On Tue, 2004-10-12 at 18:57, Roland Dreier wrote: > Do you mean that the mthca driver should snoop on receive completions > on QP0 and look for traps with SLID==0? This seems pretty awkward > (since the MAD layer is going to lose those receive completions, how > do we handle freeing the MAD memory, etc., etc) and doesn't really > seem practical. The last time we discussed this I thought we reached > the conclusion that the MAD layer should give the low-level driver a > look at every MAD before doing agent dispatch, eg: > http://article.gmane.org/gmane.linux.drivers.openib/4264
It seems like the thread ended with an unanswered question. The answer appears to be that process_mad was used. Is that what we want to do for OpenIB ? Is this to be done for all MADs, all SMPs, or only certain ones ? If certain ones, is there a flag to indicate which ones need to be fed to the driver ? Does that avoid the issue of more than 1 consumer for a particular MAD ? That was another "rule" that I thought we achieved consensus on. It looks like the consumed flag takes care of this. Would there also need to be a send routine provided (by the MAD layer) which does not require an agent registration ? If so, is there a way to ensure that the driver is the only component to use this routine ? -- Hal _______________________________________________ openib-general mailing list [EMAIL PROTECTED] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
