[openib-general] Re: [PATCH] opensm: observe PartitionEnforcementCap of zero

[Hal Rosenstock]

On Thu, 2006-03-30 at 13:04, Rolf Manderscheid wrote:
> Hi Hal,
> > I presume this is with the trunk version, right ?
> Right.
> >
> > This appears like it would get past the ERR 4108 but wouldn't OpenSM
> > then insert the block several lines down in
> > osm_port.c::osm_physp_set_pkey_tbl where it calls osm_pkey_tbl_set ?
> >   
> The error is reported upon receipt of the SubnSet(PKeyTable) response.  
> The problem
> is that opensm is sending that set in the first place.  The patch to 
> shorten the initial vector length
> prevents opensm from doing the set inappropriately so we won't actually 
> get to the
> osm_pkey_rcv_process that calls osm_physp_set_pkey_tbl (which reports 
> error 4108 and
> calls osm_pkey_tbl_set).

One more thing:

With this patch, I have noticed the following:

Anafa based switch
   0: 0x0000 0xffff 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
   8: 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
  16: 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
  24: 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
32 pkeys capacity for this port

AnafaII based switch
   0: 0xffff 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000 0x0000
8 pkeys capacity for this port

So for some unknown reason, the default pkey was shifted in the pkey
table with this patch in. I'm not sure how that happened.

-- Hal

> > Also, is PortInfo:PartitionEnforcementInbound/Outbound set properly (0)
> > for these ports which don't support partition enforcement ? (That might
> > be another issue).
> >   
> Yes, those components are set to zero.  Actually, it looks like they're 
> unconditionally set to zero.
> 
>     Rolf
>