On Thu, 2006-05-11 at 01:48, Jason Gunthorpe wrote: > On Wed, May 10, 2006 at 09:56:58PM -0700, Roland Dreier wrote: > > Hal> Huh ? In this case, aren't the subnet prefixes are required > > Hal> to be different ? > > > > It's kind of a crazy thing to do but I don't see anything in the IB > > spec that forbids two subnets with the same subnet prefix, or any > > reason why a router couldn't route between them. The SMs would just > > have to be smart enough to return the LID of the router for paths to > > ports on the other subnet, and the routers would have to have explicit > > routes rather than forwarding based on just GID prefix. > > Hmm, this is an interesting point, you can do this in IP land using > host routes. > > How about this - the Path record (and related) SA responses include > the Hop Limit fields and the spec says: > > 8.3.6 Hop Limit: [..] Setting this value to 0 or 1 will ensure that > the packet will not be forwarded beyond the local subnet. > > So, it is within the spec to use HopLmt >= 2 as the GRH required flag.
That would be a simpler check but HopLimit is not a required component of PathRecord but I think this may not be sufficient as just because a HopLimit >= 2 doesn't mean that a packet would be forwarded off subnet. > I'd propose that the combination of a non-link-local prefix and a >= 2 > Hop Limit should force a GRH. SM's that do not support routers should > always fill in 0 for HopLmt. Why is a request with just a non link local prefix (with HopLimit wildcarded) not sufficient ? -- Hal > Jason _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
