From owner-freebsd-stable Mon Jun 19 4:21:36 2000 Delivered-To: freebsd-stable@freebsd.org Received: from atlrel2.hp.com (atlrel2.hp.com [156.153.255.202]) by hub.freebsd.org (Postfix) with ESMTP id 7911037BCB4 for ; Mon, 19 Jun 2000 04:21:08 -0700 (PDT) (envelope-from steve@sse0691.bri.hp.com) Received: from l3107mxr.atl.hp.com (l3107mxr.atl.hp.com [15.19.254.19]) by atlrel2.hp.com (Postfix) with ESMTP id 3706321CF; Mon, 19 Jun 2000 07:21:07 -0400 (EDT) Received: from sse0691.bri.hp.com (sse0691.bri.hp.com [15.144.0.53]) by l3107mxr.atl.hp.com (Postfix) with ESMTP id E02654FD93; Mon, 19 Jun 2000 06:21:01 -0400 (EDT) Received: (from steve@localhost) by sse0691.bri.hp.com (8.9.3/8.9.3) id MAA27379; Mon, 19 Jun 2000 12:23:00 +0100 (BST) (envelope-from steve) Date: Mon, 19 Jun 2000 12:22:57 +0100 From: Steve Roome To: Cy Schubert - ITSD Open Systems Group Cc: freebsd-stable@freebsd.org Subject: Re: ATA DMA Problem Message-ID: <20000619122255.A27262@moose.bri.hp.com> References: <200006171506.e5HF6sg00904@cwsys.cwsent.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 1.0.1i In-Reply-To: <200006171506.e5HF6sg00904@cwsys.cwsent.com>; from Cy.Schubert@uumail.gov.bc.ca on Sat, Jun 17, 2000 at 08:06:22AM -0700 Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG On Sat, Jun 17, 2000 at 08:06:22AM -0700, Cy Schubert - ITSD Open Systems Group wrote: > I've discovered an interesting problem as follows: > > Jun 17 05:16:59 cwsys /kernel: ad2: HARD READ ERROR blk# > 4999424ata1-master: WAR > NING: WAIT_READY active=ATA_ACTIVE_ATA > Jun 17 05:16:59 cwsys /kernel: ad2: DMA problem fallback to PIO mode > Jun 17 05:16:59 cwsys last message repeated 15 times This probably isn't much help or even the same problem, but there's a possibility it's the same thing, so I'm gonna follow up anyway. I started getting hard read errors on a 6 month old 34Gb IBM disk shortly after installing FreeBSD 4.0 as well. However this happened on a mounted DOS partition and that since became unusable - DOS reported it as a bad block after that as well. I found that it would only be visible under FreeBSD using dd like shown (although I didn't try changing blocksize.) The FAT filesystem was mountable from windows and DOS, except that once DOS scandisk had run it obviously cleaned up a bit to work round it. However, using an IBM fixit disk I couldn't find anything wrong with the disk - even though DOS scandisk has since wrecked it's filesystem and made dual boot to windows impossible. (C:\WINDOWS\SYSTEM died). In my case I beleive I probably have a problem with the disk, however if this is happenning is it possible that a similar driver problem could have been at fault ? Or, more likely, have we both got broken disks. Steve > 1. Possible heat problem, bs=1024k probably drives the disk harder than > the default bs=512. Interestingly, 3.4 didn't experience this > problem. > Maybe 4.0's new ATA driver is more efficient, driving the disk > harder > and causing the disk's controller board to run a little hotter than > it > did under 3.4. (most) This seems unlikely and I've had similar western digital disks that used to be used very heavily without any problems for a prolonged periods without any trouble. > 2. Possible disk/controller/DMA timing issue, e.g. hardware problem > excluding HDA failure. > > 3. Possible FreeBSD ATA/DMA driver bug. (least) > Other things to consider : 4. You actually have a bad block. 5. You're drive/controller/driver combination can't actually perform to spec, and with UDMA at full pelt the disk ends up getting back to the controller with an "I can't keep up", which is interpreted as a hard read error. > Any comments would be welcome. I'm going to send off for a replacement disk anyway, as I think option 4 is reasonably likely. Steve To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message