sound: deadlock between snd_pcm_oss_write/snd_pcm_oss_mmap

[Dmitry Vyukov]

Hello,

I've got the following deadlock report while running syzkaller fuzzer:

[ INFO: possible circular locking dependency detected ]
4.5.0-rc1+ #305 Not tainted
-------------------------------------------------------
syz-executor/14254 is trying to acquire lock:
 (&runtime->oss.params_lock){+.+.+.}, at: [<ffffffff8528a504>]
snd_pcm_oss_change_params+0xd4/0x3540 sound/core/oss/pcm_oss.c:852

but task is already holding lock:
 (&mm->mmap_sem){++++++}, at: [<ffffffff816b267c>] vm_mmap_pgoff

which lock already depends on the new lock.


the existing dependency chain (in reverse order) is:

-> #1 (&mm->mmap_sem){++++++}:
       [<ffffffff8145b9dc>] lock_acquire+0x1dc/0x430
kernel/locking/lockdep.c:3587
       [<ffffffff816def51>] __might_fault+0x141/0x1d0 mm/memory.c:3802
       [<     inline     >] copy_from_user ./arch/x86/include/asm/uaccess.h:714
       [<     inline     >] snd_pcm_oss_write1 sound/core/oss/pcm_oss.c:1376
       [<ffffffff852940f0>] snd_pcm_oss_write+0x250/0x700
sound/core/oss/pcm_oss.c:2694
       [<ffffffff817b90b3>] __vfs_write+0x113/0x480 fs/read_write.c:528
       [<ffffffff817bab47>] vfs_write+0x167/0x4a0 fs/read_write.c:577
       [<     inline     >] SYSC_write fs/read_write.c:624
       [<ffffffff817bde31>] SyS_write+0x111/0x220 fs/read_write.c:616
       [<ffffffff866531b6>] entry_SYSCALL_64_fastpath+0x16/0x7a
arch/x86/entry/entry_64.S:185

-> #0 (&runtime->oss.params_lock){+.+.+.}:
       [<     inline     >] check_prev_add kernel/locking/lockdep.c:1855
       [<     inline     >] check_prevs_add kernel/locking/lockdep.c:1960
       [<     inline     >] validate_chain kernel/locking/lockdep.c:2146
       [<ffffffff8145807b>] __lock_acquire+0x31eb/0x4700
kernel/locking/lockdep.c:3208
       [<ffffffff8145b9dc>] lock_acquire+0x1dc/0x430
kernel/locking/lockdep.c:3587
       [<     inline     >] __mutex_lock_common kernel/locking/mutex.c:518
       [<ffffffff8664891c>] mutex_lock_interruptible_nested+0xbc/0xbe0
kernel/locking/mutex.c:647
       [<ffffffff8528a504>] snd_pcm_oss_change_params+0xd4/0x3540
sound/core/oss/pcm_oss.c:852
       [<ffffffff8528f01d>] snd_pcm_oss_mmap+0x3dd/0x4c0
sound/core/oss/pcm_oss.c:2807
       [<ffffffff81705747>] mmap_region+0x897/0x1010 mm/mmap.c:1624
       [<ffffffff81706614>] do_mmap+0x754/0x990 mm/mmap.c:1403
       [<     inline     >] do_mmap_pgoff include/linux/mm.h:1982
       [<ffffffff816b26af>] vm_mmap_pgoff+0x15f/0x1b0 mm/util.c:328
       [<     inline     >] SYSC_mmap_pgoff mm/mmap.c:1453
       [<ffffffff816ff85a>] SyS_mmap_pgoff+0x34a/0x580 mm/mmap.c:1411
       [<     inline     >] SYSC_mmap arch/x86/kernel/sys_x86_64.c:95
       [<ffffffff811aeeb6>] SyS_mmap+0x16/0x20 arch/x86/kernel/sys_x86_64.c:86
       [<ffffffff866531b6>] entry_SYSCALL_64_fastpath+0x16/0x7a
arch/x86/entry/entry_64.S:185

other info that might help us debug this:

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  lock(&mm->mmap_sem);
                               lock(&runtime->oss.params_lock);
                               lock(&mm->mmap_sem);
  lock(&runtime->oss.params_lock);

 *** DEADLOCK ***

1 lock held by syz-executor/14254:
 #0:  (&mm->mmap_sem){++++++}, at: [<ffffffff816b267c>]
vm_mmap_pgoff+0x12c/0x1b0 mm/util.c:327

stack backtrace:
CPU: 2 PID: 14254 Comm: syz-executor Not tainted 4.5.0-rc1+ #305
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
 00000000ffffffff ffff88003214f780 ffffffff82be11ad ffffffff8959ac60
 ffffffff8959ac60 ffffffff89573f60 ffff88003214f7d0 ffffffff814512a8
 ffff8800333cdf00 ffff8800333ce742 0000000000000000 ffff8800333ce720
Call Trace:
 [<     inline     >] __dump_stack lib/dump_stack.c:15
 [<ffffffff82be11ad>] dump_stack+0x6f/0xa2 lib/dump_stack.c:50
 [<ffffffff814512a8>] print_circular_bug+0x288/0x340
kernel/locking/lockdep.c:1228
 [<     inline     >] check_prev_add kernel/locking/lockdep.c:1855
 [<     inline     >] check_prevs_add kernel/locking/lockdep.c:1960
 [<     inline     >] validate_chain kernel/locking/lockdep.c:2146
 [<ffffffff8145807b>] __lock_acquire+0x31eb/0x4700 kernel/locking/lockdep.c:3208
 [<ffffffff8145b9dc>] lock_acquire+0x1dc/0x430 kernel/locking/lockdep.c:3587
 [<     inline     >] __mutex_lock_common kernel/locking/mutex.c:518
 [<ffffffff8664891c>] mutex_lock_interruptible_nested+0xbc/0xbe0
kernel/locking/mutex.c:647
 [<ffffffff8528a504>] snd_pcm_oss_change_params+0xd4/0x3540
sound/core/oss/pcm_oss.c:852
 [<ffffffff8528f01d>] snd_pcm_oss_mmap+0x3dd/0x4c0 sound/core/oss/pcm_oss.c:2807
 [<ffffffff81705747>] mmap_region+0x897/0x1010 mm/mmap.c:1624
 [<ffffffff81706614>] do_mmap+0x754/0x990 mm/mmap.c:1403
 [<     inline     >] do_mmap_pgoff include/linux/mm.h:1982
 [<ffffffff816b26af>] vm_mmap_pgoff+0x15f/0x1b0 mm/util.c:328
 [<     inline     >] SYSC_mmap_pgoff mm/mmap.c:1453
 [<ffffffff816ff85a>] SyS_mmap_pgoff+0x34a/0x580 mm/mmap.c:1411
 [<     inline     >] SYSC_mmap arch/x86/kernel/sys_x86_64.c:95
 [<ffffffff811aeeb6>] SyS_mmap+0x16/0x20 arch/x86/kernel/sys_x86_64.c:86


On commit 26cd83670f2f5a3d5b5514a1f7d96567cdb9558b.