porting to world radio chips from conexant

alfred hitch alfred.hitch at gmail.com
Tue Jun 7 02:59:11 EDT 2005


Hi All,

just following up on last post:

when is portEnabled set to 1, (which seems to be causing last error I am having)

I see in a grep on code that 
its being done at practically 2 places only:
1) wpa.c in rsn_preauth_init() .. which I can ignore for wpa-eap I am testing
2) in wpa_supplicant.c:
wpa_supplicant_event() -> association event ..

So, shall I presume that association event generation is must for wpa-eap ?

Till now, I was porting very minimal functions to get wpa-psk working,
so please point to me if my understanding of code isnt correct.

Cheers,
Alfred

On 6/7/05, alfred hitch <alfred.hitch at gmail.com> wrote:
> Hi Jouni / All,
> 
> 1) Well first a good news I have been able to get wpa-psk working with
> conexant worldradio cards,
> shall release code some time by end of month (need to check / sanitise
> copyright issues ).
> Wasnt that difficult in the end, interfaces and code of supplicant was
> pretty neat I must admire.
> 
> Now a problem -;)
> 
> 2) I thot WPA-EAP should be a breeze now, but I am finding some
> difficulty in state machine to go ahead.
> I think there is some parameter I am missing, can some one help me here please.
> 
> Observation: a simple conf file which forces for WPA-EAP only, and
> even no server certificate validation (simplest cases to get it
> working)
> wpa_supplicant starts and receives a eapol message also.
> ethereal logs also show that ap had accepted association request and
> had sent EAP-Request identity
> 
> supplicant traces show that no state machine transition apparently
> happened .. it didnt go into connecting stage from disconnected ..
> 
> No replies are given back to AP from supplicant and so poor AP chap
> keeps on trying sending again request identity again and again ..
> 
> Now, why is supplicant not replying back, no traces at all infact from the
> eapol_sm_rx_eapol function:
> just the eapol message dump
> and EAPOL-Received EAP packet frame.
> 
> ->I dont see any function entry signatures being hit ..
> ->eapol message dumps look as per ethereal no message corruption also.
> -> I put a trace in eapol_sm_ex_eapol func.before it calls eapol_sm_step()
> { char buf[4096];
>    eapol_sm_get_status(sm, buf, 4096, 1);
>    wpa_printf(MSG_INFO, "eapol_sm_get_status  returned %s",buf);
>     wpa_printf(MSG_INFO,"sm->userLogoff %d sm->logoffSent %d
> sm->initialize %d sm->portEnabled %d sm->portControl %d sm->sPortMode
> %d",sm->userLogoff, sm->logoffSent,sm->initialize,
> sm->portEnabled,sm->portControl, sm->sPortMode);
> }
> 
> All the values on second line were zero and PAE state was reported as
> disconnected only.
> 
> After this I see in SM_STEP(SUPP_PAE) function:
>  else if (((sm->portControl == Auto) &&
>                   (sm->sPortMode != sm->portControl)) ||
>                  sm->initialize || !sm->portEnabled)
>                   SM_ENTER_GLOBAL(SUPP_PAE, DISCONNECTED);
> 
> always being enter'ed and so it wasnt going to case SUPP_PAE_CONNECTING:
> which is below it
> (Is my expectation also correct infact).
> 
> This it appears to me makes no state transition
> 
> Cheers,
> Alfred
> PS: attached conf file being used.
> 
> 
>



More information about the HostAP mailing list