]> git.ipfire.org Git - thirdparty/linux.git/commit
pwrseq: core: fix use-after-free in pwrseq_debugfs_seq_next()
authorWentao Liang <vulab@iscas.ac.cn>
Tue, 16 Jun 2026 15:10:49 +0000 (15:10 +0000)
committerBartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
Mon, 22 Jun 2026 08:01:08 +0000 (10:01 +0200)
commit257595adf9dac15ae1edd9d07753fbc576a7583d
treec9852d24d5a1b6d3a7b0ec7ebcc13c7155c7d139
parent31e6aeafcdde965aa10e10e93ee186520555ec3d
pwrseq: core: fix use-after-free in pwrseq_debugfs_seq_next()

pwrseq_debugfs_seq_next() declares 'next' with __free(put_device),
which causes put_device() to be called on the returned pointer when
the variable goes out of scope.  This results in a use-after-free
since the seq_file framework receives a pointer whose reference has
already been dropped.

Simply removing __free(put_device) would fix the UAF but would leak
the reference acquired by bus_find_next_device(), as stop() only
calls up_read(&pwrseq_sem) and never releases the device reference.

Fix this by making the reference counting consistent across all
seq_file callbacks, matching the standard pattern used by PCI and
SCSI:

- start(): use get_device() so it returns a referenced pointer.
- next(): explicitly put_device(curr) to release the previous
  device's reference (no NULL check needed - the seq_file framework
  only calls next() while the previous return was non-NULL).
- stop(): put_device(data) to release the last iterated device's
  reference, with a NULL guard since stop() may be called with NULL
  when start() returned NULL or next() reached end-of-sequence.

Cc: stable@vger.kernel.org
Fixes: 249ebf3f65f8 ("power: sequencing: implement the pwrseq core")
Signed-off-by: Wentao Liang <vulab@iscas.ac.cn>
Link: https://patch.msgid.link/20260616151049.1705503-1-vulab@iscas.ac.cn
Signed-off-by: Bartosz Golaszewski <bartosz.golaszewski@oss.qualcomm.com>
drivers/power/sequencing/core.c