On Tue, 2008-06-24 at 10:59 +0200, Johannes Berg wrote:

quoted

[ INFO: possible circular locking dependency detected ]
2.6.26-rc6-wl #57
-------------------------------------------------------
iw/2763 is trying to acquire lock:
 (genl_mutex){--..}, at: [<ffffffff81216e4c>] ctrl_dumpfamily

+0x31/0xf7

quoted

but task is already holding lock:
 (nlk->cb_mutex){--..}, at: [<ffffffff812137c8>] netlink_dump

+0x40/0x19d

quoted

which lock already depends on the new lock.

None of those is anything nl80211 has influence over, afaict.

Agreed. I can follow the two scenarios below. Apparently it's a ABBA
locking issue. Just don't know which should be the correct sequence for
using with the two locks.

Thanks,
-yi

quoted

the existing dependency chain (in reverse order) is:

-> #1 (nlk->cb_mutex){--..}:
       [<ffffffff810518cb>] __lock_acquire+0xa4c/0xc28
       [<ffffffff81215774>] netlink_dump_start+0xd9/0x169
       [<ffffffff81051e5d>] lock_acquire+0x51/0x6c
       [<ffffffff81215774>] netlink_dump_start+0xd9/0x169
       [<ffffffff8104eab1>] debug_mutex_lock_common+0x16/0x23
       [<ffffffff8127c622>] mutex_lock_nested+0xde/0x23f
       [<ffffffff81215774>] netlink_dump_start+0xd9/0x169
       [<ffffffff81216d7e>] genl_rcv_msg+0xae/0x14b
       [<ffffffff8127c774>] mutex_lock_nested+0x230/0x23f
       [<ffffffff81216cd0>] genl_rcv_msg+0x0/0x14b
       [<ffffffff8121483a>] netlink_rcv_skb+0x34/0x7c
       [<ffffffff81216fe1>] genl_rcv+0x21/0x30
       [<ffffffff812145ec>] netlink_unicast+0x1f7/0x287
       [<ffffffff811f7eb7>] __alloc_skb+0x84/0x132
       [<ffffffff81214d60>] netlink_sendmsg+0x274/0x287
       [<ffffffff811f1b49>] sock_sendmsg+0xe2/0xff
       [<ffffffff8104582f>] autoremove_wake_function+0x0/0x2e
       [<ffffffff81050878>] mark_held_locks+0x49/0x67
       [<ffffffff811f1d7d>] sys_sendmsg+0x217/0x28a
       [<ffffffff8107de49>] __do_fault+0x35d/0x3a5
       [<ffffffff8100bf10>] tracesys+0xd5/0xda
       [<ffffffffffffffff>] 0xffffffffffffffff

-> #0 (genl_mutex){--..}:
       [<ffffffff81050226>] print_circular_bug_header+0xc8/0xcf
       [<ffffffff810517d0>] __lock_acquire+0x951/0xc28
       [<ffffffff81216e4c>] ctrl_dumpfamily+0x31/0xf7
       [<ffffffff81051e5d>] lock_acquire+0x51/0x6c
       [<ffffffff81216e4c>] ctrl_dumpfamily+0x31/0xf7
       [<ffffffff8104eab1>] debug_mutex_lock_common+0x16/0x23
       [<ffffffff8127c622>] mutex_lock_nested+0xde/0x23f
       [<ffffffff81050a10>] trace_hardirqs_on+0xf3/0x117
       [<ffffffff81216e4c>] ctrl_dumpfamily+0x31/0xf7
       [<ffffffff812137e3>] netlink_dump+0x5b/0x19d
       [<ffffffff81214ef3>] netlink_recvmsg+0x180/0x2a8
       [<ffffffff811f1a48>] sock_recvmsg+0xf0/0x10f
       [<ffffffff8104582f>] autoremove_wake_function+0x0/0x2e
       [<ffffffff81050878>] mark_held_locks+0x49/0x67
       [<ffffffff811f295c>] sys_recvmsg+0x15c/0x254
       [<ffffffff81067ac3>] audit_syscall_entry+0x12d/0x160
       [<ffffffff8127d6da>] trace_hardirqs_on_thunk+0x35/0x3a
       [<ffffffff8100bf10>] tracesys+0xd5/0xda
       [<ffffffffffffffff>] 0xffffffffffffffff

--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html