Hi all,
 
I tried the DFS branch version of the direver and I experienced the same problem, at a given moment the number of  interrupts on wifi0 is stuck and the system is very slow (during these tests the link is heavily loaded).
 
I have also tried to use the driver (DFS branch and trunk version) in monitor mode on both side of the link in order to avoid ad-hoc mode problems but I experienced the same problem..
 
Is there a way to restart the interrupts without removing the module and reloading it?? I tried the function ath_change_mtu in order to call the ath_reset ... but nothing has changed...
 
I would know also how does the system continue to work without interrupts?? The system is very slow  but is not stuck, I can ping the other end of the link with a delay of 50 ms while the delay is usually a limited to 1 - 2 ms. Are there some tasks scheduled periodically to analyze the incoming packets? 
 
Thanks you very much!!
 
Regards,
 
Lorenzo  
 
 
 
2009/1/22 Louay Sakka <louay.sakka@linowave.com>
Actually there is quit simpler scenario that can cause this, both on
madwifi-free or any previous version I test so fat (0.9.4 or 0.9.3.3)

Just have the madwifi in AP mode, and connect two Laptop over the
wireless enable ap_bridge ,then scp big files from one to another.

You will see plenty of stuck beacon, and scp will halt.

wget work just fine.

This is tried on LinoWave MIPS board, but I think it will show on any
other platform.


Louay


Pavel Roskin wrote:
> On Thu, 2009-01-22 at 19:41 +0100, Lorenzo Bianconi wrote:
>>  Hi Pavel,
>>
>> thank you for the readiness with which you answered.
>>
>> From dmesg I did not notice anything wrong. At this moment I need to
>> use  the trunk version because this is the latest version included in
>> OpenWrt ... to try the DFS branch I need more time..
>>
>> I tried to remove the module and to reload it and the the number of
>>  interrupts on wifi0 has started to grow again.
>>
>> How is that possible??
>
> I believe the ad-hoc support is quite poor.  I tried to reproduce that
> problem and got messages like this in the kernel log:
>
> [79699.892949] wifi0: ath_bstuck_tasklet: Stuck beacon; resetting
> (beacon miss count: 120)
> [79699.912846] wifi0: ath_bstuck_tasklet: Stuck beacon; resetting
> (beacon miss count: 114)
> [79699.932766] wifi0: ath_bstuck_tasklet: Stuck beacon; resetting
> (beacon miss count: 119)
>
> I didn't even try to simulate the radar pulse.
>

--
Louay Sakka

LinoWave Corp
1001-101 Champlain Av L7L5Z4
Burlington On Canada
Tel: 12892880060
Fax: 12892880061
Cell:12898882248
www.linowave.com