This adds a regression test for the following kernel patch:
b4678df184b3 ("errseq: Always report a writeback error once")
This is motivated by some rather odd behavior done by the PostgreSQL
project. The main database writers will offload the fsync calls to a
separate process, which can open files after a writeback error has
already occurred.
This used to work with older kernels that reported the error to only
one fd, but with the errseq_t changes we lost the ability to see
errors that occurred before the open. The above patch restores that
behavior.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The tests _require_test_lsattr, but don't actually use lsattr. They
use chattr +i/-i, so require the appropriate command.
_require_test_lsattr checks the FS_IOC_GETFLAGS ioctl on a directory
and that is not supported in overlayfs. _require_chattr checks the
ioctl on a file, which is supported in overlayfs, so this change
makes the tests run and pass on overlayfs.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The 3 tests that _require_xfs_io_command "flink", actually require
O_TMPFILE support and flink command, but the former is far unlikely
to be missing. The test btrfs/058 doesn't even use the flink
command.
When running these tests on a filesystem that does not support
O_TMPFILE (e.g. overlayfs) the result is not very infomative:
generic/004 1s ... [not run] xfs_io flink failed (old kernel/wrong fs?)
Decouple the requirements for "flink" command and "-T" command line
flag and require the former explicitly in tests that use it.
As a result the report is now more informative:
generic/004 1s ... [not run] O_TMPFILE is not supported
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Fuzzing has recently uncovered a couple of conditions where we don't
detect corruptions that reallocate already allocated inodes. This
test exercises those cases, and checks that we shut down the
filesystem appropriately when such a corruption occurs.
Signed-Off-By: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
There are a bunch of tests that are run by the auto group that are
marked dangerous. This was done because the test exercised a crash
or other fatal error that has since been fixed. Remove the dangerous
tag from the auto tests that pass on a 4.17-rc3 kernel as they are
not dangerous anymore.
Signed-Off-By: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Multiple origin references to the same lower file from upper files
that are not hardlinks will falsely return the same st_ino/st_dev
for two different overlay files and will cause 'diff' to falsely
report that content of files is the same when it is not.
This test checks that overlayfs detects and fails lookup of a
multiply referenced origin.
The check for multiply referenced origin was a by-product of kernel
commit 31747eda41ef ("ovl: hash directory inodes for fsnotify")
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test description says: "This is the same test as generic/156 except
that we also check the inode reflink flag.", only since commit
f1c3fee ("xfs: remove NOCOW_FL testing from test") it is almost
exactly the same test as generic/156.
almost - because since xfs/132 diverged from generic/156 the change
911efb0 ("reflink: change to relative margins") is only applied to
the original test.
Anyway, there doesn't seem to be a reason to keep this duplicated
test anymore, so remove it.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The blocktrash fuzz tests for xfs will try to mount and write to the
filesystem after corrupting it. However, the mount may not necessarily
succeed, in which case we must not write junk to the root filesystem.
Use the new _try_scratch_mount to guard against that.
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>
Add fsstress to the pile of things that we race with rmap repair to
ensure that the rmap repair isolates the filesystem correctly while it
is doing its repairs.
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>
Create malicious . and .. entries (you didn't see the zero-width
joiners at the end, did you?) in a directory to see if scrub will pick
them up.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test if a filesystem will allow us to create names with easily
confusable unicode sequences (character spoofing) and, if on XFS,
whether or not xfs_scrub will notice.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Since we've rewriting the xfs_scrub Unicode name scanner to use libicu
to detect potential spoof names, change our check for unicode-enabled
name scanning xfs_scrub to look for libicu instead of libunistring,
adjust the golden output to reflect the new library's detection
capabilities and make sure we get all the scrub output by invoking with
-v.
Note that this requires xfsprogs 4.16 or newer; since xfs_scrub is (for
now) an experimental program, we don't care about breaking backwards
compatibility.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Try injecting a Unicode directional override character in the middle of
a name to see if the fs can handle it / xfs_scrub will complain.
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Kanda Motohiro reported that expanding a tiny xattr into a large
xattr fails on XFS because we remove the tiny xattr from a shortform
fork and then try to re-add it after converting the fork to extents
format having not removed the ATTR_REPLACE flag. This fails because
the attr is no longer present, causing a fs shutdown.
[Eryu: introduce function "fail" and use it where appropriate]
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=199119
Reported-by: kanda.motohiro@gmail.com
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>
Test how the "insert range" fallocate operation interacts with the
maximum file size (s_maxbytes).
- Shift extents past the max file size (exposes an ext4 bug).
- Increase i_size past the max file size (exposes an xfs bug).
Signed-off-by: Eric Biggers <ebiggers@google.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
POSIX requires that record locks are preserved across an execve(2).
But currently the locks are released if process is multithreaded at
the time that execve is called.
As Jeff Layton wrote in his patch:
"
In that case, we'll end up unsharing the files_struct but the locks
will still have their fl_owner set to the address of the old one.
Eventually, when the other threads die and the last reference to the
old files_struct is put, any POSIX locks get torn down since it
looks like a close occurred on them.
The result is that all of your open files will be intact with none
of the locks you held before execve.
"
Add a new regression test for this particular case.
[Eryu: rewrite commit log and test description]
Signed-off-by: Xiong Zhou <xzhou@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Use common helper functions where needed. By doing this it improves
code readability and debugging of it.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
These tests check for constant inode number on copy up with
nonsamefs layer configuration. This problem is fixes only when
opting-in with the xino=on mount option, so let the tests enable the
mount option on new kernels and notrun on old kernels.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
This test renames a merge directory so it needs to enable
redirect_dir feature, which is not enabled by default.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
_overlay_check_scratch_dirs needs to base scratch fs to be mounted,
so only unmount overlay before check.
Remove redundant definition of upperdir/workdir path, which also
uses hardcoded path instead of the config vars OVL_UPPER/OVL_WORK.
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The XFS filestreams allocator caches dir inode -> agno mappings in
an MRU mechanism that holds elements in memory for an amount of time
and then cleans up expired elements in the background. The elements
typically held inode pointers without holding a reference to the
associated inode. This means that if the inode is reclaimed before
an expired entry is cleaned up, the MRU reaper can access freed
memory and cause a panic.
Test for this problem by performing continuous filestreams
allocations under short-lived parent directory inodes. This will
produce KASAN use-after-free splats if enabled during the test.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
This test case takes a long time to complete at the default
LOAD_FACTOR=1, so reduce the nr_extents to 256, so for larger
systems it can still use higher LOAD_FACTOR.
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
btrfs-progs patch[1] replaced read(2) write(2) with splice(2) and
caused the append-redirect to stop working.
Before:
btrfs send /btrfs/ro_send > /dev/null
At subvol /btrfs/ro_snap
btrfs send /btrfs/ro_send >> /dev/null
At subvol /btrfs/ro_snap
After:
btrfs send /btrfs/ro_send > /dev/null
At subvol /btrfs/ro_snap
btrfs send /btrfs/ro_send >> /dev/null
At subvol /btrfs/ro_snap
ERROR: failed to read stream from kernel: Invalid argument
Further in the test case the line..
btrfs/130
::
_run_btrfs_util_prog send $SCRATCH_MNT/ro_snap > /dev/null 2>&1
which intended to redirect send output to /dev/null, but ended up
append redirect to the $seqres.full file. And so this test case
failed as 'Invalid argument' for sometime now.
Still as append of a btrfs send output doesn't make sense, so fix
the fstests.
Also adds logs going into $seqres.full.
[1]
ba23855cdc8961bbaef1fcad4854d494cdb3afd3
btrfs-progs: send: use splice syscall instead of read/write to transfer buffer
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Jeff Layton
Darrick J. Wong
Amir Goldstein
Dave Chinner
Eric Biggers
Xiong Zhou
Anand Jain
Brian Foster