From owner-cvs-all@FreeBSD.ORG Thu Sep 30 19:24:46 2004 Return-Path: Delivered-To: cvs-all@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 26A4416A4CE; Thu, 30 Sep 2004 19:24:46 +0000 (GMT) Received: from electra.cse.Buffalo.EDU (electra.cse.Buffalo.EDU [128.205.32.2]) by mx1.FreeBSD.org (Postfix) with ESMTP id B874043D39; Thu, 30 Sep 2004 19:24:45 +0000 (GMT) (envelope-from kensmith@cse.Buffalo.EDU) Received: from electra.cse.Buffalo.EDU (localhost [127.0.0.1]) i8UJOiTH020847; Thu, 30 Sep 2004 15:24:44 -0400 (EDT) Received: (from kensmith@localhost) by electra.cse.Buffalo.EDU (8.12.10/8.12.9/Submit) id i8UJOhPd020846; Thu, 30 Sep 2004 15:24:43 -0400 (EDT) Date: Thu, 30 Sep 2004 15:24:43 -0400 From: Ken Smith To: Poul-Henning Kamp Message-ID: <20040930192443.GE19077@electra.cse.Buffalo.EDU> References: <200409301438.i8UEcx89022771@repoman.freebsd.org> <48971.1096571367@critter.freebsd.dk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <48971.1096571367@critter.freebsd.dk> User-Agent: Mutt/1.4.1i cc: cvs-src@FreeBSD.org cc: src-committers@FreeBSD.org cc: cvs-all@FreeBSD.org cc: Ken Smith Subject: Re: cvs commit: src/sys/sparc64/sparc64 machdep.c X-BeenThere: cvs-all@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: CVS commit messages for the entire tree List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 30 Sep 2004 19:24:46 -0000 On Thu, Sep 30, 2004 at 09:09:27PM +0200, Poul-Henning Kamp wrote: > In message <200409301438.i8UEcx89022771@repoman.freebsd.org>, Ken Smith writes: > >kensmith 2004-09-30 14:38:59 UTC > > > > FreeBSD src repository > > > > Modified files: > > sys/sparc64/sparc64 machdep.c > > Log: > > This along with v1.6 of counter.c fixes some timecounter issues on > > MP machines (hopefully). CPU timers are OK on UP machines but we > > don't keep the timers in sync on MP machines so if the CPU's timer > > is chosen as the primary timecounter it's possible for time to > > not be monotonically increasing because different CPU's counters > > may be used at different times. But the CPU's counters are otherwise > > one of the higher quality counters available. So, on UP machines > > we'll use a relatively high quality value but on MP machines we'll > > use a quality that should prevent the CPU's counters from being chosen. > > The 'microsecond clock' is present in all suns so we don't leave any > machines without timecounters, right ? > Funny you should ask. Testing the first version of this patch was how we found the bug in counter.c - the machine I booted the test kernel with happened to decide the counter-timer wasn't suitable either due to the bug and it came up with no timecounters. You should see what happens to a machine that's not keeping time. :-) Marius has device maps for virtually all known Sparc hardware and he said all of the MP machines do have the counter-timer available in them. Only the IIi and IIe processor based systems don't and they're only UP machines. We are drastically low on timecounter-capable devices on sparc64 though, it seems the counter-timer and the CPU are about all there is. If this makes you nervous and if you have any suggestions I'm all ears. -- Ken Smith - From there to here, from here to | kensmith@cse.buffalo.edu there, funny things are everywhere. | - Theodore Geisel |