VirtualBox

Ticket #10751 (new defect)

Opened 3 years ago

Last modified 3 years ago

OEL5 guest panic on init_tsc_clocksource

Reported by: asmca Owned by:
Priority: major Component: VM control
Version: VirtualBox 4.1.18 Keywords: OEL, UEK, panic, init_tsc_clocksource
Cc: Guest type: Linux
Host type: Linux

Description

I was running Fedora 15 with VirtualBox-4.0-4.0.10_72479_fedora15-1.i686. After upgrade os kernel of oel5 guest to 2.6.32-300.29.1.el5uek, it failed to reboot again, panic.

I upgrade VirtualBox Version too, it failed yet... Following in my environment:

HOST


$uname -a Linux suse.rac 2.6.43.8-1.fc15.i686 #1 SMP Mon Jun 4 20:28:56 UTC 2012 i686 i686 i386 GNU/Linux

VirtualBox-4.1-4.1.18_78361_fedora15-1.i686.rpm

Guest OS


As described in attched snapshot, running 2.6.32-200.20.1.el5uek is fine. running 2.6.32-300.29.1.el5uek failed. "oel_guest_kernel_selection.png"

Kernel Panic on init_tsc_clocksource


Described as attched pic: "panic_screenshot.png"

Attachments

oel_guest_kernel_selection.png Download (11.0 KB) - added by asmca 3 years ago.
screenshot of guest os kernel lists. failed on 2.6.32-300
panic_screenshot.png Download (23.2 KB) - added by asmca 3 years ago.
kernel panic screenshot, on init_tsc_clocksource
VBox.log.2_6_32_300_fail.log Download (51.3 KB) - added by asmca 3 years ago.
VBox Log failed on 2.6.32-300, till the panic
sc_running_2_6_32_200_fine.png Download (226.0 KB) - added by asmca 3 years ago.
this pic is not about the panic, os can boot well on 2.6.32-200 kernel.
VBox.log.2_6_32_200.running.fine.log Download (65.9 KB) - added by asmca 3 years ago.
this log is not on panic, just on 2.6.32-200 kernel and running fine.
SEL5.vbox Download (24.1 KB) - added by asmca 3 years ago.
configure file of the guest machine, OEL5.
VBox.log.HPET_enabled.log Download (55.1 KB) - added by asmca 3 years ago.
After enable HPET, successfully boot the new kernel.

Change History

Changed 3 years ago by asmca

screenshot of guest os kernel lists. failed on 2.6.32-300

Changed 3 years ago by asmca

kernel panic screenshot, on init_tsc_clocksource

Changed 3 years ago by asmca

VBox Log failed on 2.6.32-300, till the panic

Changed 3 years ago by asmca

this pic is not about the panic, os can boot well on 2.6.32-200 kernel.

Changed 3 years ago by asmca

this log is not on panic, just on 2.6.32-200 kernel and running fine.

Changed 3 years ago by asmca

configure file of the guest machine, OEL5.

comment:1 Changed 3 years ago by asmca

logs uploaded,

comment:2 Changed 3 years ago by asmca

Well another issue is that I always wait a long time when boot the EL5 node. <at stage showed on screen: sc_running_2_6_32_200_fine.png > Whoever can help to look at this issue may look at

attachment sc_running_2_6_32_200_fine.png attachment VBox.log.2_6_32_200.running.fine.log attachment SEL5.vbox

Thanks a lot, -s

comment:3 Changed 3 years ago by frank

You already enabled the I/O-APIC in the VM settings. Could you do two additional tests:

  • enable the the HPET (VBoxManage modifyvm VM_NAME --hpet on) => does the VM boot?
  • if that does not help, could you enable the ICH9 chipset (experimental!) (VBoxManage modifyvm VM_NAME --chipset ich9) => does the VM boot?

comment:4 Changed 3 years ago by asmca

Hi Frank,

I just enabled HPET and it successfully boots now.

attach the VBox.log.HPET_enabled.log

Changed 3 years ago by asmca

After enable HPET, successfully boot the new kernel.

comment:5 Changed 3 years ago by asmca

Hi Frank,

When you're free would you have a look at the new log, I still dont know why boot needs a lot of time,

after <00:00:20> to <00:02:12>

some 2min waiting for the boot screen continue... is it a issue that my HOST machine low or some configuration side problem?

Thanks, -s

comment:6 Changed 3 years ago by frank

That might be indeed your host. This is a Pentium-D which is actually a Dual Pentium 4 and this CPU behaves not very well during virtualization. Also 2GB of host RAM are more the lower limit. There is one thing you could try: Add

divider=10

to the GRUB command line of the guest kernel. I'm not 100% sure if this option is still evaluated by the UEK kernel but it's worth a try.

comment:7 Changed 3 years ago by asmca

Thanks Frank,

When I try divider=10 on 2.6.32-200 kernel, it seems no obvious performance improvement. and I hit the above kernel panic again when enable divider=10 on 2.6.32-300 kernel.

So I think it may because of the poor machine hardware issue.

BTW, is there any bestpractice or document on the configuration/development?

Regards, -s

Note: See TracTickets for help on using tickets.

www.oracle.com
ContactPrivacy policyTerms of Use