From owner-freebsd-current Mon Jan 11 21:52:21 1999 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id VAA00214 for freebsd-current-outgoing; Mon, 11 Jan 1999 21:52:21 -0800 (PST) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from st-lcremean.tidalwave.net (host-e186.tidalwave.net [208.213.203.186] (may be forged)) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id VAA00167 for ; Mon, 11 Jan 1999 21:51:58 -0800 (PST) (envelope-from lee@st-lcremean.tidalwave.net) Received: (from lee@localhost) by st-lcremean.tidalwave.net (8.9.1/8.8.8) id AAA68888; Tue, 12 Jan 1999 00:49:45 -0500 (EST) (envelope-from lee) Message-ID: <19990112004945.A68879@tidalwave.net> Date: Tue, 12 Jan 1999 00:49:45 -0500 From: Lee Cremeans To: Andrew Atrens Cc: current@FreeBSD.ORG Subject: Re: potential ide_pci.c bugster Reply-To: lcremean@tidalwave.net References: <19990111161251.A63501@tidalwave.net> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.93.2i In-Reply-To: ; from Andrew Atrens on Tue, Jan 12, 1999 at 12:18:27AM -0500 X-OS: FreeBSD 3.0-CURRENT X-Evil: microsoft.com Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Tue, Jan 12, 1999 at 12:18:27AM -0500, Andrew Atrens wrote: > > Lee, > > Around the middle of December you (actually Mike on behalf of you) > committed a patch to pci_ide.c which has since been hosing my system: [snip] > With this commit, my system freezes on boot with DMA timeout messages when > it tries to mount root on wd1. It's a total showstopper. > > I disabled DMA (ala wdc0 flags) and everything worked, though of course > more slowly. :) So, instead I chose to back out your change, and wait to > see if anyone else fell in the pit. Since the mailing list seems to be > pretty silent on the topic, maybe it's just me :) ... > > I've got an ABIT motherboard with TX chipset, bussed at 83MHz, that seems ^^^^^^^^^^ ^^^^^ There's the problem. UDMA drives tend to be very finicky about their timing, and since their timing is derived from the system PCI clock, overclocking to 83 MHz makes the frequency on the cable 41.5 MHz -- which your Quantum EL drive (and also my Seagate Medalist Pro 9140, for that matter) chokes on. The only thing you can really do to get reliable UDMA is to downclock...first 75, then 66 MHz. The bigger problem, though, is that ide_pci is a little too gung-ho about UDMA...if it sees a drive that supports it, it'll go ahead and set it up, totally unaware of any bus-speed issues, and with no way to tell it to use a lesser DMA mode. I'm going to ask on -current about adding a flag to force off UDMA on certain units, making them fall back to multiword DMA 2 (which is nowhere near as finicky). -- +--------------------------------------------------------------------+ | Lee Cremeans -- Manassas, VA, USA (WakkyMouse on DALnet and WTnet)| | lcremean@tidalwave.net| http://st-lcremean.tidalwave.net/~lee | To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message