From owner-freebsd-stable  Mon Oct 30 10:53:54 2000
Delivered-To: freebsd-stable@freebsd.org
Received: from mmap.nyct.net (mmap.nyct.net [216.44.109.243])
	by hub.freebsd.org (Postfix) with ESMTP id 7268237B4CF
	for <freebsd-stable@FreeBSD.ORG>; Mon, 30 Oct 2000 10:53:50 -0800 (PST)
Received: by mmap.nyct.net (Postfix, from userid 1000)
	id EAE22F9A0; Mon, 30 Oct 2000 13:52:47 -0500 (EST)
Date: Mon, 30 Oct 2000 13:52:47 -0500
From: Michael Bacarella <mbac@mmap.nyct.net>
To: freebsd-stable@FreeBSD.ORG
Subject: followup: Re: vm_page_remove() problem..
Message-ID: <20001030135247.A6740@mmap.nyct.net>
References: <20001027111723.A9057@mmap.nyct.net>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
User-Agent: Mutt/1.0.1i
In-Reply-To: <20001027111723.A9057@mmap.nyct.net>; from mbac on Fri, Oct 27, 2000 at 11:17:23AM -0400
Sender: owner-freebsd-stable@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

We compiled a non-SMP version of the same kernel and this problem has gone
away, or at least hasn't happened in 3 days when it used to happen twice
a day or so for 2 weeks. I realize a crash dump would be far more useful but
it would not be appreciated if I intentionally crashed a production machine.

To keep this short, I'll provide additional information on demand if
anyone is interested.

Thanks

On Fri, Oct 27, 2000 at 11:17:23AM -0400, Michael Bacarella wrote:
> This keeps happening to one of our multiprocessor servers. About twice
> a day.
> 
> panic: vm_page_remove(): page not found in hash
> mp_lock = 01000001; cpuid = 1; lapic.id = 01000000
> boot() called on cpu#1
> 
> syncing disks... 68 68 68 68 68 68 68 68 68 68 68 68 68 68 68 68 68 68 68 68 
> giving up on 67 buffers
> Uptime: 21h28m53s
> Automatic reboot in 15 seconds - press a key on the console to abort
> Rebooting...
> cpu_reset called on cpu#1
> cpu_reset: Stopping other CPUs
> cpu_reset: Restarting BSP
> cpu_reset_proxy: Grabbed mp lock for BSP
> cpu_reset_proxy: Stopped CPU 1
> 
> FreeBSD bsd10.nyct.net 4.1.1-STABLE FreeBSD 4.1.1-STABLE #0: Fri Oct 20 15:58:40 EDT 2000     myj@bsd6.nyct.net:/usr/obj/usr/src/sys/NYCT  i386
> 
> We've tweaked some variables in response to this (maxusers?) but it
> doesn't seem to do the trick.
> 
> It happens most when I do something memory intensive (like stopping and
> restarting apache, and all several-hundred children), but it really does
> happen quite randomly.
> 
> My wild uneducated guess is that both processors are calling vm_page_remove()
> on a page and the one that doesn't happen first ends up panic'ing because
> it can't find the page anymore.
> 
> We're in the process of trying it with a non-SMP kernel, but I figure I'd
> put this out early in any case as it's obviously a bug of some kind. If
> I really find myself at the end of a rope, I'm going to look to see if it's
> a hardware problem.

-- 
Michael Bacarella <mbac@mmap.nyct.net>	;finger address for public key
GPG Key Fingerprint: B4E4 82F5 BCAC AB83 E6F7  B5AA 933E 2A75 79A4 A9C1


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-stable" in the body of the message