>Should there be some conditionalization of any non IBA compliant code so >it is only turned on if someone really wants this ?
That is doable. AFIK, the only non-compliance is using the permissive LIDs in the CM REQ. I don't believe this causes any interoperability issues with fully compliant code. It should just result in a rejected connection request. >I presume this is to be replaced by the real code some time in the >future once the IBA spec for these router issues is decided. Yes. The changes to the RDMA CM and passive side IB CM will likely need to be replaced (patches 2 & 3). The active side IB CM changes (patch 1) may be okay unless wording to the 1.2 spec changes. (Patch 1 could also be used to support non-reversible paths at some point, but more work may be needed.) - Sean _______________________________________________ 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
