From owner-freebsd-current  Sun Dec 19  8:32:28 1999
Delivered-To: freebsd-current@freebsd.org
Received: from freebsd.dk (freebsd.dk [212.242.42.178])
	by hub.freebsd.org (Postfix) with ESMTP id ACCB214F94
	for <freebsd-current@FreeBSD.ORG>; Sun, 19 Dec 1999 08:32:24 -0800 (PST)
	(envelope-from sos@freebsd.dk)
Received: (from sos@localhost)
	by freebsd.dk (8.9.3/8.9.1) id RAA22826;
	Sun, 19 Dec 1999 17:32:17 +0100 (CET)
	(envelope-from sos)
From: Soren Schmidt <sos@freebsd.dk>
Message-Id: <199912191632.RAA22826@freebsd.dk>
Subject: Re: ATA: more Promise Ultra wedges
In-Reply-To: <14428.10180.553556.127603@grasshopper.cs.duke.edu> from Andrew Gallatin at "Dec 18, 1999 07:56:15 pm"
To: gallatin@cs.duke.edu (Andrew Gallatin)
Date: Sun, 19 Dec 1999 17:32:16 +0100 (CET)
Cc: freebsd-current@FreeBSD.ORG
X-Mailer: ELM [version 2.4ME+ PL54 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 8bit
Sender: owner-freebsd-current@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

It seems Andrew Gallatin wrote:
> 
> Søren,
> 
> It looks like I spoke to soon when I said the world was safe for
> Promise Ultra users:
> 
> ad3: ad_timeout: lost disk contact - resetting
> ata4: resetting devices .. ad3: HARD WRITE ERROR blk# 6594768ad3: DMA problem en
> countered, fallback to PIO mode
> ad3: DMA problem encountered, fallback to PIO mode
> done
> ad1: UDMA CRC READ ERROR blk# 10522095 retrying
> ad3: ad_timeout: lost disk contact - resetting
> ata4: resetting devices .. done

Is the above all info, or is it snipped somehow, its looks a bit wierd..

> At this point the machine is unpingable & will not respond to a break
> on the console.  This is with a ccd stripe set, striped across 4
> Maxtor "Diamondmax" drives attached one per channel to 2 Promise Ultra
> cards.  `The kernel sources are dated slightly before the build time in
> the below boot messages.  (I'd have given you verbose messages, but
> this is a transcript from the serial console logs & the machine is
> wedged solid right now). I'm running with a timeout of 30 seconds as I
> was hoping to avoid a 'lost contact - resetting' situation as all hell
> breaks loose when those appear.
> 
> BTW, I'd really like a tunable or some way to prevent a permanent
> fallback to PIO.  I'm more than willing to tolerate one hard error per
> week or so on a disk which sees 10s of gigabytes of data read &
> written between errors.  

Hmm, I'll think about what we can do here...

> The driver was much more stable back in July when (I guess) you just
> ignored errors.  Using a July kernel, this machine will stay up for
> months with nothing but the occasional:
> 
> ad3: status=51 error=84
> ad_interrupt: hard error
> 
> It never looses contact, never wedges.  Oh for the good old days..

Hmm, back then you would get hosed data without knowing it, you
feel better taht way ? I dont :) but I need to solve the hanging
problem, I'll agree on that, anybody having a couple of bad disks
I could borrow :)

-Søren


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