linux-next.git/include/linux/fs, branch master

fs: look up superblocks via the device table in fs_holder_ops

2026-06-29T08:31:53+00:00

Switch the fs_holder_ops callbacks from recovering the single owning superblock out of bdev->bd_holder to walking the device-to-superblock table and acting on every superblock registered for the device. The holder argument becomes purely the block layer's exclusivity token and is no longer needed by the fs specific callbacks. All devices opened with fs_holder_ops are registered by now: the main device since setup_bdev_super() switched to fs_bdev_file_open_by_dev() and the extra devices (xfs log and realtime devices, btrfs member devices, the ext4 external journal) since the preceding per-filesystem conversions. So no event is lost in the switchover. The walk uses a refcount-pinning cursor: each step takes a reference on the entry via sd_ref and resumes from its sd_node. Unlinking an entry is deferred to the last unpin, so a cursor never resumes from a removed node. mark_dead and sync only need the passive reference the entry holds plus s_umount, which they take with super_lock_shared(). freeze and thaw additionally need an active reference and acquire it with get_active_super(), which waits for the superblock to be born before taking s_active. Taking s_active before the superblock is born would pin a still-mounting superblock so a racing mount that aborts could never drop s_active to zero and reach SB_DYING, deadlocking the wait for SB_BORN. This is how filesystems_freeze() and filesystems_thaw() acquire it too. One semantic change: when no live superblock uses the device anymore (the holder is dying or was never registered), fs_bdev_freeze() and fs_bdev_thaw() now return 0 - freeze after syncing the block device - where they used to return -EINVAL. The freeze-deny release path moves to the table in the same switchover. A device made unfreezable for a btrfs membership change must drop its table entry before re-allowing freezing; otherwise a freeze racing the release reaches the superblock through the still-registered entry and is stranded once the release unlinks it. Split fs_bdev_unregister() out of fs_bdev_file_release() - the inverse of fs_bdev_register() - so btrfs_release_device_allow_freeze() can drop the {dev, sb} entry, re-allow freezing on the still-open device, then close it. Re-allowing only after the entry is gone keeps a racing freeze from reaching the superblock, and doing it while the file is still open avoids touching the block device after the close. btrfs previously yielded bd_holder before re-allowing, which this commit makes irrelevant to freeze resolution. Link: https://patch.msgid.link/20260616-work-super-bdev_holder_global-v2-12-7df6b864028e@kernel.org Signed-off-by: Christian Brauner (Amutable)

fs: add dedicated block device open helpers for filesystems

2026-06-29T08:31:52+00:00

Add fs_bdev_file_open_by_{dev,path}() and fs_bdev_file_release(). They open the device with fs_holder_ops and register a claim in the device-to-superblock table. Claims on the same (device, superblock) pair share one entry, so when a filesystem claims a device it already uses (xfs with its log on the data device), no second entry is added and each superblock will be acted on once. The holder argument remains purely the block layer's exclusivity token: a superblock, or a file_system_type for a device shared by several superblocks of that type. The shared case only becomes usable once the fs_holder_ops callbacks resolve superblocks through the table instead of bdev->bd_holder. Convert the main device, setup_bdev_super() and kill_block_super(), over: the open finds the entry registered by sget_fc() and claims it again. cramfs and romfs bypass kill_block_super() so they can handle MTD mounts and release the main device with a plain bdev_fput(), which would leave the claim behind: the (dev, sb) entry would never be unregistered and the passive reference it holds would keep the superblock alive forever. Convert their release paths in the same step. The frozen-device check stays in setup_bdev_super() for the primary device and is added to fs_bdev_register() for new claims, i.e. every additional device a filesystem opens through the helpers. Only a (device, superblock) pair the superblock claimed earlier may be reopened while frozen (xfs with its log on the data device): the freeze already covers that superblock through the existing claim, so nothing escapes it. Without the setup_bdev_super() check a device frozen before the mount even started (dm lock_fs, loop) could be mounted and written to (journal replay) under an active freeze, because the primary open reuses the entry registered by sget_fc() and never takes the new-claim path. Both checks read bd_fsfreeze_count only after the entry is published (by sget_fc() for the primary, by fs_bdev_register() for new claims) and pair with bdev_freeze() incrementing the count before walking the table: either the mount sees the elevated freeze count and fails with EBUSY, or the freeze finds the published entry and converges once SB_BORN is set. Link: https://patch.msgid.link/20260616-work-super-bdev_holder_global-v2-8-7df6b864028e@kernel.org Reviewed-by: Jan Kara Signed-off-by: Christian Brauner (Amutable)

fs: maintain a global device-to-superblock table

2026-06-29T08:31:52+00:00

fs_holder_ops recovers the owning superblock from bdev->bd_holder, which forces the holder to be exactly one superblock and prevents several superblocks from sharing one block device. That's what erofs is doing. As a first step introduce a global dev_t-keyed rhltable mapping each device to the superblock(s) using it. The entry is preallocated in alloc_super() and registered under sb->s_dev by the set callback through set_anon_super() and set_bdev_super(), the two helpers every set callback assigns s_dev through. Registration is the final fallible act of a set callback, so an insert failure unwinds through sget_fc()'s existing set-failure path: the fs_context keeps ownership of s_fs_info and the callers' error paths stay correct. set_anon_super() releases the anonymous dev it allocated when registration fails. Unwinding through deactivate_locked_super() instead would run kill_sb() and free s_fs_info behind the caller's back: nfs and ceph free that object through a local pointer when sget_fc() fails and would double-free. The superblock stashes the entry in sb->s_super_dev and kill_super_notify() drops the claim through it, so teardown doesn't depend on s_dev staying stable; an entry that was never registered is freed together with the superblock in destroy_super_work(). Each table entry holds a passive reference (s_passive) on its superblock, so the struct stays valid for as long as the entry is reachable. Entries are claim-counted through sd_ref: additional claims on the same (device, superblock) pair share the entry, and the unlink is deferred to the last put, so a later iteration cursor never resumes from a removed node. The table is initialized from mnt_init(): the first superblocks (the tmpfs shm mount and rootfs) are created from start_kernel() long before any initcall runs, so an initcall would be too late. The table has no readers yet; the fs_holder_ops callbacks are switched over once all devices a filesystem claims are registered. Link: https://patch.msgid.link/20260616-work-super-bdev_holder_global-v2-7-7df6b864028e@kernel.org Reviewed-by: Jan Kara Signed-off-by: Christian Brauner (Amutable)

super: convert s_count to refcount_t s_passive

2026-06-29T08:31:51+00:00

The superblock carries two counters: s_active, the active reference count that keeps the filesystem usable, and s_count, the passive reference count that merely keeps the structure itself alive. Turn the passive count into a refcount_t and rename it to s_passive to make the pairing with s_active obvious. Everything is still serialized by sb_lock, so there is no functional change; the conversion buys the usual refcount_t saturation and underflow checking. The following patches start dropping passive references without holding sb_lock and make the device-to-superblock table hold one passive reference per registered entry, which a plain integer cannot support. Link: https://patch.msgid.link/20260616-work-super-bdev_holder_global-v2-2-7df6b864028e@kernel.org Reviewed-by: Jan Kara Signed-off-by: Christian Brauner (Amutable)

Merge tag 'fuse-update-7.2' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse

2026-06-18T15:50:52+00:00

Pull fuse updates from Miklos Szeredi: - Fix lots of bugs, most from the late 6.x era, but some going back to 2.6.x - Add subsystems (io-uring, passthrough) and respective maintainers (Bernd, Joanne and Amir) - Separate transport and fs layers (Miklos) - Don't block on cat /dev/fuse (Joanne) - Perform some refactoring in fuse-uring (Joanne) - Don't use bounce-buffer for READDIR reply in virtio-fs (Matthew Ochs) - Clean up documentation (Randy) - Improve tracing (Amir) - Extend page cache invalidation after DIO (Cheng Ding) - Invalidate readdir cache on epoch change (Jun Wu) - Misc cleanups * tag 'fuse-update-7.2' of git://git.kernel.org/pub/scm/linux/kernel/git/mszeredi/fuse: (81 commits) fuse-uring: clear ent->fuse_req in commit_fetch error path fuse-uring: use named constants for io-uring iovec indices fuse-uring: refactor setting up copy state for payload copying fuse-uring: use enum types for header copying fuse-uring: refactor io-uring header copying from ring fuse-uring: refactor io-uring header copying to ring fuse-uring: separate next request fetching from sending logic fuse: invalidate readdir cache on epoch bump virtio-fs: avoid double-free on failed queue setup fuse: invalidate page cache after DIO and async DIO writes fuse: set ff->flock only on success fuse: clean up interrupt reading fuse: remove stray newline in fuse_dev_do_read() fuse: use READ_ONCE in fuse_chan_num_background() fuse: dax: Move long delayed work on system_dfl_long_wq fuse: add fuse_request_sent tracepoint fuse: Add SPDX ID lines to some files fuse: use QSTR() instead of QSTR_INIT() in fuse_get_dentry fuse: convert page array allocation to kcalloc() fuse: use current creds for backing files ...

fuse: invalidate page cache after DIO and async DIO writes

2026-06-15T12:06:20+00:00

This fixe does page cache invalidation after DIO and async DIO writes for both O_DIRECT and FOPEN_DIRECT_IO cases. Commit b359af8275a9 ("fuse: Invalidate the page cache after FOPEN_DIRECT_IO write") fixed xfstests generic/209 for DIO writes in the FOPEN_DIRECT_IO path. DIO writes without FOPEN_DIRECT_IO are already handled by generic_file_direct_write(). However, async DIO writes (xfstests generic/451) remain unhandled. After this fix: - Async write with FUSE_ASYNC_DIO: invalidate in fuse_aio_invalidate_worker() - Otherwise (Sync or async write without FUSE_ASYNC_DIO): - With FOPEN_DIRECT_IO: invalidate in fuse_direct_write_iter() - Without FOPEN_DIRECT_IO: invalidate in generic_file_direct_write() Workqueue is required for async write invalidation to prevent deadlock: calling it directly in the I/O end routine (which is in fuse worker thread context) can block on a folio lock held by a buffered I/O thread waiting for the same fuse worker thread. Co-developed-by: Jingbo Xu Signed-off-by: Jingbo Xu Signed-off-by: Cheng Ding Reviewed-by: Jingbo Xu Signed-off-by: Miklos Szeredi

Merge tag 'pull-dcache' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs

2026-06-14T22:45:31+00:00

Pull dcache updates from Al Viro: - d_alloc_parallel() API change (Neil's with my changes) - NORCU fixes - Reorganization and simplification of dentry eviction logic - Simplifying rcu_read_lock() scopes in fs/dcache.c - Secondary roots work - getting rid of NFS fake root dentries and dealing with remaining shrink_dcache_for_umount() and shrink_dentry_list() races - making cursors NORCU (surprisingly easy) * tag 'pull-dcache' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (22 commits) make cursors NORCU nfs: get rid of fake root dentries wind ->s_roots via ->d_sib instead of ->d_hash shrink_dentry_tree(): unify the calls of shrink_dentry_list() shrinking rcu_read_lock() scope in d_alloc_parallel() d_walk(): shrink rcu_read_lock() scope document dentry_kill() adjust calling conventions of lock_for_kill(), fold __dentry_kill() into dentry_kill() Document rcu_read_lock() use in select_collect2() Shift rcu_read_{,un}lock() inside fast_dput() simplify safety for lock_for_kill() slowpath fold lock_for_kill() and __dentry_kill() into common helper fold lock_for_kill() into shrink_kill() shrink_dentry_list(): start with removing from shrink list d_prune_aliases(): make sure to skip NORCU aliases kill d_dispose_if_unused() make to_shrink_list() return whether it has moved dentry to list select_collect(): ignore dentries on shrink lists if they have positive refcounts find_acceptable_alias(): skip NORCU aliases with zero refcount fix a race between d_find_any_alias() and final dput() of NORCU dentries ...

Merge tag 'vfs-7.2-rc1.procfs' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs

2026-06-14T22:37:58+00:00

Pull procfs updates from Christian Brauner: - Revamp fs/filesystems.c The file was a mess with a hand-rolled linked list in desperate need of a cleanup. The filesystems list is now RCU-ified, /proc files can be marked permanent from outside fs/proc/, and the string emitted when reading /proc/filesystems is pre-generated and cached instead of pointer-chasing and printfing entry by entry on every read. The file is read frequently because libselinux reads it and is linked into numerous frequently used programs (even ones you would not suspect, like sed!). Scalability also improves since reference maintenance on open/close is bypassed. open+read+close cycle single-threaded (ops/s): before: 442732 after: 1063462 (+140%) open+read+close cycle with 20 processes (ops/s): before: 606177 after: 3300576 (+444%) A follow-up patch adds missing unlocks in some corner cases and tidies things up. - Relax the mount visibility check for subset=pid mounts When procfs is mounted with subset=pid, all static files become unavailable and only the dynamic pid information is accessible. In that case there is no point in imposing the full mount visibility restrictions on the mounter - everything that can be hidden in procfs is already inaccessible. These restrictions prevented procfs from being mounted inside rootless containers since almost all container implementations overmount parts of procfs to hide certain directories. As part of this /proc/self/net is only shown in subset=pid mounts for CAP_NET_ADMIN, reconfiguring subset=pid is rejected, the SB_I_USERNS_VISIBLE superblock flag is replaced with an FS_USERNS_MOUNT_RESTRICTED filesystem flag, fully visible mounts are recorded in a list, and the mount restrictions are finally documented. - Protect ptrace_may_access() with exec_update_lock in procfs Most uses of ptrace_may_access() in procfs should hold exec_update_lock to avoid TOCTOU issues with concurrent privileged execve() (like setuid binary execution). This fixes the easy cases - the owner and visibility checks and the FD link permission checks - with the gnarlier ones to follow later. * tag 'vfs-7.2-rc1.procfs' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs: fs: fix ups and tidy ups to /proc/filesystems caching proc: protect ptrace_may_access() with exec_update_lock (FD links) proc: protect ptrace_may_access() with exec_update_lock (part 1) docs: proc: add documentation about mount restrictions proc: handle subset=pid separately in userns visibility checks proc: prevent reconfiguring subset=pid proc: subset=pid: Show /proc/self/net only for CAP_NET_ADMIN fs: cache the string generated by reading /proc/filesystems sysfs: remove trivial sysfs_get_tree() wrapper fs: RCU-ify filesystems list fs: move SB_I_USERNS_VISIBLE to FS_USERNS_MOUNT_RESTRICTED proc: allow to mark /proc files permanent outside of fs/proc/ namespace: record fully visible mounts in list

wind ->s_roots via ->d_sib instead of ->d_hash

2026-06-05T04:34:56+00:00

shrink_dcache_for_umount() is supposed to handle the possibility of some of the dentries to be evicted being in other threads shrink lists; it either kills them, leaving an empty husk to be freed by the owner of shrink list whenever it gets around to that, or it waits for the eviction in progress to get completed. That relies upon dentry remaining attached to the tree until the eviction reaches dentry_unlist() and its ->d_sib gets removed from the list. Unfortunately, the secondary roots are linked via ->d_hash, rather than ->d_sib and they become removed from that list before their inode references are dropped. If shrink_dentry_list() from another thread ends up evicting one of the secondary roots and gets to that point in dentry_kill() when shrink_dcache_for_umount() is looking for secondary roots, the latter will *not* notice anything, possibly leading to warnings about busy inodes at umount time and all kinds of breakage after that. Moreover, shrink_dcache_for_umount() walks the list of secondary roots with no protection whatsoever, so it might end up calling dget() on a dentry that already passed through lockref_mark_dead(&dentry->d_lockref); ending up with corrupted refcount and possible UAF. AFAICS, the most straightforward way to deal with that would be to have secondary roots linked via ->d_sib rather than ->d_hash; then they would remain on the list until killed, and we could use d_add_waiter() machinery to wait for eviction in progress. Changes: * secondary roots look the same as ->s_root from d_unhashed() and d_unlinked() POV now. * secondary roots are represented as "no parent, but on ->d_sib" instead of "no parent, but on ->d_hash". * since ->d_sib is a plain hlist, we protect it with per-superblock spinlock (sb->s_roots_lock) instead of the LSB of the head pointer (for non-root dentries it would be protected by ->d_lock of parent). * __d_obtain_alias() uses ->d_sib for linkage when allocating a secondary root. * d_splice_alias_ops() detects splicing of a secondary root and removes it from the list before calling __d_move(). * dentry_unlist() detects eviction of a secondary root and removes it from the list; no need to play the games for d_walk() sake, since the latter is not going to look for the next sibling of those anyway. * ___d_drop() doesn't care about ->s_roots anymore. * shrink_dcache_for_umount() uses proper locking for access to the list of secondary roots and if it runs into one that is in the middle of eviction waits for that to finish. Signed-off-by: Al Viro

writeback: use a per-sb counter to drain inode wb switches at umount

2026-05-22T10:06:35+00:00

Tracking in-flight inode wb switches with a single global counter (isw_nr_in_flight) plus a synchronize_rcu() based wait in cgroup_writeback_umount() forces every umount to take a global hit whenever any other superblock on the system has wb switches in flight, even if the superblock being unmounted has none of its own. Replace the global synchronize_rcu()/flush_workqueue() pair with a per-sb counter, s_isw_nr_in_flight, plus three small helpers: - cgroup_writeback_pin(sb) - increment counter - cgroup_writeback_unpin(sb) - decrement and wake drainer if last - cgroup_writeback_drain(sb) - wait for counter to reach zero The wiring is: - inode_prepare_wbs_switch() pins before checking SB_ACTIVE and grabbing the inode; failure paths unpin before returning. A lockless SB_ACTIVE check at the top of the function lets us skip the atomic_inc/smp_mb dance once SB_ACTIVE has been cleared (it is monotonic and never set back). - process_inode_switch_wbs() unpins after the matching iput(). - cgroup_writeback_umount() drains the per-sb counter via wait_var_event(). The smp_mb() pair between inode_prepare_wbs_switch() and cgroup_writeback_umount() keeps the SB_ACTIVE / counter ordering: either the umounter sees a non-zero counter and waits, or the switcher sees SB_ACTIVE cleared and aborts before grabbing the inode. The global isw_nr_in_flight is left in place, since it is still used to throttle in-flight switches via WB_FRN_MAX_IN_FLIGHT. The rcu_read_lock() extension in inode_switch_wbs() and cleanup_offline_cgwb() that the race fix added is no longer needed and is reverted; the synchronize_rcu() that the race fix added to cgroup_writeback_umount() is dropped as well. The following numbers were measured on a 16 vCPU QEMU guest with 4 background superblocks each churning "create memcg -> write 1 MiB -> rmdir memcg" to keep the global isw_nr_in_flight non-zero. Latencies are wall-clock around umount(8); only the target sb's umount is measured. Target sb runs its own cgwb churn: p50 p95 p99 max global synchronize_rcu() 67.6 ms 88.3 ms 88.3 ms 96.8 ms per-sb counter (this) 7.9 ms 10.0 ms 10.0 ms 10.1 ms Idle target umount latency under cross-sb cgwb-switch pressure: p50 p95 p99 max global synchronize_rcu() 62.7 ms 95.4 ms 108.1 ms 108.6 ms per-sb counter (this) 5.3 ms 6.9 ms 7.4 ms 7.4 ms no-pressure baseline 4.9 ms 5.9 ms 6.3 ms 6.7 ms 8 concurrent umounts of idle sbs under the same pressure: p50 p95 max global synchronize_rcu() 61.3 ms 99.5 ms 113.7 ms per-sb counter (this) 8.1 ms 9.1 ms 9.5 ms In-kernel cgroup_writeback_umount() time across the same run (bpftrace, ~340 calls covering all scenarios): global synchronize_rcu() 12371 ms total (~36 ms / call) per-sb counter (this) 1.37 ms total ( ~4 us / call) Suggested-by: Christian Brauner Link: https://lore.kernel.org/r/177910456953.488929.2169908940676707307.b4-review@b4 Reviewed-by: Jan Kara Signed-off-by: Baokun Li Link: https://patch.msgid.link/20260521095016.2791354-4-libaokun@linux.alibaba.com Acked-by: Tejun Heo Signed-off-by: Christian Brauner (Amutable)