On Wed, Aug 12, 2009 at 5:03 AM, Sasha Khapyorsky <[email protected]>wrote:
> On 09:43 Fri 07 Aug , Hal Rosenstock wrote: > > > > so use TRUE/FALSE rather than IB_INVALID_PARMETER > > > > Signed-off-by: Hal Rosenstock <[email protected]> > > --- > > diff --git a/opensm/opensm/osm_mcast_tbl.c > b/opensm/opensm/osm_mcast_tbl.c > > index 82850be..38c06c1 100644 > > --- a/opensm/opensm/osm_mcast_tbl.c > > +++ b/opensm/opensm/osm_mcast_tbl.c > > @@ -273,7 +273,7 @@ osm_mcast_tbl_get_block(IN osm_mcast_tbl_t * const > p_tbl, > > mlid_start_ho = (uint16_t) (block_num * IB_MCAST_BLOCK_SIZE); > > > > if (mlid_start_ho + IB_MCAST_BLOCK_SIZE - 1 > p_tbl->max_mlid_ho) > > - return (IB_INVALID_PARAMETER); > > + return (TRUE); > > In this case p_block array is not initialized, so just returning 'TRUE' > is not a good idea. That's how it's handled now in the code. > > > Actually if we are hitting this case it can indicate an inconsistent > mcast_tbl initialization That's one possibility. The other would be a SA query for a specific block that is not validated first. > - I would suggest to rework this part and likely to drop this check at all. Dropping it should be OK as long as all callers validate prior to calling (and it looks like they do). -- Hal > > > Sasha > _______________________________________________ > general mailing list > [email protected] > http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general > > To unsubscribe, please visit > http://openib.org/mailman/listinfo/openib-general >
_______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
