The frequent reason why your kernel can't boot is because of a improper created initrd image. Here is a small description of what you can do if you encounter a similar problem.
Contents[hide] |
[edit]What is initrd image
Your boot loader usually supports
initrd instruction. For example, in GRUB:OpenVZ (2.6.8-022stab077)
root (hd0,0)
kernel /vmlinuz-2.6.8-022stab077 ro root=LABEL=/ console=tty0
initrd /initrd-2.6.8-022stab077.img
GRUB loads initrd-2.6.8-022stab077.img file at a certain address in memory. When kernel boots, it checks for initrd image, and if it exists starts init script that resides on this image. init script is usually written in nash (a sort of bash-like shell, just smaller). When init script on initrd image is finished, kernel usually calls standard System V init process (/sbin/init, etc.)
[edit]Why initrd image is necessary
Suppose your root partion resides on some SCSI device and driver for this SCSI devices is compiled as a kernel module. Of course this module is required at boot time to have access to the root partion — but it is not in the kernel. Thus the need for an initrd image.
Additionally after udev subsystem become common, somebody has to start udev to create device nodes. This is initrd's duty too.
[edit]Typical problem
Consider a real problem. After booting the kernel we get the following:
... Creating root device mkrootdev: label / not found Mounting root filesystem mount: error 2 mounting ext3 mount: error 2 mounting none Switching to new root switchroot: mount failed: 22 umount /initrd/dev failed: 2 Kernel panic - not sysncing: Attempted to kill init!
This can appear if there is no module loaded for device, where root partion resides. To solve the problem, extract the initrd image.
[edit]Extracting initrd image
Initrd image is just cpio-gzip archive. So to extract it:
$ mkdir initrd $ cd initrd $ gzip -dc /boot/initrd-2.6.16-026test014.4-smp.cpio | cpio -id $ ls -1 bin dev etc init initrd-2.6.16-026test014.4-smp.cpio lib loopfs proc sbin sys sysroot
[edit]Analyzing init script
$ cat init #!/bin/nash mount -t proc /proc /proc setquiet echo Mounted /proc filesystem echo Mounting sysfs mount -t sysfs none /sys echo Creating /dev mount -o mode=0755 -t tmpfs none /dev mknod /dev/console c 5 1 mknod /dev/null c 1 3 mknod /dev/zero c 1 5 mkdir /dev/pts mkdir /dev/shm echo Starting udev /sbin/udevstart echo -n "/sbin/hotplug" > /proc/sys/kernel/hotplug echo "Loading mptbase.ko module" insmod /lib/mptbase.ko echo "Loading mptscsih.ko module" insmod /lib/mptscsih.ko /sbin/udevstart echo Creating root device mkrootdev /dev/root umount /sys echo Mounting root filesystem mount -o defaults --ro -t ext3 /dev/root /sysroot mount -t tmpfs --bind /dev /sysroot/dev echo Switching to new root switchroot /sysroot umount /initrd/dev
We can see that init tries to load modules
mptbase.ko and mptscsih.ko. Check for presense of these modules on initrd image:$ ls -1 ./lib/ mptbase.ko mptscsih.ko
So they are here... But on the node in question there is a device supported by driver in another module:
mptspi.ko! After adding it to the image and into init script everything should work.[edit]Creating initrd
We just have to cpio and gzip directory cpio:
$ find ./ | cpio -H newc -o > /boot/new-initrd.cpio 1354 blocks $ cd /boot $ gzip new-initrd.cpio $ mv new-initrd.cpio.gz new-initrd.img
Next, try to boot your kernel with newly created initrd image.
[edit]Who create initrd by default?
Usually there is an
mkdinitrd package installed, that allows to create initrd image. You can use this program, it has a lot of options. OpenVZ kernel RPM-package (and “make install” target too) uses this program to create an initial (default) initrd image.
