[Openib-windows] [IBAL][Patch] reference counter for IOC obj in IOC PnP manager

[Yossi Leybovich]

> -----Original Message-----
> From: Fabian Tillier [mailto:ftillier.sst at gmail.com] 
> Sent: Thursday, November 23, 2006 7:06 AM
> To: Yossi Leybovich
> Cc: Alex Estrin; openib-windows at openib.org
> Subject: Re: [Openib-windows] [IBAL][Patch] reference counter 
> for IOC obj in IOC PnP manager
> 
> Hi Yossi,
> 
> This reply is about the IOC PnP sweep problem...
> 
> On 11/17/06, Yossi Leybovich <sleybo at mellanox.co.il> wrote:
> > Hi
> >
> > 1. cases that AL start new sweep before the former sweep 
> finished IBAL 
> > will not start new sweep if there are queries outstanding
> >
> > Index: core/al/kernel/al_ioc_pnp.c
> > ===================================================================
> > --- core/al/kernel/al_ioc_pnp.c (revision 1797)
> > +++ core/al/kernel/al_ioc_pnp.c (working copy)
> > @@ -1223,7 +1223,7 @@
> >        case IB_PNP_SM_CHANGE:
> >        case IB_PNP_PORT_ACTIVE:
> >                /* Initiate a sweep - delay a bit to allow 
> the ports to 
> > come up. */
> > -               if( g_ioc_poll_interval )
> > +               if( g_ioc_poll_interval && !gp_ioc_pnp->query_cnt)
> >                {
> >                        cl_status = cl_timer_start( 
> > &gp_ioc_pnp->sweep_timer, 250 );
> >                        CL_ASSERT( cl_status == CL_SUCCESS );
> 
> Is this the only change related to the sweep issue you ran into?
> 
> This doesn't fix the port down event for which Alex provided a patch.
> It probably makes sense to apply Alex's patch, unless you 
> think it causes a regression.
> 

I am thinking of using the query_cnt (of the mad_srv on the sweep
result) to ref count the IOC and IOU.
>From what I understood from the code in case that the
result->svc->qury_cnt is 0 we should not have any outstanding 
Queries .Is that right assumption?


> - Fab
>