In generic/459, if the DISCARD of the thin device somehow fails with
this message:
device-mapper: thin: Data device (dm-1) discard unsupported: Disabling discard passdown.
Then we can end up with arbitrary gunk in the thin device. This
causes mkfs to fail because it's afraid to format the device. Don't
be afraid, just zap it. FWIW mkfs.xfs thinks that the thinp device
has an xfs external log because sometimes the thinp device just
happen to be backed by the log of the previous test's scratch fs.
Fix this by making the _mkfs_dev helper always format the device,
per Eryu Guan's suggestion.
[eguan: fix subject to reflect the fix is in common/rc]
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Fix test failures with new inode pointer verifiers... and also make
sure that the running xfs actually supports realtime rmap. (This
should stop the current crop of weird test failures since nobody has
rtrmap yet anyway...)
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Remove the inode-paths check from _check_xfs_test_fs because we don't
support inode paths, xfsprogs doesn't have a xfs_{check,repair}_ipaths
tool, and it's broken anyway because we ignore _check_xfs_filesystem
(which tells whether or not the filesystem is even still mounted).
[eguan: also fix the return value of _check_xfs_test_fs]
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Add some helper functions to require that we can reload a given
module, and add a helper to actually do that. Refactor the existing
users to use the generics.
We need to hoist completely the behaviors of the old btrfs module
helper because we need to confirm before starting the test that we
actually can remove the module.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Make the xfs_scrub output that gets recorded to $seqres.full follow
the format of xfs_repair checks.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Move all the requirements checking for xfs_scrub into a helper function.
Make sure the helper properly detects the presence of the scrub ioctl
and situations where we can't run scrub (e.g. norecovery).
Refactor the existing three xfs_scrub call sites to use the helper to
check if it's appropriate to run scrub.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
When we're doing the second online scrub (to figure out if the
repair did any good) we shouldn't let that second scrub preen the
filesystem in any way. If scrub finds things it can't/won't preen
that turns into a nonzero return code which gets reported
(incorrectly) as a failure.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
The upcoming xfs_scrub tool will have the ability to warn about
suspicious UTF-8 normalization collisions. We want generic/45[34]
to be able to test this functionality, but to do that we have to
forcibly set the codeset to UTF-8 via LC_ALL since the rest of
xfstests only uses LC_ALL=C.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Occasionally speculative preallocation kicks in when writing files
to a filesystem under test. These preallocations consume quota and
/usually/ aren't around after we drop_caches, but there's nothing to
guarantee that they actually have, so the quota reports will be
different before and after the fs remount, causing sporadic test
failures in generic/{23[123],270}.
We now have xfs_spaceman which can instruct XFS to forcibly remove
the speculative preallocations. This fixes the sporadic failures,
at least for XFS.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Found it when test btrfs delalloc accounting overflow, Fix offset
error. We will fill in the gaps between the created extents, then
outstanding extents will all be merged into 1.
Signed-off-by: Robbie Ko <robbieko@synology.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
No agreement on how to fix it in the foreseeable future. So remove
it from auto group to prevent newbie tester from spending days
waiting it to finish.
Reported-by: Po-Hsu Lin <po-hsu.lin@canonical.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Nikolay Borisov <nborisov@suse.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
_overlay_scratch_unmount is not supposed to be invoked directly by
tests. Also, since unmounting scratch fs is optional, this commit
removes invocation of _overlay_scratch_unmount.
Signed-off-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Currently running 'btrfs device delete' can end up with losing data
raid profile (if any), this test is to reproduce the problem.
The fix is
"Btrfs: avoid losing data raid profile when deleting a device"
Signed-off-by: Liu Bo <bo.li.liu@oracle.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
XFS had a bug that resulted in an unexpected NULL buffer during
unlink of an inode with a multi-level attr fork tree. This occurred
due to a stale reference to content in a released/reclaimed buffer.
Use the XFS buffer LRU reference count error injection tag to
recreate the conditions for the bug. Create a file with a
multi-level attr fork tree and then unlink it with buffer caching
disabled.
Commit f35c5e10c6ed ("xfs: reinit btree pointer on attr tree
inactivation walk") fixed the bug.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
This commit adds a test to verify consistent d_ino feature when
the overlayfs instance is composed of two different underlying
filesystem instances.
For example,
$ mount -t xfs /dev/loop0 /mnt/test
$ mount -t xfs /dev/loop1 /mnt/scratch
$ mkdir /mnt/scratch/upper
$ mkdir /mnt/scratch/work
$ mount -t overlay overlay -o lowerdir=/mnt/test \
-o upperdir=/mnt/scratch/upper \
-o workdir=/mnt/scratch/work /mnt/merge
The goal of this test is to verify that the inode numbers returned by
readdir(3) (i.e. dirent->d_ino) are consistent with inode numbers
returned by stat(2) (i.e. stat->st_ino) in all the below listed cases,
- Parent's (i.e. "..") d_ino must always be calculated because a
pure dir can be residing inside a merged dir.
- d_ino for "." must always be calculated because the present
directory can have a copy-up origin.
- Verify d_ino of '.' and '..' before and after dir becomes impure.
While at it also verify if trusted.overlay.impure xattr is
set/reset appropriately and invalidation of readdir cache.
- Verify copied up file's (inside a impure dir) d_ino.
- Verify invalidation of readdir cache.
- Verify d_ino values corresponding to "." and ".." entries of a
pure lower dir.
- Verify d_ino of ".." entry of a merged dir.
- Verify pure lower residing in dir which has another lower layer
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Chandan Rajendra <chandan@linux.vnet.ibm.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
These tests appear to mix buffered and O_DIRECT I/O; easily triggered
on both xfs and ext4.
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
With upstream commit 332391a9935d ("fs: Fix page cache inconsistency
when mixing buffered and AIO DIO"), ext4 (and probably other
non-iomap based filesystems) need a _check_dmesg() whitelist entry
for a new WARN in dio_complete() -- generic/095 and 247 trigger a
dio_complete() splat pretty reliably for me.
Add dio_complete() entry to _filter_xfs_dmesg(), rename it and use
it unconditionally in generic/095, 224, 247 and 446.
[eguan: source common/filter too in generic/446]
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Move this XFS-specific _filter_xfs_dmesg() to common/filter so all
tests could use it. It will be renamed & made more generic and used
by more tests in later patches.
[eguan: add commit log]
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
This fixes test failure with kernel v4.14-rc4 and default index=off
configuration.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
generic/166 is takes way too long to run on iscsi disks - over an
*hour* on flash based iscsi targets. In comparison, it takes 18s to
run on a pmem device.
The issue is that it takes 3-4s per file write cycle on slow disks,
and it does a thousand write cycles. The problem is taht reflink is
so much faster than the write cycle that it's doing many more
snapshots on slow disks than fast disks, and this slows it down even
more.
e.g. the pmem system that takes 18s to run does just under 1000
snapshots - roughly one per file write. 20 minutes into the iscsi
based test, it's only done ~300 write cycles but there are almost
10,000 snapshots been taken. IOWs, we're doing 30 snapshots a file
write, not ~1.
Fix this by rate limiting snapshots to at most 1 per whole file
write. This reduces the number of snapshots taken on fast devices by
~50% (runtime on pmem device went from 18s -> 8s) but reduced it to
1000 on slow devices and reduced runtime from 3671s to just 311s.
Signed-Off-By: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Test generic/422 hardcodes in the output file how much space files it
creates are supposed to take up on disk. This doesn't work when
additional space is allocated for extended attributes for ACLs or
SELinux labels.
Instead, record the actual space used in generic/422.full, and only
check if the writeback changes the space used.
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
In this commit a new test case is added to test that i_size races
don't occur under dio reads/writes. We add a program in /src dir,
which has a writer to issue some append dio writes. Meanwhile it
has a reader in this test to do some dio reads. As we expect,
reader should read nothing or data with 'a'. But it might read some
data with '0'.
The bug can be reproduced by this test case [1].
1. http://patchwork.ozlabs.org/patch/311761/
This ostensibly tests commit:
9fe55eea7 Fix race when checking i_size on direct i/o read
Update by Eric Sandeen:
- update to recent xfstests
- update commit log
Update by Eryu Guan:
- add aio-dio support to the test and add 'aio' group
- add ability to test different alignments
- move test from src/ to src/aio-dio-regress/
- add .gitignore entry
- rebase against latest xfstests with various minor fixes & cleanups
- update commit log
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Zheng Liu <wenqing.lz@taobao.com>
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Run delalloc writes & append writes & non-data-integrity syncs
concurrently to test the race between block map change vs writeback.
This is to cover an XFS bug that data could be written to wrong
block and delay allocated blocks are leaked because the block map
was changed due to the removal of speculative allocated eofblocks
when writeback is in progress.
And this test partially mimics what lustre-racer[1] test does, using
which this bug was first found.
[1] https://git.hpdd.intel.com/?p=fs/lustre-release.git;a=tree;f=lustre/tests/racer;hb=HEAD
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
The test uses a scratch device, so add _require_scratch so
the test will be skipped instead of failing when $SCRATCH_DEV
has not been specified.
Signed-off-by: Hans Holmberg <hans.holmberg@cnexlabs.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Eryu Guan <eguan@redhat.com>
Darrick J. Wong
Robbie Ko
Qu Wenruo
Chandan Rajendra
Hou Tao
Liu Bo
Brian Foster
Ilya Dryomov
Amir Goldstein
Dave Chinner
Andreas Gruenbacher
Zheng Liu
Eryu Guan
Hans Holmberg