Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 26 Feb 2009 23:58:18 +0200
From:      Alexander Motin <mav@FreeBSD.org>
To:        Manfred Antar <null@pozo.com>
Cc:        Gavin Atkinson <gavin@FreeBSD.org>, FreeBSD-Current <freebsd-current@freebsd.org>
Subject:   Re: Fatal trap 18 on Current 1386 since Saturday 21 Feb 2009
Message-ID:  <49A7107A.5080208@FreeBSD.org>
In-Reply-To: <200902260208.n1Q280jg001474@pozo.com>
References:  <200902250259.n1P2xvYh001449@pozo.com> <1235559230.15704.4.camel@buffy.york.ac.uk> <200902251410.n1PEAla5010605@pozo.com> <1235573928.15704.10.camel@buffy.york.ac.uk> <200902251508.n1PF8ORo001737@pozo.com> <1235576175.15704.15.camel@buffy.york.ac.uk> <200902260208.n1Q280jg001474@pozo.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Manfred Antar wrote:
> At 07:36 2/25/2009, Gavin Atkinson wrote:
>> My suspicion is that it is this commit:
>>
>> http://docs.freebsd.org/cgi/getmsg.cgi?fetch=1637857+0+archive/2009/svn-src-all/20090222.svn-src-all 
>>
>>
>> Might be worthwhile trying the source from just before, and just after
>> it and seeing if that is the cause.
> 
> Yes that's it I put:
> 
> /sys/dev/ata/ata-all.c
> /sys/dev/ata/ata-all.h
> /sys/dev/ata/ata-pci.h
> /sys/dev/ata/ata-sata.c
> /sys/dev/ata/chipsets/ata-promise.c
> 
> from before the changes on the 21st in current /sys and kernel builds 
> and boots fine
> My original date of the 14th was wrong, I looked at the calendar thought 
> it was the 14th
> but actually uname -a of the last working kernel was the 21st 7am.

Looking on on your first and later posts I would say that this commit 
may happen not a root of your problem. It may just trigger something else.

Problem begins when some of your drives on ata1 channel timeouts request 
that leads to several controller reinits. I think that durung that 
driver somehow gets wrong drive parameters that leads to division by 
zero error inside ata_tf_write().

I have just committed one patch that probably does not fix the problem, 
but change process a bit. Could you try it and send me all ata related 
verbose messages starting from controller detection.

-- 
Alexander Motin



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?49A7107A.5080208>