wpa_supplicant error when trying to use EAP-PEAP/GTC

[Pedro Ramalhais]

On Tue, 2004-10-12 at 14:32, Jouni Malinen wrote:
> On Tue, Oct 12, 2004 at 03:23:48PM +0200, Henrik Brix Andersen wrote:
> 
> > I was told by Pedro Ramalhais in #ipw2100 on freenode that the patch
> > would not work with wpa_supplicant-0.2.5 since it relied on a
> > misbehavior of wpa_supplicant-0.2.4 that had been corrected in 0.2.5.
> 
> Oh, I did not know about that kind of "feature".. Getting some details
> of this might be useful if any help is needed for getting this to work
> with 0.2.5.

Well, it was a misbehaviour in driver_wext with drivers that expected
ssid_len+1 when setting the essid, and that is now fixed in hostAP CVS
and 0.2.5. So driver_ipw2100.c did this:

if (wpa_driver_wext_set_ssid(ifname, ssid, ssid_len+1) < 0)

which should now be changed to

if (wpa_driver_wext_set_ssid(ifname, ssid, ssid_len) < 0)

I haven't tested current hostAP(wpa_supp) CVS or wpa_supp 0.2.5 with
that change, but i intend to test it soon so i can submit you the patch.
I'm currently testing the new "to be released" ipw2100 driver, and that
will also need some changes to driver_ipw2100.c in wpa_supp.

> > The patch applies fine (with offsets in a couple of hunks). I have been
> > unable to locate the original patch online but I have a copy of it in my
> > portage overlay at
> > http://dev.gentoo.org/~brix/files/overlay/net-wireless/wpa-supplicant/files/wpa-supplicant-0.2.4-ipw2100.patch.gz
> > 
> > Are there any plans to include the patch in wpa_supplicant?
> 
> I have been waiting for someone to submit that patch for inclusion into
> development branch. I don't have ipw2100 card myself, so I have not been
> testing this myself. If there is a known issue with the current patch,
> it sounds like that should be resolved first. Anyway, I'm ok with adding
> this to wpa_supplicant once someone tells me it is ready for this..

I was waiting for James Ketrenos to decide on who shoud submit it, and
it took a while to get that answer, so i'll be submitting it.
You can just get that 0.2.4 patch and modify the ssid_len+1 line to
ssid_len.
Henrik Brix Andersen just tested it with 0.2.5 and it worked, so it
should also work in CVS.

Thanks!
-- 
Pedro Ramalhais <ramalhais at serrado.net>