On Sun, 2007-11-18 at 17:05 +0000, Sasha Khapyorsky wrote:
> On 12:52 Sat 17 Nov , Hal Rosenstock wrote:
> > On Thu, 2007-11-15 at 11:03 +0200, Sasha Khapyorsky wrote:
> > > Remove not really used (but leaked) umad_ca field from ibumad vendor
> > > structure.
> > >
> > > Signed-off-by: Sasha Khapyorsky <[EMAIL PROTECTED]>
> > > ---
> > > opensm/include/vendor/osm_vendor_ibumad.h | 1 -
> > > opensm/libvendor/osm_vendor_ibumad.c | 18 ------------------
> > > 2 files changed, 0 insertions(+), 19 deletions(-)
> > >
> > > diff --git a/opensm/include/vendor/osm_vendor_ibumad.h
> > > b/opensm/include/vendor/osm_vendor_ibumad.h
> > > index 743b393..84fd21a 100644
> > > --- a/opensm/include/vendor/osm_vendor_ibumad.h
> > > +++ b/opensm/include/vendor/osm_vendor_ibumad.h
> > > @@ -158,7 +158,6 @@ typedef struct _osm_vendor {
> > > osm_bind_handle_t agents[UMAD_CA_MAX_AGENTS];
> > > char ca_names[OSM_UMAD_MAX_CAS][UMAD_CA_NAME_LEN];
> > > vendor_match_tbl_t mtbl;
> > > - umad_ca_t umad_ca;
> > > umad_port_t umad_port;
> > > pthread_mutex_t cb_mutex;
> > > pthread_mutex_t match_tbl_mutex;
> > > diff --git a/opensm/libvendor/osm_vendor_ibumad.c
> > > b/opensm/libvendor/osm_vendor_ibumad.c
> > > index 240a97b..9e186d5 100644
> > > --- a/opensm/libvendor/osm_vendor_ibumad.c
> > > +++ b/opensm/libvendor/osm_vendor_ibumad.c
> > > @@ -703,24 +703,6 @@ osm_vendor_open_port(IN osm_vendor_t * const p_vend,
> > > }
> > >
> > > /* Port found, try to open it */
> > > - if (umad_get_ca(p_vend->ca_names[ca], &p_vend->umad_ca) < 0) {
> > > - osm_log(p_vend->p_log, OSM_LOG_ERROR,
> > > - "osm_vendor_open_port: ERR 542A: "
> > > - "umad_get_ca() failed\n");
> > > - goto Exit;
> > > - }
> > > -
> > > - /* Validate that node is an IB node type */
> > > - if (p_vend->umad_ca.node_type < 1 || p_vend->umad_ca.node_type > 3) {
> >
> > Where is this check done now ? This was put in to skip iWARP nodes. Have
> > you tried that configuration ?
>
> With iWARP? No.
>
> Actually if it is important
It needs to work with a mix of iWARP and IB adapters in the machine.
> we can put this check back and to use just
> local umad_ca variable which will be freed locally. Sounds good?
Yes, that looks like it would work fine.
-- 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
