From owner-freebsd-current Fri Feb 13 18:46:12 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id SAA10318 for freebsd-current-outgoing; Fri, 13 Feb 1998 18:46:12 -0800 (PST) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from smtp01.primenet.com (smtp01.primenet.com [206.165.6.131]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id SAA10260 for ; Fri, 13 Feb 1998 18:46:05 -0800 (PST) (envelope-from tlambert@usr06.primenet.com) Received: (from daemon@localhost) by smtp01.primenet.com (8.8.8/8.8.8) id TAA07513; Fri, 13 Feb 1998 19:46:00 -0700 (MST) Received: from usr06.primenet.com(206.165.6.206) via SMTP by smtp01.primenet.com, id smtpd007482; Fri Feb 13 19:45:59 1998 Received: (from tlambert@localhost) by usr06.primenet.com (8.8.5/8.8.5) id TAA12922; Fri, 13 Feb 1998 19:45:57 -0700 (MST) From: Terry Lambert Message-Id: <199802140245.TAA12922@usr06.primenet.com> Subject: Re: Working (apparently) soft-update code available. To: chuckr@glue.umd.edu (Chuck Robey) Date: Sat, 14 Feb 1998 02:45:57 +0000 (GMT) Cc: julian@whistle.com, current@FreeBSD.ORG In-Reply-To: from "Chuck Robey" at Feb 13, 98 08:27:59 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > > look in: > > ftp://ftp.freebsd.org/pub/FreeBSD/incoming/softupdates4.tgz > > > > first extract the README and read it before extracting everything. > > this should patch cleanly against last night's -current. > > Julian, is this safe for an SMP system? As long as you are using the Big Giant Lock, it should be safe. The syncer process that replaces updated runs as a kernel process; this means that it's a process in the kernel address space and with a seperate stack. In theory, this will prevent you from coming in an doing something while it's doing something. But this theory depends on the Big Giant Lock in the scheduler. A pushdown below the system call trap layer could make this dangerous, unless done exactly right. Heed Julian's warning and run it on an expendable partition. I would be very interested in seeing a three-process thrash on a two processor machine (or an N+1 process thrash on an N processor machine) to see if you can kick it in the groin through the scheduler (SMP FreeBSD does not have CPU affinity in the form of seperate scheduler queues with a roll count for the idle process to determine the right time to do load migration). "We didn't make the Big Giant Lock, we only made it Big and Giant!" -- FreeBSD/SMP Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message