From owner-freebsd-current@FreeBSD.ORG Wed Jul 23 16:28:31 2003 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 5626B37B401 for ; Wed, 23 Jul 2003 16:28:31 -0700 (PDT) Received: from renaissance.homeip.net (m49.net81-67-14.noos.fr [81.67.14.49]) by mx1.FreeBSD.org (Postfix) with ESMTP id 64DB043F3F for ; Wed, 23 Jul 2003 16:28:28 -0700 (PDT) (envelope-from anthony.ginepro@laposte.net) Received: by renaissance.homeip.net (Postfix, from userid 1001) id E232F2053; Thu, 24 Jul 2003 01:28:26 +0200 (CEST) Date: Thu, 24 Jul 2003 01:28:26 +0200 From: Anthony Ginepro To: "Georg-W. Koltermann" Message-ID: <20030723232826.GA54677@renaissance.homeip.net> References: <1058998043.628.77.camel@bat.localnet> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1058998043.628.77.camel@bat.localnet> User-Agent: Mutt/1.4.1i X-Operating-System: FreeBSD 5.1-RELEASE i386 cc: freebsd-current@freebsd.org Subject: Re: still data corruption with 5.1-R on Intel Pentium 4 X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 23 Jul 2003 23:28:31 -0000 > There have been threads about data corruption in RAM on P4 and other > i386 machines on this list. I also observed the problem, on my laptop > with 5.0-R. It seemed to go away with 5.1-R, on the laptop. > > Recently I upgraded my home PC which is a P4 2.0A from 4.8-R to 5.1-R. > No problems at first. > > Until I ran "portsdb -Uu". I got a couple mysterious SIG4 and SIG11. > Just to be sure I rebooted and tried again, same result. I rebooted > another time, this time with ACPI disabled, and tried again, still the > same. > > Then I rebuilt my kernel with options DISABLE_PSE and DISABLE_PG_G, as > suggested by Terry in the old threads: Bingo, everything is fine now, no > more SIG4 or SIG11 during portsdb -Uu. > > It seems the workarounds that we have in 5.1-R are not effective enough. > What do you think? I'm having some SIG4 too when doing the massive operation "portsdb -Uu". I don't mind too much because it's usually only two make process which die however I would like to know if DISABLE_PSE and DISABLE_PG_G impacts performance or not and if it's specific to BSD. I tried lately rebuilding world and kernel with CPU_TYPE=p3 in order to be sure it was not due to gcc over-optimizing for p4, we will see. Anthony.