[openib-general] question regarding GRH flag in ib_ah_attr

[Hal Rosenstock]

On Sun, 2006-05-14 at 15:30, Jason Gunthorpe wrote:
> On Sun, May 14, 2006 at 07:40:25AM -0400, Hal Rosenstock wrote:
> > > > Not always true in terms of local subnet (multicast and management MAD
> > > > response exceptions).
> > > 
> > > Yes, but these are well specified. Multicast must always have a GRH.
> > > MAD requests are covered under my scenario above and MAD responses
> > > to MAD requests with GRH's are specified to use the GRH and set the
> > > HopLimit = 0xFF.
> > 
> > Where does the spec say HopLmt needs to be 0xFF for multicast ?
> 
> I ment that the spec says a MAD response with a GRH should have 0xFF
> for HopLmt. (13.5.4.4)

Right; from the MAD response rules.

> I'd expect the Multicast HopLmt to come from the SA, just like in the
> unicast case.

OK; that's what I thought.

> > Off subnet is either determined by the prefix comparison or HopLimit >=2
> > in the response from the SA. The latter is implied by C8-16 on p. 229.
> 
> The only possible downside of using HopLimit, that I can see, is
> compatability with existing SA's. Do all existing SA's set HopLmt to 0
> or 1 in path record responses? (Since no SA's support routers,
> that would be correct..)

I would argue that the implementations would not be conformant if that
were not the case currently.

> Scope should not be a problem because the SA can follow whatever
> scope based rules might exist and then set HopLimit properly.

Sure, the SA would certainly use the scope to know whether it needs to
go beyond the local subnet for path resolution (both unicast and
multicast).

> FWIW, my vote would be to use HopLimit, since that lets the SA
> tell the client if it should use a GRH. With prefix comparison GRH
> usage is not under the control of the SA - so it is less flexable.

Makes sense to me (now)...

-- Hal

> Jason