linux-stable.git/fs/xfs, branch master

Merge tag 'vfs-7.2-rc2.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs

2026-07-03T15:48:05+00:00

Pull vfs fixes from Christian Brauner:

 - netfs:

    - fix the decision when to disallow write-streaming with fscache in
      use, handling of asynchronous cache object creation, a double fput
      in cachefiles, clearing S_KERNEL_FILE without the inode lock held,
      page extraction bugs in the iov_iter helpers (a potential
      underflow, a missing allocation failure check, a memory leak, and
      a folio offset miscalculation), writeback error and ENOMEM
      handling, DIO write retry for filesystems without a
      ->prepare_write() method, and the replacement of the wb_lock mutex
      with a bit lock plus writethrough collection offload so that
      multiple asynchronous writebacks don't interfere with each other.

    - Fix the barriering when walking the netfs subrequest list during
      retries as it was possible to see a subrequest that was just added
      by the application thread.

 - iomap:

    - Change iomap to submit read bios after each extent instead of
      building them up across extents. The old behavior was considered
      problematic for a while and now caused an actual erofs bug.

    - Guard the ioend io_size EOF trim in iomap against underflow when a
      concurrent truncate moves EOF below the start of the ioend,
      wrapping io_size to a huge value.

 - overlayfs

    - Fix a stale overlayfs comment about the locking order.

    - Store the linked-in upper dentry instead of the disconnected
      O_TMPFILE dentry during overlayfs tmpfile copy-up. With a FUSE or
      virtiofs upper layer ->d_revalidate() would try to look up "/" in
      the workdir and fail, causing persistent ESTALE errors that broke
      dpkg and apt.

 - vfs-bpf:

   Have the bpf_real_data_inode() kfunc take a struct file instead of a
   dentry so it is usable from the bprm_check_security, mmap_file, and
   file_mprotect hooks, and rename it from bpf_real_inode() to make the
   data-inode semantics explicit. The kfunc landed this cycle so the
   change is safe.

 - afs:

   NULL pointer dereferences in the callback service and in
   afs_get_tree(), several memory and refcount leaks, missing locking
   around the dynamic root inode numbers and premature cell exposure
   through /afs, a netns destruction hang caused by a misplaced
   increment of net->cells_outstanding, a bulk lookup malfunction caused
   by the dir_emit() API change, inode (re)initialisation issues, and
   assorted smaller fixes to error codes, seqlock handling, and debug
   output.

 - vfs:

   Refuse O_TMPFILE creation with an unmapped fsuid or fsgid and add a
   selftest for it.

 - vboxsf:

   Add Jori Koolstra as vboxsf maintainer, taking over from Hans de
   Goede.

 - dio:

   Release the pages attached to a short atomic dio bio; the REQ_ATOMIC
   size check error path leaked them.

 - procfs:

   Only bump the parent directory link count when registering
   directories in procfs. Registering regular files inflated the count
   and leaked a link on every create and remove cycle.

 - minix:

   Avoid an unsigned overflow in the minix bitmap block count
   calculation that let crafted images with huge inode or zone counts
   pass superblock validation and crash the kernel during mount.

 - cachefiles:

   Fix a double unlock in the cachefiles nomem_d_alloc error path left
   over from the start_creating() conversion.

 - fat:

   Stop fat from reading directory entries past the 0x00
   end-of-directory marker. If the trailing on-disk slots aren't
   zero-filled the driver surfaced arbitrary garbage as directory
   entries.

 - freexvfs:

   Don't BUG() on unknown typed-extent types in freevxfs, reachable via
   ioctl(FIBMAP) on a crafted image; fail with an I/O error instead.

 - orangefs:

   Keep the readdir entry size 64-bit in orangefs fill_from_part().
   Truncating it to __u32 bypassed the bounds check and led to
   out-of-bounds reads triggerable by the userspace client.

 - xfs:

   Fix the error unwind in xfs_open_devices() which released the rt
   device file twice and left dangling buftarg pointers behind that were
   freed again when the failed mount was torn down.

 - exec:

   Fix an off-by-one in the comment documenting the maximum binfmt
   rewrite depth in exec_binprm(). The code allows five rewrites, not
   four; restricting the code would break userspace so the comment is
   fixed instead.

 - file handles:

   Reject detached mounts in capable_wrt_mount(). A detached mount can
   be dissolved concurrently, leaving a NULL mount namespace that
   open_by_handle_at() would dereference.

* tag 'vfs-7.2-rc2.fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs: (57 commits)
  netfs: Fix barriering when walking subrequest list
  iomap: submit read bio after each extent
  fuse: call fuse_send_readpages explicitly from fuse_readahead
  iomap: consolidate bio submission
  fhandle: reject detached mounts in capable_wrt_mount()
  netfs: Fix DIO write retry for filesystems without a ->prepare_write()
  netfs: Fix folio state after ENOMEM whilst under writeback iteration
  netfs: Fix writeback error handling
  netfs: Fix writethrough to use collection offload
  netfs: Replace wb_lock with a bit lock for asynchronicity
  netfs: Fix kdoc warning
  scatterlist: Fix offset in folio calc in extract_xarray_to_sg()
  iov_iter: Remove unused variable in kunit_iov_iter.c
  iov_iter: Fix a memory leak in iov_iter_extract_user_pages()
  iov_iter: Fix missing alloc fail check in iov_iter_extract_bvec_pages()
  iov_iter: Fix potential underflow in iov_iter_extract_xarray_pages()
  cachefiles: Fix file burial to take lock when unsetting S_KERNEL_FILE
  cachefiles: Fix double fput
  netfs: Fix netfs_create_write_req() to handle async cache object creation
  netfs: Fix decision whether to disallow write-streaming due to fscache use
  ...

iomap: consolidate bio submission

2026-07-01T13:26:47+00:00

Add a iomap_bio_submit_read_endio helper factored out of
iomap_bio_submit_read to that all ->submit_read implementations for
iomap_read_ops that use iomap_bio_read_folio_range can shared the
logic.

Right now that logic is mostly trivial, but already has a bug for XFS
because the XFS version is too trivial:  file system integrity validation
needs a workqueue context and thus can't happen from the default iomap
bi_end_io I/O handler.  Unfortunately the iomap refactoring just before
fs integrity landed moved code around here and the call go misplaced,
meaning it never got called.  The PI information still is verified by
the block layer, but the offloading is less efficient (and the future
userspace interface can't get at it).

Fixes: 0b10a370529c ("iomap: support T10 protection information")
Cc: stable@vger.kernel.org # v7.1
Signed-off-by: Christoph Hellwig 
Link: https://patch.msgid.link/20260629121750.3392300-2-hch@lst.de
Acked-by: Namjae Jeon 
Reviewed-by: "Darrick J. Wong" 
Reviewed-by: Joanne Koong 
Signed-off-by: Christian Brauner (Amutable)

xfs: fix the error unwind in xfs_open_devices()

2026-07-01T13:26:27+00:00

Since the rt and log block devices are closed in xfs_free_buftarg() the
buftarg owns the device file. The error unwind does not respect that:
when the log buftarg allocation fails, out_free_rtdev_targ frees the rt
buftarg - releasing rtdev_file - and then falls through to
out_close_rtdev and releases it a second time.

The unwind also leaves mp->m_rtdev_targp and mp->m_ddev_targp pointing
to the freed buftargs. The failed mount continues into
deactivate_locked_super() -> xfs_kill_sb() -> xfs_mount_free(), which
frees them again.

Clear the buftarg pointers once the unwind freed them and clear
rtdev_file once the rt buftarg owns it, so nothing is released twice.

Reachable when a buftarg allocation fails after the data buftarg was
set up: an I/O error in sync_blockdev() or an allocation failure in
xfs_init_buftarg() while mounting with external rt and log devices.

Link: https://patch.msgid.link/20260616-work-super-bdev_holder_global-v2-1-7df6b864028e@kernel.org
Fixes: 41233576e9a4 ("xfs: close the RT and log block devices in xfs_free_buftarg")
Signed-off-by: Christian Brauner (Amutable)

xfs: simplify __xfs_buf_ioend

2026-07-01T10:29:50+00:00

__xfs_buf_ioend can only resubmit the buffer for asynchronous
writes, which means the retry handling xfs_buf_iowait is not needed.

Because of this can stop returning a value from __xfs_buf_ioend and
just release the buffer for async I/O that does not require retries.

Also drop the __-prefix now that the semantics are straight forward.

Signed-off-by: Christoph Hellwig 
Reviewed-by: Carlos Maiolino 
Reviewed-by: "Darrick J. Wong" 
Signed-off-by: Carlos Maiolino

xfs: fix handling of synchronous errors in xfs_buf_submit

2026-07-01T10:29:50+00:00

Synchronous readers and writers already run __xfs_buf_ioend from
xfs_buf_iowait after being woken through bp->b_iowait, so we
should not call it here, which can lead to double completions.

Fixes: 4b90de5bc0f5 ("xfs: reduce context switches for synchronous buffered I/O")
Signed-off-by: Christoph Hellwig 
Reviewed-by: Carlos Maiolino 
Reviewed-by: "Darrick J. Wong" 
Signed-off-by: Carlos Maiolino

xfs: remove xfs_buf_ioend

2026-07-01T10:29:50+00:00

There are two callers of xfs_buf_ioend, one of which always has the
XBF_ASYNC flag set.  Open code the logic in both callers to prepare for a
bug fix.

Signed-off-by: Christoph Hellwig 
Reviewed-by: Carlos Maiolino 
Reviewed-by: "Darrick J. Wong" 
Signed-off-by: Carlos Maiolino

xfs: improve the xfs_buf_ioend_fail calling convention

2026-07-01T10:29:50+00:00

Move setting the ASYNC flag into xfs_buf_ioend_fail, assert that the
buffer is locked as expected, and drop the confusing _ioend in the
name.

Signed-off-by: Christoph Hellwig 
Reviewed-by: Carlos Maiolino 
Reviewed-by: "Darrick J. Wong" 
Signed-off-by: Carlos Maiolino

xfs: use null daddr for unset first bad log block

2026-07-01T09:39:32+00:00

xlog_do_recovery_pass() may return before setting first_bad.  The caller
must distinguish that case from an error at a valid log block, including
block zero after the log wraps.

Initialize first_bad to XFS_BUF_DADDR_NULL and test it explicitly before
treating the error as a torn write.

Fixes: 7088c4136fa1 ("xfs: detect and trim torn writes during log recovery")
Suggested-by: Darrick J. Wong 
Reported-by: syzbot+b7dfbed0c6c2b5e9fd34@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=b7dfbed0c6c2b5e9fd34
Cc: stable@vger.kernel.org # v4.5
Signed-off-by: Yousef Alhouseen 
Reviewed-by: "Darrick J. Wong" 
Signed-off-by: Carlos Maiolino

xfs: fix memory leak in xfs_dqinode_metadir_create()

2026-07-01T09:37:54+00:00

If xfs_metadir_create() fails in xfs_dqinode_metadir_create(), the current
code returns directly, leaking the allocated update and transaction state.
If the subsequent commit fails, the caller-owned inode reference is left
behind.

Fix this memory leak by routing the create failure path through
xfs_metadir_cancel().  For both create and commit failures, finish and
release any inode returned to the caller, mirroring the unwind pattern in
xfs_metadir_mkdir().

The bug was first flagged by an experimental analysis tool we are
developing for kernel memory-management bugs while analyzing
v6.13-rc1. The tool is still under development and is not yet publicly
available. Manual inspection confirms that the bug is still
present in v7.1.1.

An x86_64 allyesconfig build showed no new warnings. Runtime validation
used kprobe fault injection during `mount -o uquota` on a metadir XFS
image. Injecting xfs_metadir_create() reproduced the old active-update path
that left mount stuck later in mount setup; after this change, the same
injection reported cancel_hits=1 and irele_hits=1. Injecting
xfs_metadir_commit() exercised the old inode-reference leak path; after
this change, it reported irele_hits=1.

Fixes: e80fbe1ad8ef ("xfs: use metadir for quota inodes")
Cc: stable@vger.kernel.org # v6.13
Signed-off-by: Dawei Feng 
Reviewed-by: "Darrick J. Wong" 
Signed-off-by: Carlos Maiolino

xfs: release dquot buffer after dqflush failure

2026-07-01T09:35:50+00:00

xfs_qm_dqpurge() gets a locked buffer from xfs_dquot_use_attached_buf().
If xfs_qm_dqflush() fails, the error path skips xfs_buf_relse() and then
calls xfs_dquot_detach_buf(), which tries to lock the same buffer again.

Release the buffer after xfs_qm_dqflush() returns so the error path drops
the caller hold and unlocks the buffer before the dquot is detached,
matching the other dqflush callers.

Fixes: a40fe30868ba ("xfs: separate dquot buffer reads from xfs_dqflush")
Cc: stable@vger.kernel.org # v6.13+
Signed-off-by: Yingjie Gao 
Reviewed-by: "Darrick J. Wong" 
Signed-off-by: Carlos Maiolino