| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The volume mutex does not protect against anything in this case, the
comment about scrub is right but not related to locking and looks
confusing. The comment in btrfs_find_device_missing_or_by_path is wrong
and confusing too.
The device_list_mutex is not held here to protect device lookup, but in
this case device replace cannot run in parallel with device removal (due
to exclusive op protection), so we don't need further locking here.
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
The function __cancel_balance name is confusing with the cancel
operation of balance and it really resets the state of balance back to
zero. The unset_balance_control helper is called only from one place and
simple enough to be inlined.
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Replace a WARN_ON with a proper check and message in case something goes
really wrong and resumed balance cannot set up its exclusive status.
The check is a user friendly assertion, I don't expect to ever happen
under normal circumstances.
Also document that the paused balance starts here and owns the exclusive
op status.
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
The device replace is paused by unmount or read only remount, and
resumed on next mount or write remount.
The exclusive status should be checked properly as it's a global
invariant and we must not allow 2 operations run. In this case, the
balance can be also paused and resumed under same conditions. It's
always checked first so dev-replace could see the EXCL_OP already taken,
BUT, the ioctl would never let start both at the same time.
Replace the WARN_ON with message and return 0, indicating no error as
this is purely theoretical and the user will be informed. Resolving that
manually should be possible by waiting for the other operation to finish
or cancel the paused state.
Reviewed-by: Anand Jain <[email protected]>
Reviewed-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Make the clearning visible in the callers so we can pair it with the
test_and_set part.
Signed-off-by: David Sterba <[email protected]>
|
|
Move locking and unlocking next to the BTRFS_FS_EXCL_OP bit manipulation
so it's obvious that the two happen at the same time.
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
The function logically belongs there and there's only a single caller,
no need to export it. No code changes.
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
The function will be used outside of volumes.c, the allocation
btrfs_alloc_device is also exported.
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
This is a preparatory cleanup that will make clear that the only
successful way out of btrfs_init_dev_replace_tgtdev will also set the
device_out to a valid pointer. With this guarantee, the callers can be
simplified.
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
The function is called once and is fairly small, we can merge it with
the caller.
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
This function uses fs_info::fs_devices number of time, however we
declare and use it only at the end, instead do it in the beginning of
the function and use it.
Signed-off-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
find_device() declares struct list_head *head pointer and used only once,
instead just use it directly.
Signed-off-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
__btrfs_open_devices() is un-exported drop __ prefix and rename it to
open_fs_devices().
Signed-off-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
__btrfs_close_devices() is un-exported, drop the __ prefix and rename it
to close_fs_devices().
Signed-off-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
__btrfs_open_devices() declares struct list_head *head, however head is
used only once, instead use btrfs_fs_devices::devices directly.
Signed-off-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
btrfs_fs_devices::list is the list of BTRFS fsid in the kernel, a generic
name 'list' makes it's search very difficult, rename it to fs_list.
Signed-off-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
It's provided by the transaction handle.
Signed-off-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
It's provided by the transaction handle.
Signed-off-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
It's provided by the transaction handle.
Signed-off-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
This function is called from only 1 place and is effectively a wrapper
over wait_completion/kfree. It doesn't really bring any value having
those two calls in a separate function. Just open code it and remove it.
No functional changes.
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
It can be directly referenced from the passed address_space so do that.
No functional changes.
Signed-off-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
list_empty_careful usually is a signal of something tricky going on. Its
usage in btrfs is actually not needed since both lists it's used on are
local to a function and cannot be modified concurrently. So switch to
plain list_empty. No functional changes.
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
This function is called only from btrfs_readpage and is already passed
the mapping. Simplify its signature by moving the code obtaining
reference to the extent tree in the function. No functional changes.
Signed-off-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
It's not used in the function so just remove it. No functional changes.
Signed-off-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
This function already gets the page from which the two extent trees
are referenced. Simplify its signature by moving the code getting the
trees inside the function. No functional changes.
Signed-off-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Change the behavior of rmdir(2) and allow it to delete an empty
subvolume by using btrfs_delete_subvolume() which is used by
btrfs_ioctl_snap_destroy().
This is a change in behaviour and has been requested by users. Deleting
the subvolume by ioctl requires root permissions while the rmdir way
does works with standard tools and syscalls for all users that can
access the subvolume.
The main usecase is to allow 'rm -rf /path/with/subvols' to simply work.
We were not able to find any nasty usability surprises, the intention is
to do the destructive rm. Without allowing rmdir, this would have to be
followed by the ioctl subvolume deletion, which is more of an annoyance.
Implementation details:
The required lock for @dir and inode of @dentry is already acquired in
vfs layer.
We need some check before deleting a subvolume. Permission check is done
in vfs layer, emptiness check is in btrfs_rmdir() and additional check
(i.e. neither the subvolume is a default subvolume nor send is in progress)
is in btrfs_delete_subvolume().
Note that in btrfs_ioctl_snap_destroy(), d_delete() is called after
btrfs_delete_subvolume(). For rmdir(2), d_delete() is called in vfs
layer later.
Tested-by: Goffredo Baroncelli <[email protected]>
Signed-off-by: Tomohiro Misono <[email protected]>
Reviewed-by: David Sterba <[email protected]>
[ enhance changelog ]
Signed-off-by: David Sterba <[email protected]>
|
|
Factor out the second half of btrfs_ioctl_snap_destroy() as
btrfs_delete_subvolume(), which performs some subvolume specific checks
before deletion:
1. send is not in progress
2. the subvolume is not the default subvolume
3. the subvolume does not contain other subvolumes
and actual deletion process. btrfs_delete_subvolume() requires
inode_lock for both @dir and inode of @dentry. The remaining part of
btrfs_ioctl_snap_destroy() is mainly permission checks.
Note that call of d_delete() is not included in btrfs_delete_subvolume()
as this function will also be used by btrfs_rmdir() to delete an empty
subvolume and in that case d_delete() is called in VFS layer.
As a result, btrfs_unlink_subvol() and may_destroy_subvol()
become static functions. No functional changes.
Signed-off-by: Tomohiro Misono <[email protected]>
Reviewed-by: David Sterba <[email protected]>
[ minor comment updates ]
Signed-off-by: David Sterba <[email protected]>
|
|
This is a preparation work to refactor btrfs_ioctl_snap_destroy()
and to allow rmdir(2) to delete an empty subvolume.
Signed-off-by: Tomohiro Misono <[email protected]>
Reviewed-by: David Sterba <[email protected]>
[ minor update of the function comment ]
Signed-off-by: David Sterba <[email protected]>
|
|
With commit b18253ec57c0 ("btrfs: optimize free space tree bitmap
conversion"), there are no more callers to le_test_bit(). This patch
removes le_test_bit().
Signed-off-by: Howard McLauchlan <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Presently, convert_free_space_to_extents() does a linear scan of the
bitmap. We can speed this up with find_next_{bit,zero_bit}_le().
This patch replaces the linear scan with find_next_{bit,zero_bit}_le().
Testing shows a 20-33% decrease in execution time for
convert_free_space_to_extents().
Since we change bitmap to be unsigned long, we have to do some casting
for the bitmap cursor. In le_bitmap_set() it makes sense to use u8, as
we are doing bit operations. Everywhere else, we're just using it for
pointer arithmetic and not directly accessing it, so char seems more
appropriate.
Suggested-by: Omar Sandoval <[email protected]>
Signed-off-by: Howard McLauchlan <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
le_bitmap_set() is only used by free-space-tree, so move it there and
make it static. le_bitmap_clear() is not used, so remove it.
Signed-off-by: Howard McLauchlan <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
We really want to know to which filesystem the extent map events belong,
but as it cannot be reached from the extent_map pointers, we need to
pass it down the callchain.
Reviewed-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Preparatory work to pass fs_info to btrfs_add_extent_mapping so we can
get a better tracepoint message. Extent maps do not need fs_info for
anything so we only add a dummy one without any other initialization.
Reviewed-by: Nikolay Borisov <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
The second if is really a subcase of ret being less than 0. So
introduce a generic if (ret < 0) check, and inside have another if
which explicitly handles the -ENOSPC and any other errors. No
functional changes.
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Locks should generally be released in the oppposite order they are
acquired. Generally lock acquisiton ordering is used to ensure
deadlocks don't happen. However, as becomes more complicated it's
best to also maintain proper unlock order so as to avoid possible dead
locks. This was found by code inspection and doesn't necessarily lead
to a deadlock scenario.
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Currently __endio_write_update_ordered uses labels to implement
what is essentially a simple while loop. This makes the code more
cumbersome to follow than it actually has to be. No functional
changes. No xfstest regressions were found during testing.
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Adds comments about BTRFS_FS_EXCL_OP to existing comments
about the device locks.
Signed-off-by: Anand Jain <[email protected]>
Reviewed-by: David Sterba <[email protected]>
[ minor updates ]
Signed-off-by: David Sterba <[email protected]>
|
|
It's used to print its pointer in a debug statement but doesn't really
bring any useful information to the error message.
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
The function btrfs_get_block_group_info() was introduced by the
commit 5af3e8cce8b7 ("Btrfs: make filesystem read-only when submitting
barrier fails") which used it in disk-io.c.
However, the function is only called in ioctl.c now.
Its parameter type btrfs_ioctl_space_info* is only for ioctl.
So, make it static and rename it to be original name
get_block_group_info.
No functional change.
Signed-off-by: Su Yue <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
add_pinned_bytes really cares whether the bytes being pinned are either
data or metadata. To that effect it checks whether the 'owner' argument
is less than BTRFS_FIRST_FREE_OBJECTID (256). This works because
owner can really have 2 types of values:
a) For metadata extents it holds the level at which the parent is in
the btree. This amounts to owner having the values 0-7
b) In case of modifying data extents, owner is the inode number
to which those extents belongs.
Let's make this more explicit byt converting the owner parameter to a
boolean value and either pass it directly when we know the type of
extents we are working with (i.e. in btrfs_free_tree_block). In cases
when the parent function can be called on both metadata/data extents
perform the check in the caller. This hopefully makes the interface
of add_pinned_bytes more intuitive.
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fix from David Sterba:
"A one-liner that prevents leaking an internal error value 1 out of the
ftruncate syscall.
This has been observed in practice. The steps to reproduce make a
common pattern (open/write/fync/ftruncate) but also need the
application to not check only for negative values and happens only for
compressed inlined files.
The conditions are narrow but as this could break userspace I think
it's better to merge it now and not wait for the merge window"
* tag 'for-4.17-rc6-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
Btrfs: fix error handling in btrfs_truncate()
|
|
Jun Wu at Facebook reported that an internal service was seeing a return
value of 1 from ftruncate() on Btrfs in some cases. This is coming from
the NEED_TRUNCATE_BLOCK return value from btrfs_truncate_inode_items().
btrfs_truncate() uses two variables for error handling, ret and err.
When btrfs_truncate_inode_items() returns non-zero, we set err to the
return value. However, NEED_TRUNCATE_BLOCK is not an error. Make sure we
only set err if ret is an error (i.e., negative).
To reproduce the issue: mount a filesystem with -o compress-force=zstd
and the following program will encounter return value of 1 from
ftruncate:
int main(void) {
char buf[256] = { 0 };
int ret;
int fd;
fd = open("test", O_CREAT | O_WRONLY | O_TRUNC, 0666);
if (fd == -1) {
perror("open");
return EXIT_FAILURE;
}
if (write(fd, buf, sizeof(buf)) != sizeof(buf)) {
perror("write");
close(fd);
return EXIT_FAILURE;
}
if (fsync(fd) == -1) {
perror("fsync");
close(fd);
return EXIT_FAILURE;
}
ret = ftruncate(fd, 128);
if (ret) {
printf("ftruncate() returned %d\n", ret);
close(fd);
return EXIT_FAILURE;
}
close(fd);
return EXIT_SUCCESS;
}
Fixes: ddfae63cc8e0 ("btrfs: move btrfs_truncate_block out of trans handle")
CC: [email protected] # 4.15+
Reported-by: Jun Wu <[email protected]>
Signed-off-by: Omar Sandoval <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
Pull vfs fixes from Al Viro:
"Assorted fixes all over the place"
* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
aio: fix io_destroy(2) vs. lookup_ioctx() race
ext2: fix a block leak
nfsd: vfs_mkdir() might succeed leaving dentry negative unhashed
cachefiles: vfs_mkdir() might succeed leaving dentry negative unhashed
unfuck sysfs_mount()
kernfs: deal with kernfs_fill_super() failures
cramfs: Fix IS_ENABLED typo
befs_lookup(): use d_splice_alias()
affs_lookup: switch to d_splice_alias()
affs_lookup(): close a race with affs_remove_link()
fix breakage caused by d_find_alias() semantics change
fs: don't scan the inode cache before SB_BORN is set
do d_instantiate/unlock_new_inode combinations safely
iov_iter: fix memory leak in pipe_get_pages_alloc()
iov_iter: fix return type of __pipe_get_pages()
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux
Pull btrfs fixes from David Sterba:
"We've accumulated some fixes during the last week, some of them were
in the works for a longer time but there are some newer ones too.
Most of the fixes have a reproducer and fix user visible problems,
also candidates for stable kernels. They IMHO qualify for a late rc,
though I did not expect that many"
* tag 'for-4.17-rc5-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/kdave/linux:
btrfs: fix crash when trying to resume balance without the resume flag
btrfs: Fix delalloc inodes invalidation during transaction abort
btrfs: Split btrfs_del_delalloc_inode into 2 functions
btrfs: fix reading stale metadata blocks after degraded raid1 mounts
btrfs: property: Set incompat flag if lzo/zstd compression is set
Btrfs: fix duplicate extents after fsync of file with prealloc extents
Btrfs: fix xattr loss after power failure
Btrfs: send, fix invalid access to commit roots due to concurrent snapshotting
|
|
We set the BTRFS_BALANCE_RESUME flag in the btrfs_recover_balance()
only, which isn't called during the remount. So when resuming from
the paused balance we hit the bug:
kernel: kernel BUG at fs/btrfs/volumes.c:3890!
::
kernel: balance_kthread+0x51/0x60 [btrfs]
kernel: kthread+0x111/0x130
::
kernel: RIP: btrfs_balance+0x12e1/0x1570 [btrfs] RSP: ffffba7d0090bde8
Reproducer:
On a mounted filesystem:
btrfs balance start --full-balance /btrfs
btrfs balance pause /btrfs
mount -o remount,ro /dev/sdb /btrfs
mount -o remount,rw /dev/sdb /btrfs
To fix this set the BTRFS_BALANCE_RESUME flag in
btrfs_resume_balance_async().
CC: [email protected] # 4.4+
Signed-off-by: Anand Jain <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
When a transaction is aborted btrfs_cleanup_transaction is called to
cleanup all the various in-flight bits and pieces which migth be
active. One of those is delalloc inodes - inodes which have dirty
pages which haven't been persisted yet. Currently the process of
freeing such delalloc inodes in exceptional circumstances such as
transaction abort boiled down to calling btrfs_invalidate_inodes whose
sole job is to invalidate the dentries for all inodes related to a
root. This is in fact wrong and insufficient since such delalloc inodes
will likely have pending pages or ordered-extents and will be linked to
the sb->s_inode_list. This means that unmounting a btrfs instance with
an aborted transaction could potentially lead inodes/their pages
visible to the system long after their superblock has been freed. This
in turn leads to a "use-after-free" situation once page shrink is
triggered. This situation could be simulated by running generic/019
which would cause such inodes to be left hanging, followed by
generic/176 which causes memory pressure and page eviction which lead
to touching the freed super block instance. This situation is
additionally detected by the unmount code of VFS with the following
message:
"VFS: Busy inodes after unmount of Self-destruct in 5 seconds. Have a nice day..."
Additionally btrfs hits WARN_ON(!RB_EMPTY_ROOT(&root->inode_tree));
in free_fs_root for the same reason.
This patch aims to rectify the sitaution by doing the following:
1. Change btrfs_destroy_delalloc_inodes so that it calls
invalidate_inode_pages2 for every inode on the delalloc list, this
ensures that all the pages of the inode are released. This function
boils down to calling btrfs_releasepage. During test I observed cases
where inodes on the delalloc list were having an i_count of 0, so this
necessitates using igrab to be sure we are working on a non-freed inode.
2. Since calling btrfs_releasepage might queue delayed iputs move the
call out to btrfs_cleanup_transaction in btrfs_error_commit_super before
calling run_delayed_iputs for the last time. This is necessary to ensure
that delayed iputs are run.
Note: this patch is tagged for 4.14 stable but the fix applies to older
versions too but needs to be backported manually due to conflicts.
CC: [email protected] # 4.14.x: 2b8773313494: btrfs: Split btrfs_del_delalloc_inode into 2 functions
CC: [email protected] # 4.14.x
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
[ add comment to igrab ]
Signed-off-by: David Sterba <[email protected]>
|
|
This is in preparation of fixing delalloc inodes leakage on transaction
abort. Also export the new function.
Signed-off-by: Nikolay Borisov <[email protected]>
Reviewed-by: David Sterba <[email protected]>
Reviewed-by: Anand Jain <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|
|
If a btree block, aka. extent buffer, is not available in the extent
buffer cache, it'll be read out from the disk instead, i.e.
btrfs_search_slot()
read_block_for_search() # hold parent and its lock, go to read child
btrfs_release_path()
read_tree_block() # read child
Unfortunately, the parent lock got released before reading child, so
commit 5bdd3536cbbe ("Btrfs: Fix block generation verification race") had
used 0 as parent transid to read the child block. It forces
read_tree_block() not to check if parent transid is different with the
generation id of the child that it reads out from disk.
A simple PoC is included in btrfs/124,
0. A two-disk raid1 btrfs,
1. Right after mkfs.btrfs, block A is allocated to be device tree's root.
2. Mount this filesystem and put it in use, after a while, device tree's
root got COW but block A hasn't been allocated/overwritten yet.
3. Umount it and reload the btrfs module to remove both disks from the
global @fs_devices list.
4. mount -odegraded dev1 and write some data, so now block A is allocated
to be a leaf in checksum tree. Note that only dev1 has the latest
metadata of this filesystem.
5. Umount it and mount it again normally (with both disks), since raid1
can pick up one disk by the writer task's pid, if btrfs_search_slot()
needs to read block A, dev2 which does NOT have the latest metadata
might be read for block A, then we got a stale block A.
6. As parent transid is not checked, block A is marked as uptodate and
put into the extent buffer cache, so the future search won't bother
to read disk again, which means it'll make changes on this stale
one and make it dirty and flush it onto disk.
To avoid the problem, parent transid needs to be passed to
read_tree_block().
In order to get a valid parent transid, we need to hold the parent's
lock until finishing reading child.
This patch needs to be slightly adapted for stable kernels, the
&first_key parameter added to read_tree_block() is from 4.16+
(581c1760415c4). The fix is to replace 0 by 'gen'.
Fixes: 5bdd3536cbbe ("Btrfs: Fix block generation verification race")
CC: [email protected] # 4.4+
Signed-off-by: Liu Bo <[email protected]>
Reviewed-by: Filipe Manana <[email protected]>
Reviewed-by: Qu Wenruo <[email protected]>
[ update changelog ]
Signed-off-by: David Sterba <[email protected]>
|
|
Incompat flag of LZO/ZSTD compression should be set at:
1. mount time (-o compress/compress-force)
2. when defrag is done
3. when property is set
Currently 3. is missing and this commit adds this.
This could lead to a filesystem that uses ZSTD but is not marked as
such. If a kernel without a ZSTD support encounteres a ZSTD compressed
extent, it will handle that but this could be confusing to the user.
Typically the filesystem is mounted with the ZSTD option, but the
discrepancy can arise when a filesystem is never mounted with ZSTD and
then the property on some file is set (and some new extents are
written). A simple mount with -o compress=zstd will fix that up on an
unpatched kernel.
Same goes for LZO, but this has been around for a very long time
(2.6.37) so it's unlikely that a pre-LZO kernel would be used.
Fixes: 5c1aab1dd544 ("btrfs: Add zstd support")
CC: [email protected] # 4.14+
Signed-off-by: Tomohiro Misono <[email protected]>
Reviewed-by: Anand Jain <[email protected]>
Reviewed-by: David Sterba <[email protected]>
[ add user visible impact ]
Signed-off-by: David Sterba <[email protected]>
|
|
In commit 471d557afed1 ("Btrfs: fix loss of prealloc extents past i_size
after fsync log replay"), on fsync, we started to always log all prealloc
extents beyond an inode's i_size in order to avoid losing them after a
power failure. However under some cases this can lead to the log replay
code to create duplicate extent items, with different lengths, in the
extent tree. That happens because, as of that commit, we can now log
extent items based on extent maps that are not on the "modified" list
of extent maps of the inode's extent map tree. Logging extent items based
on extent maps is used during the fast fsync path to save time and for
this to work reliably it requires that the extent maps are not merged
with other adjacent extent maps - having the extent maps in the list
of modified extents gives such guarantee.
Consider the following example, captured during a long run of fsstress,
which illustrates this problem.
We have inode 271, in the filesystem tree (root 5), for which all of the
following operations and discussion apply to.
A buffered write starts at offset 312391 with a length of 933471 bytes
(end offset at 1245862). At this point we have, for this inode, the
following extent maps with the their field values:
em A, start 0, orig_start 0, len 40960, block_start 18446744073709551613,
block_len 0, orig_block_len 0
em B, start 40960, orig_start 40960, len 376832, block_start 1106399232,
block_len 376832, orig_block_len 376832
em C, start 417792, orig_start 417792, len 782336, block_start
18446744073709551613, block_len 0, orig_block_len 0
em D, start 1200128, orig_start 1200128, len 835584, block_start
1106776064, block_len 835584, orig_block_len 835584
em E, start 2035712, orig_start 2035712, len 245760, block_start
1107611648, block_len 245760, orig_block_len 245760
Extent map A corresponds to a hole and extent maps D and E correspond to
preallocated extents.
Extent map D ends where extent map E begins (1106776064 + 835584 =
1107611648), but these extent maps were not merged because they are in
the inode's list of modified extent maps.
An fsync against this inode is made, which triggers the fast path
(BTRFS_INODE_NEEDS_FULL_SYNC is not set). This fsync triggers writeback
of the data previously written using buffered IO, and when the respective
ordered extent finishes, btrfs_drop_extents() is called against the
(aligned) range 311296..1249279. This causes a split of extent map D at
btrfs_drop_extent_cache(), replacing extent map D with a new extent map
D', also added to the list of modified extents, with the following
values:
em D', start 1249280, orig_start of 1200128,
block_start 1106825216 (= 1106776064 + 1249280 - 1200128),
orig_block_len 835584,
block_len 786432 (835584 - (1249280 - 1200128))
Then, during the fast fsync, btrfs_log_changed_extents() is called and
extent maps D' and E are removed from the list of modified extents. The
flag EXTENT_FLAG_LOGGING is also set on them. After the extents are logged
clear_em_logging() is called on each of them, and that makes extent map E
to be merged with extent map D' (try_merge_map()), resulting in D' being
deleted and E adjusted to:
em E, start 1249280, orig_start 1200128, len 1032192,
block_start 1106825216, block_len 1032192,
orig_block_len 245760
A direct IO write at offset 1847296 and length of 360448 bytes (end offset
at 2207744) starts, and at that moment the following extent maps exist for
our inode:
em A, start 0, orig_start 0, len 40960, block_start 18446744073709551613,
block_len 0, orig_block_len 0
em B, start 40960, orig_start 40960, len 270336, block_start 1106399232,
block_len 270336, orig_block_len 376832
em C, start 311296, orig_start 311296, len 937984, block_start 1112842240,
block_len 937984, orig_block_len 937984
em E (prealloc), start 1249280, orig_start 1200128, len 1032192,
block_start 1106825216, block_len 1032192, orig_block_len 245760
The dio write results in drop_extent_cache() being called twice. The first
time for a range that starts at offset 1847296 and ends at offset 2035711
(length of 188416), which results in a double split of extent map E,
replacing it with two new extent maps:
em F, start 1249280, orig_start 1200128, block_start 1106825216,
block_len 598016, orig_block_len 598016
em G, start 2035712, orig_start 1200128, block_start 1107611648,
block_len 245760, orig_block_len 1032192
It also creates a new extent map that represents a part of the requested
IO (through create_io_em()):
em H, start 1847296, len 188416, block_start 1107423232, block_len 188416
The second call to drop_extent_cache() has a range with a start offset of
2035712 and end offset of 2207743 (length of 172032). This leads to
replacing extent map G with a new extent map I with the following values:
em I, start 2207744, orig_start 1200128, block_start 1107783680,
block_len 73728, orig_block_len 1032192
It also creates a new extent map that represents the second part of the
requested IO (through create_io_em()):
em J, start 2035712, len 172032, block_start 1107611648, block_len 172032
The dio write set the inode's i_size to 2207744 bytes.
After the dio write the inode has the following extent maps:
em A, start 0, orig_start 0, len 40960, block_start 18446744073709551613,
block_len 0, orig_block_len 0
em B, start 40960, orig_start 40960, len 270336, block_start 1106399232,
block_len 270336, orig_block_len 376832
em C, start 311296, orig_start 311296, len 937984, block_start 1112842240,
block_len 937984, orig_block_len 937984
em F, start 1249280, orig_start 1200128, len 598016,
block_start 1106825216, block_len 598016, orig_block_len 598016
em H, start 1847296, orig_start 1200128, len 188416,
block_start 1107423232, block_len 188416, orig_block_len 835584
em J, start 2035712, orig_start 2035712, len 172032,
block_start 1107611648, block_len 172032, orig_block_len 245760
em I, start 2207744, orig_start 1200128, len 73728,
block_start 1107783680, block_len 73728, orig_block_len 1032192
Now do some change to the file, like adding a xattr for example and then
fsync it again. This triggers a fast fsync path, and as of commit
471d557afed1 ("Btrfs: fix loss of prealloc extents past i_size after fsync
log replay"), we use the extent map I to log a file extent item because
it's a prealloc extent and it starts at an offset matching the inode's
i_size. However when we log it, we create a file extent item with a value
for the disk byte location that is wrong, as can be seen from the
following output of "btrfs inspect-internal dump-tree":
item 1 key (271 EXTENT_DATA 2207744) itemoff 3782 itemsize 53
generation 22 type 2 (prealloc)
prealloc data disk byte 1106776064 nr 1032192
prealloc data offset 1007616 nr 73728
Here the disk byte value corresponds to calculation based on some fields
from the extent map I:
1106776064 = block_start (1107783680) - 1007616 (extent_offset)
extent_offset = 2207744 (start) - 1200128 (orig_start) = 1007616
The disk byte value of 1106776064 clashes with disk byte values of the
file extent items at offsets 1249280 and 1847296 in the fs tree:
item 6 key (271 EXTENT_DATA 1249280) itemoff 3568 itemsize 53
generation 20 type 2 (prealloc)
prealloc data disk byte 1106776064 nr 835584
prealloc data offset 49152 nr 598016
item 7 key (271 EXTENT_DATA 1847296) itemoff 3515 itemsize 53
generation 20 type 1 (regular)
extent data disk byte 1106776064 nr 835584
extent data offset 647168 nr 188416 ram 835584
extent compression 0 (none)
item 8 key (271 EXTENT_DATA 2035712) itemoff 3462 itemsize 53
generation 20 type 1 (regular)
extent data disk byte 1107611648 nr 245760
extent data offset 0 nr 172032 ram 245760
extent compression 0 (none)
item 9 key (271 EXTENT_DATA 2207744) itemoff 3409 itemsize 53
generation 20 type 2 (prealloc)
prealloc data disk byte 1107611648 nr 245760
prealloc data offset 172032 nr 73728
Instead of the disk byte value of 1106776064, the value of 1107611648
should have been logged. Also the data offset value should have been
172032 and not 1007616.
After a log replay we end up getting two extent items in the extent tree
with different lengths, one of 835584, which is correct and existed
before the log replay, and another one of 1032192 which is wrong and is
based on the logged file extent item:
item 12 key (1106776064 EXTENT_ITEM 835584) itemoff 3406 itemsize 53
refs 2 gen 15 flags DATA
extent data backref root 5 objectid 271 offset 1200128 count 2
item 13 key (1106776064 EXTENT_ITEM 1032192) itemoff 3353 itemsize 53
refs 1 gen 22 flags DATA
extent data backref root 5 objectid 271 offset 1200128 count 1
Obviously this leads to many problems and a filesystem check reports many
errors:
(...)
checking extents
Extent back ref already exists for 1106776064 parent 0 root 5 owner 271 offset 1200128 num_refs 1
extent item 1106776064 has multiple extent items
ref mismatch on [1106776064 835584] extent item 2, found 3
Incorrect local backref count on 1106776064 root 5 owner 271 offset 1200128 found 2 wanted 1 back 0x55b1d0ad7680
Backref 1106776064 root 5 owner 271 offset 1200128 num_refs 0 not found in extent tree
Incorrect local backref count on 1106776064 root 5 owner 271 offset 1200128 found 1 wanted 0 back 0x55b1d0ad4e70
Backref bytes do not match extent backref, bytenr=1106776064, ref bytes=835584, backref bytes=1032192
backpointer mismatch on [1106776064 835584]
checking free space cache
block group 1103101952 has wrong amount of free space
failed to load free space cache for block group 1103101952
checking fs roots
(...)
So fix this by logging the prealloc extents beyond the inode's i_size
based on searches in the subvolume tree instead of the extent maps.
Fixes: 471d557afed1 ("Btrfs: fix loss of prealloc extents past i_size after fsync log replay")
CC: [email protected] # 4.14+
Signed-off-by: Filipe Manana <[email protected]>
Signed-off-by: David Sterba <[email protected]>
|