From owner-freebsd-smp Thu Oct 3 06:58:46 1996 Return-Path: owner-smp Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id GAA28196 for smp-outgoing; Thu, 3 Oct 1996 06:58:46 -0700 (PDT) Received: from spinner.DIALix.COM (root@spinner.DIALix.COM [192.203.228.67]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id GAA28188; Thu, 3 Oct 1996 06:58:40 -0700 (PDT) Received: from spinner.DIALix.COM (peter@localhost.DIALix.oz.au [127.0.0.1]) by spinner.DIALix.COM (8.8.0/8.8.0) with ESMTP id VAA06005; Thu, 3 Oct 1996 21:58:35 +0800 (WST) Message-Id: <199610031358.VAA06005@spinner.DIALix.COM> X-Mailer: exmh version 1.6.7 5/3/96 To: dyson@freebsd.org cc: freebsd-smp@freebsd.org Subject: Re: Scheduling and idle loops.. (Was Re: cvs commit: sys/kern . . In-reply-to: Your message of "Thu, 03 Oct 1996 08:23:53 EST." <199610031323.IAA00634@dyson.iquest.net> Date: Thu, 03 Oct 1996 21:58:34 +0800 From: Peter Wemm Sender: owner-smp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk "John S. Dyson" wrote: > > > > This is equal top of my "todo" list. I think we'll get the tlb > > invalidation going next though, since it's a showstopper. > > > If you need help, I intend to come up with an ifdef'ed version > of pmap that removes the 1 page invalidations. This will mean that > all you have to do is to communicate the tlbflush (invltlb) operation. > > Will that help? Probably not necessary at this stage.. For the first round, we'll try for an "either call invalidates everything" approach, then try and optimise it a bit. :-) We can do message passing if it's worth it, ie: store the address in a specific location, then have the target cpu invalidate based on that address. But I suspect the cost of an apic IPI transaction is far higher than the potential savings so it may not be worth the effort. We'd probably better worry about the basics first.. (we've gotta do an EOI somehow, I need to check the docs) > John Cheers, -Peter