Re: 3.8: possible circular locking dependency detected

From: Michael Ellerman
Date: Mon Feb 04 2013 - 08:05:27 EST

On Sun, 2013-02-03 at 21:09 -0800, Christian Kujau wrote:
> Hi,
>
> similar to what I reported earlier [0] for 3.8.0-rc1, this happens during
> "ifup wlan0" (which in effect starts wpa_supplicant to bring up a Broadcom
> b43 wifi network interface). The interface is working though and continues
> to work over several ifup/ifdown iterations.
>
> The backtrace looks awfully similar to the earlier[0] report, but this
> time it had b43* stuff in it so I thought I should report it. Full dmesg
> and .config here: http://nerdbynature.de/bits/3.8.0-rc6/

>[0] https://lkml.org/lkml/2013/1/3/543

Actually the backtrace looks very different, that was fb vs
console_lock.

This one should probably go to linux-wireless@xxxxxxxxxxxxxxx (CC-ed).

IIUI it's actually the work handling that is the problem. In
b43_wireless_core_stop() it calls cancel_work_sync() with the RTNL held,
but the work function (b43_request_firmware()) also takes the RTNL via
ieee80211_register_hw() and wiphy_register(). eg:

> [ 807.767412] CPU0 CPU1
> [ 807.768561] ---- ----
> [ 807.769690] lock(rtnl_mutex);
> [ 807.770822] process_one_work(..., &wl->firmware_load);
> [ 807.771970] lock(rtnl_mutex);
> [ 807.773115] cancel_work_sync(&wl->firmware_load);

cheers

> [ 807.693791]
> [ 807.695519] ======================================================
> [ 807.697198] [ INFO: possible circular locking dependency detected ]
> [ 807.698890] 3.8.0-rc6-00008-g8b31849 #1 Not tainted
> [ 807.700573] -------------------------------------------------------
> [ 807.702255] wpa_supplicant/4129 is trying to acquire lock:
> [ 807.703925] ((&wl->firmware_load)){+.+.+.}, at: [<c0049d1c>] flush_work+0x0/0x2b0
> [ 807.705696]
> [ 807.705696] but task is already holding lock:
> [ 807.709023] (rtnl_mutex){+.+.+.}, at: [<c04362d4>] rtnl_lock+0x1c/0x2c
> [ 807.710743]
> [ 807.710743] which lock already depends on the new lock.
> [ 807.710743]
> [ 807.715541]
> [ 807.715541] the existing dependency chain (in reverse order) is:
> [ 807.718691]
> [ 807.718691] -> #1 (rtnl_mutex){+.+.+.}:
> [ 807.721903] [<c04f69ac>] mutex_lock_nested+0x6c/0x2bc
> [ 807.723533] [<c04362d4>] rtnl_lock+0x1c/0x2c
> [ 807.725138] [<f20be0d8>] wiphy_register+0x510/0x53c [cfg80211]
> [ 807.726798] [<f64815ec>] ieee80211_register_hw+0x3f8/0x82c [mac80211]
> [ 807.728431] [<f64c6890>] b43_request_firmware+0x8c/0x198 [b43]
> [ 807.730025] [<c004b1f4>] process_one_work+0x1a4/0x498
> [ 807.731549] [<c004b8b4>] worker_thread+0x17c/0x428
> [ 807.733025] [<c0051280>] kthread+0xa8/0xac
> [ 807.734439] [<c0010990>] ret_from_kernel_thread+0x64/0x6c
> [ 807.735810]
> [ 807.735810] -> #0 ((&wl->firmware_load)){+.+.+.}:
> [ 807.738389] [<c0071b38>] lock_acquire+0x50/0x6c
> [ 807.739694] [<c0049d58>] flush_work+0x3c/0x2b0
> [ 807.740980] [<c004c30c>] __cancel_work_timer+0x94/0xec
> [ 807.742271] [<f64c748c>] b43_wireless_core_stop+0x5c/0x234 [b43]
> [ 807.743574] [<f64c76b0>] b43_op_stop+0x4c/0x88 [b43]
> [ 807.744888] [<f64a4c08>] ieee80211_stop_device+0x4c/0x8c [mac80211]
> [ 807.746240] [<f64913a8>] ieee80211_do_stop+0x2c0/0x5dc [mac80211]
> [ 807.747582] [<f64916dc>] ieee80211_stop+0x18/0x2c [mac80211]
> [ 807.748925] [<c0423d94>] __dev_close_many+0xb0/0x100
> [ 807.750257] [<c0423e10>] __dev_close+0x2c/0x4c
> [ 807.751560] [<c0427514>] __dev_change_flags+0x124/0x178
> [ 807.752868] [<c042761c>] dev_change_flags+0x1c/0x64
> [ 807.754177] [<c047db9c>] devinet_ioctl+0x69c/0x74c
> [ 807.755459] [<c047ecf8>] inet_ioctl+0xcc/0xf8
> [ 807.756709] [<c040eab0>] sock_ioctl+0x70/0x2e8
> [ 807.757948] [<c00d9f70>] do_vfs_ioctl+0xa4/0x7c0
> [ 807.759182] [<c00da6d0>] sys_ioctl+0x44/0x70
> [ 807.760407] [<c001084c>] ret_from_syscall+0x0/0x38
> [ 807.761622]
> [ 807.761622] other info that might help us debug this:
> [ 807.761622]
> [ 807.765107] Possible unsafe locking scenario:
> [ 807.765107]
> [ 807.767412] CPU0 CPU1
> [ 807.768561] ---- ----
> [ 807.769690] lock(rtnl_mutex);
> [ 807.770822] lock((&wl->firmware_load));
> [ 807.771970] lock(rtnl_mutex);
> [ 807.773115] lock((&wl->firmware_load));
> [ 807.774244]
> [ 807.774244] *** DEADLOCK ***
> [ 807.774244]
> [ 807.777405] 1 lock held by wpa_supplicant/4129:
> [ 807.778475] #0: (rtnl_mutex){+.+.+.}, at: [<c04362d4>] rtnl_lock+0x1c/0x2c
> [ 807.779628]
> [ 807.779628] stack backtrace:
> [ 807.781720] Call Trace:
> [ 807.782765] [eea2db20] [c0009160] show_stack+0x48/0x15c (unreliable)
> [ 807.784087] [eea2db60] [c04fae24] print_circular_bug+0x2b0/0x2c8
> [ 807.785169] [eea2db90] [c0071300] __lock_acquire+0x14f4/0x18c8
> [ 807.786254] [eea2dc30] [c0071b38] lock_acquire+0x50/0x6c
> [ 807.787335] [eea2dc50] [c0049d58] flush_work+0x3c/0x2b0
> [ 807.788418] [eea2dcc0] [c004c30c] __cancel_work_timer+0x94/0xec
> [ 807.789516] [eea2dcf0] [f64c748c] b43_wireless_core_stop+0x5c/0x234 [b43]
> [ 807.790629] [eea2dd20] [f64c76b0] b43_op_stop+0x4c/0x88 [b43]
> [ 807.791754] [eea2dd40] [f64a4c08] ieee80211_stop_device+0x4c/0x8c [mac80211]
> [ 807.792888] [eea2dd50] [f64913a8] ieee80211_do_stop+0x2c0/0x5dc [mac80211]
> [ 807.794016] [eea2dd80] [f64916dc] ieee80211_stop+0x18/0x2c [mac80211]
> [ 807.795124] [eea2dd90] [c0423d94] __dev_close_many+0xb0/0x100
> [ 807.796237] [eea2ddb0] [c0423e10] __dev_close+0x2c/0x4c
> [ 807.797345] [eea2ddd0] [c0427514] __dev_change_flags+0x124/0x178
> [ 807.798451] [eea2ddf0] [c042761c] dev_change_flags+0x1c/0x64
> [ 807.799564] [eea2de10] [c047db9c] devinet_ioctl+0x69c/0x74c
> [ 807.800680] [eea2de70] [c047ecf8] inet_ioctl+0xcc/0xf8
> [ 807.801792] [eea2de80] [c040eab0] sock_ioctl+0x70/0x2e8
> [ 807.802905] [eea2dea0] [c00d9f70] do_vfs_ioctl+0xa4/0x7c0
> [ 807.804025] [eea2df10] [c00da6d0] sys_ioctl+0x44/0x70
> [ 807.805145] [eea2df40] [c001084c] ret_from_syscall+0x0/0x38
> [ 807.806270] --- Exception: c01 at 0xfbf2758
> [ 807.806270] LR = 0xfbf26bc
>
> [ 850.445766] b43-phy1: Broadcom 4306 WLAN found (core revision 5)
> [ 850.469062] b43-phy1: Found PHY: Analog 2, Type 2 (G), Revision 2
> [ 850.493199] Broadcom 43xx driver loaded [ Features: P ]
> [ 850.497448] ieee80211 phy1: Selected rate control algorithm 'minstrel_ht'
> [ 860.820792] b43-phy1: Loading firmware version 410.2160 (2007-05-26 15:32:10)
>
>




--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/