ALSA: seq: Fix a potential UAF by wrong private_free call order
authorTakashi Iwai <tiwai@suse.de>
Thu, 30 Sep 2021 11:41:14 +0000 (13:41 +0200)
committerTakashi Iwai <tiwai@suse.de>
Thu, 30 Sep 2021 12:13:22 +0000 (14:13 +0200)
commit1f8763c59c4ec6254d629fe77c0a52220bd907aa
treead29f0e77dd7a5454763daeba844810931cc7082
parenteb676622846b34a751e2ff9b5910a5322a4e0000
ALSA: seq: Fix a potential UAF by wrong private_free call order

John Keeping reported and posted a patch for a potential UAF in
rawmidi sequencer destruction: the snd_rawmidi_dev_seq_free() may be
called after the associated rawmidi object got already freed.
After a deeper look, it turned out that the bug is rather the
incorrect private_free call order for a snd_seq_device.  The
snd_seq_device private_free gets called at the release callback of the
sequencer device object, while this was rather expected to be executed
at the snd_device call chains that runs at the beginning of the whole
card-free procedure.  It's been broken since the rewrite of
sequencer-device binding (although it hasn't surfaced because the
sequencer device release happens usually right along with the card
device release).

This patch corrects the private_free call to be done in the right
place, at snd_seq_device_dev_free().

Fixes: 7c37ae5c625a ("ALSA: seq: Rewrite sequencer device binding with standard bus")
Reported-and-tested-by: John Keeping <john@metanate.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20210930114114.8645-1-tiwai@suse.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
sound/core/seq_device.c