[ofa-general] vendor_id in struct ib_device_attr

[Jason Gunthorpe]

On Tue, Jun 24, 2008 at 10:51:03AM -0700, Hal Rosenstock wrote:
> On Tue, 2008-06-24 at 11:45 -0600, Jason Gunthorpe wrote:
> > On Tue, Jun 24, 2008 at 04:54:26AM -0700, Hal Rosenstock wrote:
> > > On Mon, 2008-06-23 at 14:41 -0600, Jason Gunthorpe wrote:
> > > > Personally, I feel the intention of the spec was to follow the PCI
> > > > model, it does introduce subsystem ID's for the IO controller
> > > > stuff, for instance.
> > > 
> > > Where would the subsystem IDs go ?
> > 
> > This is the stuff in the IOControllerProfile (pg 1013)
> 
> OK but that's in an optional agent; wouldn't it need to be in something
> required to be useful for this purpose ?

Oh, sorry, I don't think the subsystem ID should be something we worry
about, I was just pointing out the original authors seemed to hold a
view like the PCI SIG and had the vendor/subsystem split, at least for
some parts of the spec.

Jason