Re: [Madwifi-users] fatal interrupt

[sara <sc...@os...>]

> Let's see if we have anything in common that might give us some more
> clues.
> My hardware is Intel IXP425 using ubiquity mini PCI cards (it does not
> matter which ubiquity card I use, I get the problem with all) and
> uClinux version 2.6.17.
>
> The problem only happens for me with AP mode - I never see this at the
> STA. If I get a STA to connect to the AP, but do not pass any data
> between them then I don't get the problem. However, simply running a
> continuous ping from the STA to the AP ensures I see this error after
> only 10 minutes or less.
>
> I can reproduce this problem with the following versions of madwifi:
> - madwifi trunk r4079 (open hal)
> - madwifi trunk r3916 (closed source HAL 0.9.30.13)
> - madwifi dfs r4072 (open hal)
> - madwifi hal testing r4031
> - madwifi hal 0.10.5.6 r3861
>
> However, exact same hardware setup does not produce the error with
> madwifi 
> 0.9.4 r4073 (closed source HAL 0.9.18.0) and older revisions of 0.9.4
>
> So I wonder, is the problem cause by the HAL or madwifi? Based on the
> above it could be either (or combination of both), though it looks as if
> a case could be made for suspecting hal versions newer than 0.9.18.0
> (just what version open hal is supposed to be I don't know...). 
>
> Brett
>   
>   
Hello Brett,

I have the same problem in the AP mode. But, I am getting this issue on different
platforms like Intel IXP425 ADI Pronghorn SBC250 board and x86 soekris net4511 boards
with cm9 miniPCI cards running OpenWrt Kamikaze Linux 2.6.27 and above.

In my case, the issue can be reproduced by 
 - building the latest OpenWRT trunk with default config for the target (x86 or Intel IXP). 
 - modifying the Openwrt's madwifi (package/madwifi/Makefile) package to use r4079.

Till now, no idea of what causes this issue in general like compilation or dependencies
or configuration setups etc.,

Sara