dws@cseg.uucp (David W. Summers) (04/06/89)
I have a couple of questions for MINIX users at large: 1) I would be interested in hearing discussion of using MINIX in a protected mode environment on either the 286 or the 386. Has anyone done it? What were the problems? Does it work OK? How compatible with "standard" MINIX is it? Please forgive me if this has been discussed before, but I must have missed it. 2) It seems to be quite a job to come up with a recompilation of the KERNEL everytime something is changed. Wouldn't it be better to have a simple boot program on the first disk sector that calls a 'boot' program in the root directory which gives you several choices (maybe boot DOS or MINIX?) and does nice things like that? Then to recompile the KERNAL, all that would be needed would be to regenerate the /minix file (kind of like /unix or /xenix) instead of (re)creating the whole boot-up procedure everytime. Has anyone already done this? What are the PRO's and CON's of doing something like this? Thank you. - David Summers (dws@cseg.uucp)
Leisner.Henr@xerox.com (marty) (04/07/89)
David, You ask good questions. 1) I've done an implementation of a protected mode 286 Minix system and made substantial changes across the board. I'm going to try to understand what I did and see how it fits into 1.3/1.4 (it was based on 1.1). Bruce Evans (munnari!ditsyda.oz.au!evans@uunet.UU.NET) has done work with 286 and 386 protected mode Minix. I'm off the opinion Minix fits into a 386 machine much easier than a 286 machine (i.e if you want to do anything more than 1 text/1 data segment). 2) The Minix system is 3 distinct programs and it isn't a monolithic kernel like Unix is. I guess you would need /kernel, /mm and /fs -- there is some patching and clumsiness in coordinating the building of minix images. The biggest problem I found was I had to do a format each time I built a system -- that's one of the reasons I began booting from DOS and then switch to minix after booting. I also would like to boot in a more conventional way instead of building a magic image. marty ARPA: leisner.henr@xerox.com GV: leisner.henr NS: martin leisner:wbst139:xerox UUCP: hplabs!arisia!leisner
ugkamins@sunybcs.uucp (John Kaminski) (04/09/89)
In article <12570@louie.udel.EDU> Leisner.Henr@xerox.com (marty) writes: > [ for brevity, some stuff deleted ] > >I also would like to boot in a more conventional way instead of building a >magic image. > >marty >ARPA: leisner.henr@xerox.com >GV: leisner.henr >NS: martin leisner:wbst139:xerox >UUCP: hplabs!arisia!leisner I also have found it would sometimes be useful to start MINIX as a .COM or .EXE file. I do agree that it seems easier to contain the system in files, much like PC-DOS loads the "hidden" files IBMBIO.COM and IBMDOS.COM (heck, if you know how to look, they're not really hidden at all :^). For me, the question is whether the boot sector(s?) are large enough to hold clever enough machine code to be able to find the files properly. One criticism of mine is the fact that a file system disk, if booted, generally crashes the machine. If that's one of the extremely few things MesSDOS does better, that's one of them. Although all it does is print a somewhat stupid message prompting you to insert another disk, I think this is better than losing control, i.e., executing data (the information in the superblock). Since it was admitted in the text that the superblock is big and therefore able to store a lot of information (that could be calculated but is put there to do the calculations only once), why not use the space that way, i.e., make all disks "bootable" (but not necessarily "runnable")? P.S. -- I have received email from a coupla people saying there are people working on an Amiga version of MINIX!