[Solved] Unable find or mount root partition

whaler · 2009-10-02 13:07:01

I just installed an ASUS P6T Deluxe V2 motherboard with an Intel Core i7 cpu and Arch refuses to boot the default kernel, whereas the fallback kernel boots with no complaints.

With the default kernel the boot process stops with a message about being unable to locate the "Root device". I had this set by UUID, but changing to /dev/hda2 doesn't help (I use a separate boot partition on /dev/hda1 (sda1 in Arch parlance). Whatever can be stopping the default, but not the fallback, kernel from seeing and mounting the root partition?

When I reboot, I get a msg about "date in the future". It appears that BIOS is not updating the time. I have to set the time anew in BIOS before every cold or hot boot. I will put in a new battery and see if that helps.

Last edited by whaler (2009-10-08 06:23:28)

arkham · 2009-10-02 13:14:45

Change your uuid references to /dev/sdXy references in /boot/grub/menu.lst and /etc/fstab

whaler · 2009-10-02 14:05:10

arkham wrote:

Change your uuid references to /dev/sdXy references in /boot/grub/menu.lst and /etc/fstab

I will try that, but why is the fallback kernel booting and the default kernel not booting - with the same settings in GRUB (and of course fstab)?

whaler · 2009-10-02 16:22:55

Changing battery on the motherboard did not fix the time problem.

Arch complains that the superblock time is two hours in the future. When I go back into BIOS, it shows a time *4 hours* in the past...!

Correcting the time in BIOS allows the system to boot, but only with the fallback kernel.

lilsirecho · 2009-10-02 17:27:39

Perhaps you need to change /etc/rc.conf to "localtime" and reboot.

whaler · 2009-10-02 18:21:02

lilsirecho wrote:

Perhaps you need to change /etc/rc.conf to "localtime" and reboot.

Interesting idea. So I changed from 'utc' to 'localtime', and then BIOS - for the first time since I installed this m/b - showed a reasonably correct time. However, when booting resumed after I left BIOS, I was told the last boot time was 16 minutes in the future. OK, so I decided to wait 16 minutes and tried to boot again. This time there was no complaint and the bootup went fine.

I have still to try a cold boot. I'll do that now.

-----

OK - a cold boot went fine. Thanks for the useful tip! I thought I'd be modern and changed to 'utc' a while back...

Now for the default kernel - why can it not detect and mount the root partition??

Last edited by whaler (2009-10-02 18:29:54)

whaler · 2009-10-02 18:56:21

Changing from UUID to /dev/sdXy in GRUB and fstab does not noticably impress the default kernel - it adamantly refuses to boot.

The fallback kernel, otoh, happily accepts both.

thn81 · 2009-10-02 19:12:17

Looks like the ramdisk image for the default kernel is missing a SATA/filesystem driver needed for booting. AFAIK the fallback kernel contains all available drivers, so it boots normally. Try reinstalling the kernel (or rebuild the ramdisk with mkinitcpio). If that doesn't help, look at the output of lsmod and lspci and find out which drivers are required at boot time.

whaler · 2009-10-02 20:43:29

thn81 wrote:

Looks like the ramdisk image for the default kernel is missing a SATA/filesystem driver needed for booting. AFAIK the fallback kernel contains all available drivers, so it boots normally. Try reinstalling the kernel (or rebuild the ramdisk with mkinitcpio). If that doesn't help, look at the output of lsmod and lspci and find out which drivers are required at boot time.

I'm afraid I'm out of my depth trying this...

Can I just force a reinstall of the kernel from the repo(s)? Some handholding would be much appreciated

Edit: Running 'lspci -v' I see this for the VGA controller:

"02:00.0 VGA compatible controller: ATI Technologies Inc RV770 [Radeon HD 4850] (prog-if 00 [VGA controller])
Subsystem: Micro-Star International Co., Ltd. Device 1530
Flags: bus master, fast devsel, latency 0, IRQ 24
Memory at d0000000 (64-bit, prefetchable) [size=256M]
Memory at fbbe0000 (64-bit, non-prefetchable) [size=64K]
I/O ports at b000 [size=256]
Expansion ROM at fbbc0000 [disabled] [size=128K]
Capabilities: <access denied>"

Should there not be a driver and module here, as for every other device?

'lsmod' lists a 'radeon' module, but I still have to manually set the screen size.

Last edited by whaler (2009-10-02 21:24:31)

thn81 · 2009-10-03 12:22:39

/sbin/mkinitcpio -p kernel26

This will recreate the boot ramdisks.

Btw, did you change anything in /etc/mkinitcpio.conf or are you using a custom kernel (not the default kernel26 package)?

whaler · 2009-10-03 22:50:10

thn81 wrote:

/sbin/mkinitcpio -p kernel26
This will recreate the boot ramdisks.
Btw, did you change anything in /etc/mkinitcpio.conf or are you using a custom kernel (not the default kernel26 package)?

Thank you, thn!

I am using the default Arch kernel26 package and I have not changed anything in mkinitcpio.conf. I compared the outputs of lsmod and lspci and found only 5 modules not loaded by the fallback kernel, but none of them seem necessary for booting. Trouble is I can't tell which modules the default kernel would need to be/is not loading...

Meanwhile I seem to have locked myself out from Arch altogether. I used Parted Magic to correct a faulty partition label on a storage disk, i.e. a non-OS disk, something which led to a reordering of the disk name assignements. /dev/sdd became /dev/sda. Given that IDE/PATA is supposed to have order priority over SATA, I assume this is now the correct disk order, but Arch now refuses to boot even with the fallback kernel. I have checked that fstab and GRUB have the correct settings, and Parted Magic sees all partitions on the system and can copy to-from them, but Arch insists that the filesystem on at least one partition is beyond repair (hard to know which, since I have been unable to hit Scroll Lock at the right moment). I suspect Arch is confused and is checking a wrong partition, and I am unable to set it straight

Parted Magic still sees /sda as /sdd, as opposed to 'blkid' and 'fdisk -l'...

I think I will try to reinstall Arch, something I should have thought of from the beginning with a new motherboard. I am not impressed with ASUS, given that it cannot handle UTC and that BIOS goes back to default settings whenever I have used the Parted Magic live DVD. None of this happened with my former Gigabyte board.

Thanks to all of you for the help!

whaler · 2009-10-04 04:04:19

I have to tell you - this is getting funnier...

I just reinstalled Arch. I have three disks - two SATA of 1,5TB and one IDE of 250GB; sda, sdb and sdc, respectively, according to Archs installation prog. I have a boot partition on /sdc1, which GRUB sees as hd(0,0). So I configure GRUB and starts my new OS.

Everything goes well - GRUB finds Arch on /dev/sda1 and Arch starts to boot. Then, all of a sudden, it says that /dev/sda1 does not exist. Arch is actually sitting on that partition and telling me the partition does not exist...

This is *exactly* the same situation as with my 'old' Arch installation on /dev/sdc2. And, as with the old installation, the fallback kernel boots without a hitch.

Is there an easy way to compare which modules the two kernels will load?

whaler · 2009-10-04 07:00:40

After upgrading to the latest kernel, 2.6.30-ARCH#1, the new installation would no longer boot, saying:

quote
Waiting 10 seconds for device /dev/sda1 ...
:: Initramfs Completed - control passing to kinit
IP-Config: no devices to configure
Waiting 0 s before mounting root device ...
kjournald starting. Commit interval 5 seconds
EXT3-fs: mounted filesystem with writeback data mode.
kinit: Mounted root (ext3 filesystem) readonly.
kinit: init not found!
Kernel panic - not syncing: Attempted to kill init!
Pid: 1, comm: kinit Not tainted 2.6.20-ARCH#1
Call Trace: (etc., etc.)
unquote

The key word here is "EXT3-fs", given that the filesystem on the root partition (/dev/sda1) is Reiserfs (3). To make sure that I wasn't confused myself, I tried to mount the partition with "-t ext3" from Parted Magic. No go. With "-t reiserfs" - no problem.

Looks to me as if the latest Arch kernel is confusing EXT3 with Reiserfs. I am using EXT3 on my boot partition only; Reiserfs for root partitions, and JFS for /home and data partitions.

Not sure what to do now, except call it a day and a night, turn off a useless system and turn in.

whaler · 2009-10-08 06:22:48

I finally thought of running a memtest, which showed RAM errors.

I am considering this issue solved.

Arch Linux

#1 2009-10-02 13:07:01

[Solved] Unable find or mount root partition

#2 2009-10-02 13:14:45

Re: [Solved] Unable find or mount root partition

#3 2009-10-02 14:05:10

Re: [Solved] Unable find or mount root partition

#4 2009-10-02 16:22:55

Re: [Solved] Unable find or mount root partition

#5 2009-10-02 17:27:39

Re: [Solved] Unable find or mount root partition

#6 2009-10-02 18:21:02

Re: [Solved] Unable find or mount root partition

#7 2009-10-02 18:56:21

Re: [Solved] Unable find or mount root partition

#8 2009-10-02 19:12:17

Re: [Solved] Unable find or mount root partition

#9 2009-10-02 20:43:29

Re: [Solved] Unable find or mount root partition

#10 2009-10-03 12:22:39

Re: [Solved] Unable find or mount root partition

#11 2009-10-03 22:50:10

Re: [Solved] Unable find or mount root partition

#12 2009-10-04 04:04:19

Re: [Solved] Unable find or mount root partition

#13 2009-10-04 07:00:40

Re: [Solved] Unable find or mount root partition

#14 2009-10-08 06:22:48

Re: [Solved] Unable find or mount root partition

Board footer