[BUG]
Btrfs/022 sometimes fails with snapshot's reference mismatch with its
source.
[CAUSE]
Since commit fd08309295 ("fsstress: add the ability to create
snapshots") adds the ability for fsstress to create/delete snapshot and
subvolumes, fsstress will create new subvolumes under test dir.
For example, we could have the following subvolumes created by fsstress:
subvol a id=256
subvol b id=306
qgroupid rfer excl
-------- ---- ----
0/5 16384 16384
0/256 13914112 16384
...
0/263 3080192 2306048 << 2 *306* 048
...
0/306 13914112 16384 << 0/ *306
So when we're greping for subvolid 306, it matches qgroup 0/263 first,
which has difference size, and caused false alert.
[FIX]
Instead of greping "$subvolid" blindly, now grep "0/$subvolid" to catch
qgroupid correctly, without hitting rfer/excl values.
Suggested-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Current _btrfs_get_subvolid() can't handle the following case at all:
# btrfs subvol list $SCRATCH_MNT
ID 256 gen 9 top level 5 path subv1
ID 257 gen 7 top level 256 path subv1/subv2
ID 258 gen 8 top level 256 path subv1/subv3
ID 259 gen 9 top level 256 path subv1/subv4
If we call "_btrfs_get_subvolid $SCRATCH_MNT subv1" we will get a list
of all subvolumes, not the subvolid of subv1.
To address this problem, we go egrep to match $name which starts with a
space, and at the end of a line.
So that all other subvolumes won't hit.
Suggested-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Support 64-bit operation counts so that we can run long-soak tests for
more than 2 billion fsxops.
I figured that testcalls and simulatedopcount should both be signed
because numops is also signed. Granted, I guess numops is signed so
that we can set it to the magic value -1 and have fsx run "forever".
[Eryu: add Darrick's explanation about why we change testcalls to a
signed value.]
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
`seq X Y` will print all numbers between X and Y, including Y. Since
inode chunks contain inodes numbered from X to X+63, we need to set the
loop variables correctly.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
If we can't create the 60T sparse image for testing repair on a large fs
(such as when running on 32-bit), don't bother running the rest of the
test. This requires the actual truncate(1) command, because it returns
nonzero if the system call fails.
[Eryu: the original bug was introduced by me when committing the
original patch, which was a correct fix. Sorry about that.. ]
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
xfsprogs 5.4 prints "Discarding..." if the disk supports the trim
command. Filter this out of the output because xfs_info and friends
won't print that out.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
generic/581 passes for me, but Murphy Zhou reported that it started
failing for him. The part that failed is the part that sets the key
quota to the fsgqa user's current number of keys plus 5, then tries to
add 6 filesystem encryption keys as the fsgqa user. Adding the 6th key
unexpectedly succeeded.
What I think is happening is that because the kernel's keys subsystem
garbage-collects keys asynchronously, the quota may be freed up later
than expected after removing fscrypt keys. Thus the test is flaky.
It would be nice to fix this in the kernel, but unfortunately there
doesn't seem to be an easy fix, and the keys subsystem has always worked
this way. And it seems unlikely to cause real-world problems, as the
keys quota really just exists to prevent denial-of-service attacks.
So, for now just try to make the test more reliable by:
(1) Reduce the scope of the modified keys quota to just the part of the
test that needs it.
(2) Before getting the current number of keys for the purpose of setting
the quota, wait for any invalidated keys to be garbage-collected.
Tested with a kernel that has a 1 second sleep hacked into the beginning
of key_garbage_collector(). With that, this test fails before this
patch and passes afterwards.
Reported-by: Murphy Zhou <xzhou@redhat.com>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Acked-by: Murphy Zhou <xzhou@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
This is a test to validate that we're not adjusting up i_size before
we have the appropriate file extents on disk. We had a problem
where i_size would be adjusted up without a contiguous range of file
extents, which isn't ok without a special option enabled.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test that an incremental send operation works correctly when a file has
shared extents with itself in the send snapshot, with a hole between them,
and the file size has increased in the send snapshot.
This currently fails in 5.5-rc kernels (regression) but is fixed by a
patch that has the following subject:
Btrfs: send, fix emission of invalid clone operations within the same file
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Add a similar test to the existing one but with a file that has a
hardlink as well. This is motivated by a bug found in btrfs where
a fsync on a file that has the old name of another file results
in the logging code to hit an infinite loop. The patch that fixes
the bug in btrfs has the following subject:
"Btrfs: fix infinite loop during fsync after rename operations"
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
As of the linux kernel commit fd0ddbe2509568 ("Btrfs: send, skip
backreference walking for extents with many references"), the test is no
longer dangerous and it's fast (takes 1 to 2 seconds on a modest vm with
a debug kernel). Therefore remove it from the dangerous group and add it
to the auto and quick groups as well.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Addition of fs-specific timestamp range checking was added
in 188d20bcd1eb ("vfs: Add file timestamp range support").
Add a check for whether the kernel supports the limits check
before running the associated test.
Based on an off-list discussion, we use a simpler interim approach
until fsinfo syscall would provide fs timestamp limits info.
This isn't perfect, but works for filesystems expiring in 2038.
Suggested-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Deepa Dinamani <deepa.kernel@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
If we can't create the 60T sparse image for testing repair on a large fs
(such as when running on 32-bit), don't bother running the rest of the
test.
[Eryu: use xfs_io instead of truncate]
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Correct the logic in this test that detects failed scsi_debug
initializations. Downgrade the reaction to _notrun since the filesystem
under test did not fail, just our mockup disk.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The 32-bit calculation of the maximum filesize is incorrect. Replace it
with the formula that the kernel has used since commit 0cc3b0ec23ce
("Clarify (and fix) MAX_LFS_FILESIZE macros"). This fixes a regression
in generic/351 on 32-bit kernels.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Due to the unique structure of xfs/279 running _get_scsi_debug_dev from
a backtick from inside subshell, the "could not get scsi_debug device"
checks do not actually stop the test when modprobe scsi_debug fails.
Therefore, check the value of SCSI_DEBUG_DEV from the subshell and
_notrun the test if we couldn't get memory.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Skip this test we can't create the large sparse file needed to test
overflows.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Add disk dquot structures to the check list now that we killed the
typedef.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test that removing the SGI_ACL_FILE attr also removes the cached ACL
used for access control checking.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
When I test this case on xfs, it may fail as below:
--------------------------------------------
=== link SCRATCH_MNT/A/foo SCRATCH_MNT/bar with fsync SCRATCH_MNT/A ===
+umount: /mnt/xfstests/scratch: target is busy.
+ (In some cases useful info about processes that use
+ the device is found by lsof(8) or fuser(1))
---------------------------------------------
It fails because somethings is still using the fs when we call sync and then
try to unmount it. We can simply remove sync as the unmount is supposed to
persist the file/directory removals.
Signed-off-by: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
fsx has a closeopen option where it will close and then re-open the
file. This is handy, but what is really more useful is to drop the file
from cache completely, so add a drop_caches into this operation so that
the file is read back completely from disk to be really evil.
[Eryu: fix drop cache failure number to stay within 190]
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
I was running down a i_size problem and was missing the failure until
the next iteration of fsx operations because we do the file size check
_before_ the closeopen operation. Move it after the closeopen operation
so we can catch problems where the file gets messed up on disk.
Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Tests need the exact operations to reproduce some issues by
--replay-ops so skip tests rather than one operation if a required
operation/flag in tests is not supported.
Signed-off-by: Xiao Yang <yangx.jy@cn.fujitsu.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Also test that d_ino of readdir entries and i_ino from /proc/locks are
consistent with st_ino and that inode numbers persist after rename to
new parent, drop caches and mount cycle.
With nested xino configuration, directory st_ino is not persistent and
its st_ino/d_ino/i_ino values are not consistent, so test only non-dir
in this test.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Also test that d_ino of readdir entries and i_ino from /proc/locks are
consistent with st_ino and that inode numbers persist after rename to
new parent, drop caches and mount cycle.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Qu Wenruo
Darrick J. Wong
Eric Biggers
Josef Bacik
Filipe Manana
Deepa Dinamani
Christoph Hellwig
Yang Xu
Xiao Yang
Amir Goldstein