From owner-freebsd-hackers Mon Jan 17 13:21:32 2000 Delivered-To: freebsd-hackers@freebsd.org Received: from calvin.saturn-tech.com (calvin.saturn-tech.com [207.229.19.6]) by hub.freebsd.org (Postfix) with ESMTP id A6D6114E6E for ; Mon, 17 Jan 2000 13:20:54 -0800 (PST) (envelope-from drussell@saturn-tech.com) Received: from localhost (drussell@localhost) by calvin.saturn-tech.com (8.8.8/8.8.8) with SMTP id OAA28337; Mon, 17 Jan 2000 14:20:09 -0700 (MST) (envelope-from drussell@saturn-tech.com) Date: Mon, 17 Jan 2000 14:20:09 -0700 (MST) From: Doug Russell To: Doug Russell Cc: Thierry Herbelot , Alex , hackers@FreeBSD.ORG Subject: Re: Anyone working on a DOMEX scsi driver ? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG > > I've just bought a Mustek SCSI scanner, and it is bundled with a > > seemingly very simple SCSI controller built by a DOMEX company (from > > Taiwan) > > Throw it away. > Throw it as far as you can, then drive over it with you car. :) > > > PS : the board identifier is DMX3191D - the chip itself seems to be a > > Domex536 > > There is a Linux driver for DMX 3192U PCI SCSI in > drivers/scsi/ini9100u.c, which suggests that it might be similar to the > Initio 9100 device. What are the exact device and vendor ID numbers? > (If you don't know, just send the output of "pciconf -l".) Hmmm. Now that I look at my boards, I see that they are DMX3181LE boards with a Domex 436P chip. Your board may be "smarter". These are small ISA boards with two jumpers -- PNP/PNR and 0WS/1WS. I'm now confused. On closer inspection, the board DOES have pins at B23-B25 (IRQ3-5) and D03-D06 (IRQ10,11,12,15), which are wired to the 436P, so I may be totally out to lunch here. If it's set up properly with PNP, maybe you can give it an interrupt line. Maybe you don't want to drive over it just yet. :) I can say, I've never actually seen a DMX3181LE actually use interrupts. Installing the card and supplied drivers under NT yields 100% processor usage while scanning (you can't even move the mouse). The card/driver combo also confuses NT an hour or two after rebooting. It fails to find the scanner attached to the Domex card, AND also my CD-ROM and CD-R drives on the separate NCR-810 SCSI bus! (That isn't quite true. It can still FIND the devices on the NCR, they just take FOREVER to access, like there is a 15 second timeout on each command. Kind of like the current interrupt problem with PCMCIA ep device losing interrupts.) It confuses the whole SCSI subsystem on NT anyway. BLECH! all around. Also, from sane-mustek(5): SCSI ADAPTER TIPS Mustek SCSI scanners are typically delivered with an ISA SCSI adapter. Unfortunately, that adapter is not worth much since it is not interrupt driven. It is (sometimes) possible to get the supplied card to work, but without interrupt line, scanning will put so much load on the sys- tem, that it becomes almost unusable for other tasks. If you already have a working SCSI controller in your sys- tem, you should consider that Mustek scanners do not sup- port the SCSI-2 disconnect/reconnect protocol and hence tie up the SCSI bus while a scan is in progress. This means that no other SCSI device on the same bus can be accessed while a scan is in progress. Because the Mustek-supplied adapter is not worth much and because Mustek scanners do not support the SCSI-2 discon- nect/reconnect protocol, it is recommended to install a separate (cheap) SCSI controller for Mustek scanners. For example, ncr810 based cards are known to work fine and cost as little as fifty US dollars. I've always liked that wording. "Not worth much". :) As I said before, the 1200LS, anyway, does allow disconnect/reconnect, so the paragraphs that follow about having to set up your SCSI controller to disable disconnect/reconnect, sync negotiation, and tagged command queueing don't apply to these recent models. YMMV, of course! Later...... To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message