From owner-freebsd-questions  Fri Oct  6 03:00:49 1995
Return-Path: owner-questions
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id DAA20487
          for questions-outgoing; Fri, 6 Oct 1995 03:00:49 -0700
Received: from Root.COM (implode.Root.COM [198.145.90.17])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id DAA20478
          for <freebsd-questions@freebsd.org>; Fri, 6 Oct 1995 03:00:42 -0700
Received: from corbin.Root.COM (corbin [198.145.90.50]) by Root.COM (8.6.12/8.6.5) with ESMTP id CAA27258; Fri, 6 Oct 1995 02:59:08 -0700
Received: from localhost (localhost [127.0.0.1]) by corbin.Root.COM (8.6.12/8.6.5) with SMTP id DAA00325; Fri, 6 Oct 1995 03:00:29 -0700
Message-Id: <199510061000.DAA00325@corbin.Root.COM>
To: Dave Andersen <angio@aros.net>
cc: freebsd-questions@freebsd.org
Subject: Re: Crash questions. 
In-reply-to: Your message of "Thu, 05 Oct 95 23:18:58 MDT."
             <199510060518.XAA30988@terra.aros.net> 
From: David Greenman <davidg@Root.COM>
Reply-To: davidg@Root.COM
Date: Fri, 06 Oct 1995 03:00:28 -0700
Sender: owner-questions@freebsd.org
Precedence: bulk

>    My news machine decided that it just _needed_ to die today after 
>about 5 days of uptime without a hitch.  It's a P100, adaptec 2950 pci 
>scsi, 32mb ram, several large drives (9, 4, 1 gb, respectively), nothing 
>all that out of the ordinary.  It kernel panicked with the following 
>messages:
>
>Oct  5 23:01:41 news /kernel: kernel page directory invalid pdir=0x1a08023, 
>va=0xefbfe000
>Oct  5 23:01:41 news /kernel: kernel page directory invalid pdir=0x1a08023, 
>va=0xefbfe000
>
>Oct  5 23:01:41 news /kernel: panic: invalid kernel page directory
>Oct  5 23:01:41 news /kernel: panic: invalid kernel page directory
>
>   I'm wondering if anyone has a quick idea of what happened, or if this 
>one of those fluke things I should just shake my head about and ignore?

   Can you tell me what version of FreeBSD you're running? I think this is
caused by a page table page being paged out during a very narrow window when
a new process is being forked. I added some checks in the fork code that
should cause the machine to panic earlier if this is really the problem.
During the entire operational time of wcarchive (about 2 years), I've seen the
above panic a total of _1_ time...so I can't say that I haven't seen it, but I
do think it is very rare. I'll add it to my whiteboard so I don't forget about
it. I don't know if I'll be able to fix it before the 2.1 release, however, as
it might require an architectural change to fix...but I will look into it.

-DG