From owner-freebsd-hackers Sat May 9 15:43:06 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id PAA10946 for freebsd-hackers-outgoing; Sat, 9 May 1998 15:43:06 -0700 (PDT) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from smtp04.primenet.com (daemon@smtp04.primenet.com [206.165.6.134]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id PAA10939 for ; Sat, 9 May 1998 15:42:59 -0700 (PDT) (envelope-from tlambert@usr07.primenet.com) Received: (from daemon@localhost) by smtp04.primenet.com (8.8.8/8.8.8) id PAA28145; Sat, 9 May 1998 15:42:57 -0700 (MST) Received: from usr07.primenet.com(206.165.6.207) via SMTP by smtp04.primenet.com, id smtpd028133; Sat May 9 15:42:56 1998 Received: (from tlambert@localhost) by usr07.primenet.com (8.8.5/8.8.5) id PAA03567; Sat, 9 May 1998 15:42:56 -0700 (MST) From: Terry Lambert Message-Id: <199805092242.PAA03567@usr07.primenet.com> Subject: Re: dump/restore problem (was: Network problem with 2.2.6-RELEASE) To: tom@sdf.com (Tom) Date: Sat, 9 May 1998 22:42:56 +0000 (GMT) Cc: tlambert@primenet.com, kpielorz@caladan.tdx.co.uk, beng@lcs.mit.edu, dec@phoenix.its.rpi.edu, freebsd-hackers@FreeBSD.ORG In-Reply-To: from "Tom" at May 7, 98 11:02:50 pm X-Mailer: ELM [version 2.4 PL25] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > > options "CMD640" > > > > > > That work around the known hardware bug in about 60% of all IDE > > controllers out there (by volume). > > The percentage is a bit high. It is been a while since I've seen any > of those around. > The controller on my motherboard is an Intel 82371SB. > Since it doesn't probe as a CMD640B, the 'options "CMD640B"' is a no-op > anyhow. Linux specifies a "SERIALIZE" option rather than a "CMD640B" option. They do this because the CMD640B is not the only buggy IDE controller. One or more of Intel's is buggy as well. Unfortuantely, it's not so easy to find the test program on www.intel.com any more. The specific macrocell is the one used in the RZ1000, which ships on Intel Motherboards (except those from their Server Products Division, which always has better hardware but for apparently political reasons is never allowed to ship it anywhere). Maybe someone should look at the linux/block/ide.c version 5.11 or better and see how they do their RZ1000 autodetection? In 5.27 and above, they should look at the external rz1000.c and cmd640.c. > Besides didn't this bug just hang the machine when both channels were > accessed at once. Occasionally. More often, it silently corrupted the data, in that the transfer was marked completed, but had actually been interrupted. This would easily account for your symptoms. > > He may also want to play with the flags on his controller; also from > > LINT: > > Already all off. One other thing that occurs to me, now that you have stated that you have a 82371SB (there are no notes on the Intel page for this part; are you sure you don't have a 82371AB?). This probably means you have a Triton chipset. These chipsets do not support more than 2 simultaneous contention requests for PCI bus masters. Since the Adaptec SCSI is a bus mastering device, if you are not running the IDE controller in PIO mode, if you have any other PCI master device (ie: some ethernet controllers), then this may also be your problem. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message