mirror-linux/fs
T.J. Mercier eea5d2bb34 kernfs: Send IN_DELETE_SELF and IN_IGNORED
Currently some kernfs files (e.g. cgroup.events, memory.events) support
inotify watches for IN_MODIFY, but unlike with regular filesystems, they
do not receive IN_DELETE_SELF or IN_IGNORED events when they are
removed. This means inotify watches persist after file deletion until
the process exits and the inotify file descriptor is cleaned up, or
until inotify_rm_watch is called manually.

This creates a problem for processes monitoring cgroups. For example, a
service monitoring memory.events for memory.high breaches needs to know
when a cgroup is removed to clean up its state. Where it's known that a
cgroup is removed when all processes die, without IN_DELETE_SELF the
service must resort to inefficient workarounds such as:
  1) Periodically scanning procfs to detect process death (wastes CPU
     and is susceptible to PID reuse).
  2) Holding a pidfd for every monitored cgroup (can exhaust file
     descriptors).

This patch enables IN_DELETE_SELF and IN_IGNORED events for kernfs files
and directories by clearing inode i_nlink values during removal. This
allows VFS to make the necessary fsnotify calls so that userspace
receives the inotify events.

As a result, applications can rely on a single existing watch on a file
of interest (e.g. memory.events) to receive notifications for both
modifications and the eventual removal of the file, as well as automatic
watch descriptor cleanup, simplifying userspace logic and improving
efficiency.

There is gap in this implementation for certain file removals due their
unique nature in kernfs. Directory removals that trigger file removals
occur through vfs_rmdir, which shrinks the dcache and emits fsnotify
events after the rmdir operation; there is no issue here. However kernfs
writes to particular files (e.g. cgroup.subtree_control) can also cause
file removal, but vfs_write does not attempt to emit fsnotify events
after the write operation, even if i_nlink counts are 0. As a usecase
for monitoring this category of file removals is not known, they are
left without having IN_DELETE or IN_DELETE_SELF events generated.
Fanotify recursive monitoring also does not work for kernfs nodes that
do not have inodes attached, as they are created on-demand in kernfs.

Suggested-by: Jan Kara <jack@suse.cz>
Signed-off-by: T.J. Mercier <tjmercier@google.com>
Tested-by: syzbot@syzkaller.appspotmail.com
Acked-by: Tejun Heo <tj@kernel.org>
Link: https://patch.msgid.link/20260225223404.783173-3-tjmercier@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2026-03-12 15:51:03 +01:00
..
9p Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
adfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
affs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
afs Convert remaining multi-line kmalloc_obj/flex GFP_KERNEL uses 2026-02-22 08:26:33 -08:00
autofs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
befs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
bfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
btrfs for-7.0-rc2-tag 2026-03-03 09:08:00 -08:00
cachefiles Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
ceph Convert remaining multi-line kmalloc_obj/flex GFP_KERNEL uses 2026-02-22 08:26:33 -08:00
coda Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
configfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
cramfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
crypto Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
debugfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
devpts Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
dlm Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
ecryptfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
efivarfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
efs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
erofs erofs: fix interlaced plain identification for encoded extents 2026-02-25 17:40:58 +08:00
exfat Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
exportfs fs: add helpers name_is_dot{,dot,_dotdot} 2026-01-29 10:06:59 +01:00
ext2 Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
ext4 Convert remaining multi-line kmalloc_obj/flex GFP_KERNEL uses 2026-02-22 08:26:33 -08:00
f2fs fsverity fixes for v7.0-rc1 2026-02-22 13:12:04 -08:00
fat Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
freevxfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
fuse Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
gfs2 Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
hfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
hfsplus Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
hostfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
hpfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
hugetlbfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
iomap iomap: reject delalloc mappings during writeback 2026-03-04 14:31:56 +01:00
isofs Convert remaining multi-line kmalloc_obj/flex GFP_KERNEL uses 2026-02-22 08:26:33 -08:00
jbd2 Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
jffs2 Convert remaining multi-line kmalloc_obj/flex GFP_KERNEL uses 2026-02-22 08:26:33 -08:00
jfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
kernfs kernfs: Send IN_DELETE_SELF and IN_IGNORED 2026-03-12 15:51:03 +01:00
lockd Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
minix vfs-7.0-rc2.fixes 2026-02-25 10:34:23 -08:00
netfs netfs: Fix unbuffered/DIO writes to dispatch subrequests in strict sequence 2026-02-26 14:44:32 +01:00
nfs Convert remaining multi-line kmalloc_obj/flex GFP_KERNEL uses 2026-02-22 08:26:33 -08:00
nfs_common
nfsd nfsd-7.0 fixes: 2026-03-02 09:05:20 -08:00
nilfs2 Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
nls
notify Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
ntfs3 treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
ocfs2 Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
omfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
openpromfs
orangefs Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
overlayfs Convert 'alloc_flex' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
proc vfs-7.0-rc2.fixes 2026-02-25 10:34:23 -08:00
pstore Convert remaining multi-line kmalloc_obj/flex GFP_KERNEL uses 2026-02-22 08:26:33 -08:00
qnx4 Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
qnx6 Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
quota Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
ramfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
resctrl Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
romfs mm.git review status for linus..mm-stable 2026-02-18 20:50:32 -08:00
smb smb: client: fix oops due to uninitialised var in smb2_unlink() 2026-03-05 20:41:16 -06:00
squashfs Squashfs: check metadata block offset is within range 2026-02-24 11:13:27 -08:00
sysfs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
tests fs/tests: exec: drop duplicate bprm_stack_limits test vectors 2026-02-03 17:39:00 -08:00
tracefs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
ubifs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
udf Convert remaining multi-line kmalloc_obj/flex GFP_KERNEL uses 2026-02-22 08:26:33 -08:00
ufs Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
unicode Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
vboxsf Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
verity fsverity: add dependency on 64K or smaller pages 2026-03-02 21:05:34 -08:00
xfs xfs: add static size checks for ioctl UABI 2026-02-25 13:58:50 +01:00
zonefs Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
Kconfig
Kconfig.binfmt
Makefile vfs-7.0-rc1.nullfs 2026-02-09 13:41:34 -08:00
aio.c Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
anon_inodes.c
attr.c fs,fsverity: reject size changes on fsverity files in setattr_prepare 2026-01-29 09:39:41 -08:00
backing-file.c
bad_inode.c fs: refactor ->update_time handling 2026-01-12 14:01:32 +01:00
binfmt_elf.c rseq: slice ext: Ensure rseq feature size differs from original rseq size 2026-02-23 11:19:19 +01:00
binfmt_elf_fdpic.c Convert 'alloc_flex' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
binfmt_flat.c
binfmt_misc.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
binfmt_script.c
bpf_fs_kfuncs.c
buffer.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
char_dev.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
compat_binfmt_elf.c
coredump.c Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
d_path.c
dax.c
dcache.c vfs-7.0-rc1.misc.2 2026-02-16 13:00:36 -08:00
direct-io.c
drop_caches.c
eventfd.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
eventpoll.c vfs-7.0-rc2.fixes 2026-02-25 10:34:23 -08:00
exec.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
fcntl.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
fhandle.c Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
file.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
file_attr.c vfs-7.0-rc2.fixes 2026-02-25 10:34:23 -08:00
file_table.c Changes since last update: 2026-02-09 16:08:40 -08:00
filesystems.c sysfs(2): fs_index() argument is _not_ a pathname 2026-01-16 12:52:04 -05:00
fs-writeback.c vfs-7.0-rc2.fixes 2026-02-25 10:34:23 -08:00
fs_context.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
fs_dirent.c
fs_parser.c
fs_pin.c
fs_struct.c fs: add <linux/init_task.h> for 'init_fs' 2026-01-14 16:50:37 +01:00
fserror.c fs: report filesystem and file I/O errors to fsnotify 2026-01-13 09:58:01 +01:00
fsopen.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
init.c struct filename series 2026-02-09 16:58:28 -08:00
inode.c fsverity updates for 7.0 2026-02-12 10:41:34 -08:00
internal.h struct filename series 2026-02-09 16:58:28 -08:00
ioctl.c
kernel_read_file.c
libfs.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
locks.c NFSD 7.0 Release Notes 2026-02-12 08:23:53 -08:00
mbcache.c Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
mnt_idmapping.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
mount.h fs: add immutable rootfs 2026-01-12 16:52:09 +01:00
mpage.c
namei.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
namespace.c vfs-7.0-rc2.fixes 2026-02-25 10:34:23 -08:00
nsfs.c nsfs: tighten permission checks for handle opening 2026-02-27 22:00:11 +01:00
nullfs.c fs: add immutable rootfs 2026-01-12 16:52:09 +01:00
open.c struct filename series 2026-02-09 16:58:28 -08:00
pidfs.c vfs-7.0-rc2.fixes 2026-02-25 10:34:23 -08:00
pipe.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
pnode.c
pnode.h
posix_acl.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
proc_namespace.c
read_write.c fs: add setlease to generic_ro_fops and read-only filesystem directory operations 2026-01-12 10:55:45 +01:00
readdir.c readdir: require opt-in for d_type flags 2026-01-12 10:57:34 +01:00
remap_range.c
select.c Convert more 'alloc_obj' cases to default GFP_KERNEL arguments 2026-02-21 20:03:00 -08:00
seq_file.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
signalfd.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
splice.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
stack.c
stat.c statx: switch to CLASS(filename_maybe_null) 2026-01-16 12:52:04 -05:00
statfs.c user_statfs(): switch to CLASS(filename) 2026-01-16 12:52:04 -05:00
super.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
sync.c treewide: Replace kmalloc with kmalloc_obj for non-scalar types 2026-02-21 01:02:28 -08:00
sysctls.c
timerfd.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
userfaultfd.c Convert 'alloc_obj' family to use the new default GFP_KERNEL argument 2026-02-21 17:09:51 -08:00
utimes.c do_utimes_path(): switch to CLASS(filename_uflags) 2026-01-16 12:52:03 -05:00
xattr.c filename_...xattr(): don't consume filename reference 2026-01-16 12:52:03 -05:00