Re: Kernel WARNING: at net/core/dev.c:1330 __netif_schedule+0x2c/0x98()

[Jarek Poplawski <jarkao2@gmail.com>]

On 23-07-2008 01:04, David Miller wrote:
> From: Larry Finger <Larry.Finger@lwfinger.net>
> Date: Tue, 22 Jul 2008 13:39:08 -0500
> 
>> =============================================
>> [ INFO: possible recursive locking detected ]
>> 2.6.26-Linus-05752-g93ded9b-dirty #53
>> ---------------------------------------------
>> b43/1997 is trying to acquire lock:
>>   (_xmit_IEEE80211#2){-...}, at: [<ffffffffa028f322>] 
>> ieee80211_scan_completed+0x130/0x2e1 [mac80211]
>>
>> but task is already holding lock:
>>   (_xmit_IEEE80211#2){-...}, at: [<ffffffffa028f322>] 
>> ieee80211_scan_completed+0x130/0x2e1 [mac80211]
...
> Lockdep doesn't like that we have an array of objects (the TX queues)
> and we're iterating over them grabbing all of their locks.
> 
> Does anyone know how to teach lockdep that this is OK?

I guess, David Miller knows...:

http://permalink.gmane.org/gmane.linux.network/99784

Jarek P.

PS: if there is nothing new in lockdep the classical method would
be to change this static array:

static struct lock_class_key
netdev_xmit_lock_key[ARRAY_SIZE(netdev_lock_type)];

to

static struct lock_class_key
netdev_xmit_lock_key[ARRAY_SIZE(netdev_lock_type)][MAX_NUM_TX_QUEUES];

and set lockdep classes per queue as well. (If we are sure we don't
need lockdep subclasses anywhere this could be optimized by using
one lock_class_key per 8 queues and spin_lock_nested()).