Well, thanks guy ! It looks like the spinlocks patch did the trick. The influence of the broken ref count was reaching further than expected, I guess. At least, spinlock or something else (can't see what in the svn logs), r5642 works like a charm.
The only thing I get that looks a bit disgruntled is this: __osm_pi_rcv_process_switch_port: ERR 0F03: Unknown link state = 5, port = 0x8 But it does not seem to hurt. And the port at the other end of the link seems fine. Now, I'll try hard to break the LID assignment and I'll pester you some more if something comes up. -- Jean-Christophe Hugly <[EMAIL PROTECTED]> PANTA _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
