[openib-general] [PATCH] osm: OpenSM exits on PathRecord query with zero LID

[Eitan Zahavi]

Hi Hal,

> 1. Is this needed as an OFED 1.1 patch ?
I would leave the OFED 1.1 for now. A wrong query can still crash the SM
but I have not hear about such so-far.
> 2. Is the same thing needed for SA MultiPathRecord ?
Probably yes.


Eitan Zahavi
Senior Engineering Director, Software Architect
Mellanox Technologies LTD
Tel:+972-4-9097208
Fax:+972-4-9593245
P.O. Box 586 Yokneam 20692 ISRAEL


> -----Original Message-----
> From: Hal Rosenstock [mailto:halr at voltaire.com]
> Sent: Wednesday, December 06, 2006 1:58 PM
> To: Eitan Zahavi
> Cc: Sasha Khapyorsky; Yevgeny Kliteynik; OPENIB GENERAL
> Subject: Re: [PATCH] osm: OpenSM exits on PathRecord query with zero
LID
> 
> Hi Eitan,
> 
> On Wed, 2006-12-06 at 05:21, Eitan Zahavi wrote:
> > Hi Hal,
> >
> > This is another catch from the nightly simulator based regression.
> > Simple: if OpenSM gets a PathRecord that eventually maps into a port
> > with zero LID (either SRC or DST) if just asserts (in debug mode) on
> > getting the LFT.
> >
> > The following patch catches this error.
> 
> Thanks. Applied (only to the management git repository).
> 
> A couple of related questions:
> 1. Is this needed as an OFED 1.1 patch ?
> 2. Is the same thing needed for SA MultiPathRecord ?
> 
> -- Hal