Skip to contentSkip to conversation

kexec failed: invalid argument

2.6.22-rc4 S.Çağlar Onur 2 participants 4 messages Jun 14, 2007 – Jun 15, 2007

lore
Share & tools
lore
#0 lore raw

Hi;

With current Linus's git tree + current kexec-tools-testing [1] (also with kexec-tools-1.101) tree "kexec -e" call on reboot failed with following;

kexec failed: invalid argument

kexec document says its users responsibility to close running softwares, mounting disks in read only mode etc. and it suggest calling "kexec -e" just before the "reboot" call and talks about future merge with sysvinit's reboot and kexec, and when i did it, it will not starts new kernel and instead gaves that error.

But more supringuisly if i try to execute "kexec -e" directly (in init3 with rw disks and lots of running software) it will starts new kernel without a problem.

For reference im using following kexec line for loading the kernel
kexec -l /boot/kernel-2.6.22-rc4-CFS-v16 --append="root=/dev/sda3" --initrd=/boot/initramfs-2.6.22-rc4-CFS-v16

I'm not sure this is a regression or my(maybe me or kexec document misses some point) mistake cause this is my first attempt to use kexec, but i want to report the situation :)

[1]
http://git.kernel.org/?p=linux/kernel/git/horms/kexec-tools-testing.git;a=summary

Cheers

Signature · 4 lines
S.Çağlar Onur <[email protected]>
http://cekirdek.pardus.org.tr/~caglar/

Linux is like living in a teepee. No Windows, no Gates and an Apache in house!
#1 lore raw
Quoted 21 lines · S.Çağlar Onur

Hi;

With current Linus's git tree + current kexec-tools-testing [1] (also with kexec-tools-1.101) tree "kexec -e" call on reboot failed with following;

kexec failed: invalid argument

kexec document says its users responsibility to close running softwares, mounting disks in read only mode etc. and it suggest calling "kexec -e" just before the "reboot" call and talks about future merge with sysvinit's reboot and kexec, and when i did it, it will not starts new kernel and instead gaves that error.

But more supringuisly if i try to execute "kexec -e" directly (in init3 with rw disks and lots of running software) it will starts new kernel without a problem.

For reference im using following kexec line for loading the kernel kexec -l /boot/kernel-2.6.22-rc4-CFS-v16 --append="root=/dev/sda3" --initrd=/boot/initramfs-2.6.22-rc4-CFS-v16

Quoted 6 lines

I'm not sure this is a regression or my(maybe me or kexec document misses some point) mistake cause this is my first attempt to use kexec, but i want to report the situation :)

[1]
http://git.kernel.org/?p=linux/kernel/git/horms/kexec-tools-testing.git;a=summary

Sight unseen I'm guessing that you have a kexec aware distro that is doing something in the runlevel change scripts and thus unloading the kernel. What do: /sys/kernel/kexec_loaded and /sys/kernel/kexec_crash_loaded say?

kexec -e at any point should start a loaded kernel without problems, but it is a lot like pressing the reset button. Nothing is guaranteed to be shutdown cleanly. Now that journalling filesystems are the norm it is quite possible nothing bad will happen to you.

Eric

#2 lore raw

15 Haz 2007 Cum tarihinde, Eric W. Biederman şunları yazmıştı:

Quoted 3 lines

Sight unseen I'm guessing that you have a kexec aware distro that is doing something in the runlevel change scripts and thus unloading the kernel. What do: /sys/kernel/kexec_loaded and /sys/kernel/kexec_crash_loaded say?

Not sure, here is the pseudo output what our init system does;

- Stop running services
- Save random seed, sync clock
- Unmount mounted fs's (this includes proc/sysfs etc.)
- Call kexec -e
- Call reboot

Quoted 4 lines

kexec -e at any point should start a loaded kernel without problems, but it is a lot like pressing the reset button. Nothing is guaranteed to be shutdown cleanly. Now that journalling filesystems are the norm it is quite possible nothing bad will happen to you.

zangetsu ~ #
kexec -l /boot/kernel-2.6.22-rc4-CFS-v16 --append="root=/dev/sda3" --initrd=/boot/initramfs-2.6.22-rc4-CFS-v16
setup_linux_vesafb: 1024x768x16 @ c0000000 +300000
zangetsu ~ # cat /sys/kernel/kexec_loaded
1
zangetsu ~ # cat /sys/kernel/kexec_crash_loaded
0

Cheers

Signature · 4 lines
S.Çağlar Onur <[email protected]>
http://cekirdek.pardus.org.tr/~caglar/

Linux is like living in a teepee. No Windows, no Gates and an Apache in house!
#3 lore raw

15 Haz 2007 Cum tarihinde, S.Çağlar Onur şunları yazmıştı:

Quoted 6 lines

15 Haz 2007 Cum tarihinde, Eric W. Biederman şunları yazmıştı:

Quoted 4 lines

Sight unseen I'm guessing that you have a kexec aware distro that is doing something in the runlevel change scripts and thus unloading the kernel. What do: /sys/kernel/kexec_loaded and /sys/kernel/kexec_crash_loaded say?

Quoted 7 lines

Not sure, here is the pseudo output what our init system does;

- Stop running services
- Save random seed, sync clock
- Unmount mounted fs's (this includes proc/sysfs etc.)
- Call kexec -e
- Call reboot

Heh, i found a problem in our init system and after solving that issue kexec starts to works fine :), thanks for the tip!

Cheers

Signature · 4 lines
S.Çağlar Onur <[email protected]>
http://cekirdek.pardus.org.tr/~caglar/

Linux is like living in a teepee. No Windows, no Gates and an Apache in house!