From owner-freebsd-current  Mon Dec 11 14:24:29 2000
From owner-freebsd-current@FreeBSD.ORG  Mon Dec 11 14:24:26 2000
Return-Path: <owner-freebsd-current@FreeBSD.ORG>
Delivered-To: freebsd-current@freebsd.org
Received: from sender.ngi.de (sender.ngi.de [212.79.47.18])
	by hub.freebsd.org (Postfix) with ESMTP
	id B3DAD37B400; Mon, 11 Dec 2000 14:24:25 -0800 (PST)
Received: from Gatekeeper.FreeBSD.org (koln-3e366425.pool.mediaWays.net [62.54.100.37])
	by sender.ngi.de (Postfix) with ESMTP
	id 3F95596E52; Mon, 11 Dec 2000 23:16:34 +0100 (CET)
Received: from StefanEsser.FreeBSD.org (StefanEsser [10.0.0.1])
	by Gatekeeper.FreeBSD.org (Postfix) with ESMTP
	id D04752E; Mon, 11 Dec 2000 22:50:47 +0100 (CET)
Received: by StefanEsser.FreeBSD.org (Postfix, from userid 200)
	id A97E5145C; Mon, 11 Dec 2000 22:51:27 +0100 (CET)
Date: Mon, 11 Dec 2000 22:51:27 +0100
From: Stefan Esser <se@freebsd.org>
To: Mike Smith <msmith@FreeBSD.ORG>
Cc: dg@root.com, Mark Huizer <freebsd@dohd.org>,
	freebsd-current@FreeBSD.ORG, Stefan Esser <se@freebsd.org>
Subject: Re: PCI power states (was Re: fxp driver not reset after Windows reboot? )
Message-ID: <20001211225127.B73017@StefanEsser.FreeBSD.org>
Reply-To: Stefan Esser <se@freebsd.org>
References: <200012102045.MAA12646@implode.root.com> <200012110904.eBB94Ox03098@mass.osd.bsdi.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2.5i
In-Reply-To: <200012110904.eBB94Ox03098@mass.osd.bsdi.com>; from msmith@FreeBSD.ORG on Mon, Dec 11, 2000 at 01:04:23AM -0800
Sender: owner-freebsd-current@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

On 2000-12-11 01:04 -0800, Mike Smith <msmith@FreeBSD.ORG> wrote:
> All of these can be abstracted as PCI methods, so they won't require lots 
> of cut-n-paste in each driver:
> 
> 	pci_enable_busmaster(dev);
> 	pci_enable_io(dev, SYS_RES_IOPORT | SYS_RES_MEMORY);
> 	pci_set_powerstate(dev, PCI_POWERSTATE_D0);
> 
> Consider the above a request for review on the matter.

Shouldn't that be:

 	pci_enable_io(dev, SYS_RES_IOPORT | SYS_RES_MEMORY);
 	pci_set_powerstate(dev, PCI_POWERSTATE_D0);
	device_specific_init(dev);
 	pci_enable_busmaster(dev);

As soon as the bus-master enable bit has been set, the device may start
writing to arbitrary memory locations (in the unlikely but still possible
case, that it has been disabled in the middle of a transfer).

Cases where such behaviour caused memory corruption and warm-boot failure
have been discussed in the PCI-SIG mail list.

Regards, STefan


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