Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 12 Jul 1996 04:45:38 -0700 (PDT)
From:      Jim Dennis <jim@starshine.org>
To:        fqueries@jraynard.demon.co.uk (James Raynard)
Cc:        craigh@bugsoft.com, matt@bdd.net, freebsd-questions@freebsd.org
Subject:   Re: kernel thinks it's on sd1, when actually sd0?
Message-ID:  <199607121145.EAA02180@starshine>
In-Reply-To: <199607112035.UAA05289@jraynard.demon.co.uk> from "James Raynard" at Jul 11, 96 08:35:22 pm

next in thread | previous in thread | raw e-mail | index | archive | help
> 
> > >drive to boot from. If you get bored typing hd(1,a)/kernel every time
> > >you boot, here's how to change them so it happens automatically:-
> > ># cd /sys/i386/boot/biosboot
> > ># vi boot.c

	So I take it that there's no way to do this with a command 
	like Linux' 'rdev' (which patches the root device info into
	the kernel's binary image).  I also gather that there's there's
	no way to pass this information as "parameters" to the kernel in 
	BootEasy (similar to what LILO or LOADLIN.EXE do to the Linux 
	kernel).

	Where are the other kernel boot options documented?  I know
	about '(device,slice)/kernel.filename' (like Sun's) and about 
	'-c'  but what about -C, -v, etc (I guess -s would be "single
	user mode").

> > Hmmm - I've been trying to figure out a way to do exactly this.  I have one
> > IDE drive (all DOS) and one SCSI drive (all FreeBSD) and I'm getting tired
> > of typing "hd(1,a)/kernel" at the booteasy prompt.  I tried the steps
> > outlined above but there were several problems.  The first problem is I
...

	I'd like to solve a similar problem.  I tried to installed
	the 2.2 SNAP onto a machine here -- one 200 Mb IDE, one 2.5Gb
	SCSI (Adaptec 1542), one 32Mb DOS partition on the IDE, 
	(freed up the rest for FBSD), one 1.3 extended partition
	on the SCSI (which is split into a number of Linux partitions --
	no DOS on it).

	Currently this machine boots into DOS and uses LOADLIN.EXE to
	go into Linux (it is the machine I'm typing at now) -- where it
	functions as my uucp host, a household web server, and a 
	dial-in host and smart terminal (I call into it from 
	wherever else to read my mail etc; and I use it to dial out 
	to my other accounts).

	I'm hoping to replace this aging Slackware with a fresh BSD
	(one with a sane compilation of Taylor UUCP and cnews for 
	starters).

	The problems are as follows:

		If I use FBSD.EXE -D (the INSTALL.BAT from the CD)
		the machine hangs between the ze0 and the npx0  
		(numerical processor extension???).  This isn't 
		too bad -- the machine is an ancient 386/33 with a 
		weird ULSI brand math co-processor; my copy of 
		Norton Diags hangs in the same place.

		So I bring up the CD (booting from a freshly made
		boot floppy and using -c to disable that and various
		other unnecessary drivers).

		I go through the fdisk and and the slice editor (disklabel??) 
		to configure one 170 Mb partition into root, var, and usr 
		slices.  If I configure configure it for bad block checking
		then the system dies spectacularly when I try to commit.
		If I leave out bad block checking freebsd's install trips 
		over the block of bad sectors that really is one the drive
		(dropped the machine in a parking lot a couple of years ago
		-- right off the hotel's little luggage cart). 

		I don't mind replacing the drive really.  But I'd rather
		add another SCSI and still boot off of the IDE.  I have
		another SCSI -- but I can't bring myself to throw away
		200 Mb of perfectly good storage (never had a lick of 
		trouble with that drive; even after dropping it and 
		creating a scratch on one of the platters).

		Ideally I could get this installed and get FBSD.EXE
		to let me pass parameters to the kernel that I load.

		Is there a way?

...
> IDE drives are detected before SCSI drives by the BIOS, so the
> bootstrap code has to be on wd0.

	Technically this is only true if you configured the CMOS	
	to expect them.  It is possible to set your CMOS to 
	"none-installed" and then the BIOS extensions on your 
	SCSI adapter will have a chance to boot off of one of your 
	SCSI disks.

	I'm tempted to try this on this box here. 

	 	(DOS uses a strange way of assigning letters to 
		hard drive partitions:  first primary partition on 
		the first drive is C:, next primary partition on the
		next drive (if it exists) is D:, first logical drive
		within the (only) extended partition is the next letter
		(D: or E:) -- an so on through the logical drives on that
		partition; and finally the extended partition on the
		2nd drive is handled.  If you add a 2nd drive to an
		all DOS system -- and create just an extended partition
		on it! -- you'll get what you expect.  If you add a 
		2nd drive but put a primary partition on it then the
		drive letters for all of your logical drives in the 
		extended part. of drive 0 are all changed!)

			(I just thought I'd share that with the crowd --
			it comes from personal experience).

> -- 
> James Raynard, Edinburgh, Scotland




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199607121145.EAA02180>