Bug#4060: Update: 4060 - Kernel decompression failure.
Package: boot (boot1440.bin, install)
Version: 1.1 (kernel is 2.0.0)
Updated information...
Problem: On some systems, the compressed kernel image provided on the
installation floppy (boot1440.bin) is not decompressed properly when
read from floppy. One solution that seems to work for most users is to
*disable the internal cache*. If this does not work for you, try
disabling both the internal and external cache. This is done via the
bios setup.
This is not a fix, but a work-around. I have tested this on my system
(which has this behavior), and disabling the internal cache allows the
kernel decompression to proceed. It is my understanding (I have not had
a chance to test this) that the cache can be re-enabled once the system
has been installed.
BTW, my floppy controller is built into my Adaptec controller. I don't
have the system in front of me, but the controller is a 154x. (I just
can't remember that last digit... sorry).
Thanks to Bruce Perens <bruce@pixar.com> for looking into this.
Here are some notes from <debian-user@lists.debian.org>:
: From: "Christopher R. Hertel" <crh@nts.umn.edu>
: > I disabled the internal cache and--*poof*--the problem went away.
: > I will send a report to linux-kernel@vger.rutgers.edu.
:
: I have one other report matching this, and two anecdotes, all
: reproduced below. I suspect that BIOS floppy I/O is breaking when the
: cache is enabled, and you are thus feeding bad data to the disk
: uncompression code in the kernel. This could perhaps be a bug in
: syslinux, (the boot loader we are using) or a bug in the decompression
: code (which is in the kernel, but is sort of a 16-bit prelude to the
: actual kernel). You might try other ways of starting the kernel (LILO
: on a floppy rather than a hard disk, LOADLIN from DOS, etc.) to see if
: any of them fail with the cache enabled. That might tell you if it's
: the uncompression code or the loader that is at fault. About the only
: way to fix this is will be someone with the susceptible hardware to
: drive the process.
:
: Steve Gaarder:
: > I am installing Debian 1.1.1 on a generic clone with an AMD 486 on
: > an Opti-based motherboard. If I have the internal cache enabled in
: > setup, I get the error "invalid compressed format" after the
: > "uncompressing Linux" message. If I disable the cache, it boots
: > fine. It boots ok from the hard drive either way. Anyone know what
: > is going on?
:
: > From: Dan Bergman <dan.bergman@mailbox.swipnet.se>
: > I had problems like this a couple years back when I was installing
: > OS/2 on a Dell Machine (486,66mhz).. The installation would fail if
: > I didn't disabel both L1 and L2 caches.. But I ripped out the
: > TsengLab GFX card (VLB) and all troubles went away.. Ripping out gfx
: > card is often not a option but the Dell MB had a Cyrrus Logic Chip
: > on it...
:
: mike cotherman:
: > I had this problem when I used EDO RAM on a motherboard that did not
: > support EDO...just a thought
--
Christopher R. Hertel -)----- University of Minnesota
crh@nts.umn.edu Networking and Telecommunications Services
Reply to: