Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.anu.edu.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!spool.mu.edu!howland.reston.ans.net!blackbush.xlink.net!news.dfn.de!RRZ.Uni-Koeln.DE!zpr.uni-koeln.de!se From: se@ZPR.Uni-Koeln.DE (Stefan Esser) Newsgroups: comp.unix.bsd.freebsd.misc Subject: Re: NCR810 SCSI help needed! Date: 8 Jan 1996 00:44:29 GMT Organization: Institute for Mathematics, University of Cologne, Germany Lines: 139 Sender: se@Sysiphos (Stefan Esser) Message-ID: <4cppdd$aj3@news.rrz.uni-koeln.de> References: <4cnmn0$dk8@access4.digex.net> NNTP-Posting-Host: sysiphos.mi.uni-koeln.de To: dcmyers@access4.digex.net (David Myers) Bcc: se,jkh@freebsd.org In article <4cnmn0$dk8@access4.digex.net>, dcmyers@access4.digex.net (David Myers) writes: |> |> |> |> I picked up an NCR810-based SCSI controller the other day, and have been |> struggling with it ever since. At 68 bucks, maybe I don't have any right |> to complain, but here's what's happening. Any help would be greatly |> appreciated. I'm the current maintainer of the NCR SCSI code. And while it works just fine with just about every current hard disk drive, there are problems with cheap CDROM drives, which seem to have a tendency to violate assumptions made in the driver ... This is hard to work out: I'd need to buy a crappy CDROM drive and spend possibly tens of hours, just to allow other FreeBSD users to get the most cost effective solution. Well, I'm ready to work on this, but I don't have any need for another CDROM drive (and then even one that is known to fail :). If nobody sends me such a drive, it will be very hard to get this resolved. Getting this right will need quite a number of tries, since the NCR runs as an independent CPU, and it is the code executed by the NCR that has to be changed. Well, lets see ... |> Background: |> |> Pentium/75 system with 16 megs of RAM, 4 PCI slots. Quantum Fireball one |> gigabyte SCSI drive, NEC double-speed CDROM. No IDE devices are present; |> all I've got are the two SCSI devices and my trusty 1.44 floppy. The |> Fireball still has MacOS on it, so I'm working from floppy-only until it |> gets formatted. It's just me and DOS 6.22. The motherboard says its BIOS |> supports NCR810 and Adaptec SCSI devices. Using the BIOS set-up program, I |> tell it the main disk is SCSI. Ok. Sounds all right, so far. |> The Card: |> |> Cute little NCR810-based PCI card bought from Technoland in Santa Clara, |> California. "Made in Taiwan." The card has almost *nothing* on it -- the |> '810 chip, a 40 MHz crystal, a couple of jumpers, that's it. The manual is |> indecipherable, but I understand it supports PCI bus mastering, |> scatter/gather, full SCSI-II, etc. Yes. It indeed does. There is a small CPU, DMA controller, SCSI controller with some FIFOs and the SCSI and PCI bus logic and drivers all in this single chip. That makes those controllers a lot cheaper than say an Adaptec 2940, but at no loss in performance. A nice example of good engineering ... |> The sales guy tells me it doesn't work |> on all systems, but can't give me any particulars. it only works in motherboards with SDMS support, since there is no BIOS on the NCR card. |> The card says it must |> be placed in the busmastering PCI slot, but my motherboard manual doesn't |> tell me which one that is (I assume slot one). On just about every PCI motherboard sold today all slots support bus-master cards. Only a few early i486 chip sets were limited to two PCI bus-master cards. |> The only jumper on the card |> selects between "INTA", "INTB", "INTC", and "INTD". I have no idea what |> those mean. I'm not a PC person. It came set on INTA, but just to be |> different, I put it on INTB. Please put it back! PCI uses IntA only, except in very special situations. IntA ought to be a seperate signal for each PCI slot. You have to map each slot's IntA to some else unused IRQ in the PCI BIOS setup menu. Details ought to be explained in your mother board manual. This is very BIOS specific, and newer mother boards may in fact do it automatically, but you have to check this, anyway. |> The Symptoms: |> |> Remember, I still have to format the SCSI disk. Which means I've got to |> install some SCSI drivers so DOS can see the thing. So, I blindly copy |> over the following files onto my boot floppy, but note with some suspicion |> that they're copyright *1993* by NCR Corporation: |> |> DOSCAM.SYS |> CDROM.SYS |> MSCDEX.EXE You don't need any of them for DOS access to your hard disk. They are only required for the CDROM. Newer versions of these drivers are available from many FTP servers (as well as SDMS upgrades). |> I figured that these would act as the high-level DOS drivers, which would |> work in conjunction with my motherboard's SCSI BIOS to let met get access |> to the Fireball. Once the Fireball and the CDROM are on-line, I slip in the |> FreeBSD CD and off I go. Well, no. I get a message saying that the NCR |> BIOS would not load, because "SDMS BIOS not found." Now, I don't know what |> "SDMS BIOS" is, but my motherboard is supposed to be able to drive the '810 |> chip. What happened? I've got an indecipherable SCSI manual, an |> indecipherable motherboard manual, and almost no PC knowledge. Hmm, you got a SDMS BIOS not found ... That is bad. Please verify the boot messages. There ought to be some line containing the word SDMS and then a list of SCSI devices (your Fireball and the CDROM). If those don't get displayed, then there is no SDMS! You will be able to use the NCR controller from FreeBSD without SDMS, but you won't be able to boot from a SCSI diskk drive. |> I have no idea what to do next. Do any of you? My only goal is to get |> FreeBSD running; I couldn't care less about this MS-DOS nonsense, and would |> happily skip DOS-level formatting if it can somehow be avoided. Thanks in |> advance. Well, in fact you don't need to do a DOS level format! But without SDMS, you've got some problem, because the primary and secondary boot loaders rely on BIOS support. Regards, STefan -- Stefan Esser, Zentrum fuer Paralleles Rechnen Tel: +49 221 4706021 Universitaet zu Koeln, Weyertal 80, 50931 Koeln FAX: +49 221 4705160 ============================================================================== http://www.zpr.uni-koeln.de/~se <se@ZPR.Uni-Koeln.DE>