From owner-freebsd-stable@FreeBSD.ORG Wed Apr 29 09:48:51 2009 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 1747D106564A for ; Wed, 29 Apr 2009 09:48:51 +0000 (UTC) (envelope-from chris#@1command.com) Received: from mail.1command.com (dsl081-172-045.sea1.dsl.speakeasy.net [64.81.172.45]) by mx1.freebsd.org (Postfix) with ESMTP id C431D8FC0A for ; Wed, 29 Apr 2009 09:48:50 +0000 (UTC) (envelope-from chris#@1command.com) Received: from mail.1command.com (localhost.1command.com [127.0.0.1]) by mail.1command.com (8.13.3/8.13.3) with ESMTP id n3T9mg7w051890 for ; Wed, 29 Apr 2009 02:48:48 -0700 (PDT) (envelope-from chris#@1command.com) Received: (from www@localhost) by mail.1command.com (8.13.3/8.13.3/Submit) id n3T9mgOK051889 for freebsd-stable@freebsd.org; Wed, 29 Apr 2009 02:48:42 -0700 (PDT) (envelope-from chris#@1command.com) Received: from hitme.hitometer.net (hitme.hitometer.net [64.81.172.194]) by webmail.1command.com (H.R. Communications Messaging System) with HTTP; Wed, 29 Apr 2009 02:48:42 -0700 Message-ID: <20090429024842.sbgmnyjl28ko0kog@webmail.1command.com> X-Priority: 3 (Normal) Date: Wed, 29 Apr 2009 02:48:42 -0700 From: Chris H To: freebsd-stable@freebsd.org MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format="flowed" Content-Disposition: inline Content-Transfer-Encoding: 7bit User-Agent: H.R. Communications Internet Messaging System (HCIMS) 4.1 Professional (not for redistribution) / UNIX Subject: ATA DMA issues - are there problems lurking in 7.1? X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 29 Apr 2009 09:48:51 -0000 Greetings, I have several Tyan S2518UGN (Thunder LE-T) server boards that I've been running for a few yrs. I've always used the SCSI on them, so never really used the ATA ports. I once experimented dual-booting NT2003 & FBSD on one of the ATA ports about a year ago. But ran into the problem I'm running into now. Back then I just blamed it on the drive (even tho it was brand new, and ran NT without complaints). But FreeBSD complains. So this is the story now; I'm trying it again - different drive, and using XP-SP3 & BSD. XP installs/runs w/o a hitch. I then install BSD on the second half of the drive, and make an entry into the XP boot loader. Reboot into the fresh BSD install - now the trouble begins... I periodically receive the following messages: Apr 29 01:07:45 udns kernel: ata0: reiniting channel .. Apr 29 01:07:45 udns kernel: ata0: reset tp1 mask=03 ostat0=58 ostat1=50 Apr 29 01:07:45 udns kernel: ata0: stat0=0x50 err=0x01 lsb=0x00 msb=0x00 Apr 29 01:07:45 udns kernel: ata0: stat1=0x00 err=0x01 lsb=0x14 msb=0xeb Apr 29 01:07:45 udns kernel: ata0: reset tp2 stat0=50 stat1=00 devices=0x9 Apr 29 01:07:45 udns kernel: ad0: setting PIO4 on ROSB4 chip Apr 29 01:07:45 udns kernel: ad0: setting UDMA33 on ROSB4 chip Apr 29 01:07:45 udns kernel: acd0: setting PIO4 on ROSB4 chip Apr 29 01:07:45 udns kernel: ata0: reinit done .. Apr 29 01:07:45 udns kernel: ad0: TIMEOUT - READ_DMA retrying (1 retry left) LBA=18520542 Apr 29 01:07:45 udns kernel: Apr 29 01:07:50 udns kernel: ata0: reiniting channel .. Apr 29 01:07:50 udns kernel: ata0: reset tp1 mask=03 ostat0=58 ostat1=50 Apr 29 01:07:50 udns kernel: ata0: stat0=0x50 err=0x01 lsb=0x00 msb=0x00 Apr 29 01:07:50 udns kernel: ata0: stat1=0x00 err=0x01 lsb=0x14 msb=0xeb Apr 29 01:07:50 udns kernel: ata0: reset tp2 stat0=50 stat1=00 devices=0x9 Apr 29 01:07:50 udns kernel: ad0: setting PIO4 on ROSB4 chip Apr 29 01:07:50 udns kernel: ad0: setting UDMA33 on ROSB4 chip Apr 29 01:07:50 udns kernel: acd0: setting PIO4 on ROSB4 chip Apr 29 01:07:50 udns kernel: ata0: reinit done .. Apr 29 01:07:50 udns kernel: ad0: TIMEOUT - READ_DMA retrying (0 retries left) LBA=18520542 The board has a Serverworks/AMI BIOS (current flash). I'm running an 80 wire ribbon on a UDMA166 capable drive (first attampt on a 40Gig, second on a 13Gig). I can't imagine what would cause this. The kernel is GENERIC - as I won't be able to build world/kernel until I can resolve this. Is the kernel being too aggressive - eg; attempting to use the drive's capability when the BIOS only provides UDMA33? Anyway, until I can resolve this I'll be forced to /attempt/ to pool a bunch of SCSI drives into a single (software) RAID0. Will geom(8) and gmirror(8) do this for me? The drives are not all identical (size/manufacturer), but are U160's. But I was hoping to create one large drive out of them. Doable? Pointers/advice/etc GREATLY appreciated. Thank you for all your time and consideration. --Chris