We are experimenting with OFED 1.2 (alpha1) and have dumped the SL2VL tables for both a switch port and an HCA port using the smpqueury command:
switch: # SL2VL table: Lid 103 # SL: | 0| 1| 2| 3| 4| 5| 6| 7| 8| 9|10|11|12|13|14|15| ports: in 0, out 1: | 0| 1| 2| 3| 0| 1| 2| 3| 0| 1| 2| 3| 0| 1| 2| 3| ports: in 1, out 1: | 0| 1| 2| 3| 0| 1| 2| 3| 0| 1| 2| 3| 0| 1| 2| 3| ... HCA: # SL2VL table: Lid 37 # SL: | 0| 1| 2| 3| 4| 5| 6| 7| 8| 9|10|11|12|13|14|15| ports: in 0, out 0: | 3| 2| 1| 0| 3| 2| 1| 0| 3| 2| 1| 0| 3| 2| 1| 0| I would have expected the behavior that the switch describes - a one to one mapping mod the supported number of VLs. But I can't explain why the HCA VLs are in reverse order to the SL. If this were a host endian issue I would have expected both to behave the same. Can someone explain what is going on? Dale _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
