Bug 7598
| Summary: | SMP kernel hangs up after load processor.ko on UP system | ||
|---|---|---|---|
| Product: | ACPI | Reporter: | Yauhen Kharuzhy (jekhor) |
| Component: | Power-Processor | Assignee: | Zhang Rui (rui.zhang) |
| Status: | REJECTED INSUFFICIENT_DATA | ||
| Severity: | normal | CC: | acpi-bugzilla, sam |
| Priority: | P2 | ||
| Hardware: | i386 | ||
| OS: | Linux | ||
| Kernel Version: | 2.6.19 | Subsystem: | |
| Regression: | --- | Bisected commit-id: | |
| Attachments: |
Full config
acpidump from acer 9300 laptop dmesg of boot on 2.6.20 kernel (from ubuntu fesity) which doesn't hang dmesg (without noacpi !!) for 2.6.20 (from ubuntu feisty) which doesn't hang dmesg for 2.6.22 without noacpi |
||
|
Description
Yauhen Kharuzhy
2006-11-30 07:46:41 UTC
Created attachment 9688 [details]
Full config
Hi, Yauhen Can the system work well ? Yes, kernels >= 2.6.21 work well. 2.6.22, 2.6.23 both hang for me a short while after processor.ko loads. They hang so tightly that even alt-sysreq-* fail to do anything. Many details are posted here: https://bugs.launchpad.net/linux/+bug/144030 The easiest way to reproduce the hang on-demand is to: insmod /somewhere-else/processor.ko /etc/init.d/gdm start BTW I forgot to mention this occurs with SMP or UP kernels on a UP machine. and in fact while in an X session, insmod processor.ko will hang the system right away. Can you try following two boot options and let me know if any helps - "max_cstate=1" - "max_cstate=2" Also, is 2.6.16 last known kernel that did not have the problem? Can you give dmesg with that kernel (with ACPI_PROCESSOR loaded). Also attach acpidump output from your system please. (in latest pmtools here - http://www.kernel.org/pub/linux/kernel/people/lenb/acpi/utils/ ) Created attachment 14386 [details]
acpidump from acer 9300 laptop
Here's the latest acpidump of my acer 9300 laptop that also hangs soon after boot
with max_cstate=1 or max_cstate=2 ubuntu still hangs (no alt-sysreq) as soon as GDM starts. Created attachment 14388 [details]
dmesg of boot on 2.6.20 kernel (from ubuntu fesity) which doesn't hang
The ubuntu feisty fawn kernel does not hang; here is a dmesg from that kernel.
Created attachment 14389 [details]
dmesg (without noacpi !!) for 2.6.20 (from ubuntu feisty) which doesn't hang
this time I booted the fesity kernel without noacpi
Created attachment 14390 [details]
dmesg for 2.6.22 without noacpi
This dmesg is the hanging kernel, before it hung.
(modprobe processor.ko from an X session is a sure way to hang it.
It takes a while to hang without X.)
I'm not getting a problem with 2.6.24-3 s part of Ubuntu Hardy Heron alpha 3. Sadly the upgrade failed an I had to fresh install making debuging the previous kernel sort of difficult... however I will make a fresh feisty->gutsy installation to debug this further if you want. Otherwise... it seems "fixed" in 2.6.24 I think this has come back with 2.6.24-12 as part of Ubuntu Hardy Heron updates Is this still a problem with 2.6.25. can you please verify if the problem still exists with 2.6.26? Is there a particular 2.6.26 set of debs you want me to try? I'll need the nvidia to go with it. Sam no, but you'd better try a clean kernel.org kernel. Would you be happy with an intrepid ibex test? Sam Hi, Sam Will you please add the boot option of "hpet=off pci=nobios" on the latest kernel and see whether the problem still exists? Thanks. Sorry the correct boot option should be "hpet=disable pci=nobios".("hpet=off" is not correct option).
Thanks.
I'll try that on 2.6.24 but it may take a while to gather enough statistical evidence as more recent kernels crashed less often. I'll put those in the boot conf today. Thanks for the tip. Sam -----Original Message----- From: bugme-daemon@bugzilla.kernel.org Sent: Tuesday, July 22, 2008 7:34 AM To: sam@liddicott.com Subject: [Bug 7598] SMP kernel hangs up after load processor.ko on UP system http://bugzilla.kernel.org/show_bug.cgi?id=7598 ------- Comment #21 from yakui.zhao@intel.com 2008-07-21 23:34 ------- Sorry the correct boot option should be "hpet=disable pci=nobios".("hpet=off" is not correct option). Thanks. > ------- Comment #21 from yakui.zhao@intel.com 2008-07-21 23:34 -------
> Sorry the correct boot option should be "hpet=disable pci=nobios".("hpet=off"
> is not correct option).
> Thanks.
>
Did You want me to try this on 2.6.26 or 2.6.24 ?
Sam
Hi, Sam Either of 2.6.24 and 2.6.26 is OK. sam, any updates? With a intrepid kernel and the two boot options I was asked to use there have been no hangs. It's been enough for me to be statistically sure that these 2 boot options made a difference. Sam -----Original Message----- From: bugme-daemon@bugzilla.kernel.org Sent: 10 September 2008 03:34 To: sam@liddicott.com Subject: [Bug 7598] SMP kernel hangs up after load processor.ko on UP system http://bugzilla.kernel.org/show_bug.cgi?id=7598 rui.zhang@intel.com changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |NEEDINFO ------- Comment #25 from rui.zhang@intel.com 2008-09-09 19:34 ------- sam, any updates? does a latest kernel work for you without any boot option? There are a lot of time/cstate updates recently. no response from the bug reporter. sam, please re-open the bug if the problem still exists in the latest kernel, e.g. 2.6.27 |