Commit 2f4c63e
sound: use-after-free in snd_seq_queue_alloc
On Wed, 08 Feb 2017 10:41:14 +0100,
Dmitry Vyukov wrote:
>
> Hello,
>
>
> I've got the following report while running syzkaller fuzzer on
> 8b1b41e:
>
> BUG: KASAN: use-after-free in snd_seq_queue_alloc+0x663/0x690
> sound/core/seq/seq_queue.c:200 at addr ffff880086ba1d00
> Read of size 4 by task syz-executor2/31851
> CPU: 2 PID: 31851 Comm: syz-executor2 Not tainted 4.10.0-rc5+ torvalds#201
> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
>
> Call Trace:
> __asan_report_load4_noabort+0x3e/0x40 mm/kasan/report.c:327
> snd_seq_queue_alloc+0x663/0x690 sound/core/seq/seq_queue.c:200
> snd_seq_ioctl_create_queue+0xad/0x310 sound/core/seq/seq_clientmgr.c:1508
> snd_seq_ioctl+0x2da/0x4d0 sound/core/seq/seq_clientmgr.c:2130
> vfs_ioctl fs/ioctl.c:43 [inline]
> do_vfs_ioctl+0x1bf/0x1790 fs/ioctl.c:683
> SYSC_ioctl fs/ioctl.c:698 [inline]
> SyS_ioctl+0x8f/0xc0 fs/ioctl.c:689
> entry_SYSCALL_64_fastpath+0x1f/0xc2
>
> Allocated:
> PID = 31851
> [<ffffffff83483a17>] kzalloc include/linux/slab.h:490 [inline]
> [<ffffffff83483a17>] queue_new sound/core/seq/seq_queue.c:113 [inline]
> [<ffffffff83483a17>] snd_seq_queue_alloc+0x107/0x690
> sound/core/seq/seq_queue.c:191
> [<ffffffff834748dd>] snd_seq_ioctl_create_queue+0xad/0x310
> sound/core/seq/seq_clientmgr.c:1508
> [<ffffffff8347878a>] snd_seq_ioctl+0x2da/0x4d0
> sound/core/seq/seq_clientmgr.c:2130
> [<ffffffff81aa41cf>] vfs_ioctl fs/ioctl.c:43 [inline]
> [<ffffffff81aa41cf>] do_vfs_ioctl+0x1bf/0x1790 fs/ioctl.c:683
> [<ffffffff81aa582f>] SYSC_ioctl fs/ioctl.c:698 [inline]
> [<ffffffff81aa582f>] SyS_ioctl+0x8f/0xc0 fs/ioctl.c:689
> [<ffffffff841c9c81>] entry_SYSCALL_64_fastpath+0x1f/0xc2
>
> Freed:
> PID = 31854
> [<ffffffff81a0b5c3>] kfree+0xd3/0x250 mm/slab.c:3822
> [<ffffffff834817e0>] queue_delete+0x90/0xb0 sound/core/seq/seq_queue.c:156
> [<ffffffff834826cc>] snd_seq_queue_delete+0x3c/0x50
> sound/core/seq/seq_queue.c:213
> [<ffffffff8347480a>] snd_seq_ioctl_delete_queue+0x6a/0x90
> sound/core/seq/seq_clientmgr.c:1534
> [<ffffffff8347878a>] snd_seq_ioctl+0x2da/0x4d0
> sound/core/seq/seq_clientmgr.c:2130
> [<ffffffff81aa41cf>] vfs_ioctl fs/ioctl.c:43 [inline]
> [<ffffffff81aa41cf>] do_vfs_ioctl+0x1bf/0x1790 fs/ioctl.c:683
> [<ffffffff81aa582f>] SYSC_ioctl fs/ioctl.c:698 [inline]
> [<ffffffff81aa582f>] SyS_ioctl+0x8f/0xc0 fs/ioctl.c:689
>
>
>
> Looking at the code:
>
> int snd_seq_queue_alloc(int client, int locked, unsigned int info_flags)
> {
> struct snd_seq_queue *q;
>
> q = queue_new(client, locked);
> if (q == NULL)
> return -ENOMEM;
> q->info_flags = info_flags;
> if (queue_list_add(q) < 0) {
> queue_delete(q);
> return -ENOMEM;
> }
> snd_seq_queue_use(q->queue, client, 1); /* use this queue */
> return q->queue;
> }
>
> After queue_list_add(q) q can be deleted by another thread, so
> snd_seq_queue_use(q->queue, client, 1) already potentially operates on
> deleted object.
A good catch! The fix patch is below.
thanks,
Takashi
-- 8< --
From: Takashi Iwai <tiwai@suse.de>
Subject: [PATCH] ALSA: seq: Fix race at creating a queue
When a sequencer queue is created in snd_seq_queue_alloc(),it adds the
new queue element to the public list before referencing it. Thus the
queue might be deleted before the call of snd_seq_queue_use(), and it
results in the use-after-free error, as spotted by syzkaller.
The fix is to reference the queue object at the right time.
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>1 parent 8d8c9ae commit 2f4c63e
1 file changed
+20
-13
lines changed| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
181 | 181 | | |
182 | 182 | | |
183 | 183 | | |
| 184 | + | |
| 185 | + | |
184 | 186 | | |
185 | 187 | | |
186 | 188 | | |
| |||
192 | 194 | | |
193 | 195 | | |
194 | 196 | | |
| 197 | + | |
195 | 198 | | |
196 | 199 | | |
197 | 200 | | |
198 | 201 | | |
199 | | - | |
200 | 202 | | |
201 | 203 | | |
202 | 204 | | |
| |||
502 | 504 | | |
503 | 505 | | |
504 | 506 | | |
505 | | - | |
506 | | - | |
507 | | - | |
508 | | - | |
509 | | - | |
510 | | - | |
| 507 | + | |
| 508 | + | |
511 | 509 | | |
512 | | - | |
513 | | - | |
514 | | - | |
515 | | - | |
516 | | - | |
517 | | - | |
518 | 510 | | |
519 | 511 | | |
520 | 512 | | |
| |||
529 | 521 | | |
530 | 522 | | |
531 | 523 | | |
| 524 | + | |
| 525 | + | |
| 526 | + | |
| 527 | + | |
| 528 | + | |
| 529 | + | |
| 530 | + | |
| 531 | + | |
| 532 | + | |
| 533 | + | |
| 534 | + | |
| 535 | + | |
| 536 | + | |
| 537 | + | |
| 538 | + | |
532 | 539 | | |
533 | 540 | | |
534 | 541 | | |
| |||
0 commit comments