From owner-freebsd-current@FreeBSD.ORG Sat May 3 13:01:01 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 03B6637B401 for ; Sat, 3 May 2003 13:01:01 -0700 (PDT) Received: from gw.catspoiler.org (217-ip-163.nccn.net [209.79.217.163]) by mx1.FreeBSD.org (Postfix) with ESMTP id 55D0243FAF for ; Sat, 3 May 2003 13:01:00 -0700 (PDT) (envelope-from truckman@FreeBSD.org) Received: from FreeBSD.org (scratch.catspoiler.org [192.168.101.3]) by gw.catspoiler.org (8.12.9/8.12.9) with ESMTP id h43K0pM7020832; Sat, 3 May 2003 13:00:56 -0700 (PDT) (envelope-from truckman@FreeBSD.org) Message-Id: <200305032000.h43K0pM7020832@gw.catspoiler.org> Date: Sat, 3 May 2003 13:00:51 -0700 (PDT) From: Don Lewis To: marcus@marcuscom.com In-Reply-To: <1051978452.52870.12.camel@shumai.marcuscom.com> MIME-Version: 1.0 Content-Type: TEXT/plain; charset=us-ascii cc: smoberly@karamazov.org cc: current@FreeBSD.org Subject: Re: [FIXED?] More on the blkfree panic 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: Sat, 03 May 2003 20:01:01 -0000 On 3 May, Joe Marcus Clarke wrote: > On Sat, 2003-05-03 at 05:42, Don Lewis wrote: >> Interesting ... I'm having problems reproducing this problem. >> >> The first report of this bug landed in my inbox at about 15:12 UTC on >> April 27. Someone else who reported the bug claimed to have cvsup'ed at >> about 10:00 UTC on that day. >> >> I just did >> cvs -R update -Pd -D'4/27/2003 14:00:00 UTC' >> make buildworld >> make buildkernel >> make installkernel >> and rebooted. >> >> I haven't managed to trigger the problem with a simultaneous buildworld >> and untaring ports.tgz. I even suspected that this problem might be >> triggered by free memory pressure so I added >> hw.physmem="128M" >> to /boot/loader.conf and rebooted. I've pushed the machine heavily into >> paging, but it still seems to be stable, but painfully sluggish. > > I was also able to reproduce this, though not as reliably as my > localhost NFS mount trick, by doing a make extract in > x11/XFree86-4-libraries. This was on a lightly loaded P4 with 1 GB of > RAM and UFS1 file systems. I tried running this in a loop for a while, but still no sign of a problem. The system is an Athlon XP with 1 GB of RAM, a single UFS1 file system on a 10K RPM Seagate Cheetah SCSI drive with write-caching disabled. I can't try the NFS mount trick unless I rebuild the kernel to disable the DEBUG_VFS_LOCKS option. With this option enabled the NFS server code fails the vnode lock assertions very quickly. BTW, that puppy makes some very interesting humming sounds when the memory is cranked down and it is paging heavily, at around 240 I/O's per second ;-) On 3 May, Joe Marcus Clarke wrote: > Just as an update. After Alan's recent VM commits, the problem seems to > be resolved. I have been running my localhost NFS tinderbox build for a > while, and the machine has not rebooted...yet. Yeah, that vfs_subr.c commit sure looks suspicious. Maybe I've just been lucky not to have run into this problem since I've been running with the bad version of this file all along.