[openib-general] [PATCH] opensm: osm_sa_path_record: mcast destination detection fix

[Hal Rosenstock]

On Mon, 2006-08-21 at 13:59, Sasha Khapyorsky wrote:
> On 13:45 Mon 21 Aug     , Hal Rosenstock wrote:
> > On Mon, 2006-08-21 at 13:22, Sasha Khapyorsky wrote:
> > > Return error when mcast destination is not consistently indicated.
> > > 
> > > Signed-off-by: Sasha Khapyorsky <sashak at voltaire.com>
> > 
> > Thanks. Applied (to both trunk and 1.1) with the following minor changes
> > below:
> > 
> > >  osm/opensm/osm_sa_path_record.c |   16 +++++++++++-----
> > >  1 files changed, 11 insertions(+), 5 deletions(-)
> > > 
> > > diff --git a/osm/opensm/osm_sa_path_record.c b/osm/opensm/osm_sa_path_record.c
> > > index caa9f32..6b0fb28 100644
> > > --- a/osm/opensm/osm_sa_path_record.c
> > > +++ b/osm/opensm/osm_sa_path_record.c
> > > @@ -1486,7 +1486,7 @@ __osm_pr_match_mgrp_attributes(
> > >  
> > >  /**********************************************************************
> > >   **********************************************************************/
> > > -static boolean_t
> > > +static int
> > >  __osm_pr_rcv_check_mcast_dest(
> > >    IN osm_pr_rcv_t*         const p_rcv,
> > >    IN const osm_madw_t*     const p_madw )
> > > @@ -1494,7 +1494,7 @@ __osm_pr_rcv_check_mcast_dest(
> > >    const ib_path_rec_t*     p_pr;
> > >    const ib_sa_mad_t*       p_sa_mad;
> > >    ib_net64_t               comp_mask;
> > > -  boolean_t                is_multicast = FALSE;
> > > +  unsigned                 is_multicast = 0;
> > >  
> > >    OSM_LOG_ENTER( p_rcv->p_log, __osm_pr_rcv_check_mcast_dest );
> > >  
> > > @@ -1514,11 +1514,13 @@ __osm_pr_rcv_check_mcast_dest(
> > >    {
> > >      if( cl_ntoh16( p_pr->dlid ) >= IB_LID_MCAST_START_HO &&
> > >          cl_ntoh16( p_pr->dlid ) <= IB_LID_MCAST_END_HO )
> > > -      is_multicast = TRUE;
> > > -    else if( is_multicast )
> > > +      is_multicast = 1;
> > > +    else if( is_multicast ) {
> > >        osm_log( p_rcv->p_log, OSM_LOG_ERROR,
> > >                 "__osm_pr_rcv_check_mcast_dest: ERR 1F12: "
> > >                 "PathRecord request indicates MGID but not MLID\n" );
> > > +      return -1;
> > 
> > I made this go through the exit so the routine end log message is put
> > into the log.
> 
> Right.
> 
> Now there is 'is_multicast = -1' - you may want to change is_multicast
> type to int (now it is unsigned).

Thanks. Just did that (to both trunk and 1.1).

-- Hal

> 
> > 
> > > +    }
> > >    }
> > >  
> > >   Exit:
> > > @@ -1693,6 +1695,7 @@ osm_pr_rcv_process(
> > >    cl_qlist_t               pr_list;
> > >    ib_net16_t               sa_status;
> > >    osm_port_t*              requester_port;
> > > +  int ret;
> > >  
> > >    OSM_LOG_ENTER( p_rcv->p_log, osm_pr_rcv_process );
> > >  
> > > @@ -1737,7 +1740,10 @@ osm_pr_rcv_process(
> > >    cl_plock_acquire( p_rcv->p_lock );
> > >  
> > >    /* Handle multicast destinations separately */
> > > -  if( __osm_pr_rcv_check_mcast_dest( p_rcv, p_madw ) )
> > > +  if( (ret = __osm_pr_rcv_check_mcast_dest( p_rcv, p_madw )) < 0)
> > 
> > I added a send of SA status error for IB_MAD_STATUS_INVALID_FIELD here
> > as well as an unlock.
> 
> Sure.
> 
> Sasha
> 
> > 
> > > +    goto Exit;
> > > +
> > > +  if(ret > 0)
> > >      goto McastDest;
> > >  
> > >    osm_log( p_rcv->p_log, OSM_LOG_DEBUG,
> > 
> > -- Hal
> >