kernel/git/next/linux-next.git/tools/testing/selftests/filesystems, branch master

Merge branch 'fs-next' of linux-next

2026-09-14T14:06:34+00:00

Merge branch 'vfs.all' of https://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs.git

2026-09-14T10:04:56+00:00

selftests/epoll: fix race condition in multi-waiter wakeup tests

2026-09-14T04:33:27+00:00

In tests with multiple concurrent waiters on edge-triggered epoll instances where an emitter writes to multiple sockets (epoll16, epoll56, epoll58): When the emitter performs its first write(), ep_poll_callback() fires and wakes up both waiters because one waiter uses epoll_wait() and the other one uses poll(). This translates to different wait queues, ep->wq for epoll and ep->poll_wait for poll/select, which are both awoken by the kernel because of that single write. Next, both waiter threads invoke epoll_wait(), but since there is only one event, only one epoll_wait() will return non-zero because of the edge-triggered mode being used (in level-triggered mode, the kernel would re-queue the event because of remaining unread data). Since the second waiter sees an empty ready list, it does not increment ctx.count and the test fails spuriously with ctx.count == 1 instead of 2. Emitter (CPU 0) Thread 0 (CPU 1) Thread 1 (CPU 2) =============== ================ ================ epoll_wait(e0, -1) poll(e0, -1) [on e0->wq] [on e0->poll_wait] write(sfd[1]) | +--(Kernel wakes BOTH e0->wq and e0->poll_wait via callback)--+ | | | wakes up wakes up | | epoll_wait() reaps e1 poll() returns 1 | | (e1 removed via ET) (wants event) | | e0->rdllist is EMPTY | | | count++ (count = 1) v | | epoll_wait(e0, 0) | | sees EMPTY list! | | returns 0! | | thread exits | v | write(sfd[3]) | (event arrives too late!) v EXPECT_EQ(count, 2) <-- SPURIOUS FAILURE! Introduce waiter_entry1ap_loop() to retry poll() if the initial epoll_wait(..., 0) yielded no events. This ensures the thread waits for the subsequent write rather than failing immediately. Apply this helper in epoll16, epoll56, and for both waiter threads in epoll58. Link: https://lore.kernel.org/20260828-selftest-epoll-fix-race-v2-1-953ab57fd60a@codasip.com Fixes: f2728fe80cef ("selftests: add epoll selftests") Signed-off-by: Florian Schmaus Signed-off-by: Andrew Morton Cc: Heiher Cc: Roman Penyaev Cc: Shuah Khan Cc: Christian Brauner

Merge branch 'vfs-7.4.misc' into vfs.all

2026-09-11T13:14:18+00:00

Merge branch 'vfs-7.4.lookup' into vfs.all

2026-09-11T13:01:43+00:00

Signed-off-by: Christian Brauner

Merge branch 'vfs-7.4.kernfs' into vfs.all

2026-09-11T13:01:43+00:00

Signed-off-by: Christian Brauner

selftests: cover kernfs file handles and same-parent rename

2026-09-10T07:36:48+00:00

The kernfs tests only reach kernfs through lookup and readdir. Two paths are not covered: file handles, which find a node without a lookup through its parent, and rename of a node that keeps its parent. Add three tests: - decode a file handle, live and after the node is gone; - decode while the node is being removed; - look up a name while an interface is renamed, which renames its /sys/class/net entry with the parent unchanged. Two small fixes while here. ns_tag_isolates_class_net now counts only symlinks, because bonding adds a bonding_masters attribute that is not a device. A failing mkdtemp() now skips instead of aborting. Signed-off-by: Shakeel Butt Link: https://patch.msgid.link/20260905191613.3143937-2-shakeel.butt@linux.dev Signed-off-by: Christian Brauner (Amutable)

bfs: remove the filesystem

2026-09-04T10:43:07+00:00

BFS (not to be confused with BeFS, the BeOS filesystem) is the filesystem UnixWare uses for its boot partition. It has extremely limited functionality, only supporting contiguous files and no subdirectories. Even though the bfs driver is very small and is unlikely to cause future maintenance problems, given that the only type of data stored on such a partition is likely to be kernels and bootloader settings, there is very little reason anyone would want to access it from Linux. Other old Unix filesystems (efs, freevxfs) have been removed recently, and bfs is highly unlikely to have any users, so remove it as well. Retain the UAPI header to be safe. Also update the former maintainer's email address in CREDITS. Signed-off-by: Ethan Nelson-Moore Link: https://patch.msgid.link/20260904082954.48738-1-enelsonmoore@gmail.com Acked-by: Jan Kara Acked-by: Lorenzo Stoakes (ARM) Acked-by: Tigran Aivazian Signed-off-by: Christian Brauner (Amutable)

selftests: cover kernfs dentry revalidation

2026-09-04T08:06:57+00:00

kernfs_test only checked two xattr calls. It did not touch the part of kernfs that userspace hits most: the dentry cache. The kernel adds and removes kernfs nodes on its own, with no VFS call behind it, and removal cannot always unhash the dentries that are already cached. So ->d_revalidate() is what keeps the cache honest, let's add tests for it. Add tests that make the kernel change kernfs directly, instead of going through create and unlink: - writing cgroup.subtree_control makes the kernel add and remove files in every child cgroup. That checks two things: a cached negative dentry is dropped once the file appears, and a cached positive one is dropped once the file goes away. - renaming a network interface renames the sysfs node under a dentry that is already cached. - sysfs mounted in a new network namespace must list that namespace's interfaces and not the parent's, which covers the KERNFS_NS tagging code. - many lookups running against concurrent mkdir and rmdir. A lookup may only succeed or fail with an errno that means "it is gone". - open across removal, readdir returning each entry once, and a telldir()/seekdir() round trip, which uses the name hash cookie in kernfs_dir_pos(). Test the other direction too: walking dentries that are already cached must not drop them. That is not just a slowdown. d_invalidate() calls detach_mounts(), so a dentry dropped by mistake also unmounts whatever is mounted under that directory. Both filesystems are needed. cgroup2 has no ->rename and is not namespace tagged. sysfs cannot be mounted with a set of children we control. The config fragment stops a kernel built with kselftest-merge from passing while it runs almost nothing. To confirm the tests really catch a broken backstop, kernfs_dir_changed() was stubbed to return false. 8 of the 11 tests then failed. Signed-off-by: Shakeel Butt Link: https://patch.msgid.link/20260902014050.499002-1-shakeel.butt@linux.dev Signed-off-by: Christian Brauner (Amutable)

selftests/fuse: add test for FUSE_HAS_SYNCFS privilege gating

2026-09-02T07:36:07+00:00

Add a selftest that talks the raw FUSE protocol over /dev/fuse (rather than via libfuse, which negotiates INIT internally) so it can both choose whether to advertise FUSE_HAS_SYNCFS and directly observe whether a FUSE_SYNCFS opcode is forwarded by the kernel. Three cases are covered: T1: host-root mount, server sets FUSE_HAS_SYNCFS -> FUSE_SYNCFS must reach the server. T2: host-root mount, server does not opt in -> FUSE_SYNCFS must not be sent (back-compat). T3: server opts in but opened /dev/fuse without CAP_SYS_ADMIN while still in the initial user namespace -> FUSE_SYNCFS must be withheld. This is the case that distinguishes gating on the server's privilege from gating on the mount's user namespace. Signed-off-by: Jimmy Zuber Assisted-by: Claude:claude-opus-4-8 [Claude-Code] Signed-off-by: Miklos Szeredi