[ofa-general] Re: [PATCH] opensm: PortInfo set decision flow simplification

[Sasha Khapyorsky]

On 11:24 Mon 12 Nov     , Hal Rosenstock wrote:
> > diff --git a/opensm/opensm/osm_link_mgr.c b/opensm/opensm/osm_link_mgr.c
> > index 19d03d9..b151c76 100644
> > --- a/opensm/opensm/osm_link_mgr.c
> > +++ b/opensm/opensm/osm_link_mgr.c
> > @@ -389,15 +389,12 @@ __osm_link_mgr_set_physp_pi(IN osm_link_mgr_t * const p_mgr,
> >  	   b. got_set_resp on the physical port is FALSE. This means we haven't
> >  	   seen this port before - need to send PortInfoSet to it.
> >  	 */
> > -	if (send_set ||
> > -	    (osm_node_get_type(p_node) != IB_NODE_TYPE_SWITCH
> > -	     && p_physp->got_set_resp == FALSE)
> > -	    || (osm_node_get_type(p_node) == IB_NODE_TYPE_SWITCH
> > -		&& port_num == 0 && p_physp->got_set_resp == FALSE)
> > -	    || (osm_node_get_type(p_node) == IB_NODE_TYPE_SWITCH
> > -		&& port_num != 0
> > -		&& (p_mgr->p_subn->first_time_master_sweep == TRUE
> > -		    || p_physp->got_set_resp == FALSE))) {
> > +	if (p_physp->got_set_resp == FALSE
> > +	    || (osm_node_get_type(p_node) == IB_NODE_TYPE_SWITCH && port_num
> > +		&& p_mgr->p_subn->first_time_master_sweep == TRUE))
> 
> This doesn't look logically the same to me. I think it sets send_set in
> some cases where it wasn't before.

Could you elaborate? When?

Sasha