Targeted fuzzing tests which destroy various pieces of filesystem or
allocation group metadata; the tests look for (a) kernel detection of
corruption, (b) xfs_repair repair of said corruption, and (c)
post-repair fs usability.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Introduce tests for XFS and ext4 which format a filesystem, populate
it, then uses blocktrash and e2fuzz to corrupt the metadata. The FS
is remounted, modified, and unmounted. Following that, xfs_repair or
e2fsck are run until it no longer finds errors to correct, after which
the FS is mounted yet again and exercised to see if there are any
errors remaining.
The XFS test requires an xfs_db that can handle blocktrash and v5
filesystems.
The ext4 test requires metadata_csum support in e2fsprogs.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
XFS had a bug which lead to spurious checksum failures during
verification of log records during recovery. This occurred when the
filesystem was mounted for recovery with a different log buffer size
(via the 'logbsize=...' mount option from when the filesystem crashed.
Create a regression test that dirties the log using one particular log
buffer size, shuts down the fs and attempts recovery using a larger log
buffer size.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Tests xfs_db's ability to change & restore UUIDs on V5 filesystems,
and tests xfs_copy's ability to change the UUID on the copy.
Update to _filter_uuid is so that it will catch the UUID output
from xfs_admin -u, which is slightly different than the regexp it
was expecting.
This requires new userspace which knows how to change the UUID on
a V5 superblock.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
XFS dynamic inode allocation has a fundamental limitation in that an
inode chunk requires a contiguous extent of a minimum size. Depending on
the level of free space fragmentation, inode allocation can fail with
ENOSPC where the filesystem might not be near 100% usage.
The sparse inodes feature was implemented to provide an inode allocation
strategy that maximizes the ability to allocate inodes under free space
fragmentation. This test fragments free space and verifies that
filesystems that support sparse inode allocation can allocate a minimum
percentage of inodes on the fs.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Commit bbe051c841d5 ("xfs: disallow ro->rw remount on norecovery mount")
disabled rw remount on norecovery ro mount, this test makes sure the
behavior is correct.
Signed-off-by: Eryu Guan <eguan@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
In certain cases, the extent size hints can cause maximum extent
size overflows resulting in extent tree corruptions. This test
exercises the original reproducer, and another corner case
demonstrated to expose problems on 1k block size filesystems.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
--
Version 2:
- TESTDIR->TEST_DIR
- append output to $seqres.full
xfs_repair compares attr names in the root namespace to
two special/reserved names, "SGI_ACL_FILE" and "SGI_ACL_DEFAULT"
and if the value in them aren't valid acls, flags this as
an inconsistency.
However, due to various bugs, xfs_repair may only compare
a smaller portion of the on-disk value; hence either
substrings or superstrings may match, and false-positive
corruption will be detected. This test checks for those
false positives; i.e. the ACL names created in this test
may cause xfs_repair to "fix" them, but it should not.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Eryu Guan <eguan@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
This patch moves the generic testcases defined in xfs into tests/generic/.
xfs/085 -> generic/052
xfs/086 -> generic/054
xfs/087 -> generic/055
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The secondary superblock verification in xfs_repair was subject to a bug
that unnecessarily leads to a brute force superblock scan if the last
superblock in the fs happens to be corrupt. Normally, xfs_repair handles
one-off superblock corruption gracefully using a heuristic that finds
the most consistent superblock content across the set of secondary
superblocks.
Create a regression test for xfs_repair that corrupts the last
superblock in the fs. Verify the superblock is updated from the
previously verified sb content and a brute force scan is not initiated.
In the event of failure, detect that a brute force scan has started and
abort the repair in order to fail the test quickly.
To support the test, extend the xfs_repair filter to handle corrupted
superblock repair output and provide generic test output for arbitrary
AG counts.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Adds a new test that checks for correct behaviour of
XFS_IOC_FSSETXATTR for directories: extent sizes should always be
settable on a directory, even if the directory already has allocated
extents.
Signed-off-by: Iustin Pop <iustin@k1024.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
This tests xfs_copy, copying all combinations of sector
size and block size possible on the platform running the
test.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The bulkstat mechanism is used by xfsdump and other tools to index all
inodes allocated in a filesystem. Run some simple tests with varying
inode layouts and bulkstat batch sizes to verify bulkstat finds the
expected number of inodes.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
This test creates a large-ish directory structure using
fsstress, and does a dump/restore to make sure we dump
all the files.
Without the fix for the regression caused by:
c7cb51d xfs: fix error handling at xfs_inumbers
we will see failures like:
-xfsrestore: 486 directories and 1590 entries processed
+xfsrestore: 30 directories and 227 entries processed
as it fails to process all inodes.
I think that existing tests have a much smaller set of files,
and so don't trip the bug.
I don't do a file-by-file comparison here, because for some
reason the diff output gets garbled; this test only checks
that we've dumped & restored the correct number of files.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
XFS buffered I/O writeback has a subtle race condition that leads to
stale data exposure if the filesystem happens to crash after delayed
allocation blocks are converted on disk and before data is written back
to said blocks.
Use file allocation commands to attempt to reproduce a related, but
slightly different variant of this problem. The associated falloc
commands can lead to partial writeback that converts an extent larger
than the range affected by falloc. If the filesystem crashes after the
extent conversion but before all other cached data is written to the
extent, stale data can be exposed.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
A buffer use after free race was discovered in the XFS log recovery
codepath if I/O failures occur during recovery. The I/O submission path
can abort the mount and release the only reference held on some buffers
before I/O completion processing (e.g., async workqueue processing)
might have completed. Badness ensues if the I/O completion path
subsequently attempts to access said buffers.
The test manufactures the race by forcing all writes to fail (via
dm-flakey) after a fixed period of time. A delay is inserted into the
mount codepath to synchronize write failures with log recovery.
Credit for discovery of the race and definition of the reproducible test
case goes to Alex Lyakas.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reported-by: Alex Lyakas <alex@zadarastorage.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Test segfault issue when repairing large xfs.
Regression test for xfsprogs commit
7f2d6b8 xfs_repair: avoid segfault if reporting progress early in repair
The original test is from the commit log, written by Eric Sandeen.
Signed-off-by: Eryu Guan <eguan@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Make sure inodes can be allocated in new space added by xfs_growfs.
Regression test for
xfs: allow inode allocations in post-growfs disk space
Signed-off-by: Eryu Guan <eguan@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Changes in the XFS logging code have lead to small leaks in the log
grant heads that consume log space slowly over time. Such problems have
gone undetected for an unnecessarily long time due to code complexity
and potential for very subtle problems. Losing only a few bytes per
logged item on a reasonably large enough fs (10s of GB) means only the
most continuously stressful workloads will cause a severe enough failure
(deadlock due to log reservation exhaustion) quickly enough to indicate
something is seriously wrong.
Recent changes in XFS export the state of the various log heads through
sysfs to aid in userspace/runtime analysis of the log. This test runs a
workload against an XFS filesystem, quiesces the fs and verifies that
the log reserve and write grant heads have not leaked any space with
respect to the current head of the physical log.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
XFS can allocate significant amounts of space to files via speculative
preallocation. Such preallocation may not be reclaimed automatically on
file close() if a file is repeatedly opened and extended. For smaller
filesystems with relatively large and slow growing files, this
preallocation can linger for some time, including contributing to out of
space conditions.
Create a situation where an fs is near out of space while several files
still have lingering, significant preallocations. Verify that new
writers reclaim the preallocated space rather than return ENOSPC. Repeat
a similar test for quota limits and EDQUOT.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Create a stress test for the free inode btree. Allocate a set of inodes
sequentually and run a hard link clone and random replacement algorithm
across the set. Background removal of the oldest directories creates a
sparse set of free inodes over time. Run an fsstress workload
concurrently to exercise the fs.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The finobt creates a duplicate subset of inode allocation metadata from
the inobt. xfs_repair should detect and repair inconsistencies in the
finobt that could be caused by bugs or corruption. This test uses xfs_db
to cause targeted corruptions in the finobt and verify repair detects
and corrects the filesystem.
In particular, the test corrupts individual finobt records to cause
inconsistency between the inode allocation count fields as well as
causing the finobt to contain a record with no free inodes.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
There is no mainline kernel support for DMAPI in XFS, and so every
time the xfstests auto group is run it throws a large number of
"[not run] Assuming DMAPI modules are not loaded". I've noted that
people are excluding the dmapi group to avoid this, so rather than
inflict pain on everyone, make hose few that need to do dmapi
testing include it specifically.
IOWs, remove the DMAPI tests from the auto group.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Dave Chinner <david@fromorbit.com>
The Q_XQUOTARM quotactl was not working properly, because
we weren't passing in proper flags. The xfs_fs_set_xstate()
ioctl handler used the same flags for Q_XQUOTAON/OFF as
well as Q_XQUOTARM, but Q_XQUOTAON/OFF look for
XFS_UQUOTA_ACCT, XFS_UQUOTA_ENFD, XFS_GQUOTA_ACCT etc,
i.e. quota type + state, while Q_XQUOTARM looks only for
the type of quota, i.e. XFS_DQ_USER, XFS_DQ_GROUP etc.
Unfortunately these flag spaces overlap a bit, so we
got semi-random results for Q_XQUOTARM; i.e. the value
for XFS_DQ_USER == XFS_UQUOTA_ACCT, etc. yeargh.
Anyway, here's a simple test that demonstrates it,
kernel patch to fix it will follow.
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
Darrick J. Wong
Brian Foster
Eric Sandeen
Eryu Guan
Dave Chinner
Jaegeuk Kim
Iustin Pop
Eric Sandeen