doing RSN with wpa_supplicant and madwifi

Sam Leffler sam at errno.com
Sun Jun 27 14:08:39 EDT 2004


On Saturday 26 June 2004 03:24 pm, Jouni Malinen wrote:
> On Sat, Jun 26, 2004 at 02:43:15PM -0700, Sam Leffler wrote:
> > Thanks, there's no reason for the net80211 layer to enforce WPA1 vs. WPA2
> > for station operation.  The iwpriv call is for configuring the
> > authenticator.
>
> OK, that sounds more logical. However, the original report mentioned
> that iwpriv command was used after starting wpa_supplicant to get RSN
> working. Is that incorrect or was that really needed to trigger
> something?

The iwpriv call specifies whether WPA1 or WPA2 is enabled.  That was really 
meant for configuring the authenticator but I overloaded it's use for station 
operation. 

driver_madwifi.c blindly enables WPA1 so the RSN (WPA2) message elements are 
discarded.  The proper fix is to permit either WPA1 or WPA2 use.  This 
requires a minor change to the net80211 layer.  Until then using iwpriv to 
toggle state after scanning is started is a workaround. 

I sent Wilfried a patch to test.  Once he gets back to me the attached change 
to driver_madwifi.c should fix things in wpa_supplicant.

	Sam
-------------- next part --------------
A non-text attachment was scrubbed...
Name: driver_madwifi.patch
Type: text/x-diff
Size: 588 bytes
Desc: not available
Url : http://lists.shmoo.com/pipermail/hostap/attachments/20040627/845de883/attachment.patch 


More information about the HostAP mailing list