From owner-freebsd-stable  Sat Mar 24 12:30:38 2001
Delivered-To: freebsd-stable@freebsd.org
Received: from earth.backplane.com (earth-nat-cw.backplane.com [208.161.114.67])
	by hub.freebsd.org (Postfix) with ESMTP id 2355A37B71A
	for <freebsd-stable@FreeBSD.ORG>; Sat, 24 Mar 2001 12:30:36 -0800 (PST)
	(envelope-from dillon@earth.backplane.com)
Received: (from dillon@localhost)
	by earth.backplane.com (8.11.2/8.9.3) id f2OKSKR05745;
	Sat, 24 Mar 2001 12:28:20 -0800 (PST)
	(envelope-from dillon)
Date: Sat, 24 Mar 2001 12:28:20 -0800 (PST)
From: Matt Dillon <dillon@earth.backplane.com>
Message-Id: <200103242028.f2OKSKR05745@earth.backplane.com>
To: Dave Tweten <tweten@nas.nasa.gov>
Cc: freebsd-stable@FreeBSD.ORG
Subject: Re: 4.3-RC Kernel Buffer Corruption (Was: 4.3-BETA makeworld of 
 current STABLE Fails)
References:  <200103240249.f2O2ng602441@gilmore.nas.nasa.gov>
Sender: owner-freebsd-stable@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

    I've gone through a diff of the -stable kernel between Feb 15 
    and March 6.  It took a while.  I had to hack cvs to allow date
    specs to include branch names (else cvs diff -D -D works off of
    HEAD).  I'll send an email with that patch a little later after
    I clean it up.

    I went through around 26786 lines of diffs.  All sorts stuff was
    changed/added, such as the kqueue stuff, M_ZERO, networking, some 
    softupdates cleanups, a minor VM fix, etc....  but most of it
    just doesn't apply to a buildworld.

    Two things stand out.

    * The allocation of memory was changed to allocate from
    the end of memory backwards rather then the beginning of memory
    forwards.

    * ATA/IDE was adjusted to support new DMA modes for VIA chipsets.


    I think there are two possibilities here:

    (1) You have bad memory somewhere and the memory ordering change
	is simply making it apparent.

    (2) Your motherboard is using the VIA chipset and the new kernels
	are using a different DMA mode then the Feb16 kernel, and 
	there is a DMA related bug.

    It should be easy to check both.  To check #2 look at the 'dmesg'
    output when booting the working kernel and when booting the new
    kernel.  Boot in bootverbose mode (I think that's -v from the bootstrap
    prompt).

    To check #1 try replacing the memory in the box.

    I will look at the changes made to the memory ordering in vm/vm_page.c
    and to zalloc later tonight, but I don't think they introduced any
    bugs.
	
						-Matt



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