If the following process returns an error,
the quota inode, not the quota file, is damaged.
(fsck_chk_quota_node-->fsck_chk_node_blk-->sanity_check_nid)
The fsck does not have a process to rebuild the quota inode.
Because sanity_check_nid is not passed, fsck->nat_area_bitmap
can not be cleared, and then the NAT of quota will be nullify
during fix_nat_entries.
During the next fsck check, the quota inode check fails
because the address of the quota inode changes to 0.
In addition, in fsck_chk_quota_files-->f2fs_filesize_update,
data is written to address 0.
Therefore, when the quota inode is corrupted, we need to rebuild it.
Signed-off-by: Wang Xiaojun <wangxiaojun11@huawei.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
In previous, we did a full scan right after kernel version is changed
and it triggered many unnecessary full scan whenever a newly built
kernel is downloaded. Plus, we are lack of a periodic health check of
on-disk filesystem images. So, I added a monthly periodic check after
a new vesion is updated.
Signed-off-by: Daeho Jeong <daehojeong@google.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
In page_symlink, reserve_new_block applies address for data block,
not for inodes. Therefore, is_inode is set to 0.
Signed-off-by: Wang Xiaojun <wangxiaojun11@huawei.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
This fixes wrong file offset updates, since we should not do (*ofs)++; for
indirect node blocks.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
For a volume using a zoned block device without conventional zones
(e.g. a NVMe ZNS drive), a regular block device must be used to store
metadata so that in-place metadata writes can be executed. The zoned
block device cannot contain any metadata blocks requiring in-place
update (e.g. bitmap blocks). When formatting a volume that contains
host managed zoned block devices, make sure that the conventional device
used to store metadata is large enough to avoid write errors.
Signed-off-by: Aravind Ramesh <aravind.ramesh@wdc.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Since the IMMUTABLE flag for compression is now changed
to F2FS_COMPRESS_RELEASED, some 'IMMUTABLE' words should be changed too:
1. sload help page
2. inline comments
Signed-off-by: Robin Hsu <robinhsu@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
This patch adds a readonly feature on small partitions by eliminating
on-disk layout for data writes such as provisioning/reserved space and
SSA.
This requires f2fs updates to understand RO feature in superblock.
Tested 4 cases:
sload.f2fs -c -a lz0 -f kernel/fs $DEV
sload.f2fs -c -a lz4 -f kernel/fs $DEV
sload.f2fs -c -r -a lz4 -f kernel/fs $DEV
sload.f2fs -c -L 3 -r -a lz4 -f kernel/fs $DEV
after:
mkfs.f2fs -O ro,compression,extra_attr -f $DEV
fsck.f2fs $DEV
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Although many other filesystems provide a tool for changing volume
label, e.g. e2label for ext filesystem, but f2fs has no way to change
volume label except set it while formatting with mkfs.f2fs.
This introduces f2fslabel, simple tool for changing label of f2fs
volume.
Signed-off-by: Dongwoo Lee <dwoo08.lee@samsung.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Current implementation, If define WITH_KERNEL_VERSION
get_kernel_uname_version just return release info, not define
WITH_KERNEL_VERSION, return release and version info.
So remove "!" in get_kernel_uname_version, make logic more clear.
Signed-off-by: lijiazi <lijiazi@xiaomi.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Fix to free node_blk when nid is 0 from dump_node_blk().
Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
With invalid node info from fsck_chk_orphan_node(), orphan_blk
and new_blk are not freed. Fix memory leak in the path.
Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
In not enough segements error path of f2fs_write_check_point_pack(),
cp_payload is not freed. Fix the error path.
Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
In 1126e38 (fsck.f2fs: add --{no-}kernel-check to bypass kernel version diff or
not) the automatic check on a kernel version change was made optional and
disabled by default on non-Android systems. This also disabled the update of
the kernel version stored in the superblock if --kernel-check is not given.
Restore the update in case a full check is forced via -y or -f to properly
reflect the version in the superblock.
Signed-off-by: Michael Lass <bevan@bi-co.net>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
This patch add an ioctl to get filename encryption mode.
Signed-off-by: Robin Hsu <robinhsu@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>
Increase debug level from 0 to 1 in migrate_block.
It seems that doesn't need to display debug message by default.
Signed-off-by: Jaehoon Chung <jh80.chung@samsung.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
As beroal <me@beroal.in.ua> reported: f2fs image can be corrupted
after below testcase:
1. truncate -s $((256*1024*1024)) img
2. mkfs.f2fs -f img
3. mount -t f2fs -o loop img /mnt
4. xfs_io -f /mnt/file -c "pwrite 0 2M" -c "fsync"
5. umount /mnt
6. resize.f2fs -s -t 262144 img
7. fsck.f2fs img
The root cause is we forgot to copy original sit/nat bitmap to
new checkpoint during rebuild_checkpoint(), fix it.
Reported-and-testd-by: beroal <me@beroal.in.ua>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
beroal reported a mount failure due to broken valid_user_blocks.
[ 6890.647749] F2FS-fs (loop0): Wrong valid_user_blocks: 16040048,
user_block_count: 10016768
From fsck,
segment_count_main [0x 9a95 : 39573]
-> 39573 * 2MB = 78GB as user space
overprov_segment_count [0x 4e29 : 20009]
-> 20009 * 2MB = 40GB as overprovisioned space which user can't see.
But,
[FSCK] valid_block_count matching with CP [Ok..] [0xf4c070]
-> 0xf4c070 = 16040048
valid_block_count [0x f4c070 : 16040048]
-> So, this is correct.
It turns out resize.f2fs gave very large and wrong overprovisioning space
result in shortage of user blocks. The root cause was f2fs_get_usable_segments()
didn't consider resize case which needs segment_count_main from new superblock.
Fixes: f8410857b7 ("f2fs-tools: zns zone-capacity support")
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
I use user namespaces & debootstrap to bootstrap a system, and then
copy that into an image using sload.f2fs. However, without an option to
preserve the owner user & group of files, some files end up being owned
by the wrong user / group (0:0), which causes all kinds of other issues
This patch adds an option -P to preserve the user and group of files.
Signed-off-by: Daniel Abrecht <public@danielabrecht.ch>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
memory which malloc for uevent don't free at get_rootdev(),
it would cause memory leak, so free it before return.
Signed-off-by: Junyong Sun <sunjy516@gmail.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Similar to .db files, these are randomly updated extremely frequently.
Signed-off-by: Park Ju Hyung <qkrwngud825@gmail.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Old blk_zone and blk_zone_report should be compatible with v2.
And, old blk_zone_report uses reserved[4] which breaks the build.
struct blk_zone_report {
__u64 sector;
__u32 nr_zones;
__u8 reserved[4];
struct blk_zone zones[0];
};
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
When formatting conventional partition with zoned one, we should align
the starting block address of next device to the zone size.
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
- mkfs.f2fs -s 4 -c second_dev first_dev
- fsck.f2fs first
Info: Segments per section = 1
Info: Sections per zone = 1
Info: sector size = 512
Info: total sectors = 69668488 (34017 MB)
Segment count (19128) mismatch with total segments from devices (19130) Can't find a valid F2FS superblock at 0x0
Segment count (19128) mismatch with total segments from devices (19130) Can't find a valid F2FS superblock at 0x1
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
To use Android defaults, you must use `-g android`. However, the man
page previously implied that you need only use `-g`.
Since you must specify `android`, I figured it would be appropriate
to name that field `default-options` in the man page. If there was ever
a reason to have a different set of default options, the `-g` option
could be used with a new `default-options` value, i.e. `-g example`.
Signed-off-by: John A. Leuenhagen <john@zlima12.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
If sload.f2fs encounters a file with nr_links > 1, it will mark it
as a possible hard link by remembering the original device and
inode. When sload.f2fs creates the file, it will check if it has
already created a file for the same original device and inode. If
so, it will add the original inode to the directory and increment
the number of links to it, instead of writing a new inode.
This allows sload.f2fs to accurately reproduce a directory tree that
contains hard links, such as those created by ostree. Without this
patch, directory trees containing hard links result in the content of
the files being duplicated.
This is version 2 of the patch; it has been rebased against the dev
branch and includes a fix from Jaegeuk Kim to avoid building data
contents twice on hard linked files.
Co-authored-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Jordan Webb <jordan@getseam.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Add F2FS compression support for sload
* Support file extension filter, either default-accept or default-deny
policy
* Support choice of compression algorithm, LZO (version 2) or LZ4
(default)
* Support custom log of cluster size
* Support minimum number of compressed blocks per cluster (default 1).
A cluster will not be compressed if the number can not be met.
* suuport -r (read-only) option
This releases compressed blocks to secure free space in advance. Note that,
all compressed files will have the immutable bit.
* Added manpage update
* Remove unecessary qbuf allocation (Jaegeuk, suggested by Satya)
Signed-off-by: Robin Hsu <robinhsu@google.com>
[Jaegeuk Kim: fix some bugs and refactor names]
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Added new commands, compress and decompress, to support
F2FS_IOC_COMPRESS_FILE and F2FS_IOC_DECOMPRESS_FILE.
Signed-off-by: Daeho Jeong <daehojeong@google.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
When error were found, we won't need to do any initialization but
just quit.
Signed-off-by: Robin Hsu <robinhsu@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
This patch adds a message when formatting the disk with compression.
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Added new commands, get_coption and set_coption, to support
F2FS_IOC_GET_COMPRESS_OPTION and F2FS_IOC_SET_COMPRESS_OPTION.
Signed-off-by: Daeho Jeong <daehojeong@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Fix a compiling error triggered by ASSERT(exp), when exp contains '%'
(e.g. integer modular operator)
Signed-off-by: Robin Hsu <robinhsu@google.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
'ret' should not have been used here: otherwise, it would be wrongly used
as the error code and then be returned from main().
Signed-off-by: Robin Hsu <robinhsu@google.com>
Reviewed-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
As Eric reported:
Commit 7a22451bc2 ("fsck.f2fs: fix to check validation of i_xattr_nid")
This commit caused a regression where 'dump.f2fs -i <inode> <device>'
now segfaults if the inode has any extended attributes.
It's because read_all_xattrs() now calls fsck_sanity_check_nid(), which
eventually dereferences f2fs_fsck::main_area_bitmap, which is NULL.
I'm not sure what was intended here.
Here's the output from gdb:
Program received signal SIGSEGV, Segmentation fault.
0x00007ffff7f750fa in f2fs_test_bit (nr=1024, p=0x0) at libf2fs.c:304
304 return (mask & *addr) != 0;
(gdb) bt
ntype=TYPE_XATTR, ni=0x7fffffffdd20) at fsck.c:449
ntype=TYPE_XATTR, ni=0x7fffffffdd20) at fsck.c:495
fsck_sanity_check_nid() should only called from fsck.f2fs context, rather
than dump.f2fs, otherwise it may cause dereferencing structure fields of
fsck incorrectly.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
