Return-path: <linux-wireless-owner@vger.kernel.org>
Received: from mail.candelatech.com ([208.74.158.172]:54273 "EHLO
	ns3.lanforge.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org
	with ESMTP id S1757781Ab0GDP7O (ORCPT
	<rfc822;linux-wireless@vger.kernel.org>);
	Sun, 4 Jul 2010 11:59:14 -0400
Message-ID: <4C30AFD0.80100@candelatech.com>
Date: Sun, 04 Jul 2010 08:59:12 -0700
From: Ben Greear <greearb@candelatech.com>
MIME-Version: 1.0
To: Vasanthakumar Thiagarajan <vasanth@atheros.com>
CC: Vasanth Thiagarajan <Vasanth.Thiagarajan@atheros.com>,
	"linux-wireless@vger.kernel.org" <linux-wireless@vger.kernel.org>
Subject: Re: ath9k doesn't clean up virtual wifis on rmmod, and crashes.
References: <4C2ED806.9000105@candelatech.com> <20100703070259.GA15479@vasanth-laptop> <4C2F6681.7070202@candelatech.com> <20100704093535.GE5053@vasanth-laptop>
In-Reply-To: <20100704093535.GE5053@vasanth-laptop>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Sender: linux-wireless-owner@vger.kernel.org
List-ID: <linux-wireless.vger.kernel.org>

On 07/04/2010 02:35 AM, Vasanthakumar Thiagarajan wrote:
> On Sat, Jul 03, 2010 at 10:04:09PM +0530, Ben Greear wrote:
>> On 07/03/2010 12:02 AM, Vasanthakumar Thiagarajan wrote:
>>> On Sat, Jul 03, 2010 at 11:56:14AM +0530, Ben Greear wrote:
>>>> It seems to me that in 2.6.34, there is no code to clean up
>>>> virtual wiphys in ath9k on rmmod.  Also, ath9k mailing list
>>>> is returning error about mis-configured DNS server.
>>>>
>>>> This is with un-modified ath9k driver and is repeatable
>>>> every time on my system (crash is often different, but
>>>> it always crashes very quickly).
>>>>
>>>> [root@atom ~]# echo add=5>   /debug/ath9k/phy1/wiphy
>>>> -bash: /debug/ath9k/phy1/wiphy: No such file or directory
>>>> [root@atom ~]# echo add=5>   /debug/ath9k/phy0/wiphy
>>>> Jul  2 23:22:19 atom kernel: phy1: Selected rate control algorithm 'ath9k_rate_control'
>>>> [root@atom ~]# Jul  2 23:22:19 atom kernel: ADDRCONF(NETDEV_UP): wlan1: link is not ready
>>>> rmmod ath9k
>>>> Jul  2 23:22:24 atom kernel: ath9k 0000:05:00.0: PCI INT A disabled
>>>> Jul  2 23:22:24 atom kernel: ath9k: Driver unloaded
>>>> [root@atom ~]# BUG: spinlock bad magic on CPU#1, iw/2877
>>>>     lock: f8a476c0, .magic: 00000000, .owner:<none>/-1, .owner_cpu: 0
>>>> Pid: 2877, comm: iw Not tainted 2.6.34 #7
>>>> Call Trace:
>>>>     [<c0730e34>] ? printk+0xf/0x13
>>>>     [<c05710d6>] spin_bug+0x7b/0x86
>>>>     [<c0440b76>] ? delayed_work_timer_fn+0x0/0x30
>>>>     [<c0571171>] do_raw_spin_lock+0x1e/0x125
>>>>     [<c042f91e>] ? scheduler_tick+0xd6/0x1c9
>>>>     [<c0440b76>] ? delayed_work_timer_fn+0x0/0x30
>>>>     [<c0732a6d>] _raw_spin_lock_irqsave+0x1b/0x20
>>>>     [<c0440b59>] __queue_work+0x12/0x2f
>>>>     [<c0440b76>] ? delayed_work_timer_fn+0x0/0x30
>>>>     [<c0440ba4>] delayed_work_timer_fn+0x2e/0x30
>>>>
>>>> I'm new to hacking on this driver..but would love to test
>>>> patches, and if someone wants to suggest a good point in
>>>> the code to remove the virtual phys, I'll make the attempt.
>>>
>>> Can you please try this patch?
>>
>> It still crashes for me, and backtraces look similar.
>>
>> Were you able to reproduce the crash?
>
> with my patch, no. Can you please give the steps to trigger
> this crash?

It happens every time if I:

echo add >   /debug/ath9k/phy0/wiphy
rmmod ath9k

I notice the patch applied with offsets.  What kernel/driver
version are you testing?  I tried .34 kernel as well as the top-of-tree
2.6.35-rcX.

Thanks,
Ben

-- 
Ben Greear <greearb@candelatech.com>
Candela Technologies Inc  http://www.candelatech.com