I guess that's just one way to make local RDMAoe subnet know that grh must be used and at the same time same code also works in real IB environment. There might be other specifications that i don't know. -James
________________________________ From: Sean Hefty [mailto:[email protected]] Sent: Thu 12/3/2009 8:55 PM To: James Yang; Tzachi Dar; [email protected] Subject: RE: [ofw] patch 2/2 Add support for RDMAoEth to the low level driver >RDMAoE always use gid for communication, thus grh_valid should be always true. >With dgid=fe80::0, ib_gid_is_link_local() will return true, and make grh_valid >FALSE. Yes - I understand that. What I don't understand is *why* a path record for RDMAoE would specify a hop count > 1 to begin with. That doesn't make sense, and a can't imagine that an RDMAoE specification would require a hop count > 1 for the local subnet case. The RDMAoE checks need to be more explicit. - Sean
_______________________________________________ ofw mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw
