Re: [2.6.38-rc5, patch] fix cfg80211_wext_siwfreq lock ordering...

[Daniel J Blueman]

On 22 February 2011 00:25, Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote:
> On Tue, 2011-02-22 at 00:11 +0800, Daniel J Blueman wrote:
>> I previously managed to reproduce a hang while scanning wireless
>> channels (reproducible with airodump-ng hopping channels); subsequent
>> lockdep instrumentation revealed a lock ordering issue [1].
>>
>> Without knowing the design intent, it looks like the locks should be
>> taken in reverse order; please comment.
>>
>> Signed-off-by: Daniel J Blueman <daniel.blueman@xxxxxxxxx>
>>
>> --- [1]
>
> Yeah, looks this way, thanks.
>
> Acked-by: Johannes Berg <johannes@xxxxxxxxxxxxxxxx>
>
>> =======================================================
>> [ INFO: possible circular locking dependency detected ]
>> 2.6.38-rc5-341cd #4
>> -------------------------------------------------------
>> airodump-ng/15445 is trying to acquire lock:
>>  (&rdev->devlist_mtx){+.+.+.}, at: [<ffffffff816b1266>]
>> cfg80211_wext_siwfreq+0xc6/0x100
>>
>> but task is already holding lock:
>>  (&wdev->mtx){+.+.+.}, at: [<ffffffff816b125c>] cfg80211_wext_siwfreq+0xbc/0x100
>>
>> which lock already depends on the new lock.
>>
>> the existing dependency chain (in reverse order) is:
>>
>> -> #1 (&wdev->mtx){+.+.+.}:
>>        [<ffffffff810a79d6>] lock_acquire+0xc6/0x280
>>        [<ffffffff816d6bce>] mutex_lock_nested+0x6e/0x4b0
>>        [<ffffffff81696080>] cfg80211_netdev_notifier_call+0x430/0x5f0
>>        [<ffffffff8109351b>] notifier_call_chain+0x8b/0x100
>>        [<ffffffff810935b1>] raw_notifier_call_chain+0x11/0x20
>>        [<ffffffff81576d92>] call_netdevice_notifiers+0x32/0x60
>>        [<ffffffff815771a4>] __dev_notify_flags+0x34/0x80
>>        [<ffffffff81577230>] dev_change_flags+0x40/0x70
>>        [<ffffffff8158587c>] do_setlink+0x1fc/0x8d0
>>        [<ffffffff81586042>] rtnl_setlink+0xf2/0x140
>>        [<ffffffff81586923>] rtnetlink_rcv_msg+0x163/0x270
>>        [<ffffffff8159d741>] netlink_rcv_skb+0xa1/0xd0
>>        [<ffffffff815867b0>] rtnetlink_rcv+0x20/0x30
>>        [<ffffffff8159d39a>] netlink_unicast+0x2ba/0x300
>>        [<ffffffff8159dd57>] netlink_sendmsg+0x267/0x3e0
>>        [<ffffffff8155e364>] sock_sendmsg+0xe4/0x110
>>        [<ffffffff8155f3a3>] sys_sendmsg+0x253/0x3b0
>>        [<ffffffff81003192>] system_call_fastpath+0x16/0x1b
>>
>> -> #0 (&rdev->devlist_mtx){+.+.+.}:
>>        [<ffffffff810a7222>] __lock_acquire+0x1622/0x1d10
>>        [<ffffffff810a79d6>] lock_acquire+0xc6/0x280
>>        [<ffffffff816d6bce>] mutex_lock_nested+0x6e/0x4b0
>>        [<ffffffff816b1266>] cfg80211_wext_siwfreq+0xc6/0x100
>>        [<ffffffff816b2fad>] ioctl_standard_call+0x5d/0xd0
>>        [<ffffffff816b3223>] T.808+0x163/0x170
>>        [<ffffffff816b326a>] wext_handle_ioctl+0x3a/0x90
>>        [<ffffffff815798d2>] dev_ioctl+0x6f2/0x830
>>        [<ffffffff8155cf3d>] sock_ioctl+0xfd/0x290
>>        [<ffffffff8117dffd>] do_vfs_ioctl+0x9d/0x590
>>        [<ffffffff8117e53a>] sys_ioctl+0x4a/0x80
>>        [<ffffffff81003192>] system_call_fastpath+0x16/0x1b
>>
>> other info that might help us debug this:
>>
>> 2 locks held by airodump-ng/15445:
>>  #0:  (rtnl_mutex){+.+.+.}, at: [<ffffffff81586782>] rtnl_lock+0x12/0x20
>>  #1:  (&wdev->mtx){+.+.+.}, at: [<ffffffff816b125c>]
>> cfg80211_wext_siwfreq+0xbc/0x100
>>
>> stack backtrace:
>> Pid: 15445, comm: airodump-ng Not tainted 2.6.38-rc5-341cd #4
>> Call Trace:
>>  [<ffffffff810a3f0a>] ? print_circular_bug+0xfa/0x100
>>  [<ffffffff810a7222>] ? __lock_acquire+0x1622/0x1d10
>>  [<ffffffff810a1f99>] ? trace_hardirqs_off_caller+0x29/0xc0
>>  [<ffffffff810a79d6>] ? lock_acquire+0xc6/0x280
>>  [<ffffffff816b1266>] ? cfg80211_wext_siwfreq+0xc6/0x100
>>  [<ffffffff810a31d7>] ? mark_held_locks+0x67/0x90
>>  [<ffffffff816d6bce>] ? mutex_lock_nested+0x6e/0x4b0
>>  [<ffffffff816b1266>] ? cfg80211_wext_siwfreq+0xc6/0x100
>>  [<ffffffff810a31d7>] ? mark_held_locks+0x67/0x90
>>  [<ffffffff816b1266>] ? cfg80211_wext_siwfreq+0xc6/0x100
>>  [<ffffffff816b1266>] ? cfg80211_wext_siwfreq+0xc6/0x100
>>  [<ffffffff816b2fad>] ? ioctl_standard_call+0x5d/0xd0
>>  [<ffffffff8157818b>] ? __dev_get_by_name+0x9b/0xc0
>>  [<ffffffff816b2f50>] ? ioctl_standard_call+0x0/0xd0
>>  [<ffffffff816b3223>] ? T.808+0x163/0x170
>>  [<ffffffff8112ddf2>] ? might_fault+0x72/0xd0
>>  [<ffffffff816b326a>] ? wext_handle_ioctl+0x3a/0x90
>>  [<ffffffff8112de3b>] ? might_fault+0xbb/0xd0
>>  [<ffffffff815798d2>] ? dev_ioctl+0x6f2/0x830
>>  [<ffffffff810a1bae>] ? put_lock_stats+0xe/0x40
>>  [<ffffffff810a1c8c>] ? lock_release_holdtime+0xac/0x150
>>  [<ffffffff8155cf3d>] ? sock_ioctl+0xfd/0x290
>>  [<ffffffff8117dffd>] ? do_vfs_ioctl+0x9d/0x590
>>  [<ffffffff8116c8ff>] ? fget_light+0x1df/0x3c0
>>  [<ffffffff8117e53a>] ? sys_ioctl+0x4a/0x80
>>  [<ffffffff81003192>] ? system_call_fastpath+0x16/0x1b
>>
>> --- [2]
>>
>> diff --git a/net/wireless/wext-compat.c b/net/wireless/wext-compat.c
>> index 3e5dbd4..d112f03 100644
>> --- a/net/wireless/wext-compat.c
>> +++ b/net/wireless/wext-compat.c
>> @@ -802,11 +802,11 @@ int cfg80211_wext_siwfreq(struct net_device *dev,
>>                       return freq;
>>               if (freq == 0)
>>                       return -EINVAL;
>> -             wdev_lock(wdev);
>>               mutex_lock(&rdev->devlist_mtx);
>> +             wdev_lock(wdev);
>>               err = cfg80211_set_freq(rdev, wdev, freq, NL80211_CHAN_NO_HT);
>> -             mutex_unlock(&rdev->devlist_mtx);
>>               wdev_unlock(wdev);
>> +             mutex_unlock(&rdev->devlist_mtx);
>>               return err;
>>       default:
>>               return -EOPNOTSUPP;

Please consider for -stable. This patch resolves the lock ordering
case my test exposed, and passes lockdep and extended testing.

Thanks,
  Daniel
-- 
Daniel J Blueman
--
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/