From owner-freebsd-hackers Mon May 28 1: 5: 6 2001 Delivered-To: freebsd-hackers@freebsd.org Received: from freebsd.dk (fw-rl0.freebsd.dk [212.242.86.114]) by hub.freebsd.org (Postfix) with ESMTP id 5C27637B422 for ; Mon, 28 May 2001 01:05:02 -0700 (PDT) (envelope-from sos@freebsd.dk) Received: (from sos@localhost) by freebsd.dk (8.11.3/8.11.3) id f4S84op02854; Mon, 28 May 2001 10:04:50 +0200 (CEST) (envelope-from sos) From: Søren Schmidt Message-Id: <200105280804.f4S84op02854@freebsd.dk> Subject: Re: Knob for ATA maximum UDMA? In-Reply-To: "from Richard Hodges at May 24, 2001 03:06:37 pm" To: Richard Hodges Date: Mon, 28 May 2001 10:03:30 +0200 (CEST) Cc: hackers@FreeBSD.ORG X-Mailer: ELM [version 2.4ME+ PL88 (25)] MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=ISO-8859-1 Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG It seems Richard Hodges wrote: On -current you can set the transfer mode to anything you like (and that the controller/device supports) using atacontrol, so no need to hack the kernel anymore :) > I was just testing out a new configuration, when I get two of > these about an hour apart, and then another today: > > ts8 /kernel: ad4: READ command timeout tag=0 serv=0 - resetting > ts8 /kernel: ata2: resetting devices .. done > > Needless to say, this is a problem. Here is the boot info: > > ts8 /kernel: ad0: 29311MB [59554/16/63] at ata0-master UDMA66 > ts8 /kernel: ad3: 29311MB [59554/16/63] at ata1-slave UDMA66 > ts8 /kernel: ad4: 29311MB [59554/16/63] at ata2-master UDMA100 > ts8 /kernel: ad5: 29311MB [59554/16/63] at ata2-slave UDMA100 > > The first two are on a VIA 82C686 bridge, the second two are HPT370. > I am wondering if UDMA100 is just too much for the cabling, and am > interested in finding out whether running them all at UDMA33 would > provide an extra safety margin. > > I see in dev/ata/ata-dma.c, ata_dmainit() checks some cable flag, and > reduces the UDMA capability to UDMA33 if neccessary. This seems like > a decent place to hardwire it for testing. Is there any other place > that also needs to be changed? > > If there actually is a difference in stability, would anyone favor > a new sysctl to put an arbitrary cap on the UDMA capabilities? As > far as I can tell, there should not be ANY performance difference > between UDMA100 and UDMA66, or even with UDMA33 if there is only > one drive per cable. > > Comments and suggestions are appreciated :-) > > -Richard > > ------------------------------------------- > Richard Hodges | Matriplex, inc. > Product Manager | 769 Basque Way > rh@matriplex.com | Carson City, NV 89706 > 775-886-6477 | www.matriplex.com > > > To Unsubscribe: send mail to majordomo@FreeBSD.org > with "unsubscribe freebsd-hackers" in the body of the message > -Søren To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message