This bug is exposed by populating a high level qgroup, and then make
it childless with old qgroup numbers, and finally do rescan.
Normally rescan should zero out all qgroups' accounting number, but
due to a kernel bug which won't mark childless qgroups dirty, their
on-disk data is never updated, thus old numbers remain and cause
qgroup corruption.
Fixed by the following kernel patch:
"btrfs: qgroup: Dirty all qgroups before rescan"
[Eryu: removed useless _filter_xfs_io]
Reported-by: Misono Tomohiro <misono.tomohiro@jp.fujitsu.com>
Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Regression test case for kernel patchset:
fs/lock: show locks info owned by dead/invisible processes
Open new fd by exec shell built-in, then require exclusive lock
by flock(1) command. Checking /proc/locks for the lock info.
This patchset post by Konstantin Khorenko <khorenko@virtuozzo.com>
has not been merged v4.18-rc8.
[Eryu: require $FLOCK_PROG and add more comments]
Signed-off-by: Xiong Murphy Zhou <xzhou@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Old xfsprogs can't change attr hdr.count to 0 on v5 filesystems, two
reasons maybe cause this issue:
1) This commit has been merged: 89baf918(xfs_db: write values into
dir/attr blocks and recalculate CRCs).
2) xfs_db write command doesn't support -d option.
That's not a real bug, so skip this test if xfs_db can't set attr
hdr.count to 0.
Signed-off-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
generic/228 changes kernel.core_pattern, so it saves it beforehand in
order to restore it. However, it adds an extra space in the process:
$ echo \"$(sysctl kernel.core_pattern | awk -F = '{print $NF}')\"
" |/usr/lib/systemd/systemd-coredump %P %u %g %s %t %c %h %e"
Instead of using awk, just use sysctl -n to only print the value.
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Since glibc 2.28 has added the statx function after commit
fd70af45528d ("Add the statx function"), stat_test.c will include
/usr/include/bits/statx.h eventually by "include <sys/stat.h>". That
causes the STATX_TYPE has already been defined before include
"status.h", then xfstests_statx will not be defined.
So make fails with the following error message:
/usr/bin/ld: /tmp/cc2Isfch.o: in function `main':
/home/luke/workspace/xfstests-dev/src/stat_test.c:690: undefined reference to `xfstests_statx'
/usr/bin/ld: /tmp/cc2Isfch.o: in function `get_reference':
/home/luke/workspace/xfstests-dev/src/stat_test.c:301: undefined reference to `xfstests_statx'
/usr/bin/ld: /tmp/cc2Isfch.o: in function `main':
/home/luke/workspace/xfstests-dev/src/stat_test.c:642: undefined reference to `xfstests_statx'
Just move out the definition of xfstests_statx between "#ifndef
STATX_TYPE" and "#endif /* STATX_TYPE */".
Signed-off-by: Lu Fengqi <lufq.fnst@cn.fujitsu.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test that if we write into an unwritten extent of a file when there
is no more space left to allocate in the filesystem and then
snapshot the file's subvolume, after a clean shutdown the data was
not lost.
This test is motivated by a bug found by Robbie Ko for which there
is a fix whose patch title is:
"Btrfs: fix unexpected failure of nocow buffered writes after
snapshotting when low on space"
Reported-by: Robbie Ko <robbieko@synology.com>
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
There's a situation where the directory structure and the inobt
thinks the inode is free, but the inode on disk thinks it is still
in use. XFS should detect it and prevent the kernel from oopsing on
lookup.
Signed-off-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Since commit a79a464d5675 ("btrfs: Allow rmdir(2) to delete an empty
subvolume"), rm -r can delete a subvolume too.
This test assumes that rm -r does not delete a subvolume.
Currently the commit does not affect the test since qgroup items
still exist after subvolume deletion, but we plan to change the
behavior and remove them along with subvolume deletion.
So update the test and keep subvolume (and qgroup item) in any kernel
version.
Signed-off-by: Misono Tomohiro <misono.tomohiro@jp.fujitsu.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test that an incremental send operation produces correct results if
a file that has a prealloc (unwritten) extent beyond its EOF gets a
hole punched in a section of that prealloc extent.
This test is motivated by a bug found in btrfs which is fixed by a
patch for the linux kernel titled:
"Btrfs: send, fix incorrect file layout after hole punching beyond eof"
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The commit b3cf8b7233 update xfs/288
to support v5 filesystem testing. That commit thought xfs_db write
command can work well with -d option on V5 XFS. But the truth is the
case doesn't use that option.
So turn to use _scratch_xfs_set_metadata_field, it will help to use
proper options for xfs_db write command.
Signed-off-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
This adds a regression test for the following series:
[PATCH v4 0/2] ext4: fix DAX dma vs truncate/hole-punch
https://lists.01.org/pipermail/linux-nvdimm/2018-July/016842.html
which adds synchronization between DAX DMA in ext4 and truncate/hole-punch.
The intention of the test is to test those specific changes, but it runs
fine both with XFS and without DAX so I've put it in the generic tests
instead of ext4 and not restricted it to only DAX configurations.
When run with v4.18-rc6 + DAX + ext4, this test will hit the following
WARN_ON_ONCE() in dax_disassociate_entry():
WARN_ON_ONCE(trunc && page_ref_count(page) > 1);
If you change this to a WARN_ON() instead, you can see that each of the
four paths being exercised in this test hits that condition many times in
the one second that the subtest is being run.
Signed-off-by: Ross Zwisler <ross.zwisler@linux.intel.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test that we are able to do send operations when one of the source
snapshots (or subvolume) has a file that is deleted while there is
still a open file descriptor for that file.
This test is motivated by a bug found in btrfs which is fixed by a patch
for the linux kernel titled:
"Btrfs: fix send failure when root has deleted files still open"
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test that if we have a file with 2 (or more) hard links in the same
parent directory, rename of the hard links, rename one of the other
hard links to the old name of the hard link we renamed before,
create a new file in the same parent directory with the old name of
second hard link we renamed, fsync fsync this new file and power
fail, we will be able to mount again the filesystem and the new file
and all hard links exist.
This test is motivated by a bug found in btrfs, where mounting the
filesystem after the power failure resulted in failure with an errno
value of EEXIST, which is fixed by a patch for the linux kernel
titled:
"Btrfs: fix mount failure after fsync due to hard link recreation"
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Latest glibc changed some rules of sorting and regexes, the usage
likes "[a-z]" maybe not only stand for lowcase letters a..z in
different locale. Similar issues include [A-Z], [0-9] and so on.
For example, in en_US.UTF-8 locale, [a-z] means aAbBcCdD...zZ, it
stands for both of uppercase and lowercase. Currently this issue
cause `make install` fails on system with new glibc.
So use POSIX character class to instead of [...] group, something
likes [:lower:], [:upper:], [:alpha:], [:alnum:], etc... are common.
Signed-off-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test that if we do a buffered write to a file, fsync it, clone a
range from another file into our file that overlaps the previously
written range, fsync the file again and then power fail, after we
mount again the filesystem, no file data was lost or corrupted.
This test is motivated by a bug found in btrfs, which is fixed by a
patch for the linux kernel titled:
"Btrfs: fix file data corruption after cloning a range and fsync"
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Filter out quota regeneration messages from xfs_repair when we check
the filesystem, because the xfs tests that encode xfs_repair output
in the golden output will fail when quotas are enabled and the
xfs_repair quota messages appear.
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>
This test tries to make the filesystem go down by setting up
dm-error and committing metadata updates. Since the test doesn't
remount the fs after it goes down to recover the log, this can
result in a dirty log being presented to the post-test filesystem
check if the filesystem is xfs and quotas are enabled.
Since this is a regression test for a NULL pointer dereference in
the kernel after the fs goes down, simply skip the post-test fsck.
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>
The xfs_db 'write -d' command has supported v5 filesystems for a few
releases now, and there's nothing about this test that require v4,
so let the test run with v5 if the user so specifies.
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>
xfs/001 fuzzes various extent fields using xfs_db. There's nothing
in it that's specific to v4 filesystems, so upgrade the test to be
able to handle v5 filesystems.
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>
These four tests check that mmap'd cow writes fail when the
filesystem goes down. For regular filesystems the msync reports
EIO, but if quotas are enabled on xfs the write itself terminates
xfs_io with a SIGBUS. We don't care how the write fails, so don't
let the SIGBUS report escape to the golden output.
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>
If a user constructs a test that loops repeatedly over below steps
on dm-thin, block allocation can fail due to discards not having
completed yet (Fixed by a685557 dm thin: handle running out of data
space vs concurrent discard):
1) fill thin device via filesystem file
2) remove file
3) fstrim
And this maybe cause a deadlock (fast device likes ramdisk can help
a lot) when racing a fstrim with a filesystem (XFS) shutdown. (Fixed
by 8c81dd46ef3c Force log to disk before reading the AGF during a
fstrim)
This case can reproduce both two bugs if they're not fixed. If only
the dm-thin bug is fixed, then the test will pass. If only the fs
bug is fixed, then the test will fail. If both of bugs aren't fixed,
the test will hang.
Signed-off-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The xfs_repair "-t" option shouldn't be used alone. An interval must
follow the -t option, or xfs_repair will report errors. And only
modify reporting interval is useless, if we don't enable ag_stride.
Signed-off-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
On some old kernel which supports COLLAPSE_RANGE and ZERO_RANGE, but
doesn't support INSERT_RANGE, this case alway fails as:
QA output created by 499
+main: filesystem does not support fallocate mode FALLOC_FL_INSERT_RANGE, disabling!
Silence is golden
fsx print one more line to break the golden image.
To fix this issue, redirect both fsx stdout and stderr to a file,
then check the return value.
Signed-off-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Fuzz every field of every structure and then try to write the
filesystem, to see how many of these writes can crash the kernel.
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>
Qu Wenruo
Xiong Murphy Zhou
Zorro Lang
Omar Sandoval
Lu Fengqi
Filipe Manana
Misono Tomohiro
Ross Zwisler
Darrick J. Wong