return -EINVAL;
}
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
if (fs_info->chunk_root->sectorsize != PAGE_SIZE) {
Btrfs: change scrub to support big blocks Scrub used to be coded for nodesize == leafsize == sectorsize == PAGE_SIZE. This is now changed to support sizes for nodesize and leafsize which are N * PAGE_SIZE. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:27 +08:00
/* not supported for data w/o checksums */
Btrfs: convert printk to btrfs_ and fix BTRFS prefix Convert all applicable cases of printk and pr_* to the btrfs_* macros. Fix all uses of the BTRFS prefix. Signed-off-by: Frank Holton <fholton@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-21 00:37:06 +08:00
btrfs_err(fs_info,
"scrub: size assumption sectorsize != PAGE_SIZE "
"(%d != %lu) fails",
Btrfs: Format PAGE_SIZE as unsigned long PAGE_SIZE is "unsigned long" everywhere, so there's no need to cast it to "unsigned long long" and format it as a 64-bit number. Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-08-20 19:20:09 +08:00
fs_info->chunk_root->sectorsize, PAGE_SIZE);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return -EINVAL;
}
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
if (fs_info->chunk_root->nodesize >
PAGE_SIZE * SCRUB_MAX_PAGES_PER_BLOCK ||
fs_info->chunk_root->sectorsize >
PAGE_SIZE * SCRUB_MAX_PAGES_PER_BLOCK) {
/*
* would exhaust the array bounds of pagev member in
* struct scrub_block
*/
Btrfs: convert printk to btrfs_ and fix BTRFS prefix Convert all applicable cases of printk and pr_* to the btrfs_* macros. Fix all uses of the BTRFS prefix. Signed-off-by: Frank Holton <fholton@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-21 00:37:06 +08:00
btrfs_err(fs_info, "scrub: size assumption nodesize and sectorsize "
"<= SCRUB_MAX_PAGES_PER_BLOCK (%d <= %d && %d <= %d) fails",
Btrfs: make the scrub page array dynamically allocated With the modified design (in order to support the devive replace procedure) it is necessary to alloc the page array dynamically. The reason is that pages are reused. At first a page is used for the bio to read the data from the filesystem, then the same page is reused for the bio that writes the data to the target disk. Since the read process and the write process are completely decoupled, this requires a new concept of refcounts and get/put functions for pages, and it requires to use newly created pages for each read bio which are freed after the write operation is finished. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 21:58:04 +08:00
fs_info->chunk_root->nodesize,
SCRUB_MAX_PAGES_PER_BLOCK,
fs_info->chunk_root->sectorsize,
SCRUB_MAX_PAGES_PER_BLOCK);
return -EINVAL;
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
mutex_lock(&fs_info->fs_devices->device_list_mutex);
dev = btrfs_find_device(fs_info, devid, NULL, NULL);
Btrfs: disallow some operations on the device replace target device This patch adds some code to disallow operations on the device that is used as the target for the device replace operation. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 01:29:28 +08:00
if (!dev || (dev->missing && !is_dev_replace)) {
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
mutex_unlock(&fs_info->fs_devices->device_list_mutex);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return -ENODEV;
}
Btrfs: don't write any data into a readonly device when scrub We should not write data into a readonly device especially seed device when doing scrub, skip those devices. Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Reviewed-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-07-24 11:37:07 +08:00
if (!is_dev_replace && !readonly && !dev->writeable) {
mutex_unlock(&fs_info->fs_devices->device_list_mutex);
rcu_read_lock();
name = rcu_dereference(dev->name);
btrfs_err(fs_info, "scrub: device %s is not writable",
name->str);
rcu_read_unlock();
return -EROFS;
}
Btrfs: avoid unnecessary scrub workers allocation We only allocate scrub workers if we pass all the necessary checks, for example, there are no operation in progress. Besides, move mutex lock protection outside of scrub_workers_get() /scrub_workers_put(). Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-10-12 02:11:12 +08:00
mutex_lock(&fs_info->scrub_lock);
Btrfs: disallow some operations on the device replace target device This patch adds some code to disallow operations on the device that is used as the target for the device replace operation. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 01:29:28 +08:00
if (!dev->in_fs_metadata || dev->is_tgtdev_for_dev_replace) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_unlock(&fs_info->scrub_lock);
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
mutex_unlock(&fs_info->fs_devices->device_list_mutex);
return -EIO;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: fix lockdep deadlock warning due to dev_replace Xfstests btrfs/011 complains about a deadlock warning, [ 1226.649039] ========================================================= [ 1226.649039] [ INFO: possible irq lock inversion dependency detected ] [ 1226.649039] 4.1.0+ #270 Not tainted [ 1226.649039] --------------------------------------------------------- [ 1226.652955] kswapd0/46 just changed the state of lock: [ 1226.652955] (&delayed_node->mutex){+.+.-.}, at: [<ffffffff81458735>] __btrfs_release_delayed_node+0x45/0x1d0 [ 1226.652955] but this lock took another, RECLAIM_FS-unsafe lock in the past: [ 1226.652955] (&fs_info->dev_replace.lock){+.+.+.} and interrupts could create inverse lock ordering between them. [ 1226.652955] other info that might help us debug this: [ 1226.652955] Chain exists of: &delayed_node->mutex --> &found->groups_sem --> &fs_info->dev_replace.lock [ 1226.652955] Possible interrupt unsafe locking scenario: [ 1226.652955] CPU0 CPU1 [ 1226.652955] ---- ---- [ 1226.652955] lock(&fs_info->dev_replace.lock); [ 1226.652955] local_irq_disable(); [ 1226.652955] lock(&delayed_node->mutex); [ 1226.652955] lock(&found->groups_sem); [ 1226.652955] <Interrupt> [ 1226.652955] lock(&delayed_node->mutex); [ 1226.652955] *** DEADLOCK *** Commit 084b6e7c7607 ("btrfs: Fix a lockdep warning when running xfstest.") tried to fix a similar one that has the exactly same warning, but with that, we still run to this. The above lock chain comes from btrfs_commit_transaction ->btrfs_run_delayed_items ... ->__btrfs_update_delayed_inode ... ->__btrfs_cow_block ... ->find_free_extent ->cache_block_group ->load_free_space_cache ->btrfs_readpages ->submit_one_bio ... ->__btrfs_map_block ->btrfs_dev_replace_lock However, with high memory pressure, tasks which hold dev_replace.lock can be interrupted by kswapd and then kswapd is intended to release memory occupied by superblock, inodes and dentries, where we may call evict_inode, and it comes to [ 1226.652955] [<ffffffff81458735>] __btrfs_release_delayed_node+0x45/0x1d0 [ 1226.652955] [<ffffffff81459e74>] btrfs_remove_delayed_node+0x24/0x30 [ 1226.652955] [<ffffffff8140c5fe>] btrfs_evict_inode+0x34e/0x700 delayed_node->mutex may be acquired in __btrfs_release_delayed_node(), and it leads to a ABBA deadlock. To fix this, we can use "blocking rwlock" used in the case of extent_buffer, but things are simpler here since we only needs read's spinlock to blocking lock. With this, btrfs/011 no more produces warnings in dmesg. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: David Sterba <dsterba@suse.com>
2015-07-17 16:49:19 +08:00
btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
Btrfs: change core code of btrfs to support the device replace operations This commit contains all the essential changes to the core code of Btrfs for support of the device replace procedure. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 20:15:27 +08:00
if (dev->scrub_device ||
(!is_dev_replace &&
btrfs_dev_replace_is_ongoing(&fs_info->dev_replace))) {
Btrfs: fix lockdep deadlock warning due to dev_replace Xfstests btrfs/011 complains about a deadlock warning, [ 1226.649039] ========================================================= [ 1226.649039] [ INFO: possible irq lock inversion dependency detected ] [ 1226.649039] 4.1.0+ #270 Not tainted [ 1226.649039] --------------------------------------------------------- [ 1226.652955] kswapd0/46 just changed the state of lock: [ 1226.652955] (&delayed_node->mutex){+.+.-.}, at: [<ffffffff81458735>] __btrfs_release_delayed_node+0x45/0x1d0 [ 1226.652955] but this lock took another, RECLAIM_FS-unsafe lock in the past: [ 1226.652955] (&fs_info->dev_replace.lock){+.+.+.} and interrupts could create inverse lock ordering between them. [ 1226.652955] other info that might help us debug this: [ 1226.652955] Chain exists of: &delayed_node->mutex --> &found->groups_sem --> &fs_info->dev_replace.lock [ 1226.652955] Possible interrupt unsafe locking scenario: [ 1226.652955] CPU0 CPU1 [ 1226.652955] ---- ---- [ 1226.652955] lock(&fs_info->dev_replace.lock); [ 1226.652955] local_irq_disable(); [ 1226.652955] lock(&delayed_node->mutex); [ 1226.652955] lock(&found->groups_sem); [ 1226.652955] <Interrupt> [ 1226.652955] lock(&delayed_node->mutex); [ 1226.652955] *** DEADLOCK *** Commit 084b6e7c7607 ("btrfs: Fix a lockdep warning when running xfstest.") tried to fix a similar one that has the exactly same warning, but with that, we still run to this. The above lock chain comes from btrfs_commit_transaction ->btrfs_run_delayed_items ... ->__btrfs_update_delayed_inode ... ->__btrfs_cow_block ... ->find_free_extent ->cache_block_group ->load_free_space_cache ->btrfs_readpages ->submit_one_bio ... ->__btrfs_map_block ->btrfs_dev_replace_lock However, with high memory pressure, tasks which hold dev_replace.lock can be interrupted by kswapd and then kswapd is intended to release memory occupied by superblock, inodes and dentries, where we may call evict_inode, and it comes to [ 1226.652955] [<ffffffff81458735>] __btrfs_release_delayed_node+0x45/0x1d0 [ 1226.652955] [<ffffffff81459e74>] btrfs_remove_delayed_node+0x24/0x30 [ 1226.652955] [<ffffffff8140c5fe>] btrfs_evict_inode+0x34e/0x700 delayed_node->mutex may be acquired in __btrfs_release_delayed_node(), and it leads to a ABBA deadlock. To fix this, we can use "blocking rwlock" used in the case of extent_buffer, but things are simpler here since we only needs read's spinlock to blocking lock. With this, btrfs/011 no more produces warnings in dmesg. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: David Sterba <dsterba@suse.com>
2015-07-17 16:49:19 +08:00
btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_unlock(&fs_info->scrub_lock);
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
mutex_unlock(&fs_info->fs_devices->device_list_mutex);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return -EINPROGRESS;
}
Btrfs: fix lockdep deadlock warning due to dev_replace Xfstests btrfs/011 complains about a deadlock warning, [ 1226.649039] ========================================================= [ 1226.649039] [ INFO: possible irq lock inversion dependency detected ] [ 1226.649039] 4.1.0+ #270 Not tainted [ 1226.649039] --------------------------------------------------------- [ 1226.652955] kswapd0/46 just changed the state of lock: [ 1226.652955] (&delayed_node->mutex){+.+.-.}, at: [<ffffffff81458735>] __btrfs_release_delayed_node+0x45/0x1d0 [ 1226.652955] but this lock took another, RECLAIM_FS-unsafe lock in the past: [ 1226.652955] (&fs_info->dev_replace.lock){+.+.+.} and interrupts could create inverse lock ordering between them. [ 1226.652955] other info that might help us debug this: [ 1226.652955] Chain exists of: &delayed_node->mutex --> &found->groups_sem --> &fs_info->dev_replace.lock [ 1226.652955] Possible interrupt unsafe locking scenario: [ 1226.652955] CPU0 CPU1 [ 1226.652955] ---- ---- [ 1226.652955] lock(&fs_info->dev_replace.lock); [ 1226.652955] local_irq_disable(); [ 1226.652955] lock(&delayed_node->mutex); [ 1226.652955] lock(&found->groups_sem); [ 1226.652955] <Interrupt> [ 1226.652955] lock(&delayed_node->mutex); [ 1226.652955] *** DEADLOCK *** Commit 084b6e7c7607 ("btrfs: Fix a lockdep warning when running xfstest.") tried to fix a similar one that has the exactly same warning, but with that, we still run to this. The above lock chain comes from btrfs_commit_transaction ->btrfs_run_delayed_items ... ->__btrfs_update_delayed_inode ... ->__btrfs_cow_block ... ->find_free_extent ->cache_block_group ->load_free_space_cache ->btrfs_readpages ->submit_one_bio ... ->__btrfs_map_block ->btrfs_dev_replace_lock However, with high memory pressure, tasks which hold dev_replace.lock can be interrupted by kswapd and then kswapd is intended to release memory occupied by superblock, inodes and dentries, where we may call evict_inode, and it comes to [ 1226.652955] [<ffffffff81458735>] __btrfs_release_delayed_node+0x45/0x1d0 [ 1226.652955] [<ffffffff81459e74>] btrfs_remove_delayed_node+0x24/0x30 [ 1226.652955] [<ffffffff8140c5fe>] btrfs_evict_inode+0x34e/0x700 delayed_node->mutex may be acquired in __btrfs_release_delayed_node(), and it leads to a ABBA deadlock. To fix this, we can use "blocking rwlock" used in the case of extent_buffer, but things are simpler here since we only needs read's spinlock to blocking lock. With this, btrfs/011 no more produces warnings in dmesg. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: David Sterba <dsterba@suse.com>
2015-07-17 16:49:19 +08:00
btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
Btrfs: avoid unnecessary scrub workers allocation We only allocate scrub workers if we pass all the necessary checks, for example, there are no operation in progress. Besides, move mutex lock protection outside of scrub_workers_get() /scrub_workers_put(). Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-10-12 02:11:12 +08:00
ret = scrub_workers_get(fs_info, is_dev_replace);
if (ret) {
mutex_unlock(&fs_info->scrub_lock);
mutex_unlock(&fs_info->fs_devices->device_list_mutex);
return ret;
}
Btrfs: disallow some operations on the device replace target device This patch adds some code to disallow operations on the device that is used as the target for the device replace operation. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 01:29:28 +08:00
sctx = scrub_setup_ctx(dev, is_dev_replace);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
if (IS_ERR(sctx)) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_unlock(&fs_info->scrub_lock);
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
mutex_unlock(&fs_info->fs_devices->device_list_mutex);
scrub_workers_put(fs_info);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
return PTR_ERR(sctx);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sctx->readonly = readonly;
dev->scrub_device = sctx;
Btrfs: fix wrong super generation mismatch when scrubbing supers We came a race condition when scrubbing superblocks, the story is: In commiting transaction, we will update @last_trans_commited after writting superblocks, if scrubber start after writting superblocks and before updating @last_trans_commited, generation mismatch happens! We fix this by checking @scrub_pause_req, and we won't start a srubber until commiting transaction is finished.(after btrfs_scrub_continue() finished.) Reported-by: Sebastian Ochmann <ochmann@informatik.uni-bonn.de> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Reviewed-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:15:19 +08:00
mutex_unlock(&fs_info->fs_devices->device_list_mutex);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: fix wrong super generation mismatch when scrubbing supers We came a race condition when scrubbing superblocks, the story is: In commiting transaction, we will update @last_trans_commited after writting superblocks, if scrubber start after writting superblocks and before updating @last_trans_commited, generation mismatch happens! We fix this by checking @scrub_pause_req, and we won't start a srubber until commiting transaction is finished.(after btrfs_scrub_continue() finished.) Reported-by: Sebastian Ochmann <ochmann@informatik.uni-bonn.de> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Reviewed-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:15:19 +08:00
/*
* checking @scrub_pause_req here, we can avoid
* race between committing transaction and scrubbing.
*/
Btrfs: wrap repeated code into scrub_blocked_if_needed() Just wrap same code into one function scrub_blocked_if_needed(). This make a change that we will move waiting (@workers_pending = 0) before we can wake up commiting transaction(atomic_inc(@scrub_paused)), we must take carefully to not deadlock here. Thread 1 Thread 2 |->btrfs_commit_transaction() |->set trans type(COMMIT_DOING) |->btrfs_scrub_paused()(blocked) |->join_transaction(blocked) Move btrfs_scrub_paused() before setting trans type which means we can still join a transaction when commiting_transaction is blocked. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Suggested-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:16:53 +08:00
__scrub_blocked_if_needed(fs_info);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
atomic_inc(&fs_info->scrubs_running);
mutex_unlock(&fs_info->scrub_lock);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (!is_dev_replace) {
Btrfs: remove scrub_super_lock holding in btrfs_sync_log() Originally, we introduced scrub_super_lock to synchronize tree log code with scrubbing super. However we can replace scrub_super_lock with device_list_mutex, because writing super will hold this mutex, this will reduce an extra lock holding when writing supers in sync log code. Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-10-25 19:12:02 +08:00
/*
* by holding device list mutex, we can
* kick off writing super in log tree sync.
*/
Btrfs: fix wrong super generation mismatch when scrubbing supers We came a race condition when scrubbing superblocks, the story is: In commiting transaction, we will update @last_trans_commited after writting superblocks, if scrubber start after writting superblocks and before updating @last_trans_commited, generation mismatch happens! We fix this by checking @scrub_pause_req, and we won't start a srubber until commiting transaction is finished.(after btrfs_scrub_continue() finished.) Reported-by: Sebastian Ochmann <ochmann@informatik.uni-bonn.de> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Reviewed-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:15:19 +08:00
mutex_lock(&fs_info->fs_devices->device_list_mutex);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
ret = scrub_supers(sctx, dev);
Btrfs: fix wrong super generation mismatch when scrubbing supers We came a race condition when scrubbing superblocks, the story is: In commiting transaction, we will update @last_trans_commited after writting superblocks, if scrubber start after writting superblocks and before updating @last_trans_commited, generation mismatch happens! We fix this by checking @scrub_pause_req, and we won't start a srubber until commiting transaction is finished.(after btrfs_scrub_continue() finished.) Reported-by: Sebastian Ochmann <ochmann@informatik.uni-bonn.de> Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Reviewed-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-04 21:15:19 +08:00
mutex_unlock(&fs_info->fs_devices->device_list_mutex);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
}
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (!ret)
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
ret = scrub_enumerate_chunks(sctx, dev, start, end,
is_dev_replace);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
wait_event(sctx->list_wait, atomic_read(&sctx->bios_in_flight) == 0);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
atomic_dec(&fs_info->scrubs_running);
wake_up(&fs_info->scrub_pause_wait);
Btrfs: cleanup scrub bio and worker wait code Just move some code into functions to make everything more readable. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 23:44:58 +08:00
wait_event(sctx->list_wait, atomic_read(&sctx->workers_pending) == 0);
btrfs scrub: add fixup code for errors on nodatasum files This removes a FIXME comment and introduces the first part of nodatasum fixup: It gets the corresponding inode for a logical address and triggers a regular readpage for the corrupted sector. Once we have on-the-fly error correction our error will be automatically corrected. The correction code is expected to clear the newly introduced EXTENT_DAMAGED flag, making scrub report that error as "corrected" instead of "uncorrectable" eventually. Signed-off-by: Jan Schmidt <list.btrfs@jan-o-sch.net>
2011-06-14 02:04:15 +08:00
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (progress)
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
memcpy(progress, &sctx->stat, sizeof(*progress));
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_lock(&fs_info->scrub_lock);
dev->scrub_device = NULL;
Btrfs: avoid unnecessary scrub workers allocation We only allocate scrub workers if we pass all the necessary checks, for example, there are no operation in progress. Besides, move mutex lock protection outside of scrub_workers_get() /scrub_workers_put(). Signed-off-by: Wang Shilong <wangsl.fnst@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-10-12 02:11:12 +08:00
scrub_workers_put(fs_info);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_unlock(&fs_info->scrub_lock);
Btrfs: scrub, fix sleep in atomic context My previous patch "Btrfs: fix scrub race leading to use-after-free" introduced the possibility to sleep in an atomic context, which happens when the scrub_lock mutex is held at the time scrub_pending_bio_dec() is called - this function can be called under an atomic context. Chris ran into this in a debug kernel which gave the following trace: [ 1928.950319] BUG: sleeping function called from invalid context at kernel/locking/mutex.c:621 [ 1928.967334] in_atomic(): 1, irqs_disabled(): 0, pid: 149670, name: fsstress [ 1928.981324] INFO: lockdep is turned off. [ 1928.989244] CPU: 24 PID: 149670 Comm: fsstress Tainted: G W 3.19.0-rc7-mason+ #41 [ 1929.006418] Hardware name: ZTSYSTEMS Echo Ridge T4 /A9DRPF-10D, BIOS 1.07 05/10/2012 [ 1929.022207] ffffffff81a22cf8 ffff881076e03b78 ffffffff816b8dd9 ffff881076e03b78 [ 1929.037267] ffff880d8e828710 ffff881076e03ba8 ffffffff810856c4 ffff881076e03bc8 [ 1929.052315] 0000000000000000 000000000000026d ffffffff81a22cf8 ffff881076e03bd8 [ 1929.067381] Call Trace: [ 1929.072344] <IRQ> [<ffffffff816b8dd9>] dump_stack+0x4f/0x6e [ 1929.083968] [<ffffffff810856c4>] ___might_sleep+0x174/0x230 [ 1929.095352] [<ffffffff810857d2>] __might_sleep+0x52/0x90 [ 1929.106223] [<ffffffff816bb68f>] mutex_lock_nested+0x2f/0x3b0 [ 1929.117951] [<ffffffff810ab37d>] ? trace_hardirqs_on+0xd/0x10 [ 1929.129708] [<ffffffffa05dc838>] scrub_pending_bio_dec+0x38/0x70 [btrfs] [ 1929.143370] [<ffffffffa05dd0e0>] scrub_parity_bio_endio+0x50/0x70 [btrfs] [ 1929.157191] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.167382] [<ffffffffa05f96bc>] rbio_orig_end_io+0x7c/0xa0 [btrfs] [ 1929.180161] [<ffffffffa05f97ba>] raid_write_parity_end_io+0x5a/0x80 [btrfs] [ 1929.194318] [<ffffffff812fa603>] bio_endio+0x53/0xa0 [ 1929.204496] [<ffffffff8130401b>] blk_update_request+0x1eb/0x450 [ 1929.216569] [<ffffffff81096e58>] ? trigger_load_balance+0x78/0x500 [ 1929.229176] [<ffffffff8144c74d>] scsi_end_request+0x3d/0x1f0 [ 1929.240740] [<ffffffff8144ccac>] scsi_io_completion+0xac/0x5b0 [ 1929.252654] [<ffffffff81441c50>] scsi_finish_command+0xf0/0x150 [ 1929.264725] [<ffffffff8144d317>] scsi_softirq_done+0x147/0x170 [ 1929.276635] [<ffffffff8130ace6>] blk_done_softirq+0x86/0xa0 [ 1929.288014] [<ffffffff8105d92e>] __do_softirq+0xde/0x600 [ 1929.298885] [<ffffffff8105df6d>] irq_exit+0xbd/0xd0 (...) Fix this by using a reference count on the scrub context structure instead of locking the scrub_lock mutex. Signed-off-by: Filipe Manana <fdmanana@suse.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-02-10 05:14:24 +08:00
scrub_put_ctx(sctx);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
return ret;
}
btrfs: return void in functions without error conditions Signed-off-by: Jeff Mahoney <jeffm@suse.com>
2012-03-01 21:56:26 +08:00
void btrfs_scrub_pause(struct btrfs_root *root)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
struct btrfs_fs_info *fs_info = root->fs_info;
mutex_lock(&fs_info->scrub_lock);
atomic_inc(&fs_info->scrub_pause_req);
while (atomic_read(&fs_info->scrubs_paused) !=
atomic_read(&fs_info->scrubs_running)) {
mutex_unlock(&fs_info->scrub_lock);
wait_event(fs_info->scrub_pause_wait,
atomic_read(&fs_info->scrubs_paused) ==
atomic_read(&fs_info->scrubs_running));
mutex_lock(&fs_info->scrub_lock);
}
mutex_unlock(&fs_info->scrub_lock);
}
btrfs: return void in functions without error conditions Signed-off-by: Jeff Mahoney <jeffm@suse.com>
2012-03-01 21:56:26 +08:00
void btrfs_scrub_continue(struct btrfs_root *root)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
struct btrfs_fs_info *fs_info = root->fs_info;
atomic_dec(&fs_info->scrub_pause_req);
wake_up(&fs_info->scrub_pause_wait);
}
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
int btrfs_scrub_cancel(struct btrfs_fs_info *fs_info)
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
{
mutex_lock(&fs_info->scrub_lock);
if (!atomic_read(&fs_info->scrubs_running)) {
mutex_unlock(&fs_info->scrub_lock);
return -ENOTCONN;
}
atomic_inc(&fs_info->scrub_cancel_req);
while (atomic_read(&fs_info->scrubs_running)) {
mutex_unlock(&fs_info->scrub_lock);
wait_event(fs_info->scrub_pause_wait,
atomic_read(&fs_info->scrubs_running) == 0);
mutex_lock(&fs_info->scrub_lock);
}
atomic_dec(&fs_info->scrub_cancel_req);
mutex_unlock(&fs_info->scrub_lock);
return 0;
}
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
int btrfs_scrub_cancel_dev(struct btrfs_fs_info *fs_info,
struct btrfs_device *dev)
btrfs: enhance transaction abort infrastructure Signed-off-by: Jeff Mahoney <jeffm@suse.com>
2012-03-02 00:24:58 +08:00
{
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_lock(&fs_info->scrub_lock);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sctx = dev->scrub_device;
if (!sctx) {
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_unlock(&fs_info->scrub_lock);
return -ENOTCONN;
}
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
atomic_inc(&sctx->cancel_req);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
while (dev->scrub_device) {
mutex_unlock(&fs_info->scrub_lock);
wait_event(fs_info->scrub_pause_wait,
dev->scrub_device == NULL);
mutex_lock(&fs_info->scrub_lock);
}
mutex_unlock(&fs_info->scrub_lock);
return 0;
}
Btrfs: minor cleanup in scrub Just a minor cleanup commit in preparation for the big block changes. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@oracle.com>
2012-03-28 02:21:26 +08:00
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
int btrfs_scrub_progress(struct btrfs_root *root, u64 devid,
struct btrfs_scrub_progress *progress)
{
struct btrfs_device *dev;
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
struct scrub_ctx *sctx = NULL;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_lock(&root->fs_info->fs_devices->device_list_mutex);
Btrfs: pass fs_info instead of root A small number of functions that are used in a device replace procedure when the operation is resumed at mount time are unable to pass the same root pointer that would be used in the regular (ioctl) context. And since the root pointer is not required, only the fs_info is, the root pointer argument is replaced with the fs_info pointer argument. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 00:03:39 +08:00
dev = btrfs_find_device(root->fs_info, devid, NULL, NULL);
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
if (dev)
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
sctx = dev->scrub_device;
if (sctx)
memcpy(progress, &sctx->stat, sizeof(*progress));
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
mutex_unlock(&root->fs_info->fs_devices->device_list_mutex);
Btrfs: rename the scrub context structure The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit is a first preparation step to adapt the scrub code to be shareable for the device replace procedure. The block device will be removed from the scrub context state structure in a later step. It used to be the source block device. The scrub code as it is used for the device replace procedure reads the source data from whereever it is optimal. The source device might even be gone (disconnected, for instance due to a hardware failure). Or the drive can be so faulty so that the device replace procedure tries to avoid access to the faulty source drive as much as possible, and only if all other mirrors are damaged, as a last resort, the source disk is accessed. The modified scrub code operates as if it would handle the source drive and thereby generates an exact copy of the source disk on the target disk, even if the source disk is not present at all. Therefore the block device pointer to the source disk is removed in a later patch, and therefore the context structure is renamed (this is the goal of the current patch) to reflect that no source block device scope is there anymore. Summary: This first preparation step consists of a textual substitution of the term "dev" to the term "ctx" whereever the scrub context is used. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-02 16:58:09 +08:00
return dev ? (sctx ? 0 : -ENOTCONN) : -ENODEV;
btrfs: scrub This adds an initial implementation for scrub. It works quite straightforward. The usermode issues an ioctl for each device in the fs. For each device, it enumerates the allocated device chunks. For each chunk, the contained extents are enumerated and the data checksums fetched. The extents are read sequentially and the checksums verified. If an error occurs (checksum or EIO), a good copy is searched for. If one is found, the bad copy will be rewritten. All enumerations happen from the commit roots. During a transaction commit, the scrubs get paused and afterwards continue from the new roots. This commit is based on the series originally posted to linux-btrfs with some improvements that resulted from comments from David Sterba, Ilya Dryomov and Jan Schmidt. Signed-off-by: Arne Jansen <sensille@gmx.net>
2011-03-08 21:14:00 +08:00
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static void scrub_remap_extent(struct btrfs_fs_info *fs_info,
u64 extent_logical, u64 extent_len,
u64 *extent_physical,
struct btrfs_device **extent_dev,
int *extent_mirror_num)
{
u64 mapped_length;
struct btrfs_bio *bbio = NULL;
int ret;
mapped_length = extent_len;
ret = btrfs_map_block(fs_info, READ, extent_logical,
&mapped_length, &bbio, 0);
if (ret || !bbio || mapped_length < extent_len ||
!bbio->stripes[0].dev->bdev) {
Btrfs: add ref_count and free function for btrfs_bio 1: ref_count is simple than current RBIO_HOLD_BBIO_MAP_BIT flag to keep btrfs_bio's memory in raid56 recovery implement. 2: free function for bbio will make code clean and flexible, plus forced data type checking in compile. Changelog v1->v2: Rename following by David Sterba's suggestion: put_btrfs_bio() -> btrfs_put_bio() get_btrfs_bio() -> btrfs_get_bio() bbio->ref_count -> bbio->refs Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:34 +08:00
btrfs_put_bbio(bbio);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
return;
}
*extent_physical = bbio->stripes[0].physical;
*extent_mirror_num = bbio->mirror_num;
*extent_dev = bbio->stripes[0].dev;
Btrfs: add ref_count and free function for btrfs_bio 1: ref_count is simple than current RBIO_HOLD_BBIO_MAP_BIT flag to keep btrfs_bio's memory in raid56 recovery implement. 2: free function for bbio will make code clean and flexible, plus forced data type checking in compile. Changelog v1->v2: Rename following by David Sterba's suggestion: put_btrfs_bio() -> btrfs_put_bio() get_btrfs_bio() -> btrfs_get_bio() bbio->ref_count -> bbio->refs Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Chris Mason <clm@fb.com>
2015-01-20 15:11:34 +08:00
btrfs_put_bbio(bbio);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
}
static int scrub_setup_wr_ctx(struct scrub_ctx *sctx,
struct scrub_wr_ctx *wr_ctx,
struct btrfs_fs_info *fs_info,
struct btrfs_device *dev,
int is_dev_replace)
{
WARN_ON(wr_ctx->wr_curr_bio != NULL);
mutex_init(&wr_ctx->wr_lock);
wr_ctx->wr_curr_bio = NULL;
if (!is_dev_replace)
return 0;
WARN_ON(!dev->bdev);
block: remove bio_get_nr_vecs() We can always fill up the bio now, no need to estimate the possible size based on queue parameters. Acked-by: Steven Whitehouse <swhiteho@redhat.com> Signed-off-by: Kent Overstreet <kent.overstreet@gmail.com> [hch: rebased and wrote a changelog] Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Ming Lin <ming.l@ssi.samsung.com> Signed-off-by: Jens Axboe <axboe@fb.com>
2015-05-19 20:31:01 +08:00
wr_ctx->pages_per_wr_bio = SCRUB_PAGES_PER_WR_BIO;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
wr_ctx->tgtdev = dev;
atomic_set(&wr_ctx->flush_all_writes, 0);
return 0;
}
static void scrub_free_wr_ctx(struct scrub_wr_ctx *wr_ctx)
{
mutex_lock(&wr_ctx->wr_lock);
kfree(wr_ctx->wr_curr_bio);
wr_ctx->wr_curr_bio = NULL;
mutex_unlock(&wr_ctx->wr_lock);
}
static int copy_nocow_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
int mirror_num, u64 physical_for_dev_replace)
{
struct scrub_copy_nocow_ctx *nocow_ctx;
struct btrfs_fs_info *fs_info = sctx->dev_root->fs_info;
nocow_ctx = kzalloc(sizeof(*nocow_ctx), GFP_NOFS);
if (!nocow_ctx) {
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
return -ENOMEM;
}
scrub_pending_trans_workers_inc(sctx);
nocow_ctx->sctx = sctx;
nocow_ctx->logical = logical;
nocow_ctx->len = len;
nocow_ctx->mirror_num = mirror_num;
nocow_ctx->physical_for_dev_replace = physical_for_dev_replace;
Btrfs: fix task hang under heavy compressed write This has been reported and discussed for a long time, and this hang occurs in both 3.15 and 3.16. Btrfs now migrates to use kernel workqueue, but it introduces this hang problem. Btrfs has a kind of work queued as an ordered way, which means that its ordered_func() must be processed in the way of FIFO, so it usually looks like -- normal_work_helper(arg) work = container_of(arg, struct btrfs_work, normal_work); work->func() <---- (we name it work X) for ordered_work in wq->ordered_list ordered_work->ordered_func() ordered_work->ordered_free() The hang is a rare case, first when we find free space, we get an uncached block group, then we go to read its free space cache inode for free space information, so it will file a readahead request btrfs_readpages() for page that is not in page cache __do_readpage() submit_extent_page() btrfs_submit_bio_hook() btrfs_bio_wq_end_io() submit_bio() end_workqueue_bio() <--(ret by the 1st endio) queue a work(named work Y) for the 2nd also the real endio() So the hang occurs when work Y's work_struct and work X's work_struct happens to share the same address. A bit more explanation, A,B,C -- struct btrfs_work arg -- struct work_struct kthread: worker_thread() pick up a work_struct from @worklist process_one_work(arg) worker->current_work = arg; <-- arg is A->normal_work worker->current_func(arg) normal_work_helper(arg) A = container_of(arg, struct btrfs_work, normal_work); A->func() A->ordered_func() A->ordered_free() <-- A gets freed B->ordered_func() submit_compressed_extents() find_free_extent() load_free_space_inode() ... <-- (the above readhead stack) end_workqueue_bio() btrfs_queue_work(work C) B->ordered_free() As if work A has a high priority in wq->ordered_list and there are more ordered works queued after it, such as B->ordered_func(), its memory could have been freed before normal_work_helper() returns, which means that kernel workqueue code worker_thread() still has worker->current_work pointer to be work A->normal_work's, ie. arg's address. Meanwhile, work C is allocated after work A is freed, work C->normal_work and work A->normal_work are likely to share the same address(I confirmed this with ftrace output, so I'm not just guessing, it's rare though). When another kthread picks up work C->normal_work to process, and finds our kthread is processing it(see find_worker_executing_work()), it'll think work C as a collision and skip then, which ends up nobody processing work C. So the situation is that our kthread is waiting forever on work C. Besides, there're other cases that can lead to deadlock, but the real problem is that all btrfs workqueue shares one work->func, -- normal_work_helper, so this makes each workqueue to have its own helper function, but only a wraper pf normal_work_helper. With this patch, I no long hit the above hang. Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Chris Mason <clm@fb.com>
2014-08-15 23:36:53 +08:00
btrfs_init_work(&nocow_ctx->work, btrfs_scrubnc_helper,
copy_nocow_pages_worker, NULL, NULL);
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
INIT_LIST_HEAD(&nocow_ctx->inodes);
btrfs: Replace fs_info->scrub_* workqueue with btrfs_workqueue. Replace the fs_info->scrub_* with the newly created btrfs_workqueue. Signed-off-by: Qu Wenruo <quwenruo@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Josef Bacik <jbacik@fb.com>
2014-02-28 10:46:17 +08:00
btrfs_queue_work(fs_info->scrub_nocow_workers,
&nocow_ctx->work);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
return 0;
}
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
static int record_inode_for_nocow(u64 inum, u64 offset, u64 root, void *ctx)
{
struct scrub_copy_nocow_ctx *nocow_ctx = ctx;
struct scrub_nocow_inode *nocow_inode;
nocow_inode = kzalloc(sizeof(*nocow_inode), GFP_NOFS);
if (!nocow_inode)
return -ENOMEM;
nocow_inode->inum = inum;
nocow_inode->offset = offset;
nocow_inode->root = root;
list_add_tail(&nocow_inode->list, &nocow_ctx->inodes);
return 0;
}
#define COPY_COMPLETE 1
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
static void copy_nocow_pages_worker(struct btrfs_work *work)
{
struct scrub_copy_nocow_ctx *nocow_ctx =
container_of(work, struct scrub_copy_nocow_ctx, work);
struct scrub_ctx *sctx = nocow_ctx->sctx;
u64 logical = nocow_ctx->logical;
u64 len = nocow_ctx->len;
int mirror_num = nocow_ctx->mirror_num;
u64 physical_for_dev_replace = nocow_ctx->physical_for_dev_replace;
int ret;
struct btrfs_trans_handle *trans = NULL;
struct btrfs_fs_info *fs_info;
struct btrfs_path *path;
struct btrfs_root *root;
int not_written = 0;
fs_info = sctx->dev_root->fs_info;
root = fs_info->extent_root;
path = btrfs_alloc_path();
if (!path) {
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
not_written = 1;
goto out;
}
trans = btrfs_join_transaction(root);
if (IS_ERR(trans)) {
not_written = 1;
goto out;
}
ret = iterate_inodes_from_logical(logical, fs_info, path,
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
record_inode_for_nocow, nocow_ctx);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (ret != 0 && ret != -ENOENT) {
Btrfs: convert printk to btrfs_ and fix BTRFS prefix Convert all applicable cases of printk and pr_* to the btrfs_* macros. Fix all uses of the BTRFS prefix. Signed-off-by: Frank Holton <fholton@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-21 00:37:06 +08:00
btrfs_warn(fs_info, "iterate_inodes_from_logical() failed: log %llu, "
"phys %llu, len %llu, mir %u, ret %d",
Btrfs: Format mirror_num as int mirror_num is always "int", hence don't cast it to "unsigned long long" and format it as a 64-bit number. Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-08-20 19:20:10 +08:00
logical, physical_for_dev_replace, len, mirror_num,
ret);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
not_written = 1;
goto out;
}
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
btrfs_end_transaction(trans, root);
trans = NULL;
while (!list_empty(&nocow_ctx->inodes)) {
struct scrub_nocow_inode *entry;
entry = list_first_entry(&nocow_ctx->inodes,
struct scrub_nocow_inode,
list);
list_del_init(&entry->list);
ret = copy_nocow_pages_for_inode(entry->inum, entry->offset,
entry->root, nocow_ctx);
kfree(entry);
if (ret == COPY_COMPLETE) {
ret = 0;
break;
} else if (ret) {
break;
}
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
out:
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
while (!list_empty(&nocow_ctx->inodes)) {
struct scrub_nocow_inode *entry;
entry = list_first_entry(&nocow_ctx->inodes,
struct scrub_nocow_inode,
list);
list_del_init(&entry->list);
kfree(entry);
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (trans && !IS_ERR(trans))
btrfs_end_transaction(trans, root);
if (not_written)
btrfs_dev_replace_stats_inc(&fs_info->dev_replace.
num_uncorrectable_read_errors);
btrfs_free_path(path);
kfree(nocow_ctx);
scrub_pending_trans_workers_dec(sctx);
}
btrfs: fix dead lock while running replace and defrag concurrently This can be reproduced by fstests: btrfs/070 The scenario is like the following: replace worker thread defrag thread --------------------- ------------- copy_nocow_pages_worker btrfs_defrag_file copy_nocow_pages_for_inode ... btrfs_writepages |A| lock_extent_bits extent_write_cache_pages |B| lock_page __extent_writepage ... writepage_delalloc find_lock_delalloc_range |B| lock_extent_bits find_or_create_page pagecache_get_page |A| lock_page This leads to an ABBA pattern deadlock. To fix it, o we just change it to an AABB pattern which means to @unlock_extent_bits() before we @lock_page(), and in this way the @extent_read_full_page_nolock() is no longer in an locked context, so change it back to @extent_read_full_page() to regain protection. o Since we @unlock_extent_bits() earlier, then before @write_page_nocow(), the extent may not really point at the physical block we want, so we have to check it before write. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-11-10 15:36:08 +08:00
static int check_extent_to_block(struct inode *inode, u64 start, u64 len,
u64 logical)
{
struct extent_state *cached_state = NULL;
struct btrfs_ordered_extent *ordered;
struct extent_io_tree *io_tree;
struct extent_map *em;
u64 lockstart = start, lockend = start + len - 1;
int ret = 0;
io_tree = &BTRFS_I(inode)->io_tree;
btrfs: drop unused parameter from lock_extent_bits We've always passed 0. Stack usage will slightly decrease. Signed-off-by: David Sterba <dsterba@suse.com>
2015-12-03 21:30:40 +08:00
lock_extent_bits(io_tree, lockstart, lockend, &cached_state);
btrfs: fix dead lock while running replace and defrag concurrently This can be reproduced by fstests: btrfs/070 The scenario is like the following: replace worker thread defrag thread --------------------- ------------- copy_nocow_pages_worker btrfs_defrag_file copy_nocow_pages_for_inode ... btrfs_writepages |A| lock_extent_bits extent_write_cache_pages |B| lock_page __extent_writepage ... writepage_delalloc find_lock_delalloc_range |B| lock_extent_bits find_or_create_page pagecache_get_page |A| lock_page This leads to an ABBA pattern deadlock. To fix it, o we just change it to an AABB pattern which means to @unlock_extent_bits() before we @lock_page(), and in this way the @extent_read_full_page_nolock() is no longer in an locked context, so change it back to @extent_read_full_page() to regain protection. o Since we @unlock_extent_bits() earlier, then before @write_page_nocow(), the extent may not really point at the physical block we want, so we have to check it before write. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-11-10 15:36:08 +08:00
ordered = btrfs_lookup_ordered_range(inode, lockstart, len);
if (ordered) {
btrfs_put_ordered_extent(ordered);
ret = 1;
goto out_unlock;
}
em = btrfs_get_extent(inode, NULL, 0, start, len, 0);
if (IS_ERR(em)) {
ret = PTR_ERR(em);
goto out_unlock;
}
/*
* This extent does not actually cover the logical extent anymore,
* move on to the next inode.
*/
if (em->block_start > logical ||
em->block_start + em->block_len < logical + len) {
free_extent_map(em);
ret = 1;
goto out_unlock;
}
free_extent_map(em);
out_unlock:
unlock_extent_cached(io_tree, lockstart, lockend, &cached_state,
GFP_NOFS);
return ret;
}
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
static int copy_nocow_pages_for_inode(u64 inum, u64 offset, u64 root,
struct scrub_copy_nocow_ctx *nocow_ctx)
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
{
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
struct btrfs_fs_info *fs_info = nocow_ctx->sctx->dev_root->fs_info;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct btrfs_key key;
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
struct inode *inode;
struct page *page;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
struct btrfs_root *local_root;
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
struct extent_io_tree *io_tree;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
u64 physical_for_dev_replace;
btrfs: fix dead lock while running replace and defrag concurrently This can be reproduced by fstests: btrfs/070 The scenario is like the following: replace worker thread defrag thread --------------------- ------------- copy_nocow_pages_worker btrfs_defrag_file copy_nocow_pages_for_inode ... btrfs_writepages |A| lock_extent_bits extent_write_cache_pages |B| lock_page __extent_writepage ... writepage_delalloc find_lock_delalloc_range |B| lock_extent_bits find_or_create_page pagecache_get_page |A| lock_page This leads to an ABBA pattern deadlock. To fix it, o we just change it to an AABB pattern which means to @unlock_extent_bits() before we @lock_page(), and in this way the @extent_read_full_page_nolock() is no longer in an locked context, so change it back to @extent_read_full_page() to regain protection. o Since we @unlock_extent_bits() earlier, then before @write_page_nocow(), the extent may not really point at the physical block we want, so we have to check it before write. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-11-10 15:36:08 +08:00
u64 nocow_ctx_logical;
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
u64 len = nocow_ctx->len;
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
unsigned long index;
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
int srcu_index;
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
int ret = 0;
int err = 0;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
key.objectid = root;
key.type = BTRFS_ROOT_ITEM_KEY;
key.offset = (u64)-1;
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
srcu_index = srcu_read_lock(&fs_info->subvol_srcu);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
local_root = btrfs_read_fs_root_no_name(fs_info, &key);
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
if (IS_ERR(local_root)) {
srcu_read_unlock(&fs_info->subvol_srcu, srcu_index);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
return PTR_ERR(local_root);
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
key.type = BTRFS_INODE_ITEM_KEY;
key.objectid = inum;
key.offset = 0;
inode = btrfs_iget(fs_info->sb, &key, local_root, NULL);
Btrfs: fix race between snapshot deletion and getting inode While running snapshot testscript created by Mitch and David, the race between autodefrag and snapshot deletion can lead to corruption of dead_root list so that we can get crash on btrfs_clean_old_snapshots(). And besides autodefrag, scrub also does the same thing, ie. read root first and get inode. Here is the story(take autodefrag as an example): (1) when we delete a snapshot or subvolume, it will set its root's refs to zero and do a iput() on its own inode, and if this inode happens to be the only active in-meory one in root's inode rbtree, it will add itself to the global dead_roots list for later cleanup. (2) after (1), the autodefrag thread may read another inode for defrag and the inode is just in the deleted snapshot/subvolume, but all of these are without checking if the root is still valid(refs > 0). So the end up result is adding the deleted snapshot/subvolume's root to the global dead_roots list AGAIN. Fortunately, we already have a srcu lock to avoid the race, ie. subvol_srcu. So all we need to do is to take the lock to protect 'read root and get inode', since we synchronize to wait for the rcu grace period before adding something to the global dead_roots list. Reported-by: Mitch Harder <mitch.harder@sabayonlinux.org> Signed-off-by: Liu Bo <bo.li.liu@oracle.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-01-29 11:22:10 +08:00
srcu_read_unlock(&fs_info->subvol_srcu, srcu_index);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (IS_ERR(inode))
return PTR_ERR(inode);
Btrfs: fix several potential problems in copy_nocow_pages_for_inode - It makes no sense that we deal with a inode in the dead tree. - fix the race between dio and page copy by waiting the dio completion - avoid the page copy vs truncate/punch hole - check if the page is in the page cache or not Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:51:00 +08:00
/* Avoid truncate/dio/punch hole.. */
wrappers for ->i_mutex access parallel to mutex_{lock,unlock,trylock,is_locked,lock_nested}, inode_foo(inode) being mutex_foo(&inode->i_mutex). Please, use those for access to ->i_mutex; over the coming cycle ->i_mutex will become rwsem, with ->lookup() done with it held only shared. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2016-01-23 04:40:57 +08:00
inode_lock(inode);
Btrfs: fix several potential problems in copy_nocow_pages_for_inode - It makes no sense that we deal with a inode in the dead tree. - fix the race between dio and page copy by waiting the dio completion - avoid the page copy vs truncate/punch hole - check if the page is in the page cache or not Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:51:00 +08:00
inode_dio_wait(inode);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
physical_for_dev_replace = nocow_ctx->physical_for_dev_replace;
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
io_tree = &BTRFS_I(inode)->io_tree;
btrfs: fix dead lock while running replace and defrag concurrently This can be reproduced by fstests: btrfs/070 The scenario is like the following: replace worker thread defrag thread --------------------- ------------- copy_nocow_pages_worker btrfs_defrag_file copy_nocow_pages_for_inode ... btrfs_writepages |A| lock_extent_bits extent_write_cache_pages |B| lock_page __extent_writepage ... writepage_delalloc find_lock_delalloc_range |B| lock_extent_bits find_or_create_page pagecache_get_page |A| lock_page This leads to an ABBA pattern deadlock. To fix it, o we just change it to an AABB pattern which means to @unlock_extent_bits() before we @lock_page(), and in this way the @extent_read_full_page_nolock() is no longer in an locked context, so change it back to @extent_read_full_page() to regain protection. o Since we @unlock_extent_bits() earlier, then before @write_page_nocow(), the extent may not really point at the physical block we want, so we have to check it before write. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-11-10 15:36:08 +08:00
nocow_ctx_logical = nocow_ctx->logical;
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
btrfs: fix dead lock while running replace and defrag concurrently This can be reproduced by fstests: btrfs/070 The scenario is like the following: replace worker thread defrag thread --------------------- ------------- copy_nocow_pages_worker btrfs_defrag_file copy_nocow_pages_for_inode ... btrfs_writepages |A| lock_extent_bits extent_write_cache_pages |B| lock_page __extent_writepage ... writepage_delalloc find_lock_delalloc_range |B| lock_extent_bits find_or_create_page pagecache_get_page |A| lock_page This leads to an ABBA pattern deadlock. To fix it, o we just change it to an AABB pattern which means to @unlock_extent_bits() before we @lock_page(), and in this way the @extent_read_full_page_nolock() is no longer in an locked context, so change it back to @extent_read_full_page() to regain protection. o Since we @unlock_extent_bits() earlier, then before @write_page_nocow(), the extent may not really point at the physical block we want, so we have to check it before write. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-11-10 15:36:08 +08:00
ret = check_extent_to_block(inode, offset, len, nocow_ctx_logical);
if (ret) {
ret = ret > 0 ? 0 : ret;
goto out;
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
while (len >= PAGE_CACHE_SIZE) {
index = offset >> PAGE_CACHE_SHIFT;
Btrfs: fix several potential problems in copy_nocow_pages_for_inode - It makes no sense that we deal with a inode in the dead tree. - fix the race between dio and page copy by waiting the dio completion - avoid the page copy vs truncate/punch hole - check if the page is in the page cache or not Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:51:00 +08:00
again:
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
page = find_or_create_page(inode->i_mapping, index, GFP_NOFS);
if (!page) {
Btrfs: convert printk to btrfs_ and fix BTRFS prefix Convert all applicable cases of printk and pr_* to the btrfs_* macros. Fix all uses of the BTRFS prefix. Signed-off-by: Frank Holton <fholton@gmail.com> Signed-off-by: Josef Bacik <jbacik@fb.com> Signed-off-by: Chris Mason <clm@fb.com>
2013-12-21 00:37:06 +08:00
btrfs_err(fs_info, "find_or_create_page() failed");
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
ret = -ENOMEM;
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
goto out;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
}
if (PageUptodate(page)) {
if (PageDirty(page))
goto next_page;
} else {
ClearPageError(page);
btrfs: fix dead lock while running replace and defrag concurrently This can be reproduced by fstests: btrfs/070 The scenario is like the following: replace worker thread defrag thread --------------------- ------------- copy_nocow_pages_worker btrfs_defrag_file copy_nocow_pages_for_inode ... btrfs_writepages |A| lock_extent_bits extent_write_cache_pages |B| lock_page __extent_writepage ... writepage_delalloc find_lock_delalloc_range |B| lock_extent_bits find_or_create_page pagecache_get_page |A| lock_page This leads to an ABBA pattern deadlock. To fix it, o we just change it to an AABB pattern which means to @unlock_extent_bits() before we @lock_page(), and in this way the @extent_read_full_page_nolock() is no longer in an locked context, so change it back to @extent_read_full_page() to regain protection. o Since we @unlock_extent_bits() earlier, then before @write_page_nocow(), the extent may not really point at the physical block we want, so we have to check it before write. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-11-10 15:36:08 +08:00
err = extent_read_full_page(io_tree, page,
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
btrfs_get_extent,
nocow_ctx->mirror_num);
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
if (err) {
ret = err;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
goto next_page;
}
Btrfs: fix several potential problems in copy_nocow_pages_for_inode - It makes no sense that we deal with a inode in the dead tree. - fix the race between dio and page copy by waiting the dio completion - avoid the page copy vs truncate/punch hole - check if the page is in the page cache or not Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:51:00 +08:00
Btrfs: fix oops when recovering the file data by scrub function We get oops while running btrfs replace start test, ------------[ cut here ]------------ kernel BUG at mm/filemap.c:608! [SNIP] Call Trace: [<ffffffffa04b36c7>] copy_nocow_pages_for_inode+0x217/0x3f0 [btrfs] [<ffffffffa04b34b0>] ? scrub_print_warning_inode+0x230/0x230 [btrfs] [<ffffffffa04b34b0>] ? scrub_print_warning_inode+0x230/0x230 [btrfs] [<ffffffffa04bb8ce>] iterate_extent_inodes+0x1ae/0x300 [btrfs] [<ffffffffa04bbab2>] iterate_inodes_from_logical+0x92/0xb0 [btrfs] [<ffffffffa04b34b0>] ? scrub_print_warning_inode+0x230/0x230 [btrfs] [<ffffffffa04b3b07>] copy_nocow_pages_worker+0x97/0x150 [btrfs] [<ffffffffa048eed4>] worker_loop+0x134/0x540 [btrfs] [<ffffffff816274ea>] ? __schedule+0x3ca/0x7f0 [<ffffffffa048eda0>] ? btrfs_queue_worker+0x300/0x300 [btrfs] [<ffffffff8106f2f0>] kthread+0xc0/0xd0 [<ffffffff8106f230>] ? flush_kthread_worker+0x80/0x80 [<ffffffff8163181c>] ret_from_fork+0x7c/0xb0 [<ffffffff8106f230>] ? flush_kthread_worker+0x80/0x80 [SNIP] RIP [<ffffffff8111f4c5>] unlock_page+0x35/0x40 RSP <ffff88010316bb98> ---[ end trace 421e79ad0dd72c7d ]--- it is because we forgot to lock the page again after we read data to the page. Fix it. Signed-off-by: Lin Feng <linfeng@cn.fujitsu.com> Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:58 +08:00
lock_page(page);
Btrfs: fix several potential problems in copy_nocow_pages_for_inode - It makes no sense that we deal with a inode in the dead tree. - fix the race between dio and page copy by waiting the dio completion - avoid the page copy vs truncate/punch hole - check if the page is in the page cache or not Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:51:00 +08:00
/*
* If the page has been remove from the page cache,
* the data on it is meaningless, because it may be
* old one, the new data may be written into the new
* page in the page cache.
*/
if (page->mapping != inode->i_mapping) {
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
unlock_page(page);
Btrfs: fix several potential problems in copy_nocow_pages_for_inode - It makes no sense that we deal with a inode in the dead tree. - fix the race between dio and page copy by waiting the dio completion - avoid the page copy vs truncate/punch hole - check if the page is in the page cache or not Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:51:00 +08:00
page_cache_release(page);
goto again;
}
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (!PageUptodate(page)) {
ret = -EIO;
goto next_page;
}
}
btrfs: fix dead lock while running replace and defrag concurrently This can be reproduced by fstests: btrfs/070 The scenario is like the following: replace worker thread defrag thread --------------------- ------------- copy_nocow_pages_worker btrfs_defrag_file copy_nocow_pages_for_inode ... btrfs_writepages |A| lock_extent_bits extent_write_cache_pages |B| lock_page __extent_writepage ... writepage_delalloc find_lock_delalloc_range |B| lock_extent_bits find_or_create_page pagecache_get_page |A| lock_page This leads to an ABBA pattern deadlock. To fix it, o we just change it to an AABB pattern which means to @unlock_extent_bits() before we @lock_page(), and in this way the @extent_read_full_page_nolock() is no longer in an locked context, so change it back to @extent_read_full_page() to regain protection. o Since we @unlock_extent_bits() earlier, then before @write_page_nocow(), the extent may not really point at the physical block we want, so we have to check it before write. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-11-10 15:36:08 +08:00
ret = check_extent_to_block(inode, offset, len,
nocow_ctx_logical);
if (ret) {
ret = ret > 0 ? 0 : ret;
goto next_page;
}
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
err = write_page_nocow(nocow_ctx->sctx,
physical_for_dev_replace, page);
if (err)
ret = err;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
next_page:
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
unlock_page(page);
page_cache_release(page);
if (ret)
break;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
offset += PAGE_CACHE_SIZE;
physical_for_dev_replace += PAGE_CACHE_SIZE;
btrfs: fix dead lock while running replace and defrag concurrently This can be reproduced by fstests: btrfs/070 The scenario is like the following: replace worker thread defrag thread --------------------- ------------- copy_nocow_pages_worker btrfs_defrag_file copy_nocow_pages_for_inode ... btrfs_writepages |A| lock_extent_bits extent_write_cache_pages |B| lock_page __extent_writepage ... writepage_delalloc find_lock_delalloc_range |B| lock_extent_bits find_or_create_page pagecache_get_page |A| lock_page This leads to an ABBA pattern deadlock. To fix it, o we just change it to an AABB pattern which means to @unlock_extent_bits() before we @lock_page(), and in this way the @extent_read_full_page_nolock() is no longer in an locked context, so change it back to @extent_read_full_page() to regain protection. o Since we @unlock_extent_bits() earlier, then before @write_page_nocow(), the extent may not really point at the physical block we want, so we have to check it before write. Signed-off-by: Gui Hecheng <guihc.fnst@cn.fujitsu.com> Tested-by: David Sterba <dsterba@suse.cz> Signed-off-by: Chris Mason <clm@fb.com>
2014-11-10 15:36:08 +08:00
nocow_ctx_logical += PAGE_CACHE_SIZE;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
len -= PAGE_CACHE_SIZE;
}
Btrfs: improve replacing nocow extents Various people have hit a deadlock when running btrfs/011. This is because when replacing nocow extents we will take the i_mutex to make sure nobody messes with the file while we are replacing the extent. The problem is we are already holding a transaction open, which is a locking inversion, so instead we need to save these inodes we find and then process them outside of the transaction. Further we can't just lock the inode and assume we are good to go. We need to lock the extent range and then read back the extent cache for the inode to make sure the extent really still points at the physical block we want. If it doesn't we don't have to copy it. Thanks, Signed-off-by: Josef Bacik <jbacik@fusionio.com> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2013-09-13 04:58:28 +08:00
ret = COPY_COMPLETE;
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
out:
wrappers for ->i_mutex access parallel to mutex_{lock,unlock,trylock,is_locked,lock_nested}, inode_foo(inode) being mutex_foo(&inode->i_mutex). Please, use those for access to ->i_mutex; over the coming cycle ->i_mutex will become rwsem, with ->lookup() done with it held only shared. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
2016-01-23 04:40:57 +08:00
inode_unlock(inode);
Btrfs: cleanup the code of copy_nocow_pages_for_inode() - It make no sense that we continue to do something after the error happened, just go back with this patch. - remove some check of copy_nocow_pages_for_inode(), such as page check after write, inode check in the end of the function, because we are sure they exist. - remove the unnecessary goto in the return value check of the write Signed-off-by: Miao Xie <miaox@cn.fujitsu.com> Signed-off-by: Josef Bacik <jbacik@fusionio.com>
2013-06-27 18:50:59 +08:00
iput(inode);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
return ret;
}
static int write_page_nocow(struct scrub_ctx *sctx,
u64 physical_for_dev_replace, struct page *page)
{
struct bio *bio;
struct btrfs_device *dev;
int ret;
dev = sctx->wr_ctx.tgtdev;
if (!dev)
return -EIO;
if (!dev->bdev) {
btrfs: switch message printers to ratelimited variants Signed-off-by: David Sterba <dsterba@suse.com>
2015-10-08 17:01:36 +08:00
btrfs_warn_rl(dev->dev_root->fs_info,
"scrub write_page_nocow(bdev == NULL) is unexpected");
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
return -EIO;
}
Btrfs: use a btrfs bioset instead of abusing bio internals Btrfs has been pointer tagging bi_private and using bi_bdev to store the stripe index and mirror number of failed IOs. As bios bubble back up through the call chain, we use these to decide if and how to retry our IOs. They are also used to count IO failures on a per device basis. Recently a bio tracepoint was added lead to crashes because we were abusing bi_bdev. This commit adds a btrfs bioset, and creates explicit fields for the mirror number and stripe index. The plan is to extend this structure for all of the fields currently in struct btrfs_bio, which will mean one less kmalloc in our IO path. Signed-off-by: Chris Mason <chris.mason@fusionio.com> Reported-by: Tejun Heo <tj@kernel.org>
2013-05-18 06:30:14 +08:00
bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
if (!bio) {
spin_lock(&sctx->stat_lock);
sctx->stat.malloc_errors++;
spin_unlock(&sctx->stat_lock);
return -ENOMEM;
}
block: Abstract out bvec iterator Immutable biovecs are going to require an explicit iterator. To implement immutable bvecs, a later patch is going to add a bi_bvec_done member to this struct; for now, this patch effectively just renames things. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Cc: "Ed L. Cashin" <ecashin@coraid.com> Cc: Nick Piggin <npiggin@kernel.dk> Cc: Lars Ellenberg <drbd-dev@lists.linbit.com> Cc: Jiri Kosina <jkosina@suse.cz> Cc: Matthew Wilcox <willy@linux.intel.com> Cc: Geoff Levand <geoff@infradead.org> Cc: Yehuda Sadeh <yehuda@inktank.com> Cc: Sage Weil <sage@inktank.com> Cc: Alex Elder <elder@inktank.com> Cc: ceph-devel@vger.kernel.org Cc: Joshua Morris <josh.h.morris@us.ibm.com> Cc: Philip Kelleher <pjk1939@linux.vnet.ibm.com> Cc: Rusty Russell <rusty@rustcorp.com.au> Cc: "Michael S. Tsirkin" <mst@redhat.com> Cc: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com> Cc: Jeremy Fitzhardinge <jeremy@goop.org> Cc: Neil Brown <neilb@suse.de> Cc: Alasdair Kergon <agk@redhat.com> Cc: Mike Snitzer <snitzer@redhat.com> Cc: dm-devel@redhat.com Cc: Martin Schwidefsky <schwidefsky@de.ibm.com> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: linux390@de.ibm.com Cc: Boaz Harrosh <bharrosh@panasas.com> Cc: Benny Halevy <bhalevy@tonian.com> Cc: "James E.J. Bottomley" <JBottomley@parallels.com> Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Cc: "Nicholas A. Bellinger" <nab@linux-iscsi.org> Cc: Alexander Viro <viro@zeniv.linux.org.uk> Cc: Chris Mason <chris.mason@fusionio.com> Cc: "Theodore Ts'o" <tytso@mit.edu> Cc: Andreas Dilger <adilger.kernel@dilger.ca> Cc: Jaegeuk Kim <jaegeuk.kim@samsung.com> Cc: Steven Whitehouse <swhiteho@redhat.com> Cc: Dave Kleikamp <shaggy@kernel.org> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Trond Myklebust <Trond.Myklebust@netapp.com> Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp> Cc: Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Ben Myers <bpm@sgi.com> Cc: xfs@oss.sgi.com Cc: Steven Rostedt <rostedt@goodmis.org> Cc: Frederic Weisbecker <fweisbec@gmail.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Len Brown <len.brown@intel.com> Cc: Pavel Machek <pavel@ucw.cz> Cc: "Rafael J. Wysocki" <rjw@sisk.pl> Cc: Herton Ronaldo Krzesinski <herton.krzesinski@canonical.com> Cc: Ben Hutchings <ben@decadent.org.uk> Cc: Andrew Morton <akpm@linux-foundation.org> Cc: Guo Chao <yan@linux.vnet.ibm.com> Cc: Tejun Heo <tj@kernel.org> Cc: Asai Thambi S P <asamymuthupa@micron.com> Cc: Selvan Mani <smani@micron.com> Cc: Sam Bradshaw <sbradshaw@micron.com> Cc: Wei Yongjun <yongjun_wei@trendmicro.com.cn> Cc: "Roger Pau Monné" <roger.pau@citrix.com> Cc: Jan Beulich <jbeulich@suse.com> Cc: Stefano Stabellini <stefano.stabellini@eu.citrix.com> Cc: Ian Campbell <Ian.Campbell@citrix.com> Cc: Sebastian Ott <sebott@linux.vnet.ibm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Minchan Kim <minchan@kernel.org> Cc: Jiang Liu <jiang.liu@huawei.com> Cc: Nitin Gupta <ngupta@vflare.org> Cc: Jerome Marchand <jmarchand@redhat.com> Cc: Joe Perches <joe@perches.com> Cc: Peng Tao <tao.peng@emc.com> Cc: Andy Adamson <andros@netapp.com> Cc: fanchaoting <fanchaoting@cn.fujitsu.com> Cc: Jie Liu <jeff.liu@oracle.com> Cc: Sunil Mushran <sunil.mushran@gmail.com> Cc: "Martin K. Petersen" <martin.petersen@oracle.com> Cc: Namjae Jeon <namjae.jeon@samsung.com> Cc: Pankaj Kumar <pankaj.km@samsung.com> Cc: Dan Magenheimer <dan.magenheimer@oracle.com> Cc: Mel Gorman <mgorman@suse.de>6
2013-10-12 06:44:27 +08:00
bio->bi_iter.bi_size = 0;
bio->bi_iter.bi_sector = physical_for_dev_replace >> 9;
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
bio->bi_bdev = dev->bdev;
ret = bio_add_page(bio, page, PAGE_CACHE_SIZE, 0);
if (ret != PAGE_CACHE_SIZE) {
leave_with_eio:
bio_put(bio);
btrfs_dev_stat_inc_and_print(dev, BTRFS_DEV_STAT_WRITE_ERRS);
return -EIO;
}
block: submit_bio_wait() conversions It was being open coded in a few places. Signed-off-by: Kent Overstreet <kmo@daterainc.com> Cc: Jens Axboe <axboe@kernel.dk> Cc: Joern Engel <joern@logfs.org> Cc: Prasad Joshi <prasadjoshi.linux@gmail.com> Cc: Neil Brown <neilb@suse.de> Cc: Chris Mason <chris.mason@fusionio.com> Acked-by: NeilBrown <neilb@suse.de>
2013-11-24 14:33:32 +08:00
if (btrfsic_submit_bio_wait(WRITE_SYNC, bio))
Btrfs: add code to scrub to copy read data to another disk The device replace procedure makes use of the scrub code. The scrub code is the most efficient code to read the allocated data of a disk, i.e. it reads sequentially in order to avoid disk head movements, it skips unallocated blocks, it uses read ahead mechanisms, and it contains all the code to detect and repair defects. This commit adds code to scrub to allow the scrub code to copy read data to another disk. One goal is to be able to perform as fast as possible. Therefore the write requests are collected until huge bios are built, and the write process is decoupled from the read process with some kind of flow control, of course, in order to limit the allocated memory. The best performance on spinning disks could by reached when the head movements are avoided as much as possible. Therefore a single worker is used to interface the read process with the write process. The regular scrub operation works as fast as before, it is not negatively influenced and actually it is more or less unchanged. Signed-off-by: Stefan Behrens <sbehrens@giantdisaster.de> Signed-off-by: Chris Mason <chris.mason@fusionio.com>
2012-11-06 18:43:11 +08:00
goto leave_with_eio;
bio_put(bio);
return 0;
}
Reference in New Issue Copy Permalink