Strange Reboot

raven.sorrow · 07-16-2009, 03:05 PM

OK, here is the situation. I've built endless kernels for a few different systems, most notably that being Gentoo, prior to that it was CORE, LFS and Slackware kernels (all using vanilla sources).

The Stats:

Code:

This system is a P4 1.6Ghz box w/ 512MB of PC800 (RAMBUS) RAM. 3 40GB HD's running in a software LVM.

The Scenario:

Code:

"/", /boot and swap are not part of the LVM. /boot is ext2, "/" is XFS (as is the rest of the LVM)

Code:

(chroot) livecd linux # pwd
/usr/src/linux
(chroot) livecd linux # grep -i etx2 .config
(chroot) livecd linux # grep -i ext2 .config
CONFIG_EXT2_FS=y
CONFIG_EXT2_FS_XATTR=y
CONFIG_EXT2_FS_POSIX_ACL=y
CONFIG_EXT2_FS_SECURITY=y
CONFIG_EXT2_FS_XIP=y
(chroot) livecd linux #
(chroot) livecd linux # grep -i xfs .config
CONFIG_XFS_FS=y
# CONFIG_XFS_QUOTA is not set
CONFIG_XFS_POSIX_ACL=y
CONFIG_XFS_RT=y
# CONFIG_XFS_DEBUG is not set
# CONFIG_VXFS_FS is not set
(chroot) livecd linux #
(chroot) livecd linux # df -h
Filesystem            Size  Used Avail Use% Mounted on
tmpfs                 976M  285M  692M  30% /
/dev/hdc              976M  285M  692M  30% /mnt/cdrom
df: `/mnt/livecd': No such file or directory
sysfs                 976M  285M  692M  30% /sys
udev                   10M  220K  9.8M   3% /dev
devpts                 10M  220K  9.8M   3% /dev/pts
/dev/hda3             976M  285M  692M  30% /
/dev/hda1             190M   23M  158M  13% /boot
/dev/mapper/vg-ebuild
                       10G  561M  9.5G   6% /ebuild
/dev/mapper/vg-home    30G  4.2M   30G   1% /home
/dev/mapper/vg-opt    5.0G  4.2M  5.0G   1% /opt
/dev/mapper/vg-usr     20G 1006M   20G   5% /usr
/dev/mapper/vg-local  5.0G  4.3M  5.0G   1% /usr/local
/dev/mapper/vg-portage
                      2.0G  303M  1.8G  15% /usr/portage
/dev/mapper/vg-var    508M   46M  462M  10% /var
/dev/mapper/vg-log    5.0G   35M  5.0G   1% /var/log
/dev/mapper/vg-tmp    508M   23M  485M   5% /var/tmp
udev                   10M  220K  9.8M   3% /dev
shm                   252M     0  252M   0% /dev/shm
shm                   252M     0  252M   0% /dev/shm
(chroot) livecd linux #

The Problem:

Code:

Grub loads and I can select a kernel just fine. The problem arises when the kernel is actually executed by grub. The box will inexplicably reboot, every time, every kernel.

My most recent kernel config file (for which the above snippets were taken from) can be found here.

If anyone has suggestions as to why this happens, by all means, please let me know ....

jhwilliams · 07-16-2009, 03:16 PM

no error message at reboot?

is your init.d/rc scripts working good?

Specify a path to the init on the kernel line in grub.

Also, wtf?:

Code:

Filesystem            Size  Used Avail Use% Mounted on
tmpfs                 976M  285M  692M  30% /
...
/dev/hda3             976M  285M  692M  30% /

raven.sorrow · 07-16-2009, 03:54 PM

Quote:

Originally Posted by jhwilliams

no error message at reboot?

None, as soon as I hit enter, the screen goes blank (assuming for a refresh from GRUB to the kernel's execution. But instead of seeing kernel output, I see my video cards BIOS.

Quote:

Originally Posted by jhwilliams

is your init.d/rc scripts working good?

System is getting far enough for the INIT scripts to take over.

Quote:

Originally Posted by jhwilliams

Specify a path to the init on the kernel line in grub.

init="" isn't a mandatory option, in this case it isn't needed as there isn't one. I'm of an older school of thought were I prefer to build in as much as I can into the kernel to prevent issues. In the past this hasn't caused any issues.

Also, wtf?:

Code:

Filesystem            Size  Used Avail Use% Mounted on
tmpfs                 976M  285M  692M  30% /
...
/dev/hda3             976M  285M  692M  30% /

[/QUOTE]

wtf indeed, I should have trimmed that. This is a Gentoo Chroot, /etc/mtab is present inside the chroot so it's picking up the livecd's mount points. I don't know why they are the same size though ....

jhwilliams · 07-16-2009, 04:20 PM

Quote:

Originally Posted by raven.sorrow

init="" isn't a mandatory option, in this case it isn't needed as there isn't one. I'm of an older school of thought were I prefer to build in as much as I can into the kernel to prevent issues. In the past this hasn't caused any issues.

Well, that's cool -- I'm not in your school of thought, but the school is integrated and accepted into the community.

Code:

920		init=		[KNL]
921				Format: <full_path>
922				Run specified binary instead of /sbin/init as init
923				process.
	...
929		initrd=		[BOOT] Specify the location of the initial ramdisk

I meant legit init ... but it sounds like you're getting there, at least. I was thinking the system was just passing through to rc6 perhaps, or that you didn't have init going atall

Try changing to just boot into runlevel 1 see if you get that far?

raven.sorrow · 07-16-2009, 06:23 PM

Quote:

Originally Posted by jhwilliams

I meant legit init ... but it sounds like you're getting there, at least. I was thinking the system was just passing through to rc6 perhaps, or that you didn't have init going atall

Try changing to just boot into runlevel 1 see if you get that far?

Well I'm not getting that far, as soon as grub passes control from the MBR to the kernel, that's when it reboots.

POST
Grub Menu
Select Kernel
Uncontrolled Reboot (as I didn't phsyically hard/soft-reboot the machine)

I did try and pass init=/bin/bb (BusyBox) using each kernel I built. Each standard kernel line and BB line all do the same thing. The reboot happened regardless of what INIT was passed or if one was even passed at all. So this leads me to think that it's the kernel itself, not the handoff from the kernel to the system RC scripts.

raven.sorrow · 07-17-2009, 01:33 PM

Ironically ... I sent and setup a similar kernel (from scratch) on a friends P4 that is not all that dissimilar then mine and it did the exact same thing. My guess is that I am selecting something that doesn't like being in the kernel .... and that's what's causing the reboots ... still at a loss though ....

jhwilliams · 07-17-2009, 05:27 PM

--[EDIT: woops, posted an unmindful response... be back later...]

raven.sorrow · 07-19-2009, 11:03 PM

Anyone else have any idea's?