Fwd: [PATCH 09/16] unit: Update event handler in WSC, eapol tests

Tuesday, 5 November 2019

Forwarding this as well, sigh. Sorry folks.

---------- Forwarded message ---------
From: Will Dietz <w(a)wdtz.org&gt;
Date: Mon, Nov 4, 2019, 9:15 PM
Subject: Re: [PATCH 09/16] unit: Update event handler in WSC, eapol tests
To: Andrew Zaborowski <andrew.zaborowski(a)intel.com&gt;

On Mon, 4 Nov 2019 21:02:07 -0600, Will Dietz <w(a)wdtz.org&gt; wrote:
...
 On Mon, 28 Oct 2019 15:05:01 +0100, Andrew Zaborowski <
andrew.zaborowski(a)intel.com&gt; wrote:
...
 >  {
 >     struct verify_data *data = user_data;
 > +   va_list args;
 > +
 > +   va_start(args, user_data);
 >
 >     switch (event) {
 >     case HANDSHAKE_EVENT_FAILED:
 > -           assert(l_get_u16(event_data) ==
 > -                           MMPDU_REASON_CODE_IEEE8021X_FAILED);
 > +           assert(va_arg(args, int) == MMPDU_REASON_CODE_IEEE8021X_FAILED);
...
 >             data->eapol_failed = true;
 >             break;
 > +   case HANDSHAKE_EVENT_EAP_NOTIFY:
 > +   {
 > +           const struct wsc_credential *cred;
 > +
 > +           assert(va_arg(args, unsigned int) !=
 > +                   EAP_WSC_EVENT_CREDENTIAL_OBTAINED);

 This condition appears to have been inverted accidentally :3.
 I noticed because this causes the test to (always) fail.
 Compare to the logic this replaces:

 > -
 > -   if (event != EAP_WSC_EVENT_CREDENTIAL_OBTAINED)
 > -           assert(false);
 > -

 I'm not in a position to properly submit a patch currently,
 but if there's agreement on this please go ahead and fix :).
 (flipping the '!=' to '==' in the assert does the trick, etc.)

 ~Will 
Nit while visiting the file: the va_start calls don't have matching
va_end calls, which are present in wsc.c which test-wsc.c appears based
on.

Not a big deal, but happened to notice while inspecting.

~Will

2021

2020

2019