Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 18 Jul 1996 18:30:20 -0700 (PDT)
From:      Jim Dennis <jim@starshine.org>
To:        questions@freebsd.org
Subject:   text = 0xe3000 -   -- and locks:  Why?
Message-ID:  <199607190130.SAA00179@starshine>

next in thread | raw e-mail | index | archive | help


	Okay all,

	I give in.  What's the magic incantation I have to type
	into my system to get it to boot?

	Here's the situation:

	I have a 16Mb 386DX33 with a ULSI 387 Match-coprocessor, a
	QuickPath Portfolio 550E I/O card (4serial, 2 IDE channels, 
	4 floppy, 2 parallel, and a game port) one 220M WD IDE hard
	drive, on 2.5G DEC SCSI hard drive, an Adaptec 1542C
	SCSI host adapter, and an old Texel double speed CD-ROM
	(and a 1Mb Tseng ET4000 SVGA card).

	I have a 32Mb DOS partition on the IDE drive. Linux is 
	installed an extended partition of the DEC (the SCSI HD)
	it is booted via LOADLIN.EXE.  All of that works like a 
	charm (I'm typing on it now).

	So far, under the old Linux installation, I have configured
	uucp, dial in terminal, dial out terminal, and some fax stuff
	(mgetty).  It is my intent to convert this system to FreeBSD,
	bring up Taylor, get a working copy of cnews configured, and
	get dial-out and incoming PPP working, and integrate a working
	send/recieve fax gateway on it.

	I started by trying to install 2.2 SNAP on from the WC CD-ROM.
	
	First problem:

		Hangs after ze0 or zp0.

		Solution:  boot with the -c option, ls the configured
					devices, disable npx0
					(also used visual and disabled all of the 
				    other unecessary drivers).

	So I could get booted and into the installation script.
	Now I decided to install FreeBSD on the other 170Mb of the
	IDE drive.  It seemed like it would be easiest (I just 
	blew away the extended DOS partition that had been there).

	Second problem:

		Crashes, Panics, Hangs, and Reboots after fdisk/disklabel.

		I tried several variations on this them.  It seems like it
		gets to where it wants to write to the disk and starts
		scrolling some error (something about IOPL's???) and flickering
		the screen until it reboots itself.

		I can't read the messages in any detail.

		One time I seemed to get it to the point where I had
		selected the packages and had it die on a bad block
		(in fact it seems to bad block checking that it's dying 
		on -- there really are a couple of bad blocks on that 
		drive; they're way out at the end).
Workaround:

			Install everything on the SCSI drive (1.5 Gig free)
	
	So I created one FreeBSD partition on that. Then I split that
	into a reasonable set of slices (64M for root, 48M for swap,
	63M for /var, rest for /usr).  2.2 SNAP was still giving
	me trouble after that (I don't remember what it was right
	now  -- but it was bad enough that I switched to plan B.

	Plan B:

		Borrowed my old 2.1.0 CD set from my former employer
		(McAfee Associates) -- I don't remember if I ever 
		turned it a receipt on it but they aren't concerned --
		their systems are stable and they know where to find 
		me if they need the CD's back.
		
		Still had to use the "disable npx0" configuration option.
		Still can't install on the IDE drive.
		Got the whole thing installed on the SCSI drive with
		no error.

		Now to boot it.  My early attempts had resulted in 
		BootEasy being installed on both MBR's.  I had 
		tried (in a couple attempts out of the many) to 
		disable the IDE by removing it from the CMOS.  This 
		did force the BIOS to try booting off of the SCSI --
		but didn't resolve the main problem.

		I press F5 (other drive),
		and F2 (second primary partition on the SCSI drive)
		and type -c or "kernel.GENERIC -c" to attempt to 
		load the kernel from the hard drive (so I can tell
		it to ignore my "not quite compatible" math co).
		(I've also tried removing the math co physically
		from the system).

		What I get now is a message like:

			sd(1,a)/kernel @ 0x100000
			text = 0xe3000 -

		(the "twirl-a-gig" spins for a second and tops right there).

		I tried booting, loading the Fixit disk/shell, and copying a
		different kernel onto the root partition on sd0.  I've tried
		several variations of boots and parameters (including physically
		unplugging the IDE drive).

		Throughout every attempt at configuring this I've been
		able to boot into DOS and thence to Linux (and get my 
		mail).  All of the mail you've seen from me here has been
		coming from this system.

	I really don't want to resort to plan c.

	Plan C:

		Give up on FreeBSD for this box.  Install a new Linux
		(Debian 1.1?) and use it instead.

	So:

		What can I do to get this thing up and running
		(at least long enough to build a kernel and get
		it copied off to my DOS partition where I can use
		FBSD.EXE (???) or a LOADLIN clone to load it)?

		How do I create, get, a boot, root set that gets
		me to a shell prompt (the fixit shell seems a little
		too limited and *wayyyy toooo slooooowww*?

		What am I doing wrong?

		What are the rest of those boot parameters for
		(-vCs etc)?

		I wish I could be more detailed about the exact error messages
		I'm getting.  I kept figuring that this would be a simple 
		installation -- so I didn't take notes (like I would have if
		this wasn't my home machine).

	I should note that part of this process was before I bought
	the Portfolio and the Tseng video.  I was getting largely the 
	same symptoms with my old 512K PVGA (Paradise clone) and 
	my my ratty old Wonder Systems "super I/O" card, and with
	my alternative Promise IDE and "Startech" 2 port serial card"
	back in there.  I chose to replace the video card since
	X was unusable on the old one (just too slow and too ugly)
	-- and I do need to *occasionally* see things in Netscape and/or
	Mosaic.  I chose to replace the I/O card since their was an 
	I/O conflict that would even kill Linux if I touched COM3
	(/dev/ttyS2) on the old set up -- and the 16450 UARTS weren't 
	keeping up with my file transfers via the 28.8 modem.

	Since I've upgraded the I/O card Linux has been very stable
	(and it would go months at a time with no problem even before
	so long as I didn't touch that one com port).

	Plan D is to keep this buggy old slackware installed and 
	upgrade each component piecemeal (starting with cnews).

	(I knew that I'd get punished for daring to call myself a 
	"consultant" -- argh!).

Jim Dennis,
Starshine Technical Services



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199607190130.SAA00179>