[ofw] patch: [ibal] Allow cm to send a lap request after afailed (by timeout) lap request.

[Tzachi Dar]

I guess that your suggestion is right but since I don't want to make a
revolution, I'll continue with this checkin.
We probably need to do the separation to a different APM state, but this
is a much bigger change.

Applied in 2486.

Thanks
Tzachi 

> -----Original Message-----
> From: Sean Hefty [mailto:sean.hefty at intel.com] 
> Sent: Thursday, October 15, 2009 6:34 AM
> To: Tzachi Dar; ofw at lists.openfabrics.org
> Subject: RE: [ofw] patch: [ibal] Allow cm to send a lap 
> request after afailed (by timeout) lap request.
> 
> >Index: 
> Q:/projMLNX_WinOF_2.0/MLNX_WinOF_2-0/core/al/kernel/al_cm_cep.c
> >===================================================================
> >--- Q:/projMLNX_WinOF_2.0/MLNX_WinOF_2-0/core/al/kernel/al_cm_cep.c 
> >(revision
> >4956)
> >+++ Q:/projMLNX_WinOF_2.0/MLNX_WinOF_2-0/core/al/kernel/al_cm_cep.c 
> >+++ (revision
> >4957)
> >@@ -2289,7 +2289,15 @@
> >    __format_drep( p_cep, NULL, 0, &p_cep->mads.drep );
> >    p_cep->state = CEP_STATE_TIMEWAIT;
> >    __insert_timewait( p_cep );
> >+   break;
> >
> >+  case CEP_STATE_LAP_SENT:
> >+   /*
> >+    * Before CEP was sent, we have been in 
> CEP_STATE_ESTABLISHED as we
> >+    * failed to send, we return to that state.
> >+    */
> >+   p_cep->state = CEP_STATE_ESTABLISHED;
> >+   break;
> >   default:
> >    break;
> >   }
> 
> I think it would be easier to track APM states separately 
> from the CEP state.
> The CEP states should match that in the spec (12.9).
> 
>