[Click] Madwifi+Click Problems

Nels Anderson nels.anderson at colorado.edu
Tue Aug 31 02:09:10 EDT 2004


Aloha,

That's what's causing that, eh?  Hmmm....I guess I just assumed it was 
something way worse, since that seemed to coincide with the driver going 
kaputt on a few different things.  I know that I was able to have it 
running using the madwifi.stripped driver with both a normal looking mac 
(which I don't care about) and I was able to bring the card up with 
different mac address (which is quite important).  I modified one of the 
perl config scripts to bring the card up with the mac specificed when 
the script was run, and that was working just fine for a while.  But 
something happened with the driver (I think, it could be something on my 
end) and now I get that error from siocsifhwaddr.  Any ideas on what 
could possibly to be done to resolve that?  Or even get some more 
debugging info than just "invalid arugment"?  Stations are also unable 
to assiociated with the the ap if it's using the madwifi.stripped driver 
I've got now, so I thought the mac weirdness was responsible for that too.

All I really need the card to do is act as an access point and be able 
to come up with a specific mac address.  Well, it needs to be using 
click too, obviously.  I can get the madwifi.stripped driver to act as a 
station, get it to fully associate with other aps, can browse using it, 
etc.  When I set it up as an access point, I can see the ap with all the 
appropriate information via kismet, but nothing will actually associate 
with it.  I grab the output from click when I'm at work tomorrow.

I'm not at Sigcomm, but I appreciate the offer.  Perhaps at some other 
point in the future.  And thanks again for the assistance, I really do 
appreciate it.

-Nels

John Bicket wrote:

>Hi Nels,
>
>This is because the madwifi.stripped driver report ath0 with a device
>type of ARPHRD_IEEE80211, and ifconfig doesn't know how to print mac
>addresses for that.  (I sent a patch to the net-tools guy, but I think
>he just ignored me).
>
>Everything else should work normally, and the address is just the first 
>6 bytes of the hwaddr in ifconfig.  (ie 00-0D-88-8A-78-E3).
>I'm not sure the stripped driver supports changing mac addresses.
>
>Let me know if this answers your question. I'm also at sigcomm now, so
>look me up if you're here.
>
>
>--john
>
>
>Nels Anderson [Nels.Anderson at colorado.edu] wrote:
>  
>
>>Aloha,
>>
>>Hello again, I'm still experiencing some problems with wifi-enable click and the
>>stripped madwifi driver for the atheros cardset.  After building the latest
>>driver and loading all the modules, ath0 comes up with the bizarre ethernet
>>address of:
>>
>>ath0   Link encap:UNSPEC  HWaddr 00-0D-88-8A-78-E3-00-00-00-00-00-00-00-00-00-00
>>
>>Also when I try and change the MAC address of the card with 'ifconfig ath0 hw
>>ether $mac' (with a proper address for $mac), it also fails with this message:
>>
>>SIOCSIFHWADDR: Invalid argument
>>
>>Anyone have any idea what might be causing these problems?
>>
>>This doesn't happen when I build and load the non-click sourceforge madwifi
>>drivers.  I actually had both of these things working normally under that
>>driver.  Now, I'm not sure if it's the stripped driver itself (I don't think it
>>is) or if it's some combination of kernel options and the stripped driver. 
>>Something the stripped madwifi uses but the sourceforge one does not.  Does
>>anyone have a 2.4.26 kernel config that they've been able to get the click-wifi
>>stuff working with?  I'm just using a stock kernel.org 2.4.26 kernel.
>>
>>If anyone has any ideas, ran across similiar problems or anything else,
>>assistance would be much appreciated.  Thanks,
>>
>>-Nels
>>    
>>


More information about the click mailing list