Steps to reproduce:
while true; do
dd if=/dev/zero of=/btrfs_dir/file count=[fs_size * 75%]
rm /btrfs_dir/file
sync
done
And we'll see dd failed because btrfs return NO_SPACE.
Reason:
Normally, btrfs_commit_transaction() call btrfs_run_delayed_iputs()
in end to free fs space for next write, but sometimes it hadn't
done work on time, because btrfs-cleaner thread get delayed-iputs
from list before, but do iput() after next write.
This is log:
[ 2569.050776] comm=btrfs-cleaner func=btrfs_evict_inode() begin
[ 2569.084280] comm=sync func=btrfs_commit_transaction() call btrfs_run_delayed_iputs()
[ 2569.085418] comm=sync func=btrfs_commit_transaction() done btrfs_run_delayed_iputs()
[ 2569.087554] comm=sync func=btrfs_commit_transaction() end
[ 2569.191081] comm=dd begin
[ 2569.790112] comm=dd func=__btrfs_buffered_write() ret=-28
[ 2569.847479] comm=btrfs-cleaner func=add_pinned_bytes() 0 + 32677888 = 32677888
[ 2569.849530] comm=btrfs-cleaner func=add_pinned_bytes() 32677888 + 23834624 = 56512512
...
[ 2569.903893] comm=btrfs-cleaner func=add_pinned_bytes() 943976448 + 21762048 = 965738496
[ 2569.908270] comm=btrfs-cleaner func=btrfs_evict_inode() end
Fix:
Make btrfs_commit_transaction() wait current running btrfs-cleaner's
delayed-iputs() done in end.
Test:
Use script similar to above(more complex),
before patch:
7 failed in 100 * 20 loop.
after patch:
0 failed in 100 * 20 loop.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
space_info's value calculation is some complex and easy to cause
bug, add WARN_ON() to help debug.
Changelog v1->v2:
Put WARN_ON()s under the ENOSPC_DEBUG mount option.
Suggested by: David Sterba <dsterba@suse.cz>
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
Bug1:
space_info->bytes_readonly was set to very large(negative) value in
btrfs_remove_block_group().
Reason:
Current code set block_group_cache->pinned = 0 in btrfs_delete_unused_bgs(),
but above space was not counted to space_info->bytes_readonly.
Then in btrfs_remove_block_group():
block_group->space_info->bytes_readonly -= block_group->key.offset;
We can see following value in trace:
btrfs_remove_block_group: pid=2677 comm=btrfs-cleaner WARNING: bytes_readonly=12582912, key.offset=134217728
Bug2:
space_info->total_bytes_pinned grow to value larger than fs size.
In a 1.2G fs, we can get following trace log:
at first:
ZL_DEBUG: add_pinned_bytes: pid=2710 comm=sync change total_bytes_pinned flags=1 869793792 + 95944704 = 965738496
after some op:
ZL_DEBUG: add_pinned_bytes: pid=2770 comm=sync change total_bytes_pinned flags=1 1780178944 + 95944704 = 1876123648
after some op:
ZL_DEBUG: add_pinned_bytes: pid=3193 comm=sync change total_bytes_pinned flags=1 2924568576 + 95551488 = 3020120064
...
Reason:
Similar to bug1, we also need to adjust space_info->total_bytes_pinned
in above code block.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
If we have any chance to make a successful write, we should not give up.
This patch adjust commit-transaction condition from:
pinned >= wanted
to
left + pinned >= wanted
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
It is another reason for NO_SPACE case.
When we found enough free space in loop and saved them to
max_hole_start/size before, and tail space contains pending extent,
origional innocent max_hole_start/size are reset in retry.
As a result, find_free_dev_extent() returns less space than it can,
and cause NO_SPACE in user program.
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
Old code bypass commit transaction when we don't have enough
pinned space, but another case is there exist freed bgs in current
transction, it have possibility to make alloc_chunk success.
This patch modify the condition to:
if (have_free_bg || have_pinned_space) commit_transaction()
Confirmed above action by printk before and after patch.
Signed-off-by: Zhao Lei <zhaolei@cn.fujitsu.com>
Signed-off-by: Chris Mason <clm@fb.com>
Commit 0d97a64e0 creates a new variable but doesn't always set it up.
This puts it back to the original method (key.offset + 1) for the cases
not covered by Filipe's new logic.
Signed-off-by: Chris Mason <clm@fb.com>
If we attempt to clone a 0 length region into a file we can end up
inserting a range in the inode's extent_io tree with a start offset
that is greater then the end offset, which triggers immediately the
following warning:
[ 3914.619057] WARNING: CPU: 17 PID: 4199 at fs/btrfs/extent_io.c:435 insert_state+0x4b/0x10b [btrfs]()
[ 3914.620886] BTRFS: end < start 4095 4096
(...)
[ 3914.638093] Call Trace:
[ 3914.638636] [<ffffffff81425fd9>] dump_stack+0x4c/0x65
[ 3914.639620] [<ffffffff81045390>] warn_slowpath_common+0xa1/0xbb
[ 3914.640789] [<ffffffffa03ca44f>] ? insert_state+0x4b/0x10b [btrfs]
[ 3914.642041] [<ffffffff810453f0>] warn_slowpath_fmt+0x46/0x48
[ 3914.643236] [<ffffffffa03ca44f>] insert_state+0x4b/0x10b [btrfs]
[ 3914.644441] [<ffffffffa03ca729>] __set_extent_bit+0x107/0x3f4 [btrfs]
[ 3914.645711] [<ffffffffa03cb256>] lock_extent_bits+0x65/0x1bf [btrfs]
[ 3914.646914] [<ffffffff8142b2fb>] ? _raw_spin_unlock+0x28/0x33
[ 3914.648058] [<ffffffffa03cbac4>] ? test_range_bit+0xcc/0xde [btrfs]
[ 3914.650105] [<ffffffffa03cb3c3>] lock_extent+0x13/0x15 [btrfs]
[ 3914.651361] [<ffffffffa03db39e>] lock_extent_range+0x3d/0xcd [btrfs]
[ 3914.652761] [<ffffffffa03de1fe>] btrfs_ioctl_clone+0x278/0x388 [btrfs]
[ 3914.654128] [<ffffffff811226dd>] ? might_fault+0x58/0xb5
[ 3914.655320] [<ffffffffa03e0909>] btrfs_ioctl+0xb51/0x2195 [btrfs]
(...)
[ 3914.669271] ---[ end trace 14843d3e2e622fc1 ]---
This later makes the inode eviction handler enter an infinite loop that
keeps dumping the following warning over and over:
[ 3915.117629] WARNING: CPU: 22 PID: 4228 at fs/btrfs/extent_io.c:435 insert_state+0x4b/0x10b [btrfs]()
[ 3915.119913] BTRFS: end < start 4095 4096
(...)
[ 3915.137394] Call Trace:
[ 3915.137913] [<ffffffff81425fd9>] dump_stack+0x4c/0x65
[ 3915.139154] [<ffffffff81045390>] warn_slowpath_common+0xa1/0xbb
[ 3915.140316] [<ffffffffa03ca44f>] ? insert_state+0x4b/0x10b [btrfs]
[ 3915.141505] [<ffffffff810453f0>] warn_slowpath_fmt+0x46/0x48
[ 3915.142709] [<ffffffffa03ca44f>] insert_state+0x4b/0x10b [btrfs]
[ 3915.143849] [<ffffffffa03ca729>] __set_extent_bit+0x107/0x3f4 [btrfs]
[ 3915.145120] [<ffffffffa038c1e3>] ? btrfs_kill_super+0x17/0x23 [btrfs]
[ 3915.146352] [<ffffffff811548f6>] ? deactivate_locked_super+0x3b/0x50
[ 3915.147565] [<ffffffffa03cb256>] lock_extent_bits+0x65/0x1bf [btrfs]
[ 3915.148785] [<ffffffff8142b7e2>] ? _raw_write_unlock+0x28/0x33
[ 3915.149931] [<ffffffffa03bc325>] btrfs_evict_inode+0x196/0x482 [btrfs]
[ 3915.151154] [<ffffffff81168904>] evict+0xa0/0x148
[ 3915.152094] [<ffffffff811689e5>] dispose_list+0x39/0x43
[ 3915.153081] [<ffffffff81169564>] evict_inodes+0xdc/0xeb
[ 3915.154062] [<ffffffff81154418>] generic_shutdown_super+0x49/0xef
[ 3915.155193] [<ffffffff811546d1>] kill_anon_super+0x13/0x1e
[ 3915.156274] [<ffffffffa038c1e3>] btrfs_kill_super+0x17/0x23 [btrfs]
(...)
[ 3915.167404] ---[ end trace 14843d3e2e622fc2 ]---
So just bail out of the clone ioctl if the length of the region to clone
is zero, without locking any extent range, in order to prevent this issue
(same behaviour as a pwrite with a 0 length for example).
This is trivial to reproduce. For example, the steps for the test I just
made for fstests:
mkfs.btrfs -f SCRATCH_DEV
mount SCRATCH_DEV $SCRATCH_MNT
touch $SCRATCH_MNT/foo
touch $SCRATCH_MNT/bar
$CLONER_PROG -s 0 -d 4096 -l 0 $SCRATCH_MNT/foo $SCRATCH_MNT/bar
umount $SCRATCH_MNT
A test case for fstests follows soon.
CC: <stable@vger.kernel.org>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Omar Sandoval <osandov@osandov.com>
Signed-off-by: Chris Mason <clm@fb.com>
If we pass a length of 0 to the extent_same ioctl, we end up locking an
extent range with a start offset greater then its end offset (if the
destination file's offset is greater than zero). This results in a warning
from extent_io.c:insert_state through the following call chain:
btrfs_extent_same()
btrfs_double_lock()
lock_extent_range()
lock_extent(inode->io_tree, offset, offset + len - 1)
lock_extent_bits()
__set_extent_bit()
insert_state()
--> WARN_ON(end < start)
This leads to an infinite loop when evicting the inode. This is the same
problem that my previous patch titled
"Btrfs: fix inode eviction infinite loop after cloning into it" addressed
but for the extent_same ioctl instead of the clone ioctl.
CC: <stable@vger.kernel.org>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Omar Sandoval <osandov@osandov.com>
Signed-off-by: Chris Mason <clm@fb.com>
While searching for extents to clone we might find one where we only use
a part of it coming from its tail. If our destination inode is the same
the source inode, we end up removing the tail part of the extent item and
insert after a new one that point to the same extent with an adjusted
key file offset and data offset. After this we search for the next extent
item in the fs/subvol tree with a key that has an offset incremented by
one. But this second search leaves us at the new extent item we inserted
previously, and since that extent item has a non-zero data offset, it
it can make us call btrfs_drop_extents with an empty range (start == end)
which causes the following warning:
[23978.537119] WARNING: CPU: 6 PID: 16251 at fs/btrfs/file.c:550 btrfs_drop_extent_cache+0x43/0x385 [btrfs]()
(...)
[23978.557266] Call Trace:
[23978.557978] [<ffffffff81425fd9>] dump_stack+0x4c/0x65
[23978.559191] [<ffffffff81045390>] warn_slowpath_common+0xa1/0xbb
[23978.560699] [<ffffffffa047f0ea>] ? btrfs_drop_extent_cache+0x43/0x385 [btrfs]
[23978.562389] [<ffffffff8104544d>] warn_slowpath_null+0x1a/0x1c
[23978.563613] [<ffffffffa047f0ea>] btrfs_drop_extent_cache+0x43/0x385 [btrfs]
[23978.565103] [<ffffffff810e3a18>] ? time_hardirqs_off+0x15/0x28
[23978.566294] [<ffffffff81079ff8>] ? trace_hardirqs_off+0xd/0xf
[23978.567438] [<ffffffffa047f73d>] __btrfs_drop_extents+0x6b/0x9e1 [btrfs]
[23978.568702] [<ffffffff8107c03f>] ? trace_hardirqs_on+0xd/0xf
[23978.569763] [<ffffffff811441c0>] ? ____cache_alloc+0x69/0x2eb
[23978.570817] [<ffffffff81142269>] ? virt_to_head_page+0x9/0x36
[23978.571872] [<ffffffff81143c15>] ? cache_alloc_debugcheck_after.isra.42+0x16c/0x1cb
[23978.573466] [<ffffffff811420d5>] ? kmemleak_alloc_recursive.constprop.52+0x16/0x18
[23978.574962] [<ffffffffa0480d07>] btrfs_drop_extents+0x66/0x7f [btrfs]
[23978.576179] [<ffffffffa049aa35>] btrfs_clone+0x516/0xaf5 [btrfs]
[23978.577311] [<ffffffffa04983dc>] ? lock_extent_range+0x7b/0xcd [btrfs]
[23978.578520] [<ffffffffa049b2a2>] btrfs_ioctl_clone+0x28e/0x39f [btrfs]
[23978.580282] [<ffffffffa049d9ae>] btrfs_ioctl+0xb51/0x219a [btrfs]
(...)
[23978.591887] ---[ end trace 988ec2a653d03ed3 ]---
Then we attempt to insert a new extent item with a key that already
exists, which makes btrfs_insert_empty_item return -EEXIST resulting in
abortion of the current transaction:
[23978.594355] WARNING: CPU: 6 PID: 16251 at fs/btrfs/super.c:260 __btrfs_abort_transaction+0x52/0x114 [btrfs]()
(...)
[23978.622589] Call Trace:
[23978.623181] [<ffffffff81425fd9>] dump_stack+0x4c/0x65
[23978.624359] [<ffffffff81045390>] warn_slowpath_common+0xa1/0xbb
[23978.625573] [<ffffffffa044ab6c>] ? __btrfs_abort_transaction+0x52/0x114 [btrfs]
[23978.626971] [<ffffffff810453f0>] warn_slowpath_fmt+0x46/0x48
[23978.628003] [<ffffffff8108a6c8>] ? vprintk_default+0x1d/0x1f
[23978.629138] [<ffffffffa044ab6c>] __btrfs_abort_transaction+0x52/0x114 [btrfs]
[23978.630528] [<ffffffffa049ad1b>] btrfs_clone+0x7fc/0xaf5 [btrfs]
[23978.631635] [<ffffffffa04983dc>] ? lock_extent_range+0x7b/0xcd [btrfs]
[23978.632886] [<ffffffffa049b2a2>] btrfs_ioctl_clone+0x28e/0x39f [btrfs]
[23978.634119] [<ffffffffa049d9ae>] btrfs_ioctl+0xb51/0x219a [btrfs]
(...)
[23978.647714] ---[ end trace 988ec2a653d03ed4 ]---
This is wrong because we should not process the extent item that we just
inserted previously, and instead process the extent item that follows it
in the tree
For example for the test case I wrote for fstests:
bs=$((64 * 1024))
mkfs.btrfs -f -l $bs -O ^no-holes /dev/sdc
mount /dev/sdc /mnt
xfs_io -f -c "pwrite -S 0xaa $(($bs * 2)) $(($bs * 2))" /mnt/foo
$CLONER_PROG -s $((3 * $bs)) -d $((267 * $bs)) -l 0 /mnt/foo /mnt/foo
$CLONER_PROG -s $((217 * $bs)) -d $((95 * $bs)) -l 0 /mnt/foo /mnt/foo
The second clone call fails with -EEXIST, because when we process the
first extent item (offset 262144), we drop part of it (counting from the
end) and then insert a new extent item with a key greater then the key we
found. The next time we search the tree we search for a key with offset
262144 + 1, which leaves us at the new extent item we have just inserted
but we think it refers to an extent that we need to clone.
Fix this by ensuring the next search key uses an offset corresponding to
the offset of the key we found previously plus the data length of the
corresponding extent item. This ensures we skip new extent items that we
inserted and works for the case of implicit holes too (NO_HOLES feature).
A test case for fstests follows soon.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
Near the end of close_ctree, we're calling btrfs_free_block_rsv
to free up the orphan rsv. The problem is this call updates the
space_info, which has already been freed.
This adds a new __ function that directly calls kfree instead of trying
to update the space infos.
Signed-off-by: Chris Mason <clm@fb.com>
We loop through all of the dirty block groups during commit and write
the free space cache. In order to make sure the cache is currect, we do
this while no other writers are allowed in the commit.
If a large number of block groups are dirty, this can introduce long
stalls during the final stages of the commit, which can block new procs
trying to change the filesystem.
This commit changes the block group cache writeout to take appropriate
locks and allow it to run earlier in the commit. We'll still have to
redo some of the block groups, but it means we can get most of the work
out of the way without blocking the entire FS.
Signed-off-by: Chris Mason <clm@fb.com>
In order to create the free space cache concurrently with FS modifications,
we need to take a few block group locks.
The cache code also does kmap, which would schedule with the locks held.
Instead of going through kmap_atomic, lets just use lowmem for the cache
pages.
Signed-off-by: Chris Mason <clm@fb.com>
Block group cache writeout is currently waiting on the pages for each
block group cache before moving on to writing the next one. This commit
switches things around to send down all the caches and then wait on them
in batches.
The end result is much faster, since we're keeping the disk pipeline
full.
Signed-off-by: Chris Mason <clm@fb.com>
We'll need to put the io_ctl into the block_group cache struct, so
name it struct btrfs_io_ctl and move it into ctree.h
Signed-off-by: Chris Mason <clm@fb.com>
btrfs_evict_inode() needs to be more careful about stealing from the
global_rsv. We dont' want to end up aborting commit with ENOSPC just
because the evict_inode code was too greedy.
Signed-off-by: Chris Mason <clm@fb.com>
We're triggering a huge number of commits from
btrfs_async_reclaim_metadata_space. These aren't really requried,
because everyone calling the async reclaim code is going to end up
triggering a commit on their own.
Signed-off-by: Chris Mason <clm@fb.com>
When truncate starts, it allocates some space in the block reserves so
that we'll have enough to update metadata along the way.
For very large files, we can easily go through all of that space as we
loop through the extents. This changes truncate to refill the space
reservation as it progresses through the file.
Signed-off-by: Chris Mason <clm@fb.com>
As we delete large extents, we end up doing huge amounts of COW in order
to delete the corresponding crcs. This adds accounting so that we keep
track of that space and flushing of delayed refs so that we don't build
up too much delayed crc work.
This helps limit the delayed work that must be done at commit time and
tries to avoid ENOSPC aborts because the crcs eat all the global
reserves.
Signed-off-by: Chris Mason <clm@fb.com>
When we are deleting large files with large extents, we are building up
a huge set of delayed refs for processing. Truncate isn't checking
often enough to see if we need to back off and process those, or let
a commit proceed.
The end result is long stalls after the rm, and very long commit times.
During the commits, other processes back up waiting to start new
transactions and we get into trouble.
Signed-off-by: Chris Mason <clm@fb.com>
Building alpha:allmodconfig fails with
fs/btrfs/inode.c: In function 'check_direct_IO':
fs/btrfs/inode.c:8050:2: error: implicit declaration of function 'iov_iter_alignment'
due to a missing include file.
Fixes: 3737c63e1fb0 ("fs: move struct kiocb to fs.h")
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Acked-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <clm@fb.com>
This was used to make sure that a fresh btrfs from an older mkfs.btrfs,
but it also allows us to mount a buggy btrfs if this btrfs has the right
superblock head part but has something wrong with chunk tree part[1], and
after that we can hit BUG_ON()s set in the code to prevent something
impossible.
Since David has released "Btrfs progs v3.19-rc2", just remove the check,
if anyone who wants to make a fresh btrfs, please use the latest one.
[1]: http://www.spinics.net/lists/linux-btrfs/msg42358.html
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: Omar Sandoval <osandov@osandov.com>
Reviewed-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <clm@fb.com>
Orphans in the fs tree are cleaned up via open_ctree and subvolume
orphans are cleaned via btrfs_lookup_dentry -- except when a default
subvolume is in use. The name for the default subvolume uses a manual
lookup that doesn't trigger orphan cleanup and needs to trigger it
manually as well. This doesn't apply to the remount case since the
subvolumes are cleaned up by walking the root radix tree.
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Reviewed-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <clm@fb.com>
The private_data member of the Btrfs control device file
(/dev/btrfs-control) is used to hold the current transaction and needs
to be initialized to NULL to signify that no transaction is in progress.
We explicitly set the control file's private_data to NULL to be
independent of whatever value the misc subsystem initializes it to.
Backstory:
----------
The misc subsystem (which is used by /dev/btrfs-control) initializes
a file's private_data to point to the misc device when a driver has
registered a custom open file operation and initializes it to NULL
when a custom open file operation has *not* been provided.
This subtle quirk is confusing, to the point where kernel code registers
*empty* file open operations to have private_data point to the misc
device structure.
And it leads to bugs, where the addition or removal of a custom open
file operation surprisingly changes the initial contents of a file's
private_data structure.
To simplify things in the misc subsystem, a patch [1] has been proposed
to *always* set private_data to point to the misc device instead of
only doing this when a custom open file operation has been registered.
But before we can fix this in the misc subsystem itself, we need to
modify the (few) drivers that rely on this very subtle behavior.
[1] https://lkml.org/lkml/2014/12/4/939
Signed-off-by: Martin Kepplinger <martink@posteo.de>
Signed-off-by: Tom Van Braeckel <tomvanbraeckel@gmail.com>
Signed-off-by: Chris Mason <clm@fb.com>
fiemap_fill_next_extent returns 0 on success, -errno on error, 1 if this was
the last extent that will fit in user array. If 1 is returned, the return
value may eventually returned to user space, which should not happen, according
to manpage of ioctl.
Signed-off-by: Chengyu Song <csong84@gatech.edu>
Reviewed-by: David Sterba <dsterba@suse.cz>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>
Due to insufficient check in btrfs_is_valid_xattr, this unexpectedly
works:
$ touch file
$ setfattr -n user. -v 1 file
$ getfattr -d file
user.="1"
ie. the missing attribute name after the namespace.
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=94291
Reported-by: William Douglas <william.douglas@intel.com>
CC: <stable@vger.kernel.org> # 2.6.29+
Signed-off-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <clm@fb.com>
While committing a transaction we free the log roots before we write the
new super block. Freeing the log roots implies marking the disk location
of every node/leaf (metadata extent) as pinned before the new super block
is written. This is to prevent the disk location of log metadata extents
from being reused before the new super block is written, otherwise we
would have a corrupted log tree if before the new super block is written
a crash/reboot happens and the location of any log tree metadata extent
ended up being reused and rewritten.
Even though we pinned the log tree's metadata extents, we were issuing a
discard against them if the fs was mounted with the -o discard option,
resulting in corruption of the log tree if a crash/reboot happened before
writing the new super block - the next time the fs was mounted, during
the log replay process we would find nodes/leafs of the log btree with
a content full of zeroes, causing the process to fail and require the
use of the tool btrfs-zero-log to wipeout the log tree (and all data
previously fsynced becoming lost forever).
Fix this by not doing a discard when pinning an extent. The discard will
be done later when it's safe (after the new super block is committed) at
extent-tree.c:btrfs_finish_extent_commit().
Fixes: e688b7252f (Btrfs: fix extent pinning bugs in the tree log)
CC: <stable@vger.kernel.org>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
We can get into inconsistency between inodes and directory entries
after fsyncing a directory. The issue is that while a directory gets
the new dentries persisted in the fsync log and replayed at mount time,
the link count of the inode that directory entries point to doesn't
get updated, staying with an incorrect link count (smaller then the
correct value). This later leads to stale file handle errors when
accessing (including attempt to delete) some of the links if all the
other ones are removed, which also implies impossibility to delete the
parent directories, since the dentries can not be removed.
Another issue is that (unlike ext3/4, xfs, f2fs, reiserfs, nilfs2),
when fsyncing a directory, new files aren't logged (their metadata and
dentries) nor any child directories. So this patch fixes this issue too,
since it has the same resolution as the incorrect inode link count issue
mentioned before.
This is very easy to reproduce, and the following excerpt from my test
case for xfstests shows how:
_scratch_mkfs >> $seqres.full 2>&1
_init_flakey
_mount_flakey
# Create our main test file and directory.
$XFS_IO_PROG -f -c "pwrite -S 0xaa 0 8K" $SCRATCH_MNT/foo | _filter_xfs_io
mkdir $SCRATCH_MNT/mydir
# Make sure all metadata and data are durably persisted.
sync
# Add a hard link to 'foo' inside our test directory and fsync only the
# directory. The btrfs fsync implementation had a bug that caused the new
# directory entry to be visible after the fsync log replay but, the inode
# of our file remained with a link count of 1.
ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/foo_2
# Add a few more links and new files.
# This is just to verify nothing breaks or gives incorrect results after the
# fsync log is replayed.
ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/foo_3
$XFS_IO_PROG -f -c "pwrite -S 0xff 0 64K" $SCRATCH_MNT/hello | _filter_xfs_io
ln $SCRATCH_MNT/hello $SCRATCH_MNT/mydir/hello_2
# Add some subdirectories and new files and links to them. This is to verify
# that after fsyncing our top level directory 'mydir', all the subdirectories
# and their files/links are registered in the fsync log and exist after the
# fsync log is replayed.
mkdir -p $SCRATCH_MNT/mydir/x/y/z
ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/x/y/foo_y_link
ln $SCRATCH_MNT/foo $SCRATCH_MNT/mydir/x/y/z/foo_z_link
touch $SCRATCH_MNT/mydir/x/y/z/qwerty
# Now fsync only our top directory.
$XFS_IO_PROG -c "fsync" $SCRATCH_MNT/mydir
# And fsync now our new file named 'hello', just to verify later that it has
# the expected content and that the previous fsync on the directory 'mydir' had
# no bad influence on this fsync.
$XFS_IO_PROG -c "fsync" $SCRATCH_MNT/hello
# Simulate a crash/power loss.
_load_flakey_table $FLAKEY_DROP_WRITES
_unmount_flakey
_load_flakey_table $FLAKEY_ALLOW_WRITES
_mount_flakey
# Verify the content of our file 'foo' remains the same as before, 8192 bytes,
# all with the value 0xaa.
echo "File 'foo' content after log replay:"
od -t x1 $SCRATCH_MNT/foo
# Remove the first name of our inode. Because of the directory fsync bug, the
# inode's link count was 1 instead of 5, so removing the 'foo' name ended up
# deleting the inode and the other names became stale directory entries (still
# visible to applications). Attempting to remove or access the remaining
# dentries pointing to that inode resulted in stale file handle errors and
# made it impossible to remove the parent directories since it was impossible
# for them to become empty.
echo "file 'foo' link count after log replay: $(stat -c %h $SCRATCH_MNT/foo)"
rm -f $SCRATCH_MNT/foo
# Now verify that all files, links and directories created before fsyncing our
# directory exist after the fsync log was replayed.
[ -f $SCRATCH_MNT/mydir/foo_2 ] || echo "Link mydir/foo_2 is missing"
[ -f $SCRATCH_MNT/mydir/foo_3 ] || echo "Link mydir/foo_3 is missing"
[ -f $SCRATCH_MNT/hello ] || echo "File hello is missing"
[ -f $SCRATCH_MNT/mydir/hello_2 ] || echo "Link mydir/hello_2 is missing"
[ -f $SCRATCH_MNT/mydir/x/y/foo_y_link ] || \
echo "Link mydir/x/y/foo_y_link is missing"
[ -f $SCRATCH_MNT/mydir/x/y/z/foo_z_link ] || \
echo "Link mydir/x/y/z/foo_z_link is missing"
[ -f $SCRATCH_MNT/mydir/x/y/z/qwerty ] || \
echo "File mydir/x/y/z/qwerty is missing"
# We expect our file here to have a size of 64Kb and all the bytes having the
# value 0xff.
echo "file 'hello' content after log replay:"
od -t x1 $SCRATCH_MNT/hello
# Now remove all files/links, under our test directory 'mydir', and verify we
# can remove all the directories.
rm -f $SCRATCH_MNT/mydir/x/y/z/*
rmdir $SCRATCH_MNT/mydir/x/y/z
rm -f $SCRATCH_MNT/mydir/x/y/*
rmdir $SCRATCH_MNT/mydir/x/y
rmdir $SCRATCH_MNT/mydir/x
rm -f $SCRATCH_MNT/mydir/*
rmdir $SCRATCH_MNT/mydir
# An fsck, run by the fstests framework everytime a test finishes, also detected
# the inconsistency and printed the following error message:
#
# root 5 inode 257 errors 2001, no inode item, link count wrong
# unresolved ref dir 258 index 2 namelen 5 name foo_2 filetype 1 errors 4, no inode ref
# unresolved ref dir 258 index 3 namelen 5 name foo_3 filetype 1 errors 4, no inode ref
status=0
exit
The expected golden output for the test is:
wrote 8192/8192 bytes at offset 0
XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
wrote 65536/65536 bytes at offset 0
XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
File 'foo' content after log replay:
0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa
*
0020000
file 'foo' link count after log replay: 5
file 'hello' content after log replay:
0000000 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
*
0200000
Which is the output after this patch and when running the test against
ext3/4, xfs, f2fs, reiserfs or nilfs2. Without this patch, the test's
output is:
wrote 8192/8192 bytes at offset 0
XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
wrote 65536/65536 bytes at offset 0
XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
File 'foo' content after log replay:
0000000 aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa aa
*
0020000
file 'foo' link count after log replay: 1
Link mydir/foo_2 is missing
Link mydir/foo_3 is missing
Link mydir/x/y/foo_y_link is missing
Link mydir/x/y/z/foo_z_link is missing
File mydir/x/y/z/qwerty is missing
file 'hello' content after log replay:
0000000 ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
*
0200000
rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x/y/z': No such file or directory
rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x/y': No such file or directory
rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/x': No such file or directory
rm: cannot remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/foo_2': Stale file handle
rm: cannot remove '/home/fdmanana/btrfs-tests/scratch_1/mydir/foo_3': Stale file handle
rmdir: failed to remove '/home/fdmanana/btrfs-tests/scratch_1/mydir': Directory not empty
Fsck, without this fix, also complains about the wrong link count:
root 5 inode 257 errors 2001, no inode item, link count wrong
unresolved ref dir 258 index 2 namelen 5 name foo_2 filetype 1 errors 4, no inode ref
unresolved ref dir 258 index 3 namelen 5 name foo_3 filetype 1 errors 4, no inode ref
So fix this by logging the inodes that the dentries point to when
fsyncing a directory.
A test case for xfstests follows.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
After looking at Liu Bo's recent patch (titled
"Btrfs: fix comp_oper to get right order") I realized the search made by
qgroup_oper_exists() was buggy because its rbtree navigation comparison
function, comp_oper_exist(), only looks at the fields bytenr and ref_root
of a tree node, ignoring the seq field completely. This was wrong because
when we insert a node into the rbtree we use comp_oper(), which takes a
decision based first on bytenr, then on seq and then on the ref_root field.
That means qgroup_oper_exists() could miss the fact that at least one
operation with given bytenr and ref_root exists.
Consider the following simple example of a 3 nodes qgroup operations
rbtree (created using comp_oper before this patch), where each node's key
is a tuple with the shape (bytenr, seq, ref_root, op):
[ (4096, 2, 20, op X) ]
/ \
/ \
[ (4096, 1, 5, op Y) ] [ (4096, 3, 10, op Z) ]
qgroup_oper_exists() when called to search for an existing operation for
bytenr 4096 and ref root 10 wouldn't find anything because it would go to
the left subtree instead of the right subtree, since comp_oper_exits()
ignores the seq field completely.
Fix this by changing the insertion navigation function to use the ref_root
field right after using the bytenr field and before using the seq field,
so that qgroup_oper_exists() / comp_oper_exist() work as expected.
This patch applies on top of the patch mentioned above from Liu.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
If we fallocate(), without the keep size flag, into an area already covered
by an extent previously fallocated, we were updating the inode's i_size but
we weren't updating the inode item in the fs/subvol tree. A following umount
+ mount would result in a loss of the inode's size (and an fsync would miss
too the fact that the inode changed).
Reproducer:
$ mkfs.btrfs -f /dev/sdd
$ mount /dev/sdd /mnt
$ fallocate -n -l 1M /mnt/foobar
$ fallocate -l 512K /mnt/foobar
$ umount /mnt
$ mount /dev/sdd /mnt
$ od -t x1 /mnt/foobar
0000000
The expected result is:
$ od -t x1 /mnt/foobar
0000000 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
*
2000000
A test case for fstests follows soon.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>
The logic to detect path loops when attempting to apply a pending
directory rename, introduced in commit
f959492fc1 (Btrfs: send, fix more issues related to directory renames)
is no longer needed, and the respective fstests test case for that commit,
btrfs/045, now passes without this code (as well as all the other test
cases for send/receive).
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
If a directory's reference ends up being orphanized, because the inode
currently being processed has a new path that matches that directory's
path, make sure we evict the name of the directory from the name cache.
This is because there might be descendent inodes (either directories or
regular files) that will be orphanized later too, and therefore the
orphan name of the ancestor must be used, otherwise we send issue rename
operations with a wrong path in the send stream.
Reproducer:
$ mkfs.btrfs -f /dev/sdb
$ mount /dev/sdb /mnt
$ mkdir -p /mnt/data/n1/n2/p1/p2
$ mkdir /mnt/data/n4
$ mkdir -p /mnt/data/p1/p2
$ btrfs subvolume snapshot -r /mnt /mnt/snap1
$ mv /mnt/data/p1/p2 /mnt/data
$ mv /mnt/data/n1/n2/p1/p2 /mnt/data/p1
$ mv /mnt/data/p2 /mnt/data/n1/n2/p1
$ mv /mnt/data/n1/n2 /mnt/data/p1
$ mv /mnt/data/p1 /mnt/data/n4
$ mv /mnt/data/n4/p1/n2/p1 /mnt/data
$ btrfs subvolume snapshot -r /mnt /mnt/snap2
$ btrfs send /mnt/snap1 -f /tmp/1.send
$ btrfs send -p /mnt/snap1 /mnt/snap2 -f /tmp/2.send
$ mkfs.btrfs -f /dev/sdc
$ mount /dev/sdc /mnt2
$ btrfs receive /mnt2 -f /tmp/1.send
$ btrfs receive /mnt2 -f /tmp/2.send
ERROR: rename data/p1/p2 -> data/n4/p1/p2 failed. no such file or directory
Directories data/p1 (inode 263) and data/p1/p2 (inode 264) in the parent
snapshot are both orphanized during the incremental send, and as soon as
data/p1 is orphanized, we must make sure that when orphanizing data/p1/p2
we use a source path of o263-6-o/p2 for the rename operation instead of
the old path data/p1/p2 (the one before the orphanization of inode 263).
A test case for xfstests follows soon.
Reported-by: Robbie Ko <robbieko@synology.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
If the clone root was not readonly or the dead flag was set on it, we were
leaving without decrementing the root's send_progress counter (and before
we just incremented it). If a concurrent snapshot deletion was in progress
and ended up being aborted, it would be impossible to later attempt to
delete again the snapshot, since the root's send_in_progress counter could
never go back to 0.
We were also setting clone_sources_to_rollback to i + 1 too early - if we
bailed out because the clone root we got is not readonly or flagged as dead
we ended up later derreferencing a null pointer because we didn't assign
the clone root to sctx->clone_roots[i].root:
for (i = 0; sctx && i < clone_sources_to_rollback; i++)
btrfs_root_dec_send_in_progress(
sctx->clone_roots[i].root);
So just don't increment the send_in_progress counter if the root is readonly
or flagged as dead.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <clm@fb.com>
After we locked the root's root item, a concurrent snapshot deletion
call might have set the dead flag on it. So check if the dead flag
is set and abort if it is, just like we do for the parent root.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Chris Mason <clm@fb.com>
If we deleted xattrs from a file and fsynced the file, after a log replay
the xattrs would remain associated to the file. This was an unexpected
behaviour and differs from what other filesystems do, such as for example
xfs and ext3/4.
Fix this by, on fsync log replay, check if every xattr in the fs/subvol
tree (that belongs to a logged inode) has a matching xattr in the log,
and if it does not, delete it from the fs/subvol tree. This is a similar
approach to what we do for dentries when we replay a directory from the
fsync log.
This issue is trivial to reproduce, and the following excerpt from my
test for xfstests triggers the issue:
_crash_and_mount()
{
# Simulate a crash/power loss.
_load_flakey_table $FLAKEY_DROP_WRITES
_unmount_flakey
_load_flakey_table $FLAKEY_ALLOW_WRITES
_mount_flakey
}
rm -f $seqres.full
_scratch_mkfs >> $seqres.full 2>&1
_init_flakey
_mount_flakey
# Create out test file and add 3 xattrs to it.
touch $SCRATCH_MNT/foobar
$SETFATTR_PROG -n user.attr1 -v val1 $SCRATCH_MNT/foobar
$SETFATTR_PROG -n user.attr2 -v val2 $SCRATCH_MNT/foobar
$SETFATTR_PROG -n user.attr3 -v val3 $SCRATCH_MNT/foobar
# Make sure everything is durably persisted.
sync
# Now delete the second xattr and fsync the inode.
$SETFATTR_PROG -x user.attr2 $SCRATCH_MNT/foobar
$XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foobar
_crash_and_mount
# After the fsync log is replayed, the file should have only 2 xattrs, the ones
# named user.attr1 and user.attr3. The btrfs fsync log replay bug left the file
# with the 3 xattrs that we had before deleting the second one and fsyncing the
# file.
echo "xattr names and values after first fsync log replay:"
$GETFATTR_PROG --absolute-names --dump $SCRATCH_MNT/foobar | _filter_scratch
# Now write some data to our file, fsync it, remove the first xattr, add a new
# hard link to our file and commit the fsync log by fsyncing some other new
# file. This is to verify that after log replay our first xattr does not exist
# anymore.
echo "hello world!" >> $SCRATCH_MNT/foobar
$XFS_IO_PROG -c "fsync" $SCRATCH_MNT/foobar
$SETFATTR_PROG -x user.attr1 $SCRATCH_MNT/foobar
ln $SCRATCH_MNT/foobar $SCRATCH_MNT/foobar_link
touch $SCRATCH_MNT/qwerty
$XFS_IO_PROG -c "fsync" $SCRATCH_MNT/qwerty
_crash_and_mount
# Now only the xattr with name user.attr3 should be set in our file.
echo "xattr names and values after second fsync log replay:"
$GETFATTR_PROG --absolute-names --dump $SCRATCH_MNT/foobar | _filter_scratch
status=0
exit
The expected golden output, which is produced with this patch applied or
when testing against xfs or ext3/4, is:
xattr names and values after first fsync log replay:
# file: SCRATCH_MNT/foobar
user.attr1="val1"
user.attr3="val3"
xattr names and values after second fsync log replay:
# file: SCRATCH_MNT/foobar
user.attr3="val3"
Without this patch applied, the output is:
xattr names and values after first fsync log replay:
# file: SCRATCH_MNT/foobar
user.attr1="val1"
user.attr2="val2"
user.attr3="val3"
xattr names and values after second fsync log replay:
# file: SCRATCH_MNT/foobar
user.attr1="val1"
user.attr2="val2"
user.attr3="val3"
A patch with a test case for xfstests follows soon.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
Pull btrfs fixes from Chris Mason:
"Most of these are fixing extent reservation accounting, or corners
with tree writeback during commit.
Josef's set does add a test, which isn't strictly a fix, but it'll
keep us from making this same mistake again"
* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
Btrfs: fix outstanding_extents accounting in DIO
Btrfs: add sanity test for outstanding_extents accounting
Btrfs: just free dummy extent buffers
Btrfs: account merges/splits properly
Btrfs: prepare block group cache before writing
Btrfs: fix ASSERT(list_empty(&cur_trans->dirty_bgs_list)
Btrfs: account for the correct number of extents for delalloc reservations
Btrfs: fix merge delalloc logic
Btrfs: fix comp_oper to get right order
Btrfs: catch transaction abortion after waiting for it
btrfs: fix sizeof format specifier in btrfs_check_super_valid()
We are keeping track of how many extents we need to reserve properly based on
the amount we want to write, but we were still incrementing outstanding_extents
if we wrote less than what we requested. This isn't quite right since we will
be limited to our max extent size. So instead lets do something horrible! Keep
track of how many outstanding_extents we reserved, and decrement each time we
allocate an extent. If we use our entire reserve make sure to jack up
outstanding_extents on the inode so the accounting works out properly. Thanks,
Reported-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Josef Bacik <jbacik@fb.com>
I introduced a regression wrt outstanding_extents accounting. These are tricky
areas that aren't easily covered by xfstests as we could change MAX_EXTENT_SIZE
at any time. So add sanity tests to cover the various conditions that are
tricky in order to make sure we don't introduce regressions in the future.
Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
If we fail during our sanity tests we could get NULL deref's because we unload
the module before the dummy extent buffers are free'd via RCU. So check for
this case and just free the things directly. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
My fix
Btrfs: fix merge delalloc logic
only fixed half of the problems, it didn't fix the case where we have two large
extents on either side and then join them together with a new small extent. We
need to instead keep track of how many extents we have accounted for with each
side of the new extent, and then see how many extents we need for the new large
extent. If they match then we know we need to keep our reservation, otherwise
we need to drop our reservation. This shows up with a case like this
[BTRFS_MAX_EXTENT_SIZE+4K][4K HOLE][BTRFS_MAX_EXTENT_SIZE+4K]
Previously the logic would have said that the number extents required for the
new size (3) is larger than the number of extents required for the largest side
(2) therefore we need to keep our reservation. But this isn't the case, since
both sides require a reservation of 2 which leads to 4 for the whole range
currently reserved, but we only need 3, so we need to drop one of the
reservations. The same problem existed for splits, we'd think we only need 3
extents when creating the hole but in reality we need 4. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
Writing the block group cache will modify the extent tree quite a bit because it
truncates the old space cache and pre-allocates new stuff. To try and cut down
on the churn lets do the setup dance first, then later on hopefully we can avoid
looping with newly dirtied roots. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
Dave could hit this assert consistently running btrfs/078. This is because
when we update the block groups we could truncate the free space, which would
try to delete the csums for that range and dirty the csum root. For this to
happen we have to have already written out the csum root so it's kind of hard to
hit this case. This patch fixes this by changing the logic to only write the
dirty block groups if the dirty_cowonly_roots list is empty. This will get us
the same effect as before since we add the extent root last, and will cover the
case that we dirty some other root again but not the extent root. Thanks,
Reported-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
Direct IO can easily pass in an buffer that is greater than
BTRFS_MAX_EXTENT_SIZE, so take this into account when reserving extents in the
delalloc reservation code. Thanks,
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
My patch to properly count outstanding extents wrt MAX_EXTENT_SIZE introduced a
regression when re-dirtying already dirty areas. We have logic in split to make
sure we are taking the largest space into account but didn't have it for merge,
so it was sometimes making us think we were turning a tiny extent into a huge
extent, when in reality we already had a huge extent and needed to use the other
side in our logic. This fixes the regression that was reported by a user on
list. Thanks,
Reported-by: Markus Trippelsdorf <markus@trippelsdorf.de>
Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
Case (oper1->seq > oper2->seq) should differ with case (oper1->seq < oper2->seq).
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: David Sterba <dsterba@suse.cz>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Chris Mason <clm@fb.com>
This problem is uncovered by a test case: http://patchwork.ozlabs.org/patch/244297.
Fsync() can report success when it actually doesn't. When we
have several threads running fsync() at the same tiem and in one fsync() we
get a transaction abortion due to some problems(in the test case it's disk
failures), and other fsync()s may return successfully which makes userspace
programs think that data is now safely flushed into disk.
It's because that after fsyncs() fail btrfs_sync_log() due to disk failures,
they get to try btrfs_commit_transaction() where it finds that there is
already a transaction being committed, and they'll just call wait_for_commit()
and return. Note that we actually check "trans->aborted" in btrfs_end_transaction,
but it's likely that the error message is still not yet throwed out and only after
wait_for_commit() we're sure whether the transaction is committed successfully.
This add the necessary check and it now passes the test.
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Signed-off-by: Chris Mason <clm@fb.com>
