| View previous topic :: View next topic |
| Author |
Message |
bigfunkymo
Apprentice
Joined: 23 Jan 2004
Posts: 237
|
 Posted: Wed Apr 29, 2009 8:07 pm Post subject: WTF: SysRq + b actually causes hard freeze |
 |
|
I'm about 30 hours into trying in vain to debug this problem. It's the most insane thing I've ever seen. When I use SysRq + b to reboot my machine, it locks up tight where it will not respond to anything less than the reset button--Further SysRq commands do not respond. This includes times when the machine is in a stable state and freshly booted. I have also noticed (and I think this may be related) that instead of rebooting after a kernel panic, the system freezes in the same manner.
I have a headless machine--no video card. I'm not running X (obviously) and therefore I am also not having trouble with nvidia drivers (also obviously). It is an Intel Q6600 in a Gigabyte GA-P35-DQ6 with 8 GB of DDR2. I've tried many kernels (vanilla, gentoo, and hardened) and configurations thereof, all between 2.6.26 and 2.6.28-r7. I've disabled ACPI and APIC. Neither dmesg nor kern.log reveal any information indicative of trouble what-so-ever.
I'm having a really hard time finding any information about this sort of behavior via search engines. This is largely because searching for "sysrq freeze reboot" results in thousands of matches explaining how to use sysrq commands to solve freeze issues and force reboots. I'm not asking for anyone to solve this for me, but I want to ask if anyone else is having similar trouble and how they might be dealing with it, or if anyone had any troubleshooting advice because I don't know what can even be done next.
_________________
[No package... Grabbing a set.] |
|
| Back to top |
|
 |
pigeon768
l33t
Joined: 02 Jan 2006
Posts: 683
|
| Posted: Wed Apr 29, 2009 9:48 pm Post subject: |
|
|
| Is ACPI enabled in the kernel? |
|
| Back to top |
|
|
bigfunkymo
Apprentice
Joined: 23 Jan 2004
Posts: 237
|
| Posted: Wed Apr 29, 2009 10:04 pm Post subject: |
|
|
I've had it enabled and disabled.
_________________
[No package... Grabbing a set.] |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 23062
|
| Posted: Thu Apr 30, 2009 2:52 am Post subject: |
|
|
| I think I have seen systems that had older versions of the KVM modules loaded exhibit this behavior. If I recall correctly, there is some strange behavior associated with some reboot methods when Intel VT is enabled, so the KVM modules need a chance to halt the virtualization extensions before initiating the reboot. This may be fixed in newer versions of KVM. I do not have any systems handy that I can reboot to test that. |
|
| Back to top |
|
|
bigfunkymo
Apprentice
Joined: 23 Jan 2004
Posts: 237
|
| Posted: Thu Apr 30, 2009 8:36 am Post subject: |
|
|
| Quote: | | If I recall correctly, there is some strange behavior associated with some reboot methods when Intel VT is enabled, so the KVM modules need a chance to halt the virtualization extensions before initiating the reboot. |
Are you suggesting that I should try disabling the Intel virtualization via the BIOS configuration? You mentioned KVM module versions--if you're talking about a kernel module for virtualization, that shouldn't be an issue as I don't enable modules in my kernels. That is unless you are possibly referring to a binary module that isn't distributed with the kernel, of course (hypothetically speaking, I'm not actually aware of any such module).
_________________
[No package... Grabbing a set.] |
|
| Back to top |
|
|
Mad Merlin
Veteran
Joined: 09 May 2005
Posts: 1155
|
| Posted: Thu Apr 30, 2009 5:49 pm Post subject: |
|
|
| bigfunkymo wrote: | | Quote: | | If I recall correctly, there is some strange behavior associated with some reboot methods when Intel VT is enabled, so the KVM modules need a chance to halt the virtualization extensions before initiating the reboot. |
Are you suggesting that I should try disabling the Intel virtualization via the BIOS configuration? You mentioned KVM module versions--if you're talking about a kernel module for virtualization, that shouldn't be an issue as I don't enable modules in my kernels. That is unless you are possibly referring to a binary module that isn't distributed with the kernel, of course (hypothetically speaking, I'm not actually aware of any such module). |
There's KVM support directly in the kernel which can be builtin or as a module, and there's also KVM modules in the app-emulation/kvm ebuild (which are usually newer) which can be built for your kernel.
_________________
Game! - Where the stick is mightier than the sword! |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 23062
|
| Posted: Fri May 01, 2009 3:48 am Post subject: |
|
|
| bigfunkymo wrote: | | Are you suggesting that I should try disabling the Intel virtualization via the BIOS configuration? You mentioned KVM module versions--if you're talking about a kernel module for virtualization, that shouldn't be an issue as I don't enable modules in my kernels. That is unless you are possibly referring to a binary module that isn't distributed with the kernel, of course (hypothetically speaking, I'm not actually aware of any such module). |
If you are not using the VT extensions, you may as well turn them off. As Mad Merlin said, KVM can be a module or built in. In either case, older versions of KVM enabled VT unconditionally when the KVM code initialized. I do not recall seeing this fixed, but it has been a while since I checked. |
|
| Back to top |
|
|
|
Kernel & Hardware
