New rig freezes on boot

fopetesl · 01-28-2007, 07:07 AM

Last week my mobo combo failed on boot up

(no BIOS, no Beeps) so I replaced mobo,processor & memory.
Old system: ASUS A8N_SLi Deluxe, Athlon 3500+, 2GB GEIL DDR(1)
New system: ASUS M2V, Atlon 3800+, 1GB Kingston DDR2

Both systems with Radeon PCI Express X300, 250GB SATA.

Dual boot with XP and Mandriva 2007.

XP booted OK and after installing (new) drivers works fine, (that's how I can write this).

Mandriva will not load in any mode - the ouput from 'failsafe'

Code:

Loading sd_mod.ko module
Loading jbd.ko module
Loading ext3.ko module
Mounting /proc filesystem
Mounting sysfs
Creating device files
Mounting tempfs on /dev
Creating root device
Trying to resume from /dev/sda6
Unable to access resume device (/dev/sda6)
echo: cannot open /proc/suspend2/do-resume for write: 2
Mounting root filesystem /dev/root
mount: error 6 mounting ext3 flags default
well, retrying without option flags
mount: error 6 mounting ext3
well, retrying without any flag
mount: error 6 mounting ext3
Switching to new root
ERROR opening /dev/console!!!!: 2
unmounting old /proc
unmounting old /sys
Switchroot: mount failed: 22
Initrd finished
kernel panic - not syncing: Attempted to kill init!

Then the system freezes.

I ran 'rescue' from the install DVD which is no help except
# lsparts //.. returns nothing
# fdisk -l //.. returns nothing
# fdisk -l /dev/sda //.. returns nothing
#lsparts //.. returns nothing

If I try to mount all detected hardware:
sata_via IVT8273A installs OK
IDE/USB IVT82C586 fails with "Can't find via82Cxxx in archive"

If I try to mount I get "/mnt can't find root device".

I ran a partiton manager (from DOS?) and it sees all my partitions.

Where to now?

stress_junkie · 01-28-2007, 09:06 AM

It looks like the init process cannot handle the file system type (ext3) of the root file system. Maybe the ext3 file system code is not built into the new kernel or it is not in the initrd image. You could fix this by booting a different Linux system (on live CD for example) then using chroot to make the hard disk root partition into your current environment, then using mkinitrd to add the ext3 file system module to your initrd.

Or I could be wrong.

I'm basing my comment on several boot messages in your post about not successfully mounting the root partition. This seems to be the cause of kernel panic during boot mentioned in many posts here when people are using the 2.6 kernel.

Maybe the rescue function of the installation DVD would have an option to rebuild the initrd image or to rebuild the kernel. That would probably be the easiest solution if it is available.

I am also wondering if your root partition is on a SATA drive. If so then the kernel or initrd image require SCSI device modules. If you have put your root partition on a RAID device then you need additional modules to support that to be included in the kernel or in the initrd image and the initrd image and the kernel need to be put on a partition that is not RAID. Typically this partition would be mounted at the /boot mount point after the system is running.

Just thinking out loud.

fopetesl · 01-28-2007, 09:51 AM

Quote:

Originally Posted by stress_junkie

It looks like the init process cannot handle the file system type (ext3) of the root file system. Maybe the ext3 file system code is not built into the new kernel or it is not in the initrd image. You could fix this by booting a different Linux system (on live CD for example) then using chroot to make the hard disk root partition into your current environment, then using mkinitrd to add the ext3 file system module to your initrd.

Or I could be wrong.

I'm basing my comment on several boot messages in your post about not successfully mounting the root partition. This seems to be the cause of kernel panic during boot mentioned in many posts here when people are using the 2.6 kernel.

Maybe the rescue function of the installation DVD would have an option to rebuild the initrd image or to rebuild the kernel. That would probably be the easiest solution if it is available.

I am also wondering if your root partition is on a SATA drive. If so then the kernel or initrd image require SCSI device modules. If you have put your root partition on a RAID device then you need additional modules to support that to be included in the kernel or in the initrd image and the initrd image and the kernel need to be put on a partition that is not RAID. Typically this partition would be mounted at the /boot mount point after the system is running.

Just thinking out loud.

Ummm. Except it is 2.6.17 I think and this same HDD was running fine on the trashed Mobo. The Mobo failed on cold boot 1st thing in the morning so it's unlikely that the file system has been corrupted. Possible but not probable since the FAT32 file system is OK.

stress_junkie · 01-28-2007, 10:27 AM

Those boot messages seem pretty compelling to me regarding being unable to access the ext3 file system. Since this was working with another motherboard then we know that the kernel and initrd are okay. That causes me to question the integrity of the superblock for that partition.

It's possible that the kernel doesn't have the correct chip driver for the new motherboard.

Boot a live Linux CD and try to access that partition. See what happens.

The Mandriva DVD would probably work here. I don't have a Mandriva DVD so I can't test my suggestion. See if there is an option on the Mandriva DVD to boot an existing installed system, then select that partition.

fopetesl · 01-28-2007, 10:42 AM

Quote:

Boot a live Linux CD and try to access that partition. See what happens. The Mandriva DVD would probably work here.

Is this what you mean?

Code:

I ran 'rescue' from the install DVD which is no help except .. from the 'console':
# lsparts .. returns nothing
# fdisk -l .. returns nothing
# fdisk -l /dev/sda .. returns nothing

Maybe I should try another Live CD? Any suggestions? (Don't overlook the fact that I'm a relative

)

stress_junkie · 01-28-2007, 12:39 PM

Ooops. Yes, that would be what I had in mind.

When you said that fdisk -l reports nothing that suggests that you are using a normally privileged user account. That is weird if you are using a rescue environment. Is there some command that you can issue in the rescue environment to obtain privileges? Maybe su or sudo? (Gentoo is like that so there is some precedent in reality somewhere.)

Code:

sudo /sbin/fdisk -l

Hint: if sudo asks for a password it wants the password of the normal user account, not the root password.

You may wonder why I'm writing this since I don't use Mandriva. I just don't see anyone else jumping in to help and I don't want to leave without giving some potentially helpful ideas.

fopetesl · 01-29-2007, 06:44 AM

In 'rescue' mode it never requests a password, (or user). sudo seems to be a null?

So, I downloaded the single CD iso and ran it. Switched to a terminal and logged in as root - no password requested.
Same response to fdisk -l : nothing.

I guess the ext3 file system may be corrupted but the partitions survive intact.
Bit lost as to where to go to next

stress_junkie · 01-29-2007, 01:08 PM

Does your motherboard see the disk? The fdisk -l issued as root should have at least mentioned that the disk exists whether it has partitions or not.

Get into your motherboard's CMOS and see if it can detect the hard disk.

fopetesl · 01-30-2007, 03:47 AM

Quote:

Originally Posted by stress_junkie

Does your motherboard see the disk? The fdisk -l issued as root should have at least mentioned that the disk exists whether it has partitions or not.

Get into your motherboard's CMOS and see if it can detect the hard disk.

OK. BIOS sees the 250GB SATA disk and I can boot into grub menu.lst with options for XP and three variations of Mandriva. Can chose any one and I get a response. Not favourable if I choose Mandriva!

Note that any partitioner program (from DOS at least) sees all my partitions correctly.

Also note that I am writing this from the XP partition (FAT32) of that same HDD.

======= more info ======== (11:10 GMT) =========

It seems I have a hardware issue. I tried to run the 'patch' option from the DVD and it threw up "Found VIA VT8237A SATA 2 controller" with the query "Any other hardware?".
If I select "No"I get the response

Code:

An error occurred -
An error occurred - no valid devices were found on which to create new filesystems. Please check your hardware for the cause of this problem.

I noted that on the M2V mobo there is also a JMicron JMB363 SATA/PATA controller which I had previously discounted as a problem since I can read the DVD OK. If I select "Yes" to the above the JMB363 is not an option.
However it appears this could be a problem since:
a) it is not clear just which SATA ports it accesses
-and-
b) is not supported on kernel 2.6.17-x (it seems, however, that it is on 2.6.18 )

SO. How do I patch the kernel on a drive I cannot access.

How can I upgrade Mandriva 2007 with another Distro without trashing my programs and/or data, (/home is a separate partition, TG).

Does the name Yossarian mean anything here?

crashmeister · 01-30-2007, 05:26 AM

Well - you could use knoppix and chroot into the system to upgrade the kernel.
What I don't know is if Mandriva has a new kernel out and Knoppix is probably not playing well with rpm's but you could always download a new kernel and install that in which case knoppix might not work because you'll need gcc to do that.
Might be easier to just get a minimal install CD from gentoo and use that to chroot and install a new kernel if you don't have a problem with cli.

fopetesl · 01-30-2007, 11:21 AM

Quote:

Originally Posted by crashmeister

Well - you could use knoppix and chroot into the system to upgrade the kernel.
What I don't know is if Mandriva has a new kernel out and Knoppix is probably not playing well with rpm's but you could always download a new kernel and install that in which case knoppix might not work because you'll need gcc to do that.
Might be easier to just get a minimal install CD from gentoo and use that to chroot and install a new kernel if you don't have a problem with cli.

crash, thanks for that. I downloaded FC6 CD1. Ran chroot (FC sees my SATA ext3 partitions) and source /etc/profile. OK.
But when I run ifup eth0 I get forcedeth device eth0 does not seem to be present, delaying initialization
For some reason when I ifconfig eth0 I get warning: cannot open /proc/net/dev (No such file or directory). Limited output
What's odd is that it all worked before the catastrophe and I don't have a subdir /dev

So cannot get internet access and urpmi a later kernel

crashmeister · 01-30-2007, 12:14 PM

Calm down - one thing at a time.
Did you actually chroot into your mandriva installtion?
Does the box boot mandriva now or are you getting those errors while being chrooted or what?

Whatever - if there is no /dev dir there is no eth.
If you are really chrooting into mandrake you will need to mount the /dev and /proc directory first.

Let em know whats going on there - need be pm me and I can try to write you thru the thing.

fopetesl · 01-30-2007, 01:16 PM

Not calm? Moi? However did you get that idea?

Quote:

Did you actually chroot into your mandriva installtion?

Yes. Booted with FC6 CD #1.
Then from a CLI like so

Code:

mkdir /mnt/mdv2007
mount /dev/sda4 /mnt/mdv2007
chroot /mnt/mdv2007 /bin/bash
source /etc/profile

Quote:

Does the box boot mandriva now or are you getting those errors while being chrooted or what?

Not sure quite what you mean. If I haven't updated the Mandriva kernel, how can it boot OK? The intention is, after chroot-ing to the Mandriva installation, to initialise eth0 and connect to the internet. Then update to the latest kernel on the SATA drive.

Quote:

If you are really chrooting into mandrake you will need to mount the /dev and /proc directory first.

The /dev & /proc directories have been mounted from /dev/sda4 ( '/' partition). I can 'ls' them both. It's /proc/net/dev that seems to be missing.

Am not able to pm tonight so will do so tomorrow but am not aware of your time zone. Thanks.

crashmeister · 01-30-2007, 03:40 PM

You need to do:

mount -t proc none /mnt/mdv2007/proc
mount -o bind /dev /mnt/mdv2007/dev

before chrooting - actually in between mounting mdv2007 and chrooting.
Otherwise the dev and proc information from the running environment wont be available in the chrooted environment and things like eth wont work.

fopetesl · 01-31-2007, 06:57 AM

So this morning after some head banging I have this 1st result: neither of the two distros with 2.6.19 recognise that I have a LAN adapter on board. I plugged in a bog standard PCI NIC and can now set up an internet connection so:

Code:

mkdir /mnt/mdv2007
mount /dev/sda4 /mnt/mdv2007
mount -t proc none /mnt/mdv2007/proc
mount -o bind /dev /mnt/mdv2007/dev
chroot /mnt/mdv2007 /bin/bash
source /etc/profile

So far so good

FC6 rescue mode has already initialised eth0 via DHCP from my router, viz

Code:

route -n
Destination   Gateway      Genmask     Flags Metric Ref Use Iface
192.168.1.0    0.0.0.0   255.255.255.0  U      0     0   0 eth0
  0.0.0.0    192.168.1.1    0.0.0.0     UG     0     0   0 eth0

then I attempt to set up a host

Code:

/usr/bin/host linux.org
linux.org has address 198.182.196.8
;; Warning: Message parser reports malformed message packet
;; connection timed out; no servers could be reached

Final hurdle to overcome

(I also may have to work out why Mandy 2007 did see the on-board LAN ( and XP) but FC6 and SIDUX cannot

)