Am Mittwoch, den 10.12.2008, 16:34 -0600 schrieb ext Harry Putnam: > Dirk Heinrichs <[EMAIL PROTECTED]> writes: > > > Am Mittwoch, 10. Dezember 2008 19:53:10 schrieb Harry Putnam: > >> I happen to be stuck with a machine using an initramfs. Every of attempt > >> of mine to build a new kernel not using an initram (Or even with initram) > >> has failed with the boot routine failing to recognize my /dev/sdb3 as root. > > > > What is the error message you get? > > /dev/sdb3 is not a valid device
Hmm, can't find that message in the kernel sources... Would be good to
have the exact message you get (I know they're somtimes hard to remember
w/o copy'n'paste).
> [...] Thanks
>
> > Why do you think you need an initramfs at all? It's usually only
> > needed in very rare cases where either a driver for storage hardware
> > is needed which is not part of the kernel itself or some user space
> > things need to be done to even setup the root device before the
> > system can be booted from it (i.e. when it's encrypted).
>
> I don't want it. I've never felt a need for initramfs before.
> However with this vmware appliance. I am just keeping the original
> setup because any of my attempts to compile a newer kernel and ditch
> the initramfs have failed in the way I mentioned.
>
> Apparently some basic driver or such had been missing in all my
> attempts. even running `make' using the same `.config' that works, has
> resulted in the same failure.... the boot routine sees the root
> /dev/sdb3 and an invalid device.
>
> If I boot from the original kernel and initrd... it works find.
> Same /dev/sdb3 rocks and rolls right along.
>
> I'm attempting piece by piece to track down what it is but I'll tell
> you .. it is a flat out pain in the butt.... has been extremely
> time consuming and aggravating.... I've compiled a different kernel at
> least 25 times now... I started out with the newewst and workied back
> a few versions.... then went to the exact same version 2.6.24-r8 and
> even there my compiled bzimage still failed.
>
> The original appears to have been compiled with genkernel so I tried
> that. Put the running config:
> zcat /proc/config.gz > linux-2.6.24-r/.config
>
> into the exact same sources as the origina kernel and ran genkernel on
> it. That has also failed to produce a bootable kernel.
Did you try to examine the contents of that initrd with the one you
produce. Maybe somthing is missing there. Could you post the .config
here together with an "ls -Rl" output of the initrd contents. I suspect
the drivers mentioned by Alan are compiled as modules and put into that
initrd instead of just compiling them into the kernel. Would like to
verify that.
> > However, to get a better understanding about initramfs and how it
> > works, you should read Documentation/early-userspace/README in the
> > kernel source tree.
>
> Thanks, that looks like a far more complete write up, than what I've
> seen so far. I haven't finished it yet... it will take some study.
>
> I'm starting to wonder now if the vmappliance creator has put
> something into the initrd that isn't the standard genkernel assembly.
> Some driver that makes the boot work.
Yes, that may be the case.
> However during boot, the kernel I've built appears to fail after the
> initrd has already passed the torch on to the kernel.
Usually it's vice versa. The kernel uses the initrd as a temporary root
fs. This one sets up the real rootfs (by loading the needed driver
modules, unlocking encrypted devices or similar stuff). Finally, it
calls pivot_root to switch root devices and runs /sbin/init from the
real root device.
> And in the case of the many attempted builds and booting with no
> initrd the message is worded a little different but says the same
> thing. Apparently the initrd has some different error language.
That's because without initrd the message comes from the kernel and with
initrd it may come from the linuxrc script inside it, which is usually
handcrafted for a special purpose.
> I'll extract `exact' messages if you are willing to spend some time on
> this. But I suspect we should begin another thread with a subject
> that makes it clear this is about building a kernel for use inside a
> gentoo vmware appliance running on windows.
Sure, we can give it a try.
Bye...
Dirk
--
Dirk Heinrichs | Tel: +49 (0)162 234 3408
Configuration Manager | Fax: +49 (0)211 47068 111
Capgemini Deutschland | Mail: [EMAIL PROTECTED]
Wanheimerstraße 68 | Web: http://www.capgemini.com
D-40468 Düsseldorf | ICQ#: 110037733
GPG Public Key C2E467BB | Keyserver: www.keyserver.net
signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil
