Hi Roland, On Tue, Oct 16, 2007 at 12:34:01PM -0700, Roland Dreier wrote: > BTW, a question about all this wakeup-saving stuff. For a case like > this, would we be better off INIT_DELAYED_WORK_DEFERRABLE(), since we > don't really care about cleaning up stale address handles on a system > that is totally idle and blissfully sleeping? > > Similarly for the mthca code that polls the HCA for a catastrophic > error (internal HCA crash), do you think we should be rounding the > 5*HZ interval, or using a deferrable timer?
Good question, I've been playing it safe and just rounding timers, but it does sound like both these cases would be served fine by using INIT_DELAYED_WORK_DEFERRABLE. Want me to make a patch? Anton _______________________________________________ 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
