Re: problem compiling kernel

Bernard wrote:


2.6.30.4 does compile all right, so does 2.6.26, but 2.6.20 does not.

you find out why in the archives

Problem is that I still can't boot those I compiled, i.e. 2.6.26. because
the initrd.img is buggy. I did find something, still it is not enough to
get the process to work. In my

/etc/mkinitrd/mkinitrd.conf file, I found a modified line :

# Command to generate the initrd image.
# MKIMAGE='mkcramfs %s %s > /dev/null' this has been changed august 19,
# 2009

MKIMAGE='genromfs -d %s -f %s'

The change date has not been written by me, so this must be a conf file
that came with a recent package upgrade that I did. I tried uncommenting
the old line, commenting the new one instead. MKIMAGE='mkcramfs... became
active. What gave me this idea, is that in those error messages that I
could see at crash, it was matter of cramfs.

Well that change made mkinitrd to produce smaller images. I tried
installing them in the grub boot menu, and then, now, the boot crashes do
not happen at the same time as before... but it still crashes !

I could do nothing else than catch photos of my screen, since no log file
are recorded in such cases.

http://www.teaser.fr/~bdebreil/bootcrash1.jpg

so this is the old error, and you don't need a fix for it


and

http://www.teaser.fr/~bdebreil/bootcrash2.jpg

will show you what I got

The first crash screen is not very informative :
"
could not load '/lib/modules... no such files"

I don't think so it's as informative as it should be. It can not mount sdb2
(is it your root?)


(these files exist, but at this point in time it is not in the /boot
partition, therefore not mounted as yet). This crash came from a kernel
which I had configure to have RAID inside, not as modules. While watching
the boot logs of my working kernel, I could see that RAID was as modules.
So, I recompiled a new kernel with modules for RAID, and then boot went a
little bit further, as can be seen in the screen picture at crash :

'raid1 set md1 active with 2 out of 2 mirrors

mdadm : /dev/md1 has been started with 2 drives'

but then :

'mkdir cannot create directory '/devfs/vg00' : read only filesystem
..................
....................
failure to communicate to kernel device mapper driver
incompatible libdevmapper 1.01.00-ioctl (2005-01-17) (compat) and kernel
driver'


The problem is as far as I remember that devfs was given up ... was it
something that worked with hotplug ... I really don't remember right now,
but there was a change affecting devmapper. I think you have to read about
it, perhaps replace it and recreate initrd.

Compiling md in the kernel is the right approach to boot from raided root
without initrd. You can try this just skipping (deleteing the line in grub
temporary)

I think that this last quoted line does most explain that the tools I am
using are not appropriate.

I have good grounds to think that the problems are in my initrd.img
file... but there may also be something wrong in the compiled kernel
image.

try without initrd (with custom kernel, you can put everything you need
inside it (i.e filesystem support ide/ata etc) you then can access your
root partition and the boot process will continue from there. The initrd is
only needed to load drivers which helps you do the above.
Because you are using lvm, if not using initrd you need to compile also lvm
inside the kernel.


Could someone please tell me what tool packages to purge and what to
install instead so that I can recompile a 2.6.26 or 2.6.30 kernel that
will boot on my Debian 3.1 system with raid 1 ?

There are good howtos for upgrading from sarg -> etch and etch -> lenny.
You definitely better use udev ... devmapper is not needed anymore as far as
I know.

I did it last year ... and yes there were some troubles with the initrds ...

I could send you my scripts for building your own initrd ( I have used them
to build initrd for crypted root - before it started working in debian),
though I've already posted a 5step howto fix broken boot initrd - try
init=/bin/sh option ;-) and fix the boot by hand - you'll see what you are
missing

You could just copy over a working image and initrd (from some live cd/dvd)
edit grub and reboot - this should work.

regards


--
To UNSUBSCRIBE, email to debian-user-REQUEST@xxxxxxxxxxxxxxxx
with a subject of "unsubscribe". Trouble? Contact listmaster@xxxxxxxxxxxxxxxx

Relevant Pages

  • Re: booting from ISO image on HD?
    ... bootable ISO image should have exactly the same performances as booting ... That is, kernel image, modules, application programs and so ... >> through customized initrd files for each one. ... you can boot the kernel of your ...
    (comp.os.linux.misc)
  • Re: booting from ISO image on HD?
    ... I'm not going to write the pivot root script, ... > through customized initrd files for each one. ... I have to insmod the loop kernel module(e.g. ... The idea is that either you build what you need to boot into your ...
    (comp.os.linux.misc)
  • Re: booting from ISO image on HD?
    ... >> through customized initrd files for each one. ... > between the loopback device and the underlying file (the iso image). ... I have to insmod the loop kernel module(e.g. ... how can you boot it from an initrd??? ...
    (comp.os.linux.misc)
  • Re: booting from ISO image on HD?
    ... I don't think you can predict every concievably thing a distro initrd can do. ... the only purpose of the initrd is to 1) find out what kernel modules are ... there is a hard disk where that iso image resides. ... The normal boot mechanism starts with the bios loading a boot loader ...
    (comp.os.linux.misc)
  • Re: raid 5 on boot? root?
    ... > Is it possible to have both boot and root on RAID 5? ... > about if boot is on a normal drive and root is on RAID 5? ... > Debian kernel and the Grub boot loader. ... It is possible to have both root and boot on a raid partition. ...
    (Debian-User)