[ofa-general] RE: [ofw] saquery & osm vendor AL - ca_names missing from osm_vendor_t ?

Sasha Khapyorsky sashak at voltaire.com
Tue Feb 3 06:27:18 PST 2009


On 14:51 Mon 02 Feb     , Sean Hefty wrote:
> >>4) saquery.c is the only diags pgms (so far) which uses OpenSM MAD interfaces;
> >>the rest use libibmad.
> 
> Looking briefly at the saquery code, I don't understand the benefit to using the
> opensm vendor interfaces, versus using libibmad or even libibumad directly, and
> switching to libibumad looks doable.  (It's not clear to me that there are
> benefits to using libibmad over libibumad for saquery.)
> 
> - osm_bind_handle_t looks like it could map to a libibumad port_id (int).
> - osmv_query_sa() could map to umad_send(), followed by umad_recv() to
>   obtain the result.  (Replace osmv_query_sa with a new function.)
> - There are a couple other calls that are used to loop through all returned
>   attributes in a response MAD.  We could use the MAD attribute offset
>   directly.  (Update loops where osmv_get_query_* is called.)
> 
> Are there technical reasons why the opensm vendor library was chosen for
> saquery?

AFAIK there are no such reasons.

> Would there be any objection to changing saquery to use libibumad
> directly?  

Not from me.

Sasha



More information about the general mailing list