From owner-cvs-all Sun Jan 28 22:40:10 2001 Delivered-To: cvs-all@freebsd.org Received: from magnesium.net (toxic.magnesium.net [207.154.84.15]) by hub.freebsd.org (Postfix) with SMTP id 1004B37B69B for ; Sun, 28 Jan 2001 22:39:42 -0800 (PST) Received: (qmail 68078 invoked by uid 1142); 29 Jan 2001 06:39:41 -0000 Date: 28 Jan 2001 22:39:41 -0800 Date: Sun, 28 Jan 2001 22:39:35 -0800 From: Jason Evans To: John Baldwin Cc: cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org, Bosko Milekic Subject: Re: cvs commit: src/sys/i386/i386 mp_machdep.c Message-ID: <20010128223935.W87569@canonware.com> References: <20010127222552.S87569@canonware.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: ; from jhb@FreeBSD.org on Sun, Jan 28, 2001 at 10:19:10PM -0800 Sender: owner-cvs-all@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Sun, Jan 28, 2001 at 10:19:10PM -0800, John Baldwin wrote: > On 28-Jan-01 Jason Evans wrote: > > On Sat, Jan 27, 2001 at 10:14:25PM -0800, John Baldwin wrote: > >> Actually, what I suggested to jasone (and kind of preferred myself) was to > >> set > >> each curproc in the AP's when we initialized the idleprocs. This has the > >> benefit of being MI: > > > > Actually, I didn't forget it; the machine wouldn't boot with that change in > > place, so I took it back out. > > Did you do the check for NULL sot that you didn't change curproc on the boot > processor? If not you probably hung trying to release/recurse on Giant. > Without this or Bosko's version of the fix, the AP's have a NULL curproc when > they try to grab the ap_boot_lock and bad things can happen. I can't say for sure; I don't think the change was present at the times I generated the diffs, so I have no record of exactly what the code looked like. Chances are good though that I made exactly the mistake you describe. Jason To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe cvs-all" in the body of the message