When a metadata read is served the endio routine btree_readpage_end_io_hook
is called which eventually runs the tree-checker. If tree-checker fails
to validate the read eb then it sets EXTENT_BUFFER_CORRUPT flag. This
leads to btree_read_extent_buffer_pages wrongly assuming that all
available copies of this extent buffer are wrong and failing prematurely.
Fix this modify btree_read_extent_buffer_pages to read all copies of
the data.
This failure was exhibitted in xfstests btrfs/124 which would
spuriously fail its balance operations. The reason was that when balance
was run following re-introduction of the missing raid1 disk
__btrfs_map_block would map the read request to stripe 0, which
corresponded to devid 2 (the disk which is being removed in the test):
item 2 key (FIRST_CHUNK_TREE CHUNK_ITEM 3553624064) itemoff 15975 itemsize 112
length 1073741824 owner 2 stripe_len 65536 type DATA|RAID1
io_align 65536 io_width 65536 sector_size 4096
num_stripes 2 sub_stripes 1
stripe 0 devid 2 offset 2156920832
dev_uuid 8466c350-ed0c-4c3b-b17d-6379b445d5c8
stripe 1 devid 1 offset 3553624064
dev_uuid 1265d8db-5596-477e-af03-df08eb38d2ca
This caused read requests for a checksum item that to be routed to the
stale disk which triggered the aforementioned logic involving
EXTENT_BUFFER_CORRUPT flag. This then triggered cascading failures of
the balance operation.
Fixes: a826d6dcb3 ("Btrfs: check items for correctness as we search")
CC: stable@vger.kernel.org # 4.4+
Suggested-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There's a race between close_ctree() and cleaner_kthread().
close_ctree() sets btrfs_fs_closing(), and the cleaner stops when it
sees it set, but this is racy; the cleaner might have already checked
the bit and could be cleaning stuff. In particular, if it deletes unused
block groups, it will create delayed iputs for the free space cache
inodes. As of "btrfs: don't run delayed_iputs in commit", we're no
longer running delayed iputs after a commit. Therefore, if the cleaner
creates more delayed iputs after delayed iputs are run in
btrfs_commit_super(), we will leak inodes on unmount and get a busy
inode crash from the VFS.
Fix it by parking the cleaner before we actually close anything. Then,
any remaining delayed iputs will always be handled in
btrfs_commit_super(). This also ensures that the commit in close_ctree()
is really the last commit, so we can get rid of the commit in
cleaner_kthread().
The fstest/generic/475 followed by 476 can trigger a crash that
manifests as a slab corruption caused by accessing the freed kthread
structure by a wake up function. Sample trace:
[ 5657.077612] BUG: unable to handle kernel NULL pointer dereference at 00000000000000cc
[ 5657.079432] PGD 1c57a067 P4D 1c57a067 PUD da10067 PMD 0
[ 5657.080661] Oops: 0000 [#1] PREEMPT SMP
[ 5657.081592] CPU: 1 PID: 5157 Comm: fsstress Tainted: G W 4.19.0-rc8-default+ #323
[ 5657.083703] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.11.2-0-gf9626cc-prebuilt.qemu-project.org 04/01/2014
[ 5657.086577] RIP: 0010:shrink_page_list+0x2f9/0xe90
[ 5657.091937] RSP: 0018:ffffb5c745c8f728 EFLAGS: 00010287
[ 5657.092953] RAX: 0000000000000074 RBX: ffffb5c745c8f830 RCX: 0000000000000000
[ 5657.094590] RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff9a8747fdf3d0
[ 5657.095987] RBP: ffffb5c745c8f9e0 R08: 0000000000000000 R09: 0000000000000000
[ 5657.097159] R10: ffff9a8747fdf5e8 R11: 0000000000000000 R12: ffffb5c745c8f788
[ 5657.098513] R13: ffff9a877f6ff2c0 R14: ffff9a877f6ff2c8 R15: dead000000000200
[ 5657.099689] FS: 00007f948d853b80(0000) GS:ffff9a877d600000(0000) knlGS:0000000000000000
[ 5657.101032] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 5657.101953] CR2: 00000000000000cc CR3: 00000000684bd000 CR4: 00000000000006e0
[ 5657.103159] Call Trace:
[ 5657.103776] shrink_inactive_list+0x194/0x410
[ 5657.104671] shrink_node_memcg.constprop.84+0x39a/0x6a0
[ 5657.105750] shrink_node+0x62/0x1c0
[ 5657.106529] try_to_free_pages+0x1a4/0x500
[ 5657.107408] __alloc_pages_slowpath+0x2c9/0xb20
[ 5657.108418] __alloc_pages_nodemask+0x268/0x2b0
[ 5657.109348] kmalloc_large_node+0x37/0x90
[ 5657.110205] __kmalloc_node+0x236/0x310
[ 5657.111014] kvmalloc_node+0x3e/0x70
Fixes: 30928e9baa ("btrfs: don't run delayed_iputs in commit")
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ add trace ]
Signed-off-by: David Sterba <dsterba@suse.com>
I ran into an issue where there was some reference being held on an
inode that I couldn't track. This assert wasn't triggered, but it at
least rules out we're doing something stupid.
Reviewed-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The replace_wait and bio_counter were mistakenly added to fs_info in
commit c404e0dc2c ("Btrfs: fix use-after-free in the finishing
procedure of the device replace"), but they logically belong to
fs_info::dev_replace. Besides, bio_counter is a very generic name and is
confusing in bare fs_info context.
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This member seems to be copied from the extent_buffer locking scheme and
is at least used to assert that the read lock/unlock is properly nested.
In some way. While the _inc/_dec are called inside the read lock
section, the asserts are both inside and outside, so the ordering is not
guaranteed and we can see read/inc/dec ordered in any way
(theoretically).
A missing call of btrfs_dev_replace_clear_lock_blocking could cause
unexpected read_locks count, so this at least looks like a valid
assertion, but this will become unnecessary with later updates.
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
rb_first_cached() trades an extra pointer "leftmost" for doing the same
job as rb_first() but in O(1).
Functions manipulating href->ref_tree need to get the first entry, this
converts href->ref_tree to use rb_first_cached().
For more details about the optimization see patch "Btrfs: delayed-refs:
use rb_first_cached for href_root".
Tested-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
rb_first_cached() trades an extra pointer "leftmost" for doing the same
job as rb_first() but in O(1).
Functions manipulating href_root need to get the first entry, this
converts href_root to use rb_first_cached().
This patch is first in the sequenct of similar updates to other rbtrees
and this is analysis of the expected behaviour and improvements.
There's a common pattern:
while (node = rb_first) {
entry = rb_entry(node)
next = rb_next(node)
rb_erase(node)
cleanup(entry)
}
rb_first needs to traverse the tree up to logN depth, rb_erase can
completely reshuffle the tree. With the caching we'll skip the traversal
in rb_first. That's a cached memory access vs looped pointer
dereference trade-off that IMHO has a clear winner.
Measurements show there's not much difference in a sample tree with
10000 nodes: 4.5s / rb_first and 4.8s / rb_first_cached. Real effects of
caching and pointer chasing are unpredictable though.
Further optimzations can be done to avoid the expensive rb_erase step.
In some cases it's ok to process the nodes in any order, so the tree can
be traversed in post-order, not rebalancing the children nodes and just
calling free. Care must be taken regarding the next node.
Tested-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ update changelog from mail discussions ]
Signed-off-by: David Sterba <dsterba@suse.com>
There are two members in struct btrfs_root which indicate root's
objectid: objectid and root_key.objectid.
They are both set to the same value in __setup_root():
static void __setup_root(struct btrfs_root *root,
struct btrfs_fs_info *fs_info,
u64 objectid)
{
...
root->objectid = objectid;
...
root->root_key.objectid = objecitd;
...
}
and not changed to other value after initialization.
grep in btrfs directory shows both are used in many places:
$ grep -rI "root->root_key.objectid" | wc -l
133
$ grep -rI "root->objectid" | wc -l
55
(4.17, inc. some noise)
It is confusing to have two similar variable names and it seems
that there is no rule about which should be used in a certain case.
Since ->root_key itself is needed for tree reloc tree, let's remove
'objecitd' member and unify code to use ->root_key.objectid in all places.
Signed-off-by: Misono Tomohiro <misono.tomohiro@jp.fujitsu.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Commit e9894fd3e3 ("Btrfs: fix snapshot vs nocow writting") forced
nocow writes to fallback to COW, during writeback, when a snapshot is
created. This resulted in writes made before creating the snapshot to
unexpectedly fail with ENOSPC during writeback when success (0) was
returned to user space through the write system call.
The steps leading to this problem are:
1. When it's not possible to allocate data space for a write, the
buffered write path checks if a NOCOW write is possible. If it is,
it will not reserve space and success (0) is returned to user space.
2. Then when a snapshot is created, the root's will_be_snapshotted
atomic is incremented and writeback is triggered for all inode's that
belong to the root being snapshotted. Incrementing that atomic forces
all previous writes to fallback to COW during writeback (running
delalloc).
3. This results in the writeback for the inodes to fail and therefore
setting the ENOSPC error in their mappings, so that a subsequent
fsync on them will report the error to user space. So it's not a
completely silent data loss (since fsync will report ENOSPC) but it's
a very unexpected and undesirable behaviour, because if a clean
shutdown/unmount of the filesystem happens without previous calls to
fsync, it is expected to have the data present in the files after
mounting the filesystem again.
So fix this by adding a new atomic named snapshot_force_cow to the
root structure which prevents this behaviour and works the following way:
1. It is incremented when we start to create a snapshot after triggering
writeback and before waiting for writeback to finish.
2. This new atomic is now what is used by writeback (running delalloc)
to decide whether we need to fallback to COW or not. Because we
incremented this new atomic after triggering writeback in the
snapshot creation ioctl, we ensure that all buffered writes that
happened before snapshot creation will succeed and not fallback to
COW (which would make them fail with ENOSPC).
3. The existing atomic, will_be_snapshotted, is kept because it is used
to force new buffered writes, that start after we started
snapshotting, to reserve data space even when NOCOW is possible.
This makes these writes fail early with ENOSPC when there's no
available space to allocate, preventing the unexpected behaviour of
writeback later failing with ENOSPC due to a fallback to COW mode.
Fixes: e9894fd3e3 ("Btrfs: fix snapshot vs nocow writting")
Signed-off-by: Robbie Ko <robbieko@synology.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
This patch will introduce chunk <-> dev extent mapping check, to protect
us against invalid dev extents or chunks.
Since chunk mapping is the fundamental infrastructure of btrfs, extra
check at mount time could prevent a lot of unexpected behavior (BUG_ON).
Reported-by: Xu Wen <wen.xu@gatech.edu>
Link: https://bugzilla.kernel.org/show_bug.cgi?id=200403
Link: https://bugzilla.kernel.org/show_bug.cgi?id=200407
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Su Yue <suy.fnst@cn.fujitsu.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The exported helper just calls the static one. There's no obvious reason
to have them separate eg. for performance reasons where the static one
could be better optimized in the same unit. There's a slight decrease in
code size and stack consumption.
Signed-off-by: David Sterba <dsterba@suse.com>
Lock owner and nesting level have been unused since day 1, probably
copy&pasted from the extent_buffer locking scheme without much thinking.
The locking of device replace is simpler and does not need any lock
nesting.
Signed-off-by: David Sterba <dsterba@suse.com>
Added in 58176a9604 ("Btrfs: Add per-root block accounting and sysfs
entries") in 2007, the roots had names exported in sysfs. The code
was commented out in 4df27c4d5c ("Btrfs: change how subvolumes
are organized") and cleaned by 182608c829 ("btrfs: remove old
unused commented out code").
Signed-off-by: David Sterba <dsterba@suse.com>
The data and metadata callback implementation both use the same
function. We can remove the call indirection and intermediate helper
completely.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The data and metadata callback implementation both use the same
function. We can remove the call indirection completely.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
All implementations of the callback are trivial and do the same and
there's only one user. Merge everything together.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The end_io callbacks passed to btrfs_wq_submit_bio
(btrfs_submit_bio_done and btree_submit_bio_done) are effectively the
same code, there's no point to do the indirection. Export
btrfs_submit_bio_done and call it directly.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
After splitting the start and end hooks in a758781d4b ("btrfs:
separate types for submit_bio_start and submit_bio_done"), some of
the function arguments were dropped but not removed from the structure.
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Introduced by c6100a4b4e ("Btrfs: replace tree->mapping with
tree->private_data") to be used in run_one_async_done where it got
unused after 736cd52e0c ("Btrfs: remove nr_async_submits and
async_submit_draining").
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
EXTENT_BUFFER_DUMMY is an awful name for this flag. Buffers which have
this flag set are not in any way dummy. Rather, they are private in the
sense that are not mapped and linked to the global buffer tree. This
flag has subtle implications to the way free_extent_buffer works for
example, as well as controls whether page->mapping->private_lock is held
during extent_buffer release. Pages for an unmapped buffer cannot be
under io, nor can they be written by a 3rd party so taking the lock is
unnecessary.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ EXTENT_BUFFER_UNMAPPED, update changelog ]
Signed-off-by: David Sterba <dsterba@suse.com>
We use customized, nodesize batch value to update dirty_metadata_bytes.
We should also use batch version of compare function or we will easily
goto fast path and get false result from percpu_counter_compare().
Fixes: e2d845211e ("Btrfs: use percpu counter for dirty metadata count")
CC: stable@vger.kernel.org # 4.4+
Signed-off-by: Ethan Lien <ethanlien@synology.com>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Functions that get btrfs inode can simply reach the fs_info by
dereferencing the root and this looks a bit more straightforward
compared to the btrfs_sb(...) indirection.
If the transaction handle is available and not NULL it's used instead.
Signed-off-by: David Sterba <dsterba@suse.com>
For easier debugging, print eb->start if level is invalid. Also make
clear if bytenr found is not expected.
Signed-off-by: Su Yue <suy.fnst@cn.fujitsu.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The transaction times were changed to ktime_get_real_seconds to avoid
the y2038 overflow, but they still have a minor problem when they go
backwards or jump due to settimeofday() or leap seconds.
This changes the transaction handling to instead use ktime_get_seconds(),
which returns a CLOCK_MONOTONIC timestamp that has neither of those
problems.
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When a new extent buffer is allocated there are a few mandatory fields
which need to be set in order for the buffer to be sane: level,
generation, bytenr, backref_rev, owner and FSID/UUID. Currently this
is open coded in the callers of btrfs_alloc_tree_block, meaning it's
fairly high in the abstraction hierarchy of operations. This patch
solves this by simply moving this init code in btrfs_init_new_buffer,
since this is the function which initializes a newly allocated
extent buffer. No functional changes.
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The get_seconds() function is deprecated as it truncates the timestamp
to 32 bits. Change it to or ktime_get_real_seconds().
Signed-off-by: Allen Pais <allen.lkml@gmail.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ update changelog ]
Signed-off-by: David Sterba <dsterba@suse.com>
As verify_level_key() is checked after verify_parent_transid(), i.e.
if (verify_parent_transid())
ret = -EIO;
else if (verify_level_key())
ret = -EUCLEAN;
if parent_transid is 0, verify_parent_transid() skips verifying
parent_transid and considers eb as valid, and if verify_level_key()
reports something wrong, we're not going to know if it's caused by
corrupted metadata or non-checkecd eb (e.g. stale eb).
The stale eb can be from an outdated raid1 mirror after a degraded
mount, see eg "btrfs: fix reading stale metadata blocks after degraded
raid1 mounts" (02a3307aa9) for more details.
@parent_transid is able to tell whether the eb's generation has been
verified by the caller.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Now that we don't keep long-standing reservations for orphan items,
root->orphan_block_rsv isn't used. We can git rid of it, along with:
- root->orphan_lock, which was used to protect root->orphan_block_rsv
- root->orphan_inodes, which was used as a refcount for root->orphan_block_rsv
- BTRFS_INODE_ORPHAN_META_RESERVED, which was used to track reservations
in root->orphan_block_rsv
- btrfs_orphan_commit_root(), which was the last user of any of these
and does nothing else
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Omar Sandoval <osandov@fb.com>
Signed-off-by: David Sterba <dsterba@suse.com>
There are already 2 reports about strangely corrupted super blocks,
where csum still matches but extra garbage gets slipped into super block.
The corruption would looks like:
------
superblock: bytenr=65536, device=/dev/sdc1
---------------------------------------------------------
csum_type 41700 (INVALID)
csum 0x3b252d3a [match]
bytenr 65536
flags 0x1
( WRITTEN )
magic _BHRfS_M [match]
...
incompat_flags 0x5b22400000000169
( MIXED_BACKREF |
COMPRESS_LZO |
BIG_METADATA |
EXTENDED_IREF |
SKINNY_METADATA |
unknown flag: 0x5b22400000000000 )
...
------
Or
------
superblock: bytenr=65536, device=/dev/mapper/x
---------------------------------------------------------
csum_type 35355 (INVALID)
csum_size 32
csum 0xf0dbeddd [match]
bytenr 65536
flags 0x1
( WRITTEN )
magic _BHRfS_M [match]
...
incompat_flags 0x176d200000000169
( MIXED_BACKREF |
COMPRESS_LZO |
BIG_METADATA |
EXTENDED_IREF |
SKINNY_METADATA |
unknown flag: 0x176d200000000000 )
------
Obviously, csum_type and incompat_flags get some garbage, but its csum
still matches, which means kernel calculates the csum based on corrupted
super block memory.
And after manually fixing these values, the filesystem is completely
healthy without any problem exposed by btrfs check.
Although the cause is still unknown, at least detect it and prevent further
corruption.
Both reports have same symptoms, there's an overwrite on offset 192 of
the superblock, by 4 bytes. The superblock structure is not allocated or
freed and stays in the memory for the whole filesystem lifetime, so it's
not a use-after-free kind of error on someone else's leaked page.
As a vague point for the problable cause is mentioning of other system
freezing related to graphic card drivers.
Reported-by: Ken Swenson <flat@imo.uto.moe>
Reported-by: Ben Parsons <9parsonsb@gmail.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ add brief analysis of the reports ]
Signed-off-by: David Sterba <dsterba@suse.com>
Refactor btrfs_check_super_valid:
1) Rename it to btrfs_validate_mount_super()
Now it's more obvious when the function should be called.
2) Extract core check routine into validate_super()
Later write time check can reuse it, and if needed, we could also
use validate_super() to check each super block.
3) Add more comments about btrfs_validate_mount_super()
Mostly about what it doesn't check and when it should be called.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ rename to validate_super ]
Signed-off-by: David Sterba <dsterba@suse.com>
Move btrfs_check_super_valid() before its single caller to avoid forward
declaration.
Though such code motion is not recommended as it pollutes git history,
in this case the following patches would need to add new forward
declarations for static functions that we want to avoid.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Add a new member struct btrfs_raid_attr::bg_flag so that
btrfs_raid_array can maintain the bit map flag of the raid type, and
so we can drop btrfs_raid_group.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Currently fs_info::balance_running is 0 or 1 and does not use the
semantics of atomics. The pause and cancel check for 0, that can happen
only after __btrfs_balance exits for whatever reason.
Parallel calls to balance ioctl may enter btrfs_ioctl_balance multiple
times but will block on the balance_mutex that protects the
fs_info::flags bit.
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Mutual exclusion of device add/rm and balance was done by the volume
mutex up to version 3.7. The commit 5ac00addc7 ("Btrfs: disallow
mutually exclusive admin operations from user mode") added a bit that
essentially tracked the same information.
The status bit has an advantage over a mutex that it can be set without
restrictions of function context, so it started to be used in the
mount-time resuming of balance or device replace.
But we don't really need to track the same information in two ways.
1) After the previous cleanups, the main ioctl handlers for
add/del/resize copy the EXCL_OP bit next to the volume mutex, here
it's clearly safe.
2) Resuming balance during mount or after rw remount will set only the
EXCL_OP bit and the volume_mutex is held in the kernel thread that
calls btrfs_balance.
3) Resuming device replace during mount or after rw remount is done
after balance and is excluded by the EXCL_OP bit. It does not take
the volume_mutex at all and completely relies on the EXCL_OP bit.
4) The resuming of balance and dev-replace cannot hapen at the same time
as the ioctls cannot be started in parallel. Nevertheless, a crafted
image could trigger that and a warning is printed.
5) Balance is normally excluded by EXCL_OP and also uses own mutex to
protect against concurrent access to its status data. There's some
trickery to maintain the right lock nesting in case we need to
reexamine the status in btrfs_ioctl_balance. The volume_mutex is
removed and the unlock/lock sequence is left in place as we might
expect other waiters to proceed.
6) Similar to 5, the unlock/lock sequence is kept in
btrfs_cancel_balance to allow waiters to continue.
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: David Sterba <dsterba@suse.com>
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: stable@vger.kernel.org # 4.14.x: 2b8773313494: btrfs: Split btrfs_del_delalloc_inode into 2 functions
CC: stable@vger.kernel.org # 4.14.x
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
[ add comment to igrab ]
Signed-off-by: David Sterba <dsterba@suse.com>
Unlike previous method that tries to commit transaction inside
qgroup_reserve(), this time we will try to commit transaction using
fs_info->transaction_kthread to avoid nested transaction and no need to
worry about locking context.
Since it's an asynchronous function call and we won't wait for
transaction commit, unlike previous method, we must call it before we
hit the qgroup limit.
So this patch will use the ratio and size of qgroup meta_pertrans
reservation as indicator to check if we should trigger a transaction
commit. (meta_prealloc won't be cleaned in transaction committ, it's
useless anyway)
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When looping btrfs/074 with many cpus (>= 8), it's possible to trigger
kernel warning due to first key verification:
[ 4239.523446] WARNING: CPU: 5 PID: 2381 at fs/btrfs/disk-io.c:460 btree_read_extent_buffer_pages+0x1ad/0x210
[ 4239.523830] Modules linked in:
[ 4239.524630] RIP: 0010:btree_read_extent_buffer_pages+0x1ad/0x210
[ 4239.527101] Call Trace:
[ 4239.527251] read_tree_block+0x42/0x70
[ 4239.527434] read_node_slot+0xd2/0x110
[ 4239.527632] push_leaf_right+0xad/0x1b0
[ 4239.527809] split_leaf+0x4ea/0x700
[ 4239.527988] ? leaf_space_used+0xbc/0xe0
[ 4239.528192] ? btrfs_set_lock_blocking_rw+0x99/0xb0
[ 4239.528416] btrfs_search_slot+0x8cc/0xa40
[ 4239.528605] btrfs_insert_empty_items+0x71/0xc0
[ 4239.528798] __btrfs_run_delayed_refs+0xa98/0x1680
[ 4239.529013] btrfs_run_delayed_refs+0x10b/0x1b0
[ 4239.529205] btrfs_commit_transaction+0x33/0xaf0
[ 4239.529445] ? start_transaction+0xa8/0x4f0
[ 4239.529630] btrfs_alloc_data_chunk_ondemand+0x1b0/0x4e0
[ 4239.529833] btrfs_check_data_free_space+0x54/0xa0
[ 4239.530045] btrfs_delalloc_reserve_space+0x25/0x70
[ 4239.531907] btrfs_direct_IO+0x233/0x3d0
[ 4239.532098] generic_file_direct_write+0xcb/0x170
[ 4239.532296] btrfs_file_write_iter+0x2bb/0x5f4
[ 4239.532491] aio_write+0xe2/0x180
[ 4239.532669] ? lock_acquire+0xac/0x1e0
[ 4239.532839] ? __might_fault+0x3e/0x90
[ 4239.533032] do_io_submit+0x594/0x860
[ 4239.533223] ? do_io_submit+0x594/0x860
[ 4239.533398] SyS_io_submit+0x10/0x20
[ 4239.533560] ? SyS_io_submit+0x10/0x20
[ 4239.533729] do_syscall_64+0x75/0x1d0
[ 4239.533979] entry_SYSCALL_64_after_hwframe+0x42/0xb7
[ 4239.534182] RIP: 0033:0x7f8519741697
The problem here is, at btree_read_extent_buffer_pages() we don't have
acquired read/write lock on that extent buffer, only basic info like
level/bytenr is reliable.
So race condition leads to such false alert.
However in current call site, it's impossible to acquire proper lock
without race window.
To fix the problem, we only verify first key for committed tree blocks
(whose generation is no larger than fs_info->last_trans_committed), so
the content of such tree blocks will not change and there is no need to
get read/write lock.
Reported-by: Nikolay Borisov <nborisov@suse.com>
Fixes: 581c176041 ("btrfs: Validate child tree block's level and first key")
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Remove GPL boilerplate text (long, short, one-line) and keep the rest,
ie. personal, company or original source copyright statements. Add the
SPDX header.
Signed-off-by: David Sterba <dsterba@suse.com>
Currently if some fatal errors occur, like all IO get -EIO, resources
would be cleaned up when
a) transaction is being committed or
b) BTRFS_FS_STATE_ERROR is set
However, in some rare cases, resources may be left alone after transaction
gets aborted and umount may run into some ASSERT(), e.g.
ASSERT(list_empty(&block_group->dirty_list));
For case a), in btrfs_commit_transaciton(), there're several places at the
beginning where we just call btrfs_end_transaction() without cleaning up
resources. For case b), it is possible that the trans handle doesn't have
any dirty stuff, then only trans hanlde is marked as aborted while
BTRFS_FS_STATE_ERROR is not set, so resources remain in memory.
This makes btrfs also check BTRFS_FS_STATE_TRANS_ABORTED to make sure that
all resources won't stay in memory after umount.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Signed-off-by: David Sterba <dsterba@suse.com>
When mount fails to read trees like fs tree, checksum tree, extent
tree, etc, there is not enough information about where went wrong.
With this, messages like
"BTRFS warning (device sdf): failed to read root (objectid=7): -5"
would help us a bit.
Signed-off-by: Liu Bo <bo.liu@linux.alibaba.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
We have several reports about node pointer points to incorrect child
tree blocks, which could have even wrong owner and level but still with
valid generation and checksum.
Although btrfs check could handle it and print error message like:
leaf parent key incorrect 60670574592
Kernel doesn't have enough check on this type of corruption correctly.
At least add such check to read_tree_block() and btrfs_read_buffer(),
where we need two new parameters @level and @first_key to verify the
child tree block.
The new @level check is mandatory and all call sites are already
modified to extract expected level from its call chain.
While @first_key is optional, the following call sites are skipping such
check:
1) Root node/leaf
As ROOT_ITEM doesn't contain the first key, skip @first_key check.
2) Direct backref
Only parent bytenr and level is known and we need to resolve the key
all by ourselves, skip @first_key check.
Another note of this verification is, it needs extra info from nodeptr
or ROOT_ITEM, so it can't fit into current tree-checker framework, which
is limited to node/leaf boundary.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
For quota disabled->enable case, it's possible that at reservation time
quota was not enabled so no bytes were really reserved, while at release
time, quota was enabled so we will try to release some bytes we didn't
really own.
Such situation can cause metadata reserveation underflow, for both types,
also less possible for per-trans type since quota enable will commit
transaction.
To address this, record qgroup meta reserved bytes into
root::qgroup_meta_rsv_pertrans and ::prealloc.
So at releasing time we won't free any bytes we didn't reserve.
For DATA, it's already handled by io_tree, so nothing needs to be done
there.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Since qgroup has seperate metadata reservation types now, we can
completely get rid of the old root->qgroup_meta_rsv, which mostly acts
as current META_PERTRANS reservation type.
Signed-off-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Any time the first block group of a new type is created, we add a new
kobject to sysfs to hold the attributes for that type. Kobject-internal
allocations always use GFP_KERNEL, making them prone to fs-reclaim races.
While it appears as if this can occur any time a block group is created,
the only times the first block group of a new type can be created in
memory is at mount and when we create the first new block group during
raid conversion.
This patch adds a new list to track pending kobject additions and then
handles them after we do chunk relocation. Between relocating the
target chunk (or forcing allocation of a new chunk in the case of data)
and removing the old chunk, we're in a safe place for fs-reclaim to
occur. We're holding the volume mutex, which is already held across
page faults, and the delete_unused_bgs_mutex, which will only stall
the cleaner thread.
Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Introduced by 5cdc7ad337 ("btrfs: Replace fs_info->workers with
btrfs_workqueue.") but obsoleted by 2a4581983f ("btrfs: factor
btrfs_init_workqueues() out of open_ctree()").
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
The kernel would like to have all stack VLA usage removed[1].
Unfortunately using an integer constant variable as the size of an
array is still considered a VLA. Instead let's use directly sizeof(var)
which removes the VLA usage. Use the occasion to remove csum_size
altogether and use sizeof() also for the size passed to memcmp
[1]: https://lkml.org/lkml/2018/3/7/621
Signed-off-by: Nikolay Borisov <nborisov@suse.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Nikolay Borisov