]> www.infradead.org Git - users/hch/xfstests-dev.git/log
users/hch/xfstests-dev.git
18 months agoxfs/262: call _scratch_require_xfs_scrub scrub-check
Christoph Hellwig [Wed, 10 Jan 2024 17:44:24 +0000 (18:44 +0100)]
xfs/262: call _scratch_require_xfs_scrub

Call _scratch_require_xfs_scrub so that the test is _notrun on kernels
without online scrub support.

Signed-off-by: Christoph Hellwig <hch@lst.de>
18 months agoxfs: add a _scratch_require_xfs_scrub helper
Christoph Hellwig [Thu, 11 Jan 2024 14:10:48 +0000 (15:10 +0100)]
xfs: add a _scratch_require_xfs_scrub helper

Add a helper to call _supports_xfs_scrub with $SCRATCH_MNT and
$SCRATCH_DEV.

Signed-off-by: Christoph Hellwig <hch@lst.de>
18 months agoxfs: check that the mountpoint is actually mounted in _supports_xfs_scrub
Christoph Hellwig [Thu, 11 Jan 2024 14:10:12 +0000 (15:10 +0100)]
xfs: check that the mountpoint is actually mounted in _supports_xfs_scrub

Add a sanity check that the passed in mount point is actually mounted
to guard against actually calling _supports_xfs_scrub before
$SCRATCH_MNT is mounted.

Signed-off-by: Christoph Hellwig <hch@lst.de>
19 months agogeneric: Add integrity tests with synchronous directio
Ritesh Harjani (IBM) [Mon, 20 Nov 2023 11:19:34 +0000 (16:49 +0530)]
generic: Add integrity tests with synchronous directio

This test covers data & metadata integrity check with directio with
o_sync flag and checks the file contents & size after sudden fileystem
shutdown once the directio write is completed. ext4 directio after iomap
conversion was broken in the sense that if the FS crashes after
synchronous directio write, it's file size is not properly updated.
This test adds a testcase to cover such scenario.

Man page of open says that -
O_SYNC provides synchronized I/O file integrity completion, meaning write
operations will flush data and all associated metadata to the underlying
hardware

Reported-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Signed-off-by: "Ritesh Harjani (IBM)" <ritesh.list@gmail.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agoaio-dio-write-verify: Add sync and noverify option
Ritesh Harjani (IBM) [Mon, 20 Nov 2023 11:19:33 +0000 (16:49 +0530)]
aio-dio-write-verify: Add sync and noverify option

This patch adds -S for O_SYNC and -N for noverify option to
aio-dio-write-verify test. We will use this for integrity
verification test for aio-dio.

Signed-off-by: "Ritesh Harjani (IBM)" <ritesh.list@gmail.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agogeneric: test reading a large directory while renaming its files
Filipe Manana [Tue, 5 Dec 2023 16:39:20 +0000 (16:39 +0000)]
generic: test reading a large directory while renaming its files

Test that on a fairly large directory if we keep renaming files while
holding the directory open and doing readdir(3) calls, we don't end up
in an infinite loop.

This exercise a bug that existed in btrfs and was fixed in kernel 6.5
by commit 9b378f6ad48c ("btrfs: fix infinite directory reads").

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agogeneric: regression test of ext4_lblk_t overflow
Baokun Li [Fri, 24 Nov 2023 11:53:56 +0000 (19:53 +0800)]
generic: regression test of ext4_lblk_t overflow

Append writes to a file with logical block numbers close to 0xffffffff and
observe if a kernel crash is caused by ext4_lblk_t overflow triggering
BUG_ON at ext4_mb_new_inode_pa(). This is a regression test for commit
bc056e7163ac ("ext4: fix BUG in ext4_mb_new_inode_pa() due to overflow")

Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agobtrfs/301: require_no_compress
Boris Burkov [Mon, 20 Nov 2023 21:10:55 +0000 (13:10 -0800)]
btrfs/301: require_no_compress

btrfs/301 makes detailed size calculations to test squota edge cases
which rely on assumptions that break down with compression enabled.

Fix it by disabling the test with compression. Compression + squotas
still gets quite solid test coverage via squotas support in fsck and
normal compression enabled fstests runs.

Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agobtrfs/301: fix hardcoded subvolids
Boris Burkov [Mon, 20 Nov 2023 21:10:54 +0000 (13:10 -0800)]
btrfs/301: fix hardcoded subvolids

Hardcoded subvolids break test runs with no free-space-tree, so change
the test to use _btrfs_get_subvolid instead of assuming 256, 257, etc...

Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agooverlay/026: Fix test expectation for newer kernels
Alexander Larsson [Wed, 22 Nov 2023 15:20:13 +0000 (17:20 +0200)]
overlay/026: Fix test expectation for newer kernels

The test checks the expectaion from old kernels that set/get of
trusted.overlay.* xattrs is not supported on an overlayfs filesystem.

New kernels support set/get xattr of trusted.overlay.* xattrs, so adapt
the test to check that either both set and get work on new kernel, or
neither work on old kernel.

Signed-off-by: Alexander Larsson <alexl@redhat.com>
Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agobtrfs/048: add missing groups 'subvol', 'snapshot' and 'send'
Filipe Manana [Mon, 4 Dec 2023 15:45:11 +0000 (15:45 +0000)]
btrfs/048: add missing groups 'subvol', 'snapshot' and 'send'

This test case exercises subvolumes, snapshot and send/receive, so add
the corresponding groups to the test.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agobtrfs: add some tests to the 'compress' group
Filipe Manana [Mon, 4 Dec 2023 15:45:10 +0000 (15:45 +0000)]
btrfs: add some tests to the 'compress' group

There are several btrfs test that exercise compression in one way or
another but are not listed as part of the 'compress' group, so add them
to that group.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agogeneric/733: use correct kernel commit id
Yang Xu [Mon, 4 Dec 2023 11:34:28 +0000 (06:34 -0500)]
generic/733: use correct kernel commit id

Signed-off-by: Yang Xu <xuyang2018.jy@fujitsu.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agobtrfs/303: use quota rescan wrapper
Boris Burkov [Fri, 1 Dec 2023 21:09:45 +0000 (13:09 -0800)]
btrfs/303: use quota rescan wrapper

This new test called quota rescan directly rather than with the new
wrapper. As a result, it failed with -O squota in MKFS_OPTIONS. Using
the wrapper, it skips the rescan and passes again.

Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agogeneric/459: improve shutdown/read-only check to accommodate bcachefs
Brian Foster [Wed, 29 Nov 2023 20:16:14 +0000 (15:16 -0500)]
generic/459: improve shutdown/read-only check to accommodate bcachefs

generic/459 occasionally fails on bcachefs because the deliberately
induced I/O errors caused by exhausting the overprovisioned thin
pool can lead to filesystem shutdown. This test considers this
expected behavior on certain fs', but only checks for the ext4
remount read-only behavior. bcachefs does a similar emergency
read-only transition in response to certain I/O errors, but it
behaves more similar to an XFS shutdown and doesn't necessarily
reflect "ro" state in the mount table (unless induced by userspace).

Since the test already runs a touch command to help trigger the ext4
error handling sequence, this can be tweaked to serve double duty
and also more accurately detect read-only status on bcachefs.
Refactor into a small helper, check for touch command failure, and
consider the fs read-only if either that or the mount entry check
indicates it.

Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Bill O'Donnell <bodonnel@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agofsstress: unify verbose output format
Shiyang Ruan [Thu, 23 Nov 2023 08:58:43 +0000 (16:58 +0800)]
fsstress: unify verbose output format

Wrap offset, length within '[]' and remove meanless 't' in
do_fallocate().

Signed-off-by: Shiyang Ruan <ruansy.fnst@fujitsu.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agofstests: don't test -o norecovery in btrfs/220
Josef Bacik [Wed, 22 Nov 2023 17:15:58 +0000 (12:15 -0500)]
fstests: don't test -o norecovery in btrfs/220

This is a deprecated option and it's going away with the new mount api
patches, so remove this from the test.

Signed-off-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agotests/btrfs: add tests to the remount group
Anand Jain [Wed, 22 Nov 2023 08:55:44 +0000 (16:55 +0800)]
tests/btrfs: add tests to the remount group

Several test cases under tests/btrfs are missing from the remount
group. This patch adds the test cases that use -o remount to the remount
group.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agoxfs/604: add missing falloc test
Darrick J. Wong [Mon, 20 Nov 2023 19:35:20 +0000 (11:35 -0800)]
xfs/604: add missing falloc test

This test requires both the xfs_io falloc and fpunch commands to run.
falloc isn't supported on alwayscow=1 mounts, which means this test
fails there.

While we're at it, update the commit id since the fix was committed to
6.7-rc2.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
19 months agoxfs/601: move this to tests/generic
Darrick J. Wong [Mon, 20 Nov 2023 19:35:14 +0000 (11:35 -0800)]
xfs/601: move this to tests/generic

As per last week's discussion, xfs/601 doesn't have any xfs-specific
functionality in it.  Turn it into a generic test.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agoxfs: test refilling AGFL after lots of btree splits
Omar Sandoval [Mon, 30 Oct 2023 21:00:15 +0000 (14:00 -0700)]
xfs: test refilling AGFL after lots of btree splits

This is a regression test for patch "xfs: fix internal error from AGFL
exhaustion"), which is not yet merged. Without the fix, it will fail
with a "Structure needs cleaning" error.

Signed-off-by: Omar Sandoval <osandov@osandov.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agomisc: update xfs_io swapext usage
Darrick J. Wong [Thu, 16 Nov 2023 17:30:43 +0000 (09:30 -0800)]
misc: update xfs_io swapext usage

Since the new 'exchange range' functionality is no longer a VFS level
concept, the xfs_io swapext -v options have changed.  Update fstests to
reflect this new reality.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agomisc: privatize the FIEXCHANGE ioctl for now
Darrick J. Wong [Thu, 16 Nov 2023 17:30:37 +0000 (09:30 -0800)]
misc: privatize the FIEXCHANGE ioctl for now

I'm abandoning (for now) efforts to bring atomic file content exchanges
to the VFS.  The goal here is to reduce friction in getting online fsck
merged, so Dave and I want to take this back to being a private XFS
ioctl so we can explore with it for a while before committing it to the
stable KABI.

Shift all the existing FIEXCHANGE usage to XFS_IOC_EXCHANGE_RANGE, and
try to pick it up from xfs_fs_staging.h if the system xfslibs-dev
package has such an animal.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agoxfs/263: Remove mkfs.xfs data section argument
Yang Xu [Thu, 16 Nov 2023 07:33:15 +0000 (02:33 -0500)]
xfs/263: Remove mkfs.xfs data section argument

On machine with using raid, this case will trigger
the following error:
==== NO CRC ====
+mkfs.xfs: small data volume, ignoring data volume stripe unit 512 and stripe width 512
== Options: rw ==
== Options: usrquota,rw ==

mkfs.xfs generates this error since xfsprogs commit 42371fb36
("mkfs: ignore data blockdev stripe geometry for small filesystems").
It disables automatic detection of stripe unit and width if the
data device is less than 1GB.

But, since xfstests commit baaa392c("xfs/263: don't hardcode inode numbers in output"),
we don't care the inode number. So let's remove this data section size argument
instead of increasing this size to 1G, then we can solve this false
positive.

Signed-off-by: Yang Xu <xuyang2018.jy@fujitsu.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agofstests: btrfs: test snapshot creation with existing qgroup
Qu Wenruo [Tue, 14 Nov 2023 02:59:13 +0000 (13:29 +1030)]
fstests: btrfs: test snapshot creation with existing qgroup

[BUG]
There is a sysbot regression report about transaction abort during
snapshot creation, which is caused by the new timing of qgroup creation
and too strict error check.

[FIX]
The proper fix is already submitted, with the title "btrfs: do not abort
transaction if there is already an existing qgroup".

[TEST]
The new test case would reproduce the regression by:

- Create a subvolume and a snapshot of it

- Record the subvolumeid of the snapshot

- Re-create the fs
  Since btrfs won't reuse the subvolume id, we have to re-create the fs.

- Enable quota and create a qgroup with the same subvolumeid

- Create a subvolume and a snapshot of it
  For unpatched and affected kernel (thankfully no release is affected),
  the snapshot creation would fail due to aborted transaction.

- Make sure the subvolume id doesn't change for the snapshot
  There is one very hacky attempt to fix it by avoiding using the
  subvolume id, which is completely wrong and would be caught by this
  extra check.

Signed-off-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agoxfs: test unlinked inode list repair on demand
Darrick J. Wong [Mon, 13 Nov 2023 17:08:30 +0000 (09:08 -0800)]
xfs: test unlinked inode list repair on demand

Create a test to exercise recovery of unlinked inodes on a clean
filesystem.  This was definitely possible on old kernels that on an ro
mount would clean the log without processing the iunlink list.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agocommon: make helpers for ttyprintk usage
Darrick J. Wong [Mon, 13 Nov 2023 17:08:24 +0000 (09:08 -0800)]
common: make helpers for ttyprintk usage

A handful of tests write things to /dev/ttyprintk to make it easier to
pinpoint where in a test something went wrong.  This isn't entirely
robust, however, because ttyprintk is an optional feature.  In the grand
tradition of kernel design there's also a /dev/kmsg that does nearly the
same thing, is also optional, and there's no documentation spelling out
when one is supposed to use one or the other.

So.

Create a pair of helpers to append messages to the kernel log.  One
simply writes its arguments to the kernel log, and the other writes
stdin to the kernel log, stdout, and any other files specified as
arguments.

Underneath the covers, both functions will send the message to
/dev/ttyprintk if available.  If it isn't but /dev/kmsg is, they'll
send the messages there, prepending a "[U]" to emulate the only
discernable difference between ttyprintk and kmsg.

If neither are available, then either /dev or the kernel aren't allowing
us to write to the kernel log, and the messages are not logged.  The
second helper will still write the messages to stdout.

If this seems like overengineered nonsense, then yes it is.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agogeneric: test reads racing with slow reflink operations
Darrick J. Wong [Mon, 13 Nov 2023 17:08:39 +0000 (09:08 -0800)]
generic: test reads racing with slow reflink operations

XFS has a rather slow reflink operation.  While a reflink operation is
running, other programs cannot read the contents of the source file,
which is causing latency spikes.  Catherine Hoang wrote a patch to
permit reads, since the source file contents do not change.  This is a
functionality test for that patch.

[zlang@ adds `_supported_fs generic`]

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Catherine Hoang <catherine.hoang@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agocommon/btrfs: add _btrfs_get_fsid() helper
Anand Jain [Wed, 8 Nov 2023 04:28:57 +0000 (12:28 +0800)]
common/btrfs: add _btrfs_get_fsid() helper

We have two instances of reading the btrfs fsid by using the command
'btrfs filesystem show <mnt>' turn this into an easy-to-use helper
function and also use it.

Suggested-by: David Sterba <dsterba@suse.cz>
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agobtrfs/219: add to the auto group
Anand Jain [Thu, 2 Nov 2023 11:28:22 +0000 (19:28 +0800)]
btrfs/219: add to the auto group

Add this test case back to the auto group which reverts the
commit e2e7b549380a ("fstests: btrfs/219: remove it from auto group") since
the previously missing kernel commit 5f58d783fd78 ("btrfs: free device in
btrfs_close_devices for a single device filesystem") has already been
integrated.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agobtrfs/219: cloned-device mount capability update
Anand Jain [Thu, 2 Nov 2023 11:28:21 +0000 (19:28 +0800)]
btrfs/219: cloned-device mount capability update

This test case checks for failure of the cloned device mounts, which
is no longer true after the commit a5b8a5f9f835 ("btrfs: support
cloned-device mount capability"). So check for the non-presence the
temp-fsid feature and do not test for the failure of the cloned device
mount.

Reported-by: kernel test robot <oliver.sang@intel.com>
Closes: https://lore.kernel.org/oe-lkp/202310251645.5fe5495a-oliver.sang@intel.com
Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agobtrfs/219: fix _cleanup() to successful release the loop-device
Anand Jain [Thu, 2 Nov 2023 11:28:20 +0000 (19:28 +0800)]
btrfs/219: fix _cleanup() to successful release the loop-device

When we fail with the message 'We were allowed to mount when we should
have failed,' it will fail to clean up the loop devices, making it
difficult to run further test cases or the same test case again.

So we need a 2nd loop device local variable to release it. Let's
reorganize the local variables to clean them up in the _cleanup() function.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agocommon/btrfs: add helper _has_btrfs_sysfs_feature_attr
Anand Jain [Thu, 2 Nov 2023 11:28:19 +0000 (19:28 +0800)]
common/btrfs: add helper _has_btrfs_sysfs_feature_attr

With this helper, btrfs test cases can now check if a particular feature
is implemented in the kernel.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agocommon/rc: _fs_sysfs_dname fetch fsid using btrfs tool
Anand Jain [Thu, 2 Nov 2023 11:28:18 +0000 (19:28 +0800)]
common/rc: _fs_sysfs_dname fetch fsid using btrfs tool

Currently _fs_sysfs_dname gets fsid from the findmnt command however
this command provides the metadata_uuid if the device is mounted with
temp-fsid. So instead, use btrfs filesystem show command to know the fsid.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agoxfs/146: require fallocate support
Christoph Hellwig [Thu, 9 Nov 2023 16:49:47 +0000 (17:49 +0100)]
xfs/146: require fallocate support

xfs/146 uses the xfs_io falloc subcommand and thus kernel fallocate
support.  Ensure the support is present.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agofstests: generic/353 should accomodate other pwrite behaviors
Bill O'Donnell [Fri, 1 Sep 2023 16:18:16 +0000 (11:18 -0500)]
fstests: generic/353 should accomodate other pwrite behaviors

xfs_io pwrite issues a series of block size writes, but there is no
guarantee that the resulting extent(s) will be singular or contiguous.
This behavior is acceptable, but the test is flawed in that it expects
a single extent for a pwrite.

Modify test to use actual blocksize for pwrite and reflink. Also
modify it to accommodate pwrite and reflink that produce different
mapping results.

Signed-off-by: Bill O'Donnell <bodonnel@redhat.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agobtrfs: test snapshotting a subvolume that was just created
Filipe Manana [Tue, 24 Oct 2023 11:23:46 +0000 (12:23 +0100)]
btrfs: test snapshotting a subvolume that was just created

Test that snapshotting a new subvolume (created in the current transaction)
that has a btree with a height > 1, works and does not result in a fs
corruption.

This exercises a regression introduced in kernel 6.5 by the kernel commit:

  1b53e51a4a8f ("btrfs: don't commit transaction for every subvol create")

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Josef Bacik <josef@toxicpanda.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agooverlay: add test for lowerdir mount option parsing
Amir Goldstein [Mon, 23 Oct 2023 16:32:59 +0000 (19:32 +0300)]
overlay: add test for lowerdir mount option parsing

Check parsing and display of spaces and escaped colons and commans in
lowerdir mount option.

This is a regression test for two bugs introduced in v6.5 with the
conversion to new mount api.

There is another regression of new mount api related to libmount parsing
of escaped commas, but this needs a fix in libmount - this test only
verifies the fixes in the kernel, so it uses LIBMOUNT_FORCE_MOUNT2=always
to force mount(2) and kernel pasring of the comma separated options list.

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agogeneric/251: check min and max length and minlen for FSTRIM
Darrick J. Wong [Thu, 26 Oct 2023 03:21:51 +0000 (20:21 -0700)]
generic/251: check min and max length and minlen for FSTRIM

Every now and then, this test fails with the following output when
running against my development tree when configured with an 8k fs block
size:

  --- a/tests/generic/251.out 2023-07-11 12:18:21.624971186 -0700
  +++ b/tests/generic/251.out.bad 2023-10-15 20:54:44.636000000 -0700
  @@ -1,2 +1,4677 @@
   QA output created by 251
   Running the test: done.
  +fstrim: /opt: FITRIM ioctl failed: Invalid argument
  +fstrim: /opt: FITRIM ioctl failed: Invalid argument
  ...
  +fstrim: /opt: FITRIM ioctl failed: Invalid argument

Dumping the exact fstrim command lines to seqres.full produces this at
the end:

  /usr/sbin/fstrim -m 32544k -o 30247k -l 4k /opt
  /usr/sbin/fstrim -m 32544k -o 30251k -l 4k /opt
  ...
  /usr/sbin/fstrim -m 32544k -o 30255k -l 4k /opt

The count of failure messages is the same as the count as the "-l 4k"
fstrim invocations.  Since this is an 8k-block filesystem, the -l
parameter is clearly incorrect.  The test computes random -m and -l
options.

Therefore, create helper functions to guess at the minimum and maximum
length and minlen parameters that can be used with the fstrim program.
In the inner loop of the test, make sure that our choices for -m and -l
fall within those constraints.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
20 months agogeneric/251: don't snapshot $here during a test
Darrick J. Wong [Thu, 26 Oct 2023 03:12:02 +0000 (20:12 -0700)]
generic/251: don't snapshot $here during a test

Zorro complained that the next patch caused him a regression:

generic/251 249s ... [failed, exit status 1]- output mismatch (see /root/git/xfstests/results//generic/251.out.bad)
    --- tests/generic/251.out   2022-04-29 23:07:23.263498297 +0800
    +++ /root/git/xfstests/results//generic/251.out.bad 2023-10-22 14:17:07.248059405 +0800
    @@ -1,2 +1,5 @@
     QA output created by 251
     Running the test: done.
    +5838a5839
    +> aa60581221897d3d7dd60458e1cca2fa  ./results/generic/251.full
    +!!!Checksums has changed - Filesystem possibly corrupted!!!\n
    ...
    (Run 'diff -u /root/git/xfstests/tests/generic/251.out /root/git/xfstests/results//generic/251.out.bad'  to see the entire diff)
Ran: generic/251
Failures: generic/251
Failed 1 of 1 tests

The next patch writes some debugging information into $seqres.full,
which is a file underneat $RESULT_BASE.  If the test operator does not
set RESULT_BASE, it will be set to a subdir of $here by default.  Since
this test also snapshots the contents of $here before starting its loop,
any logging to $seqres.full on such a system will cause the post-copy
checksum to fail due to a mismatch.

Fix all this by copying $here to $SCRATCH_DEV and checksumming the copy
before the FITRIM stress test begins to avoid problems with $seqres.full.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs/298: fix failure when added device supports trim
Filipe Manana [Thu, 19 Oct 2023 10:06:14 +0000 (11:06 +0100)]
btrfs/298: fix failure when added device supports trim

A btrfs device add command issues a trim on the device if the device
supports trim, and then it outputs a message to stdout informing that it
performed a trim. If that happens it breaks the golden output and the
test fails like this:

   $ ./check btrfs/298
   FSTYP         -- btrfs
   PLATFORM      -- Linux/x86_64 debian0 6.6.0-rc3-btrfs-next-139+ #1 SMP PREEMPT_DYNAMIC Tue Oct  3 13:52:02 WEST 2023
   MKFS_OPTIONS  -- /dev/sdc
   MOUNT_OPTIONS -- /dev/sdc /home/fdmanana/btrfs-tests/scratch_1

   btrfs/298       - output mismatch (see /home/fdmanana/git/hub/xfstests/results//btrfs/298.out.bad)
       --- tests/btrfs/298.out 2023-10-18 23:29:06.029292800 +0100
       +++ /home/fdmanana/git/hub/xfstests/results//btrfs/298.out.bad 2023-10-19 10:54:29.693210881 +0100
       @@ -1,2 +1,3 @@
        QA output created by 298
       +Performing full device TRIM /dev/sdd (100.00GiB) ...
        Silence is golden
       ...
       (Run 'diff -u /home/fdmanana/git/hub/xfstests/tests/btrfs/298.out /home/fdmanana/git/hub/xfstests/results//btrfs/298.out.bad'  to see the entire diff)
   Ran: btrfs/298
   Failures: btrfs/298
   Failed 1 of 1 tests

Fix this by redirecting the device add's stdout to the $seqres.full file.
Any device add errors are sent to stderr, so we'll notice if errors happen
due to possible future regressions, as it will break the golden output.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agogeneric/245: Filter mv error message
Su Yue [Thu, 19 Oct 2023 01:53:56 +0000 (09:53 +0800)]
generic/245: Filter mv error message

Coreutils commit 3cb862ce5f10 ( mv: better diagnostic for 'mv dir x' failure)
was released in v9.4, changed the error message from
'mv: cannot move 'b/t' to 'a/t': Directory not empty' to
'mv: cannot overwrite 'a/t': Directory not empty' in case of
EDQUOT/EEXIST/EISDIR/EMLINK/ENOSPC/ENOTEMPTY/ETXTBSY.

The change breaks generic/245 due to the mismatched output:

generic/245 1s ... - output mismatch (see /root/xfstests-dev/results//generic/245.out.bad)
    --- tests/generic/245.out   2023-10-05 11:15:21.124295738 +0800
    +++ /root/xfstests-dev/results//generic/245.out.bad 2023-10-05 11:15:23.456315468 +0800
    @@ -1,2 +1,2 @@
    QA output created by 245
    -mv: cannot move 'TEST_DIR/test-mv/ab/aa/' to 'TEST_DIR/test-mv/aa': File exists
    +mv: cannot overwrite 'TEST_DIR/test-mv/aa': File exists
    ...
    (Run 'diff -u /root/xfstests-dev/tests/generic/245.out /root/xfstests-dev/results//generic/245.out.bad'  to see the entire diff)

Filter out and replace mv error messages to fix the test.

Signed-off-by: Su Yue <glass.su@suse.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agoxfs: add missing _require_scratch calls
Christoph Hellwig [Mon, 16 Oct 2023 13:11:03 +0000 (15:11 +0200)]
xfs: add missing _require_scratch calls

Add _require_scratch to a bunch of test that were using $SCRATCH_DEV
without that check.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agoxfs/556: call _require_dm_target later
Christoph Hellwig [Mon, 16 Oct 2023 13:11:02 +0000 (15:11 +0200)]
xfs/556: call _require_dm_target later

_require_dm_target tries to use $SCRATCH_DEV, so move it after we've
established that the configuration has a valid $SCRATCH_DEV.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agogeneric: test FALLOC_FL_UNSHARE when pagecache is not loaded
Darrick J. Wong [Mon, 9 Oct 2023 18:19:03 +0000 (11:19 -0700)]
generic: test FALLOC_FL_UNSHARE when pagecache is not loaded

Add a regression test for funsharing uncached files to ensure that we
actually manage the pagecache state correctly.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: "Ritesh Harjani (IBM)" <ritesh.list@gmail.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agogeneric/269,xfs/051: don't drop fsstress failures to stdout
Darrick J. Wong [Mon, 9 Oct 2023 18:18:45 +0000 (11:18 -0700)]
generic/269,xfs/051: don't drop fsstress failures to stdout

Prior to commit f55e46d629, these two tests would run fsstress until it
hit a failure -- ENOSPC in the case of generic/269, and EIO in the case
of xfs/051.  These errors are expected, which was why stderr was also
redirected to /dev/null.  Commit f55e46d629 removed the stderr
redirection, which has resulted in a 100% failure rate.

Fix this regression by pushing stderr stream to $seqres.full.

Fixes: f55e46d629 ("fstests: redirect fsstress' stdout to $seqres.full instead of /dev/null")
Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agogeneric/465: only complain about stale disk contents when racing directio
Darrick J. Wong [Mon, 9 Oct 2023 18:18:39 +0000 (11:18 -0700)]
generic/465: only complain about stale disk contents when racing directio

This test does a strange thing with directio -- it races a reader thread
with an appending aio writer thread and checks that the reader thread
only ever sees a (probably short) buffer containing the same contents
that are being read.

However, this has never worked correctly on XFS, which supports
concurrent readers and writers for directio.  Say you already have a
file with a single written mapping A:

AAAAAAAAAA
0        EOF

Then one thread initiates an aligned appending write:

AAAAAAAAAA---------
0        EOF      new_EOF

However, the free space is fragmented, so the file range maps to
multiple extents b and c (lowercase means unwritten here):

AAAAAAAAAAbbbbccccc
0        EOF      new_EOF

This implies separate bios for b and c.  Both bios are issued, but c
completes first.  The ioend for c will extend i_size all the way to
new_EOF.  Extent b is still marked unwritten because it hasn't completed
yet.

Next, the test reader slips in and tries to read the range between the
old EOF and the new EOF.  The file looks like this now:

AAAAAAAAAAbbbbCCCCC
0        EOF      new_EOF

So the reader sees "bbbbCCCCC" in the mapping, and the buffer returned
contains a range of zeroes followed by whatever was written to C.

For pagecache IO I would say that i_size should not be extended until
the extending write is fully complete, but the pagecache also
coordinates access so that reads and writes cannot conflict.

However, this is directio.  Reads and writes to the storage device can
be issued and acknowledged in any order.  I asked Ted and Jan about this
point, and they echoed that for directio it's expected that application
software must coordinate access themselves.

In other words, the only thing that the reader can check here is that
the filesystem is not returning stale disk contents.  Amend the test so
that null bytes in the reader buffer are acceptable.

Cc: tytso@mit.edu
Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agoxfs/178: don't fail when SCRATCH_DEV contains random xfs superblocks
Darrick J. Wong [Thu, 12 Oct 2023 15:09:22 +0000 (08:09 -0700)]
xfs/178: don't fail when SCRATCH_DEV contains random xfs superblocks

When I added an fstests config for "RAID" striping (aka MKFS_OPTIONS='-d
su=128k,sw=4'), I suddenly started seeing this test fail sporadically
with:

  --- /tmp/fstests/tests/xfs/178.out 2023-07-11 12:18:21.714970364 -0700
  +++ /var/tmp/fstests/xfs/178.out.bad 2023-07-25 22:05:39.756000000 -0700
  @@ -10,6 +10,20 @@ bad primary superblock - bad magic numbe

   attempting to find secondary superblock...
   found candidate secondary superblock...
  +unable to verify superblock, continuing...
  +found candidate secondary superblock...
  +error reading superblock 1 -- seek to offset 584115421184 failed
  +unable to verify superblock, continuing...
  +found candidate secondary superblock...
  +error reading superblock 1 -- seek to offset 584115421184 failed
  +unable to verify superblock, continuing...
  +found candidate secondary superblock...
  +error reading superblock 1 -- seek to offset 584115421184 failed
  +unable to verify superblock, continuing...
  +found candidate secondary superblock...
  +error reading superblock 1 -- seek to offset 584115421184 failed
  +unable to verify superblock, continuing...
  +found candidate secondary superblock...
  +error reading superblock 1 -- seek to offset 584115421184 failed
  +unable to verify superblock, continuing...
  +found candidate secondary superblock...
  +error reading superblock 1 -- seek to offset 584115421184 failed
  +unable to verify superblock, continuing...
  +found candidate secondary superblock...
   verified secondary superblock...
   writing modified primary superblock
   sb root inode INO inconsistent with calculated value INO

Eventually I tracked this down to a mis-interaction between the test,
xfs_repair, and the storage device.

If the device doesn't support discard, _scratch_mkfs won't zero the
entire disk to remove old dead superblocks that might have been written
by previous tests.  After we shatter the primary super, the xfs_repair
scanning code can still trip over those old supers when it goes looking
for secondary supers.

Most of the time it finds the actual AG 1 secondary super, but sometimes
it finds ghosts from previous formats.  When that happens, xfs_repair
will talk quite a bit about those failed secondaries, even if it
eventually finds an acceptable secondary sb and completes the repair.

Filter out the messages about secondary supers.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agoREADME: Update overlayfs instructions
Vyacheslav Yurkov [Sun, 1 Oct 2023 00:57:10 +0000 (02:57 +0200)]
README: Update overlayfs instructions

Overlayfs-tools and overlayfs-progs projects have been merged together.

Signed-off-by: Vyacheslav Yurkov <uvv.mail@gmail.com>
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agoxfs/{270,557,600}: update commit id for _fixed_by tag.
Darrick J. Wong [Fri, 29 Sep 2023 17:28:01 +0000 (10:28 -0700)]
xfs/{270,557,600}: update commit id for _fixed_by tag.

Update the commit id in the _fixed_by tag now that we've merged the
kernel fixes.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agocommon/rc: check error case and fail the test
Naohiro Aota [Wed, 27 Sep 2023 06:11:00 +0000 (15:11 +0900)]
common/rc: check error case and fail the test

If we place /var/lib/xfstests on a read-only filesystem, commands in
_link_out_file_named() fail to modify the files. However, they won't fail
the test. As a result, the test case fails mysteriously with only "no
qualified output" printed.

Fix it by checking the error case.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agofsx: Add a return type to aio_rw
Khem Raj [Wed, 27 Sep 2023 13:16:17 +0000 (15:16 +0200)]
fsx: Add a return type to aio_rw

Compilers complain about the function prototype otherwise

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Vyacheslav Yurkov <uvv.mail@gmail.com>
Cc: "Darrick J. Wong" <djwong@kernel.org>
Cc: Zorro Lang <zlang@redhat.com>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs/300: check existence of unshare arguments
Darrick J. Wong [Wed, 27 Sep 2023 01:42:58 +0000 (18:42 -0700)]
btrfs/300: check existence of unshare arguments

Make sure the installed unshare binary supports all the arguments that
it wants to use.  The unshare program on my system (Ubuntu 22.04)
doesn't support --map-auto, so this test fails unnecessarily.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agoxfs/018: make sure that larp mode actually works
Darrick J. Wong [Mon, 25 Sep 2023 21:42:46 +0000 (14:42 -0700)]
xfs/018: make sure that larp mode actually works

Skip this test if larp mode doesn't work.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months ago_scratch_mkfs_geom: Fix regex used for matching block size option
Chandan Babu R [Mon, 25 Sep 2023 13:48:05 +0000 (19:18 +0530)]
_scratch_mkfs_geom: Fix regex used for matching block size option

The regular expression used by _scratch_mkfs_geom() to match mkfs.xfs' block
size argument interprets the character 'b' as optional. It should actually
interpret whitespace as optional.

This causes generic/223 to fail when testing an XFS filesystem which uses an
external log device along with the -lsize option. In this case, the original
value of -lsize is replaced with the value of $blocksize.

_scratch_mkfs_sized() also uses the same incorrect regex.

Signed-off-by: Chandan Babu R <chandanbabu@kernel.org>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agogeneric: add a test to check move in mountpoints of the same export
Yongcheng Yang [Thu, 21 Sep 2023 13:43:47 +0000 (21:43 +0800)]
generic: add a test to check move in mountpoints of the same export

Add a new test to ckeck file move (rename) operation among
different mount points which are mounting to a same export.

This should be a simple test but it recently unveils an ancient
nfsd bug. Thus let's make it to be a regresstion check.

Signed-off-by: Yongcheng Yang <yoyang@redhat.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs/295: skip on zoned device as we cannot corrupt it directly
Naohiro Aota [Tue, 26 Sep 2023 14:11:47 +0000 (23:11 +0900)]
btrfs/295: skip on zoned device as we cannot corrupt it directly

We use _pwrite_byte to corrupt the root node, but such overwrite won't work
on a sequential write required zone. So, skip the test on a zoned device.

Technically, we can run this test case by checking if the physical location
lands in a conventional zone. But, the logic should be no difference than
the regular mode and I don't think it's worth doing so.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs: test scan but not register the single device fs
Anand Jain [Thu, 28 Sep 2023 09:01:30 +0000 (17:01 +0800)]
btrfs: test scan but not register the single device fs

Recently, in the kernel commit 0d9436739af2 ("btrfs: scan but don't
register device on single device filesystem"), we adopted an approach
where we scan the device to validate it. However, we do not register
it in the kernel memory since it is not required to be remembered.

However, the seed device should continue to be registered because
otherwise, the mount operation for the sprout device will fail.

This patch ensures that we honor the mount requirements and do not break
anything while making changes in this part of the code.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs/192: use append operator to output log replay results to $seqres.full
Filipe Manana [Tue, 3 Oct 2023 11:57:45 +0000 (12:57 +0100)]
btrfs/192: use append operator to output log replay results to $seqres.full

After doing log replay, btrfs/192 is overwriting the $seqres.full file
because it uses the plain ">" redirect operator, instead of an append
">>" redirect operator. As a consequence it is overriding the file and
eliminating any previous output that may be useful to debug a test
failure (such as the fsstress seed or mkfs results). So use >> instead
of >.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agofstests: redirect fsstress' stdout to $seqres.full instead of /dev/null
Filipe Manana [Tue, 3 Oct 2023 11:57:44 +0000 (12:57 +0100)]
fstests: redirect fsstress' stdout to $seqres.full instead of /dev/null

Several tests are redirecting the output of fsstress to /dev/null and this
makes it harder to debug a test failure because we have no way of knowing
what was the seed used by fsstress, as fsstress outputs the seed it uses
to stdout. Very often when such a test fails, I have to go modify to
redirect stdout to the $seqres.full file and then run it in a loop until
I find a seed that causes a failure.

So modify all tests that redirect fsstress' output to /dev/null to instead
redirect it to the $seqres.full file. Note that for some tests I've added
the style ">> $seqres.full" (with a space after >>) while for others I did
">>$seqres.full" (no space) - the reason for this was to keep style
consistency within each test case.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Acked-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs/076: fix file_size variable
Naohiro Aota [Mon, 25 Sep 2023 04:33:59 +0000 (13:33 +0900)]
btrfs/076: fix file_size variable

The file size written below is 10 MB, but the variable is set to 1 MB. Fix
it, or the test will fail.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs/283: skip if we cannot write into one extent
Naohiro Aota [Mon, 25 Sep 2023 05:55:41 +0000 (14:55 +0900)]
btrfs/283: skip if we cannot write into one extent

On the zoned mode, the extent size is limited also by
queue/zone_append_max_bytes. This breaks the assumption that the file "foo"
has a single extent and corrupts the test output.

It is difficult to support the case, so let's just skip the test in this
case.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs: skip squota incompatible tests
Boris Burkov [Thu, 28 Sep 2023 23:16:48 +0000 (16:16 -0700)]
btrfs: skip squota incompatible tests

These tests cannot succeed if mkfs enable squotas, as they either test
the specifics of qgroups behavior or they test *enabling* squotas. Skip
these in squota mode.

Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs: use new rescan wrapper
Boris Burkov [Thu, 28 Sep 2023 23:16:47 +0000 (16:16 -0700)]
btrfs: use new rescan wrapper

These tests can pass in simple quota mode if we skip the rescans via the
wrapper.

Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs: quota rescan helpers
Boris Burkov [Thu, 28 Sep 2023 23:16:46 +0000 (16:16 -0700)]
btrfs: quota rescan helpers

Many btrfs tests explicitly trigger quota rescan. This is not a
meaningful operation for simple quotas, so we wrap it in a helper that
doesn't blow up quite so badly and lets us run those tests where the
rescan is a qgroup detail.

Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs/301: new test for simple quotas
Boris Burkov [Thu, 28 Sep 2023 23:16:45 +0000 (16:16 -0700)]
btrfs/301: new test for simple quotas

Test some interesting basic and edge cases of simple quotas.

To some extent, this is redundant with the alternate testing strategy of
using MKFS_OPTIONS to enable simple quotas, running the full suite and
relying on kernel warnings and fsck to surface issues.

Signed-off-by: Boris Burkov <boris@bur.io>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs: quota mode helpers
Boris Burkov [Thu, 28 Sep 2023 23:16:44 +0000 (16:16 -0700)]
btrfs: quota mode helpers

To facilitate skipping tests depending on the qgroup mode after mkfs,
add support for figuring out the mode. This cannot just rely on the new
sysfs file, since it might not be present on older kernels.

Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agocommon: refactor sysfs_attr functions
Boris Burkov [Thu, 28 Sep 2023 23:16:43 +0000 (16:16 -0700)]
common: refactor sysfs_attr functions

Expand the has/get/require functions to allow passing a dev by
parameter, and implement the test_dev specific one in terms of the new
generic one.

Signed-off-by: Boris Burkov <boris@bur.io>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs/287: filter snapshot IDs to avoid failures when using some features
Filipe Manana [Fri, 22 Sep 2023 11:45:17 +0000 (12:45 +0100)]
btrfs/287: filter snapshot IDs to avoid failures when using some features

When running btrfs/287 with features that create extra trees or don't
the need to create some trees, such as when using the free space tree
(default for several btrfs-progs releases now) versus when not using
it (by passing -R ^free-space-tree in MKFS_OPTIONS), the test can fail
because the IDs for the two snapshots it creates changes, and the golden
output is requiring the numeric IDs of the snapshots.

For example, when disabling the free space tree, the test fails like this:

  $ MKFS_OPTIONS="-R ^free-space-tree" ./check btrfs/287
  FSTYP         -- btrfs
  PLATFORM      -- Linux/x86_64 debian0 6.6.0-rc2-btrfs-next-138+ #1 SMP PREEMPT_DYNAMIC Thu Sep 21 17:58:48 WEST 2023
  MKFS_OPTIONS  -- -R ^free-space-tree /dev/sdc
  MOUNT_OPTIONS -- /dev/sdc /home/fdmanana/btrfs-tests/scratch_1

  btrfs/287 1s ... - output mismatch (see /home/fdmanana/git/hub/xfstests/results//btrfs/287.out.bad)
      --- tests/btrfs/287.out 2023-09-22 12:39:43.060761389 +0100
      +++ /home/fdmanana/git/hub/xfstests/results//btrfs/287.out.bad 2023-09-22 12:40:54.238849251 +0100
      @@ -44,52 +44,52 @@
       Create a readonly snapshot of 'SCRATCH_MNT' in 'SCRATCH_MNT/snap1'
       Create a readonly snapshot of 'SCRATCH_MNT' in 'SCRATCH_MNT/snap2'
       resolve first extent:
      -inode 257 offset 16777216 root 257
      -inode 257 offset 8388608 root 257
      -inode 257 offset 16777216 root 256
      -inode 257 offset 8388608 root 256
      ...
      (Run 'diff -u /home/fdmanana/git/hub/xfstests/tests/btrfs/287.out /home/fdmanana/git/hub/xfstests/results//btrfs/287.out.bad'  to see the entire diff)

  HINT: You _MAY_ be missing kernel fix:
        0cad8f14d70c btrfs: fix backref walking not returning all inode refs

  Ran: btrfs/287
  Failures: btrfs/287
  Failed 1 of 1 tests

So add a filter to logical reserve calls to replace snapshot root IDs with
a logical name (snap1 and snap2).

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
21 months agobtrfs: use full subcommand name at _btrfs_get_subvolid()
Filipe Manana [Fri, 22 Sep 2023 11:45:01 +0000 (12:45 +0100)]
btrfs: use full subcommand name at _btrfs_get_subvolid()

Avoid using the shortcut "sub" for the "subvolume" command, as this is the
standard practice because such shortcuts are not guaranteed to exist in
every btrfs-progs release (they may come and go). Also make the variables
local.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: David Disseldorp <ddiss@suse.de>
Reviewed-by: Qu Wenruo <wqu@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agobtrfs/259: fix output's wrong word
Naohiro Aota [Fri, 22 Sep 2023 00:02:49 +0000 (09:02 +0900)]
btrfs/259: fix output's wrong word

It prints "File extent layout before defrag" for the both outputs, but the
latter one should be "after defrag".

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric: test new directory entries are returned after rewinding directory
Filipe Manana [Thu, 21 Sep 2023 15:16:34 +0000 (16:16 +0100)]
generic: test new directory entries are returned after rewinding directory

Test that if names are added to a directory after an opendir(3) call and
before a rewinddir(3) call, future readdir(3) calls will return the names.
This is mandated by POSIX:

  https://pubs.opengroup.org/onlinepubs/007904875/functions/rewinddir.html

This exercises a regression in btrfs which is fixed by a kernel patch that
has the following subject:

  ""btrfs: refresh dir last index during a rewinddir(3) call""

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: David Disseldorp <ddiss@suse.de>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agobtrfs/239: call fsync to create tree-log dedicated block group for zoned mode
Naohiro Aota [Thu, 21 Sep 2023 09:41:58 +0000 (18:41 +0900)]
btrfs/239: call fsync to create tree-log dedicated block group for zoned mode

Running btrfs/239 on a zoned device often fails with the following error.

  btrfs/239 5s ... - output mismatch (see /host/btrfs/239.out.bad)
      --- tests/btrfs/239.out     2023-09-21 16:56:37.735204924 +0900
      +++ /host/btrfs/239.out.bad  2023-09-21 18:22:45.401433408 +0900
      @@ -1,4 +1,6 @@
       QA output created by 239
      +/testdir/dira still exists
      +/dira does not exists
       File SCRATCH_MNT/testdir/file1 data:
       0000000 ab ab ab ab ab ab ab ab ab ab ab ab ab ab ab ab
       *
      ...

This happens because "testdir" and "dira" are not logged on the first fsync
(fsync $SCRATCH_MNT/testdir), but are written as a full commit. That
prevents updating the log on "mv" time, leaving them pre-mv state.

The full commit is induced by the creation of a new block group. On the
zoned mode, we use a dedicated block group for tree-log. That block group
is created on-demand or assigned to a metadata block group if there is
none. On the first mount of a file system, we need to create one because
there is only one metadata block group available for the regular
metadata. That creation of a new block group forces tree-log to be a full
commit on that transaction, which prevents logging "testdir" and "dira".

Fix the issue by calling fsync before the first "sync", which creates the
dedicated block group and let the files be properly logged.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agobtrfs: add missing _fixed_by_kernel_commit for a few tests
Anand Jain [Thu, 21 Sep 2023 05:22:16 +0000 (13:22 +0800)]
btrfs: add missing _fixed_by_kernel_commit for a few tests

A few tests were still using the older style of mentioning the fix in the
comment section. This patch migrates them to using
_fixed_by_kernel_commit.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agooverlay: add test for rename of lower symlink with NOATIME attr
Amir Goldstein [Thu, 21 Sep 2023 14:31:02 +0000 (17:31 +0300)]
overlay: add test for rename of lower symlink with NOATIME attr

Test for a regression in copy up of symlink that has the S_NOATIME
inode flag.

This is a regression from v5.15 reported by Ruiwen Zhao:
https://lore.kernel.org/linux-unionfs/CAKd=y5Hpg7J2gxrFT02F94o=FM9QvGp=kcH1Grctx8HzFYvpiA@mail.gmail.com/

In the bug report, the symlink has the S_NOATIME inode flag because it is
on an NFS/FUSE filesystem that sets S_NOATIME for all inodes.

The reproducer uses another technique to create a symlink with
S_NOATIME inode flag by using chattr +A inheritance on filesystems
that inherit chattr flags to symlinks.

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agobtrfs: add missing commit ids for a few tests using _fixed_by_kernel_commit
Filipe Manana [Fri, 15 Sep 2023 09:26:50 +0000 (10:26 +0100)]
btrfs: add missing commit ids for a few tests using _fixed_by_kernel_commit

The tests btrfs/288, btrfs/289 and btrfs/300 are using the "xxxx..." stub
for commit ids, as when they were submitted/merged the corresponding
btrfs patches were not yet in Linus' tree. So replace the stubs with the
commit ids.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agobtrfs/076: use _fixed_by_kernel_commit to tell the fixing kernel commit
Naohiro Aota [Fri, 15 Sep 2023 07:25:11 +0000 (16:25 +0900)]
btrfs/076: use _fixed_by_kernel_commit to tell the fixing kernel commit

The fix commit is written in the comment without a commit hash. Use
_fixed_by_kernel_commit command to describe it.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agobtrfs/076: support smaller extent size limit
Naohiro Aota [Fri, 15 Sep 2023 07:25:10 +0000 (16:25 +0900)]
btrfs/076: support smaller extent size limit

Running btrfs/076 on a zoned null_blk device will fail with the following error.

  - output mismatch (see /host/results/btrfs/076.out.bad)
      --- tests/btrfs/076.out     2021-02-05 01:44:20.000000000 +0000
      +++ /host/results/btrfs/076.out.bad 2023-09-15 01:49:36.000000000 +0000
      @@ -1,3 +1,3 @@
       QA output created by 076
      -80
      -80
      +83
      +83
      ...

This is because the default value of zone_append_max_bytes is 127.5 KB
which is smaller than BTRFS_MAX_UNCOMPRESSED (128K). So, the extent size is
limited to 126976 (= ROUND_DOWN(127.5K, 4096)), which makes the number of
extents larger, and fails the test.

Instead of hard-coding the number of extents, we can calculate it using the
max extent size of an extent. It is limited by either
BTRFS_MAX_UNCOMPRESSED or zone_append_max_bytes.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agofstests: btrfs add more tests into the scrub group
Anand Jain [Tue, 29 Aug 2023 12:32:40 +0000 (20:32 +0800)]
fstests: btrfs add more tests into the scrub group

I wanted to verify tests using the command "btrfs scrub start" and
found that there are many more test cases using "btrfs scrub start"
than what is listed in the group.list file. So, get them to the scrub
group.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agocommon/rc: make _get_max_file_size find file size on mount point
Andrey Albershteyn [Mon, 11 Sep 2023 20:06:17 +0000 (22:06 +0200)]
common/rc: make _get_max_file_size find file size on mount point

Currently, _get_max_file_size finds max file size on $TEST_DIR.
The tests/generic/692 uses this function to detect file size and
then tries to create a file on $SCRATCH_MNT.

This works fine when test and scratch filesystems have the same
block size. However, it will fail if they differ.

Make _get_max_file_size accept mount point on which to detect max
file size.

Signed-off-by: Andrey Albershteyn <aalbersh@redhat.com>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agotools/mvtest: ensure testcase is executable (755)
Shiyang Ruan [Fri, 8 Sep 2023 05:43:35 +0000 (13:43 +0800)]
tools/mvtest: ensure testcase is executable (755)

Some test cases lack executable permission ('x'). Before running each
test case, `./check` checks and grants them 'x' permission. However,
this always leads to a dirty git repo. And the absence of 'x' permission
in test cases is often overlooked during reviews.

Since maintainers use mvtest to assign new case, add this change for
convenience of maintainers.

Signed-off-by: Shiyang Ruan <ruansy.fnst@fujitsu.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agofstests: btrfs/185 update for single device pseudo device-scan
Anand Jain [Wed, 6 Sep 2023 16:24:43 +0000 (00:24 +0800)]
fstests: btrfs/185 update for single device pseudo device-scan

As we are obliterating the need for the device scan for the single device,
which will return success if the basic superblock verification passes,
even for the duplicate device of the mounted filesystem, drop the check
for the return code in this testcase and continue to verify if the device
path of the mounted filesystem remains unaltered after the scan.

Also, if the test fails, it leaves the local non-standard mount point
remained mounted, leading to further test cases failing. Call unmount
in _cleanup().

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Boris Burkov <boris@bur.io>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agooverlay: add test for persistent unique fsid
Amir Goldstein [Sun, 3 Sep 2023 07:54:11 +0000 (10:54 +0300)]
overlay: add test for persistent unique fsid

Test overlayfs fsid behavior with new mount options uuid=null/on
that were introduced in kernel v6.6:

- Test inherited upper fs fsid with mount option uuid=off/null
- Test uuid=null behavior for existing overlayfs by default
- Test persistent unique fsid with mount option uuid=on
- Test uuid=on behavior for new overlayfs by default

Signed-off-by: Amir Goldstein <amir73il@gmail.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agoxfs/270: actually test log recovery with unknown rocompat features
Darrick J. Wong [Tue, 29 Aug 2023 23:09:59 +0000 (16:09 -0700)]
xfs/270: actually test log recovery with unknown rocompat features

Make sure that log recovery will not succeed if there are unknown
rocompat features in the superblock and the log is dirty.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agoxfs/270: actually test file readability
Darrick J. Wong [Tue, 29 Aug 2023 23:09:53 +0000 (16:09 -0700)]
xfs/270: actually test file readability

Make sure we can actually read files off the ro mounted filesystem that
has an unknown rocompat feature set.

Signed-off-by: "Darrick J. Wong" <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agofstests: btrfs/261 fix failure if /var/lib/btrfs isn't writable
Anand Jain [Tue, 29 Aug 2023 12:34:06 +0000 (20:34 +0800)]
fstests: btrfs/261 fix failure if /var/lib/btrfs isn't writable

We don't need scrub status; it is okay to ignore the warnings due to
the readonly /var/lib/btrfs if any. Redirect stderr to seqres.full.
We check the scrub return status.

    +WARNING: failed to open the progress status socket at /var/lib/btrfs/scrub.progress.42fad803-d505-48f4-a04d-612dbf8bd724: Read-only file system. Progress cannot be queried
    +WARNING: failed to write the progress status file: Read-only file system. Status recording disabled

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agofstests: use btrfs check repair for repairing btrfs filesystems
Anand Jain [Mon, 21 Aug 2023 09:05:09 +0000 (17:05 +0800)]
fstests: use btrfs check repair for repairing btrfs filesystems

There are two repair functions: _repair_scratch_fs() and
_repair_test_fs(). As the names suggest, these functions are designed to
repair the filesystems SCRATCH_DEV and TEST_DEV, respectively. However,
these functions never called proper comamnd for the filesystem type btrfs.
This patch fixes it. Thx.

Signed-off-by: Anand Jain <anand.jain@oracle.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agoxfs/559: adapt to kernels that use large folios for writes
Darrick J. Wong [Tue, 29 Aug 2023 23:03:49 +0000 (16:03 -0700)]
xfs/559: adapt to kernels that use large folios for writes

The write invalidation code in iomap can only be triggered for writes
that span multiple folios.  If the kernel reports a huge page size,
scale up the write size.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agocommon: rename get_page_size to _get_page_size
Darrick J. Wong [Tue, 29 Aug 2023 23:03:43 +0000 (16:03 -0700)]
common: rename get_page_size to _get_page_size

This function does not follow the naming convention that common helpers
must start with an underscore.  Fix this.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agocommon: split _get_hugepagesize into detection and actual query
Darrick J. Wong [Tue, 29 Aug 2023 23:03:37 +0000 (16:03 -0700)]
common: split _get_hugepagesize into detection and actual query

This helper has two parts -- querying the value, and _notrun'ing the
test if huge pages aren't turned on.  Break these into the usual
_require_hugepages and _get_hugepagesize predicates so that we can adapt
xfs/559 to large folios being used for writes.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agobtrfs/282: skip test if /var/lib/btrfs isnt writable
Darrick J. Wong [Thu, 24 Aug 2023 23:47:14 +0000 (16:47 -0700)]
btrfs/282: skip test if /var/lib/btrfs isnt writable

I run fstests in a readonly container, and accidentally uninstalled the
btrfsprogs package.  When I did, this test started faililng:

  --- btrfs/282.out
  +++ btrfs/282.out.bad
  @@ -1,3 +1,7 @@
   QA output created by 282
   wrote 2147483648/2147483648 bytes at offset 0
   XXX Bytes, X ops; XX:XX:XX.X (XXX YYY/sec and XXX ops/sec)
  +WARNING: cannot create scrub data file, mkdir /var/lib/btrfs failed: Read-only file system. Status recording disabled
  +WARNING: failed to open the progress status socket at /var/lib/btrfs/scrub.progress.3e1cf8c6-8f8f-4b51-982c-d6783b8b8825: No such file or directory. Progress cannot be queried
  +WARNING: cannot create scrub data file, mkdir /var/lib/btrfs failed: Read-only file system. Status recording disabled
  +WARNING: failed to open the progress status socket at /var/lib/btrfs/scrub.progress.3e1cf8c6-8f8f-4b51-982c-d6783b8b8825: No such file or directory. Progress cannot be queried

Skip the test if /var/lib/btrfs isn't writable, or if /var/lib isn't
writable, which means we cannot create /var/lib/btrfs.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Anand Jain <anand.jain@oracle.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric/187: don't run this test on NFS
Jeff Layton [Fri, 1 Sep 2023 17:39:57 +0000 (13:39 -0400)]
generic/187: don't run this test on NFS

This test is unreliable on NFS. It fails consistently when run vs. a
server exporting btrfs, but passes when the server exports xfs. Since we
don't have any sort of attribute that we can require to test this, just
skip this one on NFS.

Also, subsume the check for btrfs into the _supported_fs check, and add
a comment for it.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric/357: don't run this test on NFS
Jeff Layton [Fri, 1 Sep 2023 17:39:56 +0000 (13:39 -0400)]
generic/357: don't run this test on NFS

NFS doesn't keep track of whether a file is reflinked or not, so it
doesn't prevent this behavior. It shouldn't be a problem for NFS anyway,
so just skip this test there.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric/294: don't run this test on NFS
Jeff Layton [Fri, 1 Sep 2023 17:39:55 +0000 (13:39 -0400)]
generic/294: don't run this test on NFS

When creating a new dentry (of any type), NFS will optimize away any
on-the-wire lookups prior to the create since that means an extra
round trip to the server. Because of that, it consistently fails this
test.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric/*: add a check for security attrs
Jeff Layton [Wed, 30 Aug 2023 10:58:52 +0000 (06:58 -0400)]
generic/*: add a check for security attrs

There are several generic tests that require "setcap", but don't check
whether the underlying fs supports security attrs. Add the appropriate
checks.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric/578: add a check to ensure that fiemap is supported
Jeff Layton [Wed, 30 Aug 2023 10:58:51 +0000 (06:58 -0400)]
generic/578: add a check to ensure that fiemap is supported

This test requires FIEMAP support.

Suggested-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agocommon/attr: fix the _require_acl test
Jeff Layton [Wed, 30 Aug 2023 10:58:50 +0000 (06:58 -0400)]
common/attr: fix the _require_acl test

_require_acl tests whether you're able to fetch the ACL from a file
using chacl, and then tests for an -EOPNOTSUPP error return.
Unfortunately, filesystems that don't support them (like NFSv4) just
return -ENODATA when someone calls getxattr for the POSIX ACL, so the
test doesn't work.

Fix the test to have chacl set an ACL on the file instead, which should
reliably fail on filesystems that don't support them.

Signed-off-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric/61[67]: support SOAK_DURATION
Darrick J. Wong [Fri, 1 Sep 2023 14:53:31 +0000 (07:53 -0700)]
generic/61[67]: support SOAK_DURATION

Now that I've finally gotten liburing installed on my test machine, I
can actually test io_uring.  Adapt these two tests to support
SOAK_DURATION so I can add it to that too.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric/650: race mount and unmount with cpu hotplug too
Darrick J. Wong [Tue, 29 Aug 2023 23:08:03 +0000 (16:08 -0700)]
generic/650: race mount and unmount with cpu hotplug too

Ritesh Harjani reported that mount and unmount can race with the xfs cpu
hotplug notifier hooks and crash the kernel, which isfixed by:

https://lore.kernel.org/linux-xfs/ZO6J4W9msOixUk05@dread.disaster.area/T/#t

Extend this test to include that.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>
22 months agogeneric/650: add SOAK_DURATION controls
Darrick J. Wong [Tue, 29 Aug 2023 23:07:58 +0000 (16:07 -0700)]
generic/650: add SOAK_DURATION controls

Make this test controllable via SOAK_DURATION, for anyone who wants to
perform a long soak test of filesystem vs. cpu hotplug.

Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Zorro Lang <zlang@redhat.com>
Signed-off-by: Zorro Lang <zlang@kernel.org>