From owner-freebsd-hackers  Sun Jan  5 04:10:18 1997
Return-Path: <owner-hackers>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id EAA25965
          for hackers-outgoing; Sun, 5 Jan 1997 04:10:18 -0800 (PST)
Received: from terminator.informatik.ba-stuttgart.de (terminator.informatik.ba-stuttgart.de [141.31.1.21])
          by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id EAA25874
          for <hackers@FreeBSD.ORG>; Sun, 5 Jan 1997 04:08:48 -0800 (PST)
Received: from helbig.informatik.ba-stuttgart.de (max66-2.informatik.ba-stuttgart.de [141.31.66.2]) by terminator.informatik.ba-stuttgart.de (8.7.6/8.7.3) with ESMTP id MAA06902 for <hackers@FreeBSD.ORG>; Sun, 5 Jan 1997 12:06:58 +0100
Received: (from wh@localhost)
          by helbig.informatik.ba-stuttgart.de (8.8.4/8.8.4)
	  id MAA21689 for hackers@FreeBSD.ORG; Sun, 5 Jan 1997 12:49:58 +0100 (MET)
From: Wolfgang Helbig <wh@MX.BA-Stuttgart.De>
Message-Id: <199701051149.MAA21689@helbig.informatik.ba-stuttgart.de>
Subject: Workaround for CMD640B flaw
To: hackers@FreeBSD.ORG
Date: Sun, 5 Jan 1997 12:49:57 +0100 (MET)
X-Mailer: ELM [version 2.4ME+ PL22 (25)]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-hackers@FreeBSD.ORG
X-Loop: FreeBSD.org
Precedence: bulk

Hi,

I attached a new disk two the secondary IDE channel as master.
The primary channel is used by another disk and an ATAPI-CDROM, both
working fine with FreeBSD 2.1.0 and 2.1.5 and -current.

Disklabeling and creating the filesystem on the new disk worked fine.
But when I started to copy files the system frooze.

With the help of FreeBSD-questions I found out, that the cause
is the CMD640B-IDE controller, that by design cannot control both
channels concurrently.

What needs to be done (and apparently is done in Linux) is to
serialize the use of the two channels (that is wdc0 and wdc1)

I tried a little changing the wd-driver but to no result.
Since the CMD640B is used on a lot of motherboards, I think we need
a workaround for this chip. May be there is an easy way to serialize? 

Any tips or help would be appreciated, since this is the first time
I meet the kernel source!

thanks,
Wolfgang Helbig