Add a test suite to verify the behavior of idmapped mounts. The test
suite also includes a range of vfs tests to verify that no regressions
are introduced by idmapped mounts. The following tests are currently
available with more to come in the future:
01. posix acls on regular and idmapped mounts
02. create operations in user namespace
03. device node creation in user namespace
04. expected ownership on idmapped mounts
05. fscaps on regular mounts
06. fscaps on idmapped mounts
07. fscaps on idmapped mounts in user namespace
08. fscaps on idmapped mounts in user namespace
with different id mappings
09. mapped fsids
10. unmapped fsids
11. cross mount hardlink
12. cross idmapped mount hardlink
13. hardlinks from idmapped mounts
14. hardlinks from idmapped mounts in user namespace
15. io_uring
16. io_uring in user namespace
17. io_uring from idmapped mounts
18. io_uring from idmapped mounts in user namespace
19. io_uring from idmapped mounts with unmapped ids
20. io_uring from idmapped mounts with unmapped ids in user namespace
21. following protected symlinks on regular mounts
22. following protected symlinks on idmapped mounts
23. following protected symlinks on idmapped mounts in user namespace
24. cross mount rename
25. cross idmapped mount rename
26. rename from idmapped mounts
27. rename from idmapped mounts in user namespace
28. symlink from regular mounts
29. symlink from idmapped mounts
30. symlink from idmapped mounts in user namespace
31. setid binaries on regular mounts
32. setid binaries on idmapped mounts
33. setid binaries on idmapped mounts in user namespace
34. setid binaries on idmapped mounts in user namespace
with different id mappings
35. sticky bit unlink operations on regular mounts
36. sticky bit unlink operations on idmapped mounts
37. sticky bit unlink operations on idmapped mounts in user namespace
38. sticky bit rename operations on regular mounts
39. sticky bit rename operations on idmapped mounts
40. sticky bit rename operations on idmapped mounts in user namespace
41. create operations in directories with setgid bit set
42. create operations in directories with setgid bit set
on idmapped mounts
43. create operations in directories with setgid bit set
on idmapped mounts in user namespace
44. verify create operations and ownership with racing threads idmapping
a mount
45. setattr truncate operations on regular mounts
46. setattr truncate operations on idmapped mounts
47. setattr truncate operations on idmapped mounts in user namespace
Here's some sample output when running with DEBUG_TRACE defined:
Cc: Amir Goldstein <amir73il@gmail.com>
Cc: Eryu Guan <guan@eryu.me>
Cc: fstests@vger.kernel.org
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Regression test to verify that creating a series of detached mounts,
attaching them to the filesystem, and unmounting them does not trigger an
integer overflow in ns->mounts causing the kernel to block any new mounts in
count_mounts() and returning ENOSPC because it falsely assumes that the
maximum number of mounts in the mount namespace has been reached, i.e. it
thinks it can't fit the new mounts into the mount namespace anymore.
The test is written in a way that it will leave the host's mount
namespace intact so we are sure to never make the host's mount namespace
unuseable!
Link: https://git.kernel.org/torvalds/c/ee2e3f50629f17b0752b55b2566c15ce8dafb557
Cc: Eryu Guan <guan@eryu.me>
Cc: David Howells <dhowells@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
xfs_admin in xfsprogs 5.11 has a bug wherein a caller who specifies an
external log device forces xfs_db to be invoked, potentially with zero
command arguments. When this happens, xfs_db will wait for input on
stdin, which causes fstests to hang. Since xfs_admin is not an
interactive tool, redirect stdin from /dev/null to prevent this issue.
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Teach _scratch_xfs_admin to support passing the realtime device to
xfs_admin so that we can actually test xfs_admin functionality with
those setups.
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Create a couple of helper functions to get and set the XFS copy on write
preallocation garbage collection interval.
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Brian Foster <bfoster@redhat.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
xfs/502 currently creates a default of 30k unlinked files per CPU.
While this completes in a reasonable amount of time on systems with
lesser numbers of CPUs, this scales poorly on high CPU count systems
that are otherwise testing smaller default filesystems. For example,
on an 80xcpu box and a 15GB (4 AG) XFS filesystem, xfs/502 requires
3 hours to complete. The same test on a 4xcpu vm (or the 80xcpu
hardware with an 80AG filesystem instead of the default of 4AGs)
completes in a little over 5 minutes. This is a rather severe
thrashing breakdown that doesn't add much value to the test
coverage.
Address this problem by scaling the file count to the AG count of
the filesystem rather than the CPU count of the test system. Since
the AG count is likely to be less than the CPU count, bump the
default scaling factor a bit from 30k per CPU to 50k per AG. From
there, larger counts can still be exercised via the global load
factor configuration.
Signed-off-by: Brian Foster <bfoster@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The minimum length space allocator (i.e.
xfs_bmap_exact_minlen_extent_alloc()) depends on the underlying
filesystem to be fragmented so that there are enough one block sized
extents available to satify space allocation requests.
xfs/{532,533,538} tests issue space allocation requests for metadata
(e.g. for blocks holding directory and xattr information). With
realtime filesystem instances, these tests would end up fragmenting
the space on realtime device. Hence minimum length space allocator
fails since the regular filesystem space is not fragmented and hence
there are no one block sized extents available.
Thus, this commit disables realtime inherit flag (if any) on root
directory so that space on data device gets fragmented rather than
realtime device.
Suggested-by: Darrick J. Wong <djwong@kernel.org>
Reported-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Chandan Babu R <chandanrlinux@gmail.com>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
xfs/{529,531,532,534,535} attempt to create test files after injecting
reduce_max_iextents error tag. Creation of test files fails when using a
multi-block directory test configuration because,
1. A directory can have a pseudo maximum extent count of 10.
2. In the worst case a directory entry creation operation can consume
(XFS_DA_NODE_MAXDEPTH + 1 + 1) * (Nr fs blocks in a single directory block)
extents.
With 1k fs block size and 4k directory block size, this evaluates to,
(5 + 1 + 1) * 4
= 7 * 4
= 28
> 10 (Pseudo maximum inode extent count).
This commit fixes the issue by creating test files before injecting
reduce_max_iextents error tag.
Reported-by: Darrick J. Wong <djwong@kernel.org>
Suggested-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Chandan Babu R <chandanrlinux@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
chown command is being executed on $testfile which is actually
deleted just before the execution of quota inode extent count
overflow test. Hence the test was not getting exercised at all.
This commit fixes the bug by using $fillerdir as the target of chown
command.
Signed-off-by: Chandan Babu R <chandanrlinux@gmail.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Write data into a file and then enable DAX on the file immediately,
the written data which is still in the buffer should be synchronized
to disk instead of discarded when the corresponding inode is evicted.
It's a regression test for commit 88149082bb8e ("fs: Handle
I_DONTCACHE in iput_final() instead of generic_drop_inode()")
Signed-off-by: Xiao Yang <yangx.jy@fujitsu.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Since commit 77573fa310d9, toggling FS_XFLAG_DAX on an existing file can make
S_DAX on the file change immediately when all applications close the file.
As a result, we don't need the redundant drop cache and umount & mount cycle.
Signed-off-by: Xiaoli Feng <xifeng@redhat.com>
Signed-off-by: Xiao Yang <yangx.jy@fujitsu.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The ltp/fsstress always fails on io_uring_queue_init() by returnning
ENOMEM. Due to io_uring accounts memory it needs under the rlimit
memlocked option, which can be quite low on some setups, especially
on 64K pagesize machine. root isn't under this restriction, but
regular users are. So only g/233 and g/270 which use $qa_user to run
fsstress are failed.
To avoid this failure, set max locked memory to unlimited before
doing fsstress, then restore it after test done.
Signed-off-by: Zorro Lang <zlang@redhat.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Otherwise, if xfs_io doesn't support this command, it will report
that filesystem fails on this testcase.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Reviewed-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Add a new test to make sure the xfs_db path command works the way the
author thinks it should.
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Non-partitionable devices, like zoned block devices, aren't showing
up in in /proc/partitions and therefore we cannot rely on it to get
a device's size.
Use blockdev --getsz to get the block device size.
Cc: Naohiro Aota <naohiro.aota@wdc.com>
Signed-off-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Test that if we set a capability on a file but not on the next files we
create, send/receive operations only apply the capability to the first
file, the one for which we have set a capability.
This is motivated by a regression that started to happen with kernel 5.8,
caused by an incompatibility between kernel commit 89efda52e6b693
("btrfs: send: emit file capabilities after chown") and a workaround for
send/receive added to btrfs-progs several years ago by btrfs-progs
commit 123a2a085027e ("btrfs-progs: receive: restore capabilities after
chown"). That workaround in btrfs-progs was added in btrfs-progs 4.1.
That kernel commit ended up fixing the bugs the btrfs-progs commit
targeted, as well as other bugs around btrfs send and capabilities.
However it did not play nice with the btrfs-progs workaround.
The kernel fix was backported to all stable kernels and the btrfs-progs
workaround was later removed from btrfs-progs by commit 81d8ea9346c6ee
("btrfs-progs: receive: remove workaround for setting capabilities"),
introduced in btrfs-progs 5.11.
The test currently fails on any environment using kernel 5.8+ and a
btrfs-progs version between 4.1 and 5.10. Updating btrfs-progs to
version 5.11 makes the test pass.
Signed-off-by: Filipe Manana <fdmanana@suse.com>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
wenli xie reported a buffer cache deadlock when an overlayfs is mounted
atop xfs and overlayfs tries to replace a single-nlink file with a
whiteout file. This test reproduces that deadlock.
Reported-by: wenli xie <wlxie7296@gmail.com>
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Combine these two helpers into a single generic function so that we can
use it in the next patch to test a regression when running overlayfs
atop xfs.
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Eryu Guan <guaneryu@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Add a couple of regression tests for "xfs: make sure the rt allocator
doesn't run off the end" and "xfs: ensure that fpunch, fcollapse, and
finsert operations are aligned to rt extent size".
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Chandan Babu R <chandanrlinux@gmail.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Casefold rename test should check if renaming a file to an equivalent
name fails as expect (e.g. renaming from "file.txt" to "FILE.TXT") and
`mv` correctly identifies that those names refers to the same file.
Currently, the test doesn't do what is expected given that it doesn't
have the file to be renamed, and `mv` returns "No such file or
directory". Fix that by creating test files and checking the correct
output.
Fixes: 12b7dddbc2 ("generic: Add tests for filename casefolding feature")
Signed-off-by: André Almeida <andrealmeid@collabora.com>
Reviewed-by: Gabriel Krisman Bertazi <krisman@collabora.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Escape SCRATCH_DEV variable so it prints its name, instead of printing
the value. If the value is "", the error message will not be very
informative.
Signed-off-by: André Almeida <andrealmeid@collabora.com>
Reviewed-by: Gabriel Krisman Bertazi <krisman@collabora.com>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
The device name of a secondary storage device isn't all that important,
but the size is.
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Eryu Guan <guaneryu@gmail.com>
Christian Brauner
Sun Ke
Darrick J. Wong
Brian Foster
Chandan Babu R
Xiao Yang
Zorro Lang
Chao Yu
Johannes Thumshirn
Filipe Manana
André Almeida