ALSA: seq: Don't break snd_use_lock_sync() loop by timeout
authorTakashi Iwai <tiwai@suse.de>
Sun, 9 Apr 2017 08:41:27 +0000 (10:41 +0200)
committerDanny Wood <danwood76@gmail.com>
Tue, 29 Jan 2019 13:16:53 +0000 (13:16 +0000)
commitf183673260cf292115bdea520f618f3e2bdd30a9
treefe3be9fb4faaceef6f5c5161fa4f15c6a25ad9ca
parentf3052fbe1232f316a3d93dcfc433a323016b9763
ALSA: seq: Don't break snd_use_lock_sync() loop by timeout

commit 4e7655fd4f47c23e5249ea260dc802f909a64611 upstream.

The snd_use_lock_sync() (thus its implementation
snd_use_lock_sync_helper()) has the 5 seconds timeout to break out of
the sync loop.  It was introduced from the beginning, just to be
"safer", in terms of avoiding the stupid bugs.

However, as Ben Hutchings suggested, this timeout rather introduces a
potential leak or use-after-free that was apparently fixed by the
commit 2d7d54002e39 ("ALSA: seq: Fix race during FIFO resize"):
for example, snd_seq_fifo_event_in() -> snd_seq_event_dup() ->
copy_from_user() could block for a long time, and snd_use_lock_sync()
goes timeout and still leaves the cell at releasing the pool.

For fixing such a problem, we remove the break by the timeout while
still keeping the warning.

Suggested-by: Ben Hutchings <ben.hutchings@codethink.co.uk>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Willy Tarreau <w@1wt.eu>
sound/core/seq/seq_lock.c