On Thursday 10 July 2008, Frans Pop wrote:
I've mailed the issue to lkml: http://lkml.org/lkml/2008/7/10/9.
My guess is that the verdict will be that upx will just have to be
taught about the changed kernel format and to not mangle the
compressed payload.
My guess was correct, but with
Frans Pop wrote:
On Thursday 10 July 2008, Joey Hess wrote:
Frans Pop wrote:
1) Loading root floppy seems fragile
Have you tried qemu? I never had this kind of apparently unreproducible
problem when developing the root loppy loader in qemu.
qemu works where Virtualbox fails, so it
On Saturday 05 July 2008, Joey Hess wrote:
Holger Wansing wrote:
I tested this file on my good old i486 machine with floppy drive.
Result:
Loading linux.ready
Probing EDD (edd=off to disable) ... ok[This is only
displayed very short time]
On Wednesday 09 July 2008, Frans Pop wrote:
On Saturday 05 July 2008, Joey Hess wrote:
Holger Wansing wrote:
I tested this file on my good old i486 machine with floppy drive.
Result:
Loading linux.ready
Probing EDD (edd=off to disable) ... ok
On Jul 9, 2008, at 3:05 PM, Frans Pop wrote:
1) Loading root floppy seems fragile
For some reason the root floppy often fails to load, depending on
which
udebs are included (i.e. probably depending on size of the initrd).
Only sometimes a floppy suddenly works.
This sounds a lot like a
On Wednesday 09 July 2008, you wrote:
On Jul 9, 2008, at 3:05 PM, Frans Pop wrote:
1) Loading root floppy seems fragile
For some reason the root floppy often fails to load, depending on
which
udebs are included (i.e. probably depending on size of the initrd).
Only sometimes a floppy
Frans Pop wrote:
Confirmed. Turns out this is related to the XEN patches that are included
in Debian's 2.6.25 kernel (and have been accepted upstream for 2.6.26).
I have compiled and successfully booted a 486 kernel from the Debian
source package with the following patches [1] disabled:
Frans Pop wrote:
1) Loading root floppy seems fragile
Have you tried qemu? I never had this kind of apparently unreproducible
problem when developing the root loppy loader in qemu.
4) Missing items in root floppy (load-floppy?)
First of all I am confused. The floppy-ng target currently
On Thursday 10 July 2008, Joey Hess wrote:
Frans Pop wrote:
1) Loading root floppy seems fragile
Have you tried qemu? I never had this kind of apparently unreproducible
problem when developing the root loppy loader in qemu.
qemu works where Virtualbox fails, so it looks like I'll be filing
On Thursday 10 July 2008, Joey Hess wrote:
What I don't understand is why this would only affect the floppy image.
Netboot and cdrom also use syslinux variants to load the same upxd
kernel image, and it works there.
Looking at a build log for netboot, we don't pack the kernel there. We
only
I'm trying to identify the change in the kernel that makes upx fail by
doing a kernel bisection. I've got a kernel halfway between .24 and .25
to boot now.
On Friday 04 July 2008, Frans Pop wrote:
The floppy syslinux config is still also broken: I cannot boot expert
mode or anything; I can
Holger Wansing wrote:
I tested this file on my good old i486 machine with floppy drive.
Result:
Loading linux.ready
Probing EDD (edd=off to disable) ... ok[This is only displayed
very short time]
Then the machine reboots.
This has something
I can build floppies with the 2.6.25 images, but in qemu I see this:
Booting from floppy...
Boot error
I've backtracked this to a specific change in a recent version of dosfstools,
and filed an RC bug.
However, immediatly upon loading linux, a fixed image crashes qemu
spectacularly:
qemu:
Hello,
On Fri, 4 Jul 2008 14:06:04 -0400 Joey Hess [EMAIL PROTECTED] wrote:
I don't know whether to file a qemu or a kernel bug here. Would someone
like to test this image in another emulator? (Or real floppy drive if there
are any actually left).
http://kitenet.net/~joey/tmp/boot.img
I
On Friday 04 July 2008, Joey Hess wrote:
I can build floppies with the 2.6.25 images, but in qemu I see this:
Booting from floppy...
Boot error
I got this same error today. Great you traced it.
I don't know whether to file a qemu or a kernel bug here. Would someone
like to test this image
15 matches
Mail list logo
