From owner-freebsd-current Thu Mar 7 11:49:52 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id LAA03530 for current-outgoing; Thu, 7 Mar 1996 11:49:52 -0800 (PST) Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id LAA03525 for ; Thu, 7 Mar 1996 11:49:47 -0800 (PST) Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id MAA14525; Thu, 7 Mar 1996 12:45:09 -0700 From: Terry Lambert Message-Id: <199603071945.MAA14525@phaeton.artisoft.com> Subject: Re: Whee! Page Fault! To: mark@linus.demon.co.uk (Mark Valentine) Date: Thu, 7 Mar 1996 12:45:09 -0700 (MST) Cc: handy@sag.space.lockheed.com, terry@lambert.org, freebsd-current@FreeBSD.ORG In-Reply-To: <199603070055.AAA01358@linus.demon.co.uk> from "Mark Valentine" at Mar 7, 96 00:55:40 am X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-current@FreeBSD.ORG Precedence: bulk > > From: "Brian N. Handy" > > Date: Wed 6 Mar, 1996 > > Subject: Re: Whee! Page Fault! > > > 0xf0103a84 _cd9660_readdir > > 0xf0103f34 _cd9660_readlink > > > > Terry scores. Recall, I was doing an 'ls' at the time of the offending > > kernel panic. > > I can reliably reproduce this (see the message "reproducible fatal trap 12" > I just mailed to this list). Must be some clue in this "coincidence", since > the cd9660 code hasn't changed since early December... > > The last CTM delta I applied and built included the change to take > sys/vm/swap_pager.c to revision 1.63. I suspect the buffer the readdir is reading into is not in core at the time, and so you get a page fault that reeenters the uiomove. This is part of the generic problem with the organization of the VOP_LOCK code that I was going on about in another thread; IMO, it shouldn't matter that you got the fault. The code needs to better handle reentrancy in some of the /sys/kern files. The CD9660 code keeps asking people to rewrite it (8-)), but the HPFS and RR stuff is so kludged in the mount code that it is nearly impossible to cleanly make a fix without rewriting most of the code. I pointed this out to Jordan before, mostly because he has a larger set of CDROM's that he can use to do testing than all of the rest of us put together (probably). Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.