On Sun, 2007-10-21 at 15:52 +0200, Sasha Khapyorsky wrote:
> On 05:53 Sun 21 Oct , Hal Rosenstock wrote:
> > On Sun, 2007-10-21 at 14:04 +0200, Sasha Khapyorsky wrote:
> > > On 16:25 Tue 16 Oct , Hal Rosenstock wrote:
> > > > infiniband-diags/perfquery.c: Support PMAs which don't support
> > > > AllPortSelect option
> > > >
> > > > Currently only support single port HCAs are supported in this mode but
> > > > this can be extended for other devices if needed
> > > >
> > > > Tested-by: Greg Kurtzer <[EMAIL PROTECTED]>
> > > > Signed-off-by: Hal Rosenstock <[EMAIL PROTECTED]>
> > >
> > > Applied. Thanks.
> > >
> > > However I have a question below.
> > >
> > > > + if (allports == 1) {
> > > > +
> > > > + /*
> > > > + * Simulate all ports support in PMA
> > > > + * Determine node type, number of (physical) ports,
> > > > + * and, if switch, whether SP0 is enhanced
> > > > + * to determine first and last port to query
> > > > + */
> > > > +
> > > > + /* For now, support single port CAs */
> > > > + if (smp_query(data, &portid, IB_ATTR_NODE_INFO, 0, 0) <
> > > > 0)
> > > > + IBERROR("smp query nodeinfo failed");
> > > > + node_type = mad_get_field(data, 0, IB_NODE_TYPE_F);
> > > > + if (node_type != IB_NODE_CA) /* NodeType other than
> > > > CA ? */
> > > > + IBERROR("smp query nodeinfo: Node type not CA");
> > >
> > > Not supporting switches and routers
> >
> > and multiport CAs
> >
> > > is temporary limitation (like
> > > all port simulation for single port CAs only), right?
> >
> > Temporary in the sense that it can be fixed but I have no current plan
> > to do so as I am unaware of any practical need for this.
>
> Probably it would be better to have cleaner error message then -
> something like "non-CA nodes are not supported yet."
To be more precise, it would be "Only single port CAs currently
supported". Feel free to change the message if you want.
> > Are you aware of such a need ?
>
> Yes, assuming "all ports" queries are useful.
Can you be more specific as to where ? Greg's email was the only
indication of where this was needed.
-- 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
