On Fri, Mar 24, 2006 at 05:43:33PM +0530, Shan Kumar wrote:
>  Hi,
> 
> I want to test kexec/kdump features in my system.
> 
> 
> 
> First kernel is 2.6.15-git12-6-default
> Second (capture) kernel is 2.6.16-rc6.
> SuSe Linux 10.0.42
> machine: i386 32 bit
> 
> I followed the following steps.
> 
> 
> 
> 1) I installed the kexec-tools-1.101-16.i386.rpm.
> 
> 2) Installed lam rpm from
> http://download.fedora.redhat.com/pub/fedora/linux/core/development/i386/Fedora/RPMS/lam-7.1.1-11.i386.rpm
> due
> to which kdump gets installed in /usr/bin
>
Why do you need this..? I have never seen this anywhere.

> 3) Update the /etc/sysconfig/kdump file
> 
>     KDUMP_KERNELVER=""
> 
>     KDUMP_RUNLEVEL="1"
> 
>     KDUMP_TRANSFER=""
> 
>     KDUMP_IMMEDIATE_REBOOT="yes"
> 
>     KDUMP_SAVEDIR="/var/log/dump"
> 
>     KDUMP_KEEP_OLD_DUMPS=5
> 
> 4) I also executed /sbin/chkcofig kdump on. It set properly.
> 
> 5) Updated the .config files for both the kernels. (files are attached)
> 
> 6) Updated the /boot/grub/menu.lst file
> 
> 
> 
>    title SUSE LINUX 10.0.42
>     root (hd0,5)
>     kernel /boot/vmlinuz root=/dev/sda6 vga=0x31a selinux=0 splash=silent
> showopts  [EMAIL PROTECTED]
>     initrd /boot/initrd
> 
> 7) make the first kernel and reboot. (/sbin/reboot [EMAIL PROTECTED] does
> not work)
> 
> 8) Executed the command
> 
>      kexec -p /usr/src/linux-2.6.16-rc6/vmlinux  --args-linux
> --elf32-core-headers
> 
> 
>      --append="root=/dev/sda6 init 1 irqpoll maxcpus=1"
> 
>  I got the result
> 
>  Cannot load /usr/src/linux-2.6.16-rc6/vmlinux
> 
> 
> 
> 
> 
>    Then I executed the command
> 
>    kexec -p /usr/src/linux-2.6.16-rc6/vmlinux
> 
> 
> 
>    I got the result
> 
> 
> 
> kexec_load failed: Cannot assign requested address
> entry       = 0x1498 flags = 1
> nr_segments = 5
> segment[0].buf   = 0x8067c88
> segment[0].bufsz = 70c0
> segment[0].mem   = 0x1000
> segment[0].memsz = 9000
> segment[1].buf   = 0x806eea0
> segment[1].bufsz = 40
> segment[1].mem   = 0xa000
> segment[1].memsz = 1000
> segment[2].buf   = 0xb763c008
> segment[2].bufsz = 1b23e8
> segment[2].mem   = 0x1000000
> segment[2].memsz = 1b3000
> segment[3].buf   = 0xb77ef008
> segment[3].bufsz = 4dc24
> segment[3].mem   = 0x11b3000
> segment[3].memsz = 4e000
> segment[4].buf   = 0xb783d008
> segment[4].bufsz = 2f086
> segment[4].mem   = 0x1202000
> segment[4].memsz = 80000
> 
> 
> 
> 9) I also executed the previous two commands with "-l" option (removed the
> -p option). But I am not sure whether the second kernel is loaded or not. I
> patched the related files for getting this result from this link.
> 
> http://article.gmane.org/gmane.comp.boot-loaders.fastboot.general/208
> 
> 
> 
> Then I make the 1st kernel and executed the previous commands. I didn't get
> any kexec_loaded file in /proc/sys/kernel dir.
> 
> 
> 
> 10) To enable the magic keys, I executed
> 
>      echo 1 > /proc/sys/kernel/sysrq     and 1 gets echoed there.
> 
>      echo c > /proc/sysrq-trigger is not working. Nothing gets echoed there.
> Also the SysRq key is not getting enabled.
> 

How did you find that SysRq keys support is not getting 
enabled?

>     How to make the SysRq key enabled and create a panic?

"echo 1 > /proc/sys/kernel/sysrq" should have enabled sysrq keys 

> 
> 
> 
> 
> Could anybody please let me know is there anything I missed out?
> 
> Do I need to do something different for the testing of kexec/kdump?
> 
> 
> 
> Could anybody please suggest me something on this?
> 
> 

I will suggest you to just go through the steps mentioned in
Documetation/kdump/kdump.txt in the kernel source tree if you
are testing kdump for mainline kernels.

The .config looks ok, but you don't need to enable
CONFIG_CRASH_DUMP for 1st kernel.

For distro specific testing please go through the documentation
(READMEs) which are generally packaged with the distro RPMs.


Thanks
Maneesh
_______________________________________________
fastboot mailing list
[email protected]
https://lists.osdl.org/mailman/listinfo/fastboot

Reply via email to