Return to BSD News archive
Xref: sserve comp.periphs.scsi:15475 comp.os.386bsd.questions:7002 comp.unix.osf.osf1:1442 Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!agate!howland.reston.ans.net!xlink.net!fauern!lrz-muenchen.de!colin.muc.de!ars.muc.de!jonas.ars.muc.de!rommel From: rommel@ars.muc.de (Kai Uwe Rommel) Newsgroups: comp.periphs.scsi,comp.os.386bsd.questions,comp.unix.osf.osf1 Subject: Re: Driver for Adaptec 274xT Distribution: world Message-ID: <2cec1a7d.jonas@jonas.ars.muc.de> Date: Fri, 19 Nov 93 02:13:51 MET References: <2c09tq$mji@homer.cs.mcgill.ca> <CGEzD8.GJH@tsoft.net> <2ccjep$n5d@u.cc.utah.edu> Organization: Private Keywords: scsi adaptec driver X-Posting-Software: UUPC/extended 1.11z inews (03Oct93 23:52) Lines: 65 terry@cs.weber.edu (A Wizard of Earth C) writes in article <2ccjep$n5d@u.cc.utah.edu>: >I called Adaptech and asked them if their 1742 EISA controller could be >configured to turn translation off. In retrospect, I should have asked >"do you know what translation is" first... .. >I explained that I wanted to boot a protected mode operating system that >shared a disk with a DOS partition on it, and they said that there was >no problem doing that. I patiently explained that, *yes* there was a >problem doing that, and tried to explain *why* there was a problem... >the following is a quote, exact as I can make it, from a suppoedly >UNIX support person at Adaptec: > >"Look, I don't care! I don't have time to have you tell me how UNIX >boots!" > >I replied that UNIX, without a great deal of additional hacking, would >*not* boot on their controllers (yes, I had tried both SVR4.02 and SCO >UNIX, in frustration!), and aked what we could do to resolve the >problem... was there a way to turn the translation off by hacking the >BIOS? I was more than willing to burn my own ROMs... ... >If they "don't have time" to let me tell them how UNIX boots for free, >they ought to hire someone to tell them for big $$$ (given the number >of Intel UNIX implementations, there's probably less than 100 of us), >because they sure as hell don't know. > >To date and to my knowledge, the *BSD implementations are the only ones >which can install correctly on translated drives without a lot of hacking, >and it's still very cumbersome. Given the small number of UNIX and UNIX >like OS's that support mounting DOS FS's, I suspect even if you could >get installed on a translated drive, you would have a hell of a time >mounting DOS FS's other than the first partition, and then you'd have to >be lucky anyway and hit a translated/untranslated cylinder boundry (ever >wonder why so many people recommend 0 and 1 as offsets?).. I may perhaps not understand all of what you talk about, but I had no problem installing FreeBSD (just a few days ago), Linux and some time ago also one of the SVR4.2 versions (Consensys) on Adaptec controllers *with* translation. No problem at all. What's the difference between telling the OS that the drive has 2607 cylinders, 16 heads and 54 sectors/track or telling it that it has 1523 cylinders, 64 heads and 32 sectors/track? Either one gives me the full size of the disk (in this case a DEC DSP-3160). Indeed, I have a problem to understand how it should work without translation for Unix, if all the cylinder/head/sector values written into the partition table by DOS or OS/2 or any other FDISK program are related to the translated geometry. And, what do you do if the physical geometry of your disk is impossible to express this way to the OS because it uses ZBR? Isn't one of the advantages of SCSI, that block devices (such as disks) on it just have addressable blocks and no longer cylinders, heads and sectors? Kai Uwe Rommel -- /* Kai Uwe Rommel Muenchen, Germany * * rommel@ars.muc.de CompuServe 100265,2651 * * rommel@informatik.tu-muenchen.de Fax +49 89 324 4524 */ DOS ... is still a real mode only non-reentrant interrupt handler, and always will be. -Russell Williams