[openib-general] [PATCH 3/3] iWARP CM

[Tom Tucker]

On Thu, 2006-03-09 at 12:32 -0800, Sean Hefty wrote:
> >> > +static int cm_conn_req_handler(struct iwcm_work *work)
> >> > +{
> >> > +	struct iw_cm_id *cm_id;
> >> > +	struct iwcm_id_private *cm_id_priv;
> >> > +	int rc;
> >> > +
> >> > +	/* If the status was not successful, ignore request */
> >> > +	if (work->event.status) {
> >> > +		printk(KERN_ERR "%s:%d Bad status=%d for connection request. "
> >> > +		       "Should be filtered by provider\n",
> >> > +		       __FUNCTION__, __LINE__,
> >> > +		       work->event.status);
> >> > +		return work->event.status;
> >> > +	}
> >>
> >> Can status really ever be non-zero here?
> >
> >Yes, this event is generated by the provider.
> 
> We're handling a new connection request.  Why would the provider report a
> failure?  The printk even states that the provider should filter this.
> 
It's just defensive coding. Should we make this 
a BUG_ON(work->event.status)?

> - Sean
>