madwifi-ng & linux 2.6.29

madwifi-ng & linux 2.6.29

[Mike (mwester)]

Now that the 2.6.29 kernel has support for some of the Atheros wifi
devices, we have a little bit of a problem.  The solution would seem to
me to be to virtualize atheros support in OE.  This would change a
number of recipes, including hostap-daemon, wpa-supplicant, and a bunch
of package and task recipes.

This assumes that hostap-daemon and wpa-supplicant won't need other
changes to support the new drivers...

So the first question to the group is who uses madwifi anymore?  SlugOS
includes drivers, but only for selected (and unusual) devices, so fixing
this is not a high priority for SlugOS, nor is maintaining compatibility
a big concern for SlugOS, I think.   Anyone else?

Mike (mwester)

Re: madwifi-ng & linux 2.6.29

[Adrien Demarez]

Hi all,

I am working with Jeremy Lainé on the boc01 architecture. Currently, we
do use Madwifi + hostapd 0.5 for our 11g Atheros minicard modules
(despite we want to switch to ath9k / ath5k when it is mature enough).
As we are still using 2.6.27 kernel and found ath9k is not mature enough
for AP in this version (at least, it does not work with our AR9160
modules), I have written a working recipe for compat-wireless (not yet
committed into OE though, please tell if anyone is interested. Beware :
it does not compile if MAC80211 or CFG80211 is defined as 'y' in the
kernel, use 'm' or 'n' instead !).
This is working, though I'd prefer using vanilla in-kernel drivers
instead of compat-wireless (so I plan to have a look to ath*k in 2.6.29
kernel very soon).

In order to use ath*k (or any drivers using nl80211) with hostapd, I
have also written a recipe for hostapd-0.6.9 (Jeremy has uploaded it
into OE) and this is working perfectly with our AR9160 modules using
ath9k from compat-wireless. I am currently testing ath5k, but I run into
a few troubles with AP mode, so until this is fixed we still need to
keep a working madwifi+hostapd-0.5.10 (I also experienced a few troubles
with madwifi + hostapd-0.6.9 so for the moment we stand with hostapd-0.5
when using madwifi, and hostapd-0.6 is not set as the default one. I
will look at this soon but getting a working ath5k/AP is more important
to me currently).
I also have to do more testing with wpa_supplicant (currently, AP mode
is my focus).

BTW, if anyone is interested, I also have written a working recipe for
madwifi using hal-0.10.5.6. I found this hal is necessary for some
recent chips like AR2417 with whom the vanilla madwifi-ng driver does
not work.

Best regards,

--
Adrien Demarez

Mike (mwester) a écrit :
> Now that the 2.6.29 kernel has support for some of the Atheros wifi
> devices, we have a little bit of a problem.  The solution would seem to
> me to be to virtualize atheros support in OE.  This would change a
> number of recipes, including hostap-daemon, wpa-supplicant, and a bunch
> of package and task recipes.
>
> This assumes that hostap-daemon and wpa-supplicant won't need other
> changes to support the new drivers...
>
> So the first question to the group is who uses madwifi anymore?  SlugOS
> includes drivers, but only for selected (and unusual) devices, so fixing
> this is not a high priority for SlugOS, nor is maintaining compatibility
> a big concern for SlugOS, I think.   Anyone else?
>
> Mike (mwester)
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>   

-- 

Adrien Demarez
Bolloré telecom
>> WiMAX is here and across Europe! See us @
http://www.wimaxoperators.org/regional/europe

Bureaux/Office : 11bis, rue Scribe | F-75009 Paris
Tel : +33 1 78 14 04 08
www.bolloretelecom.eu <http://www.bolloretelecom.eu>

------------------------------------------------------------------------

Ce message et toutes les pièces jointes (ci-après le « message ») sont
établis à l'intention exclusive de ses destinataires et sont
confidentiels. Si vous recevez ce message par erreur, merci de le
détruire et d'en avertir immédiatement l'expéditeur. Toute utilisation
de ce message non conforme à sa destination, toute diffusion ou toute
publication, totale ou partielle, est interdite, sauf autorisation
expresse. L'Internet ne permettant pas d'assurer l'intégrité de ce
message, Bolloré telecom décline toute responsabilité au titre de ce
message, dans l'hypothèse où il aurait été modifié.

------------------------------------------------------------------------

Re: madwifi-ng & linux 2.6.29

[Mike (mwester)]

Adrien Demarez wrote:
> Hi all,
> 
> I am working with Jeremy Lainé on the boc01 architecture. Currently, we
> do use Madwifi + hostapd 0.5 for our 11g Atheros minicard modules
> (despite we want to switch to ath9k / ath5k when it is mature enough).
> As we are still using 2.6.27 kernel and found ath9k is not mature enough
> for AP in this version (at least, it does not work with our AR9160
> modules), I have written a working recipe for compat-wireless (not yet
> committed into OE though, please tell if anyone is interested. Beware :
> it does not compile if MAC80211 or CFG80211 is defined as 'y' in the
> kernel, use 'm' or 'n' instead !).
> This is working, though I'd prefer using vanilla in-kernel drivers
> instead of compat-wireless (so I plan to have a look to ath*k in 2.6.29
> kernel very soon).
> 
> In order to use ath*k (or any drivers using nl80211) with hostapd, I
> have also written a recipe for hostapd-0.6.9 (Jeremy has uploaded it
> into OE) and this is working perfectly with our AR9160 modules using
> ath9k from compat-wireless. I am currently testing ath5k, but I run into
> a few troubles with AP mode, so until this is fixed we still need to
> keep a working madwifi+hostapd-0.5.10 (I also experienced a few troubles
> with madwifi + hostapd-0.6.9 so for the moment we stand with hostapd-0.5
> when using madwifi, and hostapd-0.6 is not set as the default one. I
> will look at this soon but getting a working ath5k/AP is more important
> to me currently).
> I also have to do more testing with wpa_supplicant (currently, AP mode
> is my focus).
> 
> BTW, if anyone is interested, I also have written a working recipe for
> madwifi using hal-0.10.5.6. I found this hal is necessary for some
> recent chips like AR2417 with whom the vanilla madwifi-ng driver does
> not work.
> 
> Best regards,
> 
> --
> Adrien Demarez

Wow.

I had no idea this was so complicated.

For the moment, I'm going to drop madwifi, hostapd from the SlugOS
feeds, and put a SlugOS-specific hack into wpa-supplicant.  I guess
we'll need to invest a whole lot more time to sort out the new Atheros
driver world.

Mike (mwester)

Re: madwifi-ng & linux 2.6.29

[Panayot Daskalov]

Mike (mwester <mwester <at> dls.net> writes:

> 
> Adrien Demarez wrote:
> > Hi all,
> > 
> > I am working with Jeremy Lainé on the boc01 architecture. Currently, we
> > do use Madwifi + hostapd 0.5 for our 11g Atheros minicard modules
> > (despite we want to switch to ath9k / ath5k when it is mature enough).
> > As we are still using 2.6.27 kernel and found ath9k is not mature enough
> > for AP in this version (at least, it does not work with our AR9160
> > modules), I have written a working recipe for compat-wireless (not yet
> > committed into OE though, please tell if anyone is interested. Beware :
> > it does not compile if MAC80211 or CFG80211 is defined as 'y' in the
> > kernel, use 'm' or 'n' instead !).
> > This is working, though I'd prefer using vanilla in-kernel drivers
> > instead of compat-wireless (so I plan to have a look to ath*k in 2.6.29
> > kernel very soon).
> > 
> > In order to use ath*k (or any drivers using nl80211) with hostapd, I
> > have also written a recipe for hostapd-0.6.9 (Jeremy has uploaded it
> > into OE) and this is working perfectly with our AR9160 modules using
> > ath9k from compat-wireless. I am currently testing ath5k, but I run into
> > a few troubles with AP mode, so until this is fixed we still need to
> > keep a working madwifi+hostapd-0.5.10 (I also experienced a few troubles
> > with madwifi + hostapd-0.6.9 so for the moment we stand with hostapd-0.5
> > when using madwifi, and hostapd-0.6 is not set as the default one. I
> > will look at this soon but getting a working ath5k/AP is more important
> > to me currently).
> > I also have to do more testing with wpa_supplicant (currently, AP mode
> > is my focus).
> > 
> > BTW, if anyone is interested, I also have written a working recipe for
> > madwifi using hal-0.10.5.6. I found this hal is necessary for some
> > recent chips like AR2417 with whom the vanilla madwifi-ng driver does
> > not work.
> > 
> > Best regards,
> > 
> > --
> > Adrien Demarez

Hi Adrien,

I am currently facing the issue of compiling drivers for 3Com 3CRUSBN275 USB
Wi-Fi adapter which uses the Atheros AR9001U-2NX chipset. As far as I can
understand this chipset is supported by the compat-wireless package, but I need
to cross compile it for ARM9 platform (i.MX27). So I would be very grateful if
you are willing to share your OE recipe for compat-wireless.

Of course I will give back every feedback that I can.

Thanks in advance.

Best regards,
Panayot