I bet I'm not the 1st one running into this problem.
Trying to upgrade my kernel on my well running amd64 box from 2.6.10 to
2.6.15* just fails. Booting looks fine but running sysinit shows obscure
errors like "special device /dev/hda2 does not exist" (but /dev/hda1 is
already mounted as boot partition), claims about not being able to access
the hardware clock and so on. I've checked any option in the kernel but
obviously missed something. The kernel is mostly autmatigalley generated by
make oldconfig from 2.6.10.

The old kernel runs well on exactly the same system, the system updated to
the newest ebuilds. All my other x86 boxes runs also well.

Any help?

TiA,
Jan

Jan Schubert wrote:
> I bet I'm not the 1st one running into this problem.
> Trying to upgrade my kernel on my well running amd64 box from 2.6.10 to
> 2.6.15* just fails. Booting looks fine but running sysinit shows obscure
> errors like "special device /dev/hda2 does not exist" (but /dev/hda1 is
> already mounted as boot partition), claims about not being able to access
> the hardware clock and so on. I've checked any option in the kernel but
> obviously missed something. The kernel is mostly autmatigalley generated by
> make oldconfig from 2.6.10.

Oe thing that comes inmind is that you are using devfs in your 2.6.10, since
2.6.13 kernel don't any more support it and require that you use udev.

Some experimental udev versions have been broken too, if you happen to use
unstable, I do recommend you go back on stable for udev.


//Aho

On Saturday 28 January 2006 11:59, Jan Schubert stood up and spoke the
following words to the masses in /alt.os.linux.gentoo...:/

> I bet I'm not the 1st one running into this problem.
> Trying to upgrade my kernel on my well running amd64 box from 2.6.10
> to 2.6.15* just fails. Booting looks fine but running sysinit shows
> obscure errors like "special device /dev/hda2 does not exist" (but
> /dev/hda1 is already mounted as boot partition), claims about not
> being able to access the hardware clock and so on. I've checked any
> option in the kernel but obviously missed something. The kernel is
> mostly autmatigalley generated by make oldconfig from 2.6.10.
>
> The old kernel runs well on exactly the same system, the system
> updated to the newest ebuilds. All my other x86 boxes runs also well.
>
> Any help?

If I were you, I would configure the kernel manually this time, not by
means of /oldconfig./

A lot has changed since 2.6.10 and things have been moved around in the
kernel configuration options. It's therefore quite possible that your
method of configuration has skipped something here and there.

Try a...:

make menuconfig

.... or...

make xconfig

.... or ...

make gconfig

.... and make sure that the correct options are set for your hardware
clock and so on.

I would also suggest using 2.6.15.1, which includes a number of bugfixes
versus 2.6.15.

Hope this helps... ;-)

--
With kind regards,

*Aragorn*
(Registered GNU/Linux user # 223157)

J.O. Aho wrote:
> Some experimental udev versions have been broken too, if you happen to use
> unstable, I do recommend you go back on stable for udev.

Just emerged 0.79-r1 instead of the previous 0.81-r1. Will give it a try
right now.

Thx,
Jan

--
"Lasst uns mehr Freiheit wagen!" - Angela Merkel, 30.11.2005

Aragorn wrote:

> If I were you, I would configure the kernel manually this time, not by
> means of /oldconfig./

I usually go over it afterwards. So I think I'm fine in here. But indeed,
there has been a lot of changes :-).

Thx a lot,
Jan

--
"Lasst uns mehr Freiheit wagen!" - Angela Merkel, 30.11.2005

Jan Schubert wrote:

> Just emerged 0.79-r1 instead of the previous 0.81-r1. Will give it a try
> right now.

Ah, gotya :-). Nit sure what has done the trick, but I enabled
CONFIG_HOTPLUG and set RC_DEVICE_TARBALL to yes in rc.

I'm fine now. Thx a lot,
Jan

Jan Schubert wrote:

> Jan Schubert wrote:
>
>> Just emerged 0.79-r1 instead of the previous 0.81-r1. Will give it a try
>> right now.
>
> Ah, gotya :-). Nit sure what has done the trick, but I enabled
> CONFIG_HOTPLUG and set RC_DEVICE_TARBALL to yes in rc.
>
If you don't have CONFIG_HOTPLUG enabled, udev won't work, and you won't
have any device nodes. If you have it enabled, you shouldn't need
RC_DEVICE_TARBALL.

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Greetings group,
In my case, enabling CONFIG_HOTPLUG (under General Settings) solved
the problem. I prefer not to use the device tarball mechanism, since
it seems "cleaner" to make the system work properly with udev and not
to use the device tarball.

Chris

Jan Schubert wrote:
> I bet I'm not the 1st one running into this problem.
> Trying to upgrade my kernel on my well running amd64 box from 2.6.10 to
> 2.6.15* just fails. Booting looks fine but running sysinit shows obscure
> errors like "special device /dev/hda2 does not exist" (but /dev/hda1 is
> already mounted as boot partition), claims about not being able to access
> the hardware clock and so on. I've checked any option in the kernel but
> obviously missed something. The kernel is mostly autmatigalley generated by
> make oldconfig from 2.6.10.
>
> The old kernel runs well on exactly the same system, the system updated to
> the newest ebuilds. All my other x86 boxes runs also well.
>
> Any help?
>
> TiA,
> Jan

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2 (MingW32)
Comment: GnuPT 2.7.2
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFD6uxX6ZGQ8LKA8nwRAmd8AKCrJSS7VpU+AihTkFS+Rb 2qCJmoRQCdE5h4
CCXU0VauymY0IqrZJMfEhXM=
=b0mz
-----END PGP SIGNATURE-----

Chris Head <chris2k01@hotmail.com> wrote:
>
> Greetings group,
> In my case, enabling CONFIG_HOTPLUG (under General Settings) solved
> the problem. I prefer not to use the device tarball mechanism, since
> it seems "cleaner" to make the system work properly with udev and not
> to use the device tarball.

If you want your system "cleaner", you'd be surprised at how much extra
crud is polled in by hotplug, whether needed or not. Using a device
tarball with just the devices you need /is/ the cleaner choice here,
compared to pulling in all the "might have's".

Regards,
--
*Art

On Thu, 09 Feb 2006 05:13:53 -0500, Arthur Hagen wrote:

> If you want your system "cleaner", you'd be surprised at how much extra
> crud is polled in by hotplug, whether needed or not. Using a device
> tarball with just the devices you need /is/ the cleaner choice here,
> compared to pulling in all the "might have's".

Looking at my /dev/, every device node created for me is needed. If you're
using genkernel YMMV.

--
Ben M.

To reply directly, remove all occurrences of 'remove' from the email address.