2.6.31.5 not booting

[Nido]

Hello,

After a very long gap of 6 years,I'm back working on Linux machines.. the first task I got is to compile and run 2.6.31.x.
I followed the steps mentioned in the README file in kernel source and also took referrence to some of the sites.
I am able to compile the kernel and is under /boot.
I can see files: vmlinuz-2.6.31.5, System.map-2.6.31.5,initrd-2.6.31.5.img under /boot and corresponding entries in /boot/grub/grub.conf as updated by install.sh under ../arch/x86/boot

Here is my problem:Kernel boot hangs at boot after initrd message (before decompressing kernel).
Is my problem similar to what is in the below url :https://bugzilla.redhat.com/show_bug.cgi?id=144050

I'm running RHEL AS release 4 (Nahant Update 4)2.6.9-42.ELsmp.
Please let me know how to fix it.

Thanks as always!!

[Goran Koruga]

Hello.

Do you see "BIOS EBDA" or "Linux version 2.6.x.y" line after that? Do you perhaps use "quiet" boot option?

Regards,
Goran

[Nido]

>>Do you see "BIOS EBDA" or "Linux version 2.6.x.y" line after that? Do you perhaps use "quiet" boot option?

Thanks for quicker reply.
It hangs just after initrd before decompressing kernel and don't show any message.
btw: I've loaded RHEL on VMware Machines, and checked on one of my HP and IBM Machines both shows same result :-(

-P.Naidu

[Goran Koruga]

Hi.

Then you should recheck your kernel config - those are the first things that should be printed and they occur very early in the boot.

Curious - did you forget to include support for VGA console or somesuch? Or maybe left it modularized.

Maybe you should start with RH's kernel config and tweak it when you know it works.

Regards,
Goran

[Nido]

Hi Again,
>>>Curious - did you forget to include support for VGA console or somesuch? Or maybe left it modularized.
Would vga console config affect after initrd and just before decompressing kernel ?

>>>Maybe you should start with RH's kernel config and tweak it when you know it works.
Treat me as newbie...give me some clue which settings can be checked.

Thanks, P.Naidu

[Goran Koruga]

Hi.

VGA console could affect it, because when not set (and if no other options of the same kind are enabled), display would not be available so you will not see anything.

I recommend to start with RH's kernel config, you can find it in /boot. Copy it to your kernel source dir as .config (might want to save current one), and then you can get asked about new options with:

make oldconfig

Regards,
Goran

[Nido]

Goran,
Thanks again. I'll check to run `make oldonfig` today and will get back to you.

Thanks Again, P.Naidu

[Matti_Kurkela]

The 2.6.31.5 kernel will print a message before EDD probe (in linux-2.6.31.5/arch/x86/boot/edd.c):

if (!be_quiet)
printf("Probing EDD (edd=off to disable)... ");

If you cannot see this message, your problem is probably unrelated. For example, if your hardware is a Proliant server, the console functions may have been switched to ILO2 Virtual Serial Port, so nothing will appear on the screen. (Please describe the hardware you're running.)

Please attach your kernel configuration file for analysis.

The "kernel" line(s) of your /boot/grub/grub.conf might be important too.

MK

[Nido]

Hi,

One of the machines is IBM Desktop (Intel 440bx chipset)having VMware running RHEL 4U4.
I have another laptop HP dv6000 running VMware having the same issue.

/boot/grub/grub.conf :

#boot=/dev/sda
default=1
timeout=5
splashimage=(hd0,0)/grub/splash.xpm.gz
hiddenmenu
title Red Hat Enterprise Linux AS (2.6.31.5)
root (hd0,0)
kernel /vmlinuz-2.6.31.5 ro root=/dev/VolGroup00/LogVol00 rhgb quiet
initrd /initrd-2.6.31.5.img
title Red Hat Enterprise Linux AS (2.6.9-42.ELsmp)
root (hd0,0)
kernel /vmlinuz-2.6.9-42.ELsmp ro root=/dev/VolGroup00/LogVol00 rhgb quiet
initrd /initrd-2.6.9-42.ELsmp.img
title Red Hat Enterprise Linux AS-up (2.6.9-42.EL)
root (hd0,0)
kernel /vmlinuz-2.6.9-42.EL ro root=/dev/VolGroup00/LogVol00 rhgb quiet
initrd /initrd-2.6.9-42.EL.img

Thanks for your help.
P.Naidu

[Goran Koruga]

Hi.

I asked you about quiet option initially and you do have it - remove that crap.

I hate this idiotic default in modern distributions.

Regards,
Goran

[Matti_Kurkela]

Yes. Remove the "rhgb" and "quiet" options - they offer a cool-looking start-up display when everything goes well, but when you have problems, they just hide the information that might be essential in solving the problem.

MK

[Nido]

Hello Goran/MK,

I removed rhgb and quite options while booting new kernel image manually from grub menu.
Here is the message, I'm getting when removing rhgb and quite option:

It decompresses kernel and shows:
edd=off to disable OK
booting kernel

It hangs forever while showing "booting kernel",
Also I tried changing options while make build using 'make oldconfig' but no luck yet.

Thanks, P.Naidu

[Viktor Balogh]

Be sure to include everything essential directly into your newly created kernel, do not compile these to modules! So things might here missing, e.g.: driver for the disk controller, driver for the root fs type...

[Viktor Balogh]

Did you generate the initial ramdisk? I don't know Red Hat, but in SuSE after kernel compiling I need to

# cd /boot
# mkinitrd -k vmlinuz -i initrd -M /boot/System.map

[Pnaidu]

Viktor,

>>>>Did you generate the initial ramdisk? I don't know Red Hat, but in SuSE after kernel compiling I need to .
As I know , you need not have to manually run mkinitrd for 2.6.x kernel ; but its stillvalid in 2.4.x kernel.

I tried compiling 2.6.29 kernel and got the same problem , Is it because of VMware?? There is something common which is creating same situation with different kernel images.

Your views please...

Thanks as always, P.Naidu

[Goran Koruga]

Hi.

Does the same thing happen if you use RH kernel config?

This would suggest somethig with CPU options to me.

Regards,
Goran

[Nido]

Hello Experts,
I kept the thread still alive and was working on 2.6.31 in between my other useless stuffs.

Here is the update: I've upgraded VMWARE Server to 2.0 from 1.0. and saw significant difference,I could see my system booting further miles before it crashes and dies silently.

Apart from upgrading vmware server, I have not done any changes.

Please check the beautiful attachment and let me know your view please.

As always, Thank you!!

[Matti_Kurkela]

From the kernel version 2.6.18-164.el5 I see that you've switched from RHEL 4 to RHEL 5, too. Good thinking: RHEL 5 has newer versions of system utilities, so it has a much better chance of getting the bleeding-edge 2.6.31.5 kernel running.

This is the point where the kernel would switch from the initrd temporary root-on-ramdisk to the real root filesystem.

But in your case, the real root filesystem could not be mounted. I think the messages point this out rather clearly:

> Mounting root filesystem.
> mount: could not find filesystem '/dev/root'

The rest is just the boot process failing miserably because there is no valid root filesystem available at all. When the process #1 (RedHat "nash", which would exec the real /sbin/init as soon as it becomes available) dies, the kernel notices that something is seriously amiss and displays a panic message.

There can be several reasons for this:

1.) the driver for the disk hardware that contains the root disk was not available: it was not built in the kernel, *and* it was not available as a module in initrd.

2.) the filesystem driver for the root FS was not available: it was not built in the kernel, *and* it was not available as a module in initrd.

To see what's happening, I would have to see the messages just before "Waiting for driver initialization" (the top-most line of your VMware console screenshot). If I had access to your system, I'd click on the console window, then press Shift-PageUp to see if the kernel's console backscroll feature would allow me to see any earlier messages.

If there are no messages about disk drive detection before "Waiting for driver initialization", my guess would be problem 1) from the choices I listed above.

The "make oldconfig" kernel configuration procedure isn't all-knowing: if there have been significant kernel configuration structure re-organizations (and between 2.6.18 and 2.6.31, there certainly has!), the "make oldconfig" may be unable to convert all old options to their new equivalents.

My suggestion:
Check your virtual machine properties: which virtual IDE/SCSI controller is the VM presenting to the OS?

If it's the "LSI Logic LSI53C10xx" SCSI controller, I think you'll need to enable the kernel config options CONFIG_FUSION and CONFIG_FUSION_SPI. You may have missed them because they are not among SCSI controllers: they form a device driver sub-category of their own, called "Fusion MPT device support".

For Mylex/Buslogic SCSI controller, you'll need the kernel option CONFIG_SCSI_BUSLOGIC if I'm not mistaken.

Disclaimer: I might well be wrong about these kernel options; I cannot check from a RHEL 5 VMware VM at the moment, and I'm just working with VMware documents.

MK

[Nido]

Hi,

MK: you're in my favourite list.

Screen shot says all; you're correct with all your observations, I switched to RHEL5.4 as well VMware Server 2.0 in stages.
VMware Server 2.0 allowed me to see some more messages of system boot.

Let me give a try later in the Evening, following your suggestions.
Also let me see If I can scroll up console screen to show some more messages to you.

"press Shift-PageUp" >> its in my cook-book now ; else I'll have some other options :-)

p.s. I'm doing all these to see butteFS on my plate.

Thanks once again, I'll be back with updates.

Cheers!!

[Goran Koruga]

Hello.

Perhaps you configured it to need initrd and forgot to build/include initrd image.

Regards,
Goran