Messages in this thread | | | Date | Tue, 06 Jan 2009 00:41:07 -0500 | From | yutong.zhao@utoronto ... | Subject | kexec unable to load crash kernel in X |
| |
Hi linux-kernel,
I've been trying to track a hard lock kernel panic since 2.6.24, and it persists in 2.6.28. I've set up kexec as per the instructions in /Documentation/kdump/kdump.txt. I've enabled sysrq in both the kernel, AND, in /etc/sysctl.conf
The system responds to sysrq-C when I'm in the console and before startx. Thus, properly loading the dump kernel and allowing me access to /proc/vmcore
However, when I'm in X (after startx), sysrq-c does NOT load up the crash kernel, and instead hangs system (no blinking capslock though). I see some flashing hard drive activity, yet nothing changes on the screen as it still shows a frozen snapshot of X. This is true with both nv (with proper modprobe -r nvidia), and nvidia modules.
I'm on a Thinkpad T61P Intel Core 2 Duo, 2.5GHZ, x86_64, 2GB Memory, NVIDIA QUADRO FX 570M, and note that my CONFIG_PHYSICAL_START is 0x200000, NOT 0x100000, (I wasn't sure if this should be changed as the default was 0x200000, possibly due to Core 2?). But this shouldn't matter because the kernel is RELOCATABLE(=y).. right?
I've attached my:
.config : http://dpaste.com/105844/ lspci : http://dpaste.com/105846/ lsmod : http://dpaste.com/105848/ dmesg : http://dpaste.com/105849/ grub.conf : http://dpaste.com/105850/, most relevant boot config system kernel is "kernel /boot/kernel-2.6.28-gentoo-custom root=/dev/sda3 pci=nommconfig idle=poll crashkernel=128M@16M" note that /boot/kernel-2.6.28-gentoo-custom corresponds to the bzImage used in kexec.
my kexec command is "sudo kexec -p /usr/src/linux-2.6.28-gentoo/arch/x86/boot/bzImage --append=root=/dev/sda3 1 irqpoll maxcpus=1 reset_devices"
Please assist and CC me in all messages/comments/answers.
-Yutong Zhao
| |