[Voyage-linux] Problems with alix 2d3 and madwifi on voyage-current (lenny)

Mon Feb 23 16:29:33 HKT 2009

@Juergen Adolph: If I am not mistaken, the madwifi-modules from Punky
require the new kernel from voyage 0.6.1, which is built with gcc 4.3.2. So
you need to update your kernel.

@Punky: I'll try the r3942 and let you know, though before trying older
versions I think I tried the
http://snapshots.madwifi-project.org/madwifi-trunk/madwifi-trunk-r3941-20090205.tar.gzand
I also had stuck beacons. Are you testing in an alix board too?

On Mon, Feb 23, 2009 at 9:38 AM, Kim-man 'Punky' TSE
<punkytse at punknix.com>wrote:

> Hi all,
>
> Forgot to mention, please post your constructive feedback after testing
> these different branches.
> Right now, I am running a simple AP download test using hal-testing branch
> without any stuck bacon for almost 15 hours.
>
> To me, going back to 0.9.4rc2 is not my best option, unless at the end we
> have no stable madwifi.
> Regards,
> Punky
>
>
>
> Kim-man 'Punky' TSE wrote:
>
>> Hi all,
>>
>> Apart from backport 0.9.4rc2, I have built whole lots of madwifi from
>> different branches for testing, and they are now in
>>
>> http://www.voyage.hk/dists/experimental/madwifi/
>>
>> The branches are:
>>   hal-testing (openwrt/dd-wrt) :
>> madwifi-modules-2.6.26-486-voyage_0.9.4+r3942.20090205haltesting-1+6.0-3_i386.deb
>>
>>   hal-0.10.5.6 :
>> madwifi-modules-2.6.26-486-voyage_0.9.4+r3942.20090205hal0.10.5.6-1+6.0-3_i386.deb
>>
>>   hal-free :
>> madwifi-modules-2.6.26-486-voyage_0.9.4+r3952.20090213free-1+6.0-3_i386.deb
>>
>> Regards,
>> Punky
>>
>> Juergen Adolph wrote:
>>
>>> Hi,
>>>
>>> I installed the latest pkg from 22-Feb-2009:
>>>  madwifi-modules-2.6.26-486-voyage_0.9.4~rc2-1+6.0-3_i386.deb
>>>
>>> and now, it works without any stuck beacon error - great!
>>>
>>> many thanks!
>>>
>>> regards,
>>> juergen
>>>
>>> ----- "Panagiotis Malakoudis" <malakudi at gmail.com> schrieb:
>>>
>>>
>>>
>>>> Ok, made some more progress!!!
>>>>
>>>> I patched the madwifi-source 0.9.4~rc2 with the diff file from
>>>> http://madwifi-project.org/changeset/3696/madwifi/branches/madwifi-0.9.4
>>>> so that it can compile under 2.6.26 and that fixed my "stuck beacon"
>>>> problems. I haven't seen a single "stuck beacon" with this setup. It
>>>> seems the wireless performance is quite slower than the same
>>>> combination with 2.6.23, but I am still checking on this.
>>>>
>>>>
>>>>
>>>> On Sat, Feb 21, 2009 at 11:25 PM, Panagiotis Malakoudis <
>>>> malakudi at gmail.com > wrote:
>>>>
>>>>
>>>> It certainly isn't a voyage problem. It's a madwifi problem, but since
>>>> voyage is used in many wireless applications, it becomes a voyage
>>>> problem too unfortunately.
>>>>
>>>> Juergen 's settings didn't help me. I was struggling to find a
>>>> solution, since I don't like the idea to stay in the old voyage
>>>> version.
>>>> I built voyage-kernel 2.6.23 with gcc 4.1.3 (2.6.23 doesn't compile
>>>> with 4.3.2) and tried the madwifi-source of r3772. Again, I had stuck
>>>> beacon errors and reboots.
>>>>
>>>> Then I used the madwifi-source of 0.9.4~rc2 and that fixed the
>>>> problem!!! Unfortunately, madwifi-source 0.9.4~rc2 does not compile
>>>> with 2.6.26. I get the following error:
>>>>
>>>> CC [M] /usr/src/modules/madwifi/net80211/ieee80211_power.o
>>>> /usr/src/modules/madwifi/net80211/ieee80211_power.c: In function
>>>> 'ieee80211_pwrs
>>>> ave':
>>>> /usr/src/modules/madwifi/net80211/ieee80211_power.c:240: error:
>>>> implicit declara
>>>> tion of function '__skb_append'
>>>>
>>>> __skb_append became obsolete in 2.6.26.
>>>>
>>>> So, for now I'll be using 2.6.23 + madwifi 0.9.4~rc2 and I'll keep
>>>> trying various older madwifi releases with 2.6.26 to see if I can find
>>>> a working combination.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On Sat, Feb 21, 2009 at 10:05 PM, Leigh Orf < orf at mailbag.com > wrote:
>>>>
>>>>
>>>> I just wanted to chime in as another person who has an alix 2d3
>>>> running voyage 0.6.0 who has had the same problems with the stuck
>>>> beacons, NETDEV watchdog timeout message, etc. Also, sometimes hostapd
>>>> gets stuck in an infinite loop of auth/death which is only fixed bring
>>>> ath0 down and up again. I subscribed to this list partially to see if
>>>> these problems had a solution. I do not believe it is a voyage linux
>>>> only problem. I look forward to trying out ath5k in access point mode
>>>> - it isn't there yet.
>>>>
>>>> I have managed 8 days uptime max before the board reboots itself
>>>> suddenly after a trickle of stuck beacons and NETDEV WATCHDOG: wifi0:
>>>> transmit timed out messages. I will soon update to 0.6.1.
>>>>
>>>> Leigh
>>>> --
>>>> Leigh Orf
>>>> Assistant Professor of Atmospheric Science
>>>> Room 130G Engineering and Technology
>>>> Department of Geology
>>>> Central Michigan University
>>>> Mount Pleasant, MI 48859
>>>> (989)774-1923
>>>> Amateur radio callsign: KG4ULP
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Voyage-linux mailing list
>>>> Voyage-linux at list.voyage.hk
>>>> http://list.voyage.hk/mailman/listinfo/voyage-linux
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Voyage-linux mailing list
>>>> Voyage-linux at list.voyage.hk
>>>> http://list.voyage.hk/mailman/listinfo/voyage-linux
>>>>
>>>>
>>>
>>>
>>>
>>
>>
>>
>
> --
> Regards,
> Punky
>
> Voyage Solutions (http://solution.voyage.hk)
> * Embedded Solutions and Systems
>  - Mesh Networking, Captive Portal, IP Surveillance, VoIP/PBX
>  - Network Engineering, Development Platform and Consultation
>
>
>
> _______________________________________________
> Voyage-linux mailing list
> Voyage-linux at list.voyage.hk
> http://list.voyage.hk/mailman/listinfo/voyage-linux
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://list.voyage.hk/pipermail/voyage-linux/attachments/20090223/ed1cc4c7/attachment.html>