From: Alex T. <al...@fc...> - 2002-11-05 19:01:09
|
On Tue, Nov 05, 2002 at 10:50:36AM -0700, Alex Tsariounov wrote: > On Tue, Nov 05, 2002 at 05:44:44PM +0000, John Levon wrote: > > please do and resubmit, phil convinced me we may as well ship with this > > to give it some testing > > Ok, will do. Ok, changes applied and modified patch attached. A couple of notes: Phil, I did get it wrong so I reverted back to how Intel does it (the mask calculation); John, your original question of why check all processors: we don't know which processor we are originally running on when the check is performed, so although we could check num_threads +1 processors (or something like that), it's probably simpler to just do them all at that point; also, smp_call_function() doesn't do a selective run; I changed my do_cpu_id(&apic_id) to do_cpu_id(apic_id), although ANSI allows &array, it's probably more intuitive to just use the array name. Regards, Alex |