Bug 11310 - MSI VR 330 Laptop hard locks with 2.6.24 and upper kernels
Summary: MSI VR 330 Laptop hard locks with 2.6.24 and upper kernels
Status: CLOSED CODE_FIX
Alias: None
Product: ACPI
Classification: Unclassified
Component: Other (show other bugs)
Hardware: All Linux
: P1 high
Assignee: ykzhao
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2008-08-12 07:18 UTC by Cenk Limni
Modified: 2008-11-17 17:15 UTC (History)
2 users (show)

See Also:
Kernel Version: 2.6.25
Subsystem:
Regression: Yes
Bisected commit-id:


Attachments
Pardus Normal Boot dmesg output (21.49 KB, text/plain)
2008-08-13 04:17 UTC, Cenk Limni
Details
Pardus dmesg with acpi=off parameter (17.45 KB, text/plain)
2008-08-13 04:18 UTC, Cenk Limni
Details
Ubuntu Hardy Heron Normal dmesg output (26.78 KB, text/plain)
2008-08-13 04:19 UTC, Cenk Limni
Details
Pardus dmesg with processor max_cstate=1 (21.22 KB, text/plain)
2008-08-13 04:20 UTC, Cenk Limni
Details
Ubuntu Feisty 2.6.20-17 dmesg output (20.90 KB, text/plain)
2008-09-11 05:43 UTC, GuidoI
Details
Ubuntu Hardy 2.6.22 with acpi processor built-in (20.21 KB, application/octet-stream)
2008-09-20 00:12 UTC, GuidoI
Details

Description Cenk Limni 2008-08-12 07:18:08 UTC
Latest working kernel version:2.6.23
Earliest failing kernel version:2.6.25
Distribution:Arch Linux, Pardus, OpenSuSe 11.0, gOS etc..
Hardware Environment:MSI VR330 Laptop (AMD Sempron 3600+ 2gb 667mhz kingston DDR2 RAM, Nvidia Go6100, Realtek ALC883 Sound card, Ralink rt73 wireless adapter)
Software Environment:
Problem Description:Computer hard locks at any time. Many different distrubitons made this. Not only the X server, when i boot from terminal it's locked in terminal also. When i boot the distro with "acpi=off" parameter there's no locking.

Steps to reproduce:
Comment 1 Zhang Rui 2008-08-12 18:35:43 UTC
as this is a regression,
could you please run git bisect to find out which commit introduces the problem please?
Comment 2 ykzhao 2008-08-12 20:53:17 UTC
Hi, Cenk
   Will you please try the boot option of "processor.max_cstate=1" and see whether the problem still exists? 
   Please try another boot option of "idle=halt".
   Will you please attach the output of acpidump, dmesg? (the dmesg on 2.6.23 and 2.6.25 are both required).
   Thanks.
   
Comment 3 Cenk Limni 2008-08-13 04:17:48 UTC
Created attachment 17211 [details]
Pardus Normal Boot dmesg output

This is the Pardus dmesg with no boot options
Comment 4 Cenk Limni 2008-08-13 04:18:43 UTC
Created attachment 17212 [details]
Pardus dmesg with acpi=off parameter

Pardus disto dmesg output with acpi=off option
Comment 5 Cenk Limni 2008-08-13 04:19:28 UTC
Created attachment 17213 [details]
Ubuntu Hardy Heron Normal dmesg output

Ubuntu Hardy Heron's dmesg output with no option
Comment 6 Cenk Limni 2008-08-13 04:20:42 UTC
Created attachment 17214 [details]
Pardus dmesg with processor max_cstate=1

Pardus distro dmesg output with processor max_cstate=1 option
Comment 7 Cenk Limni 2008-08-13 04:26:32 UTC
I tried "processor max_cstate=1" and "idle=halt" boot options but there's still locking. I used Pardus distro (using 2.6.25.14-104 kernel)as problematic. In Ubuntu there is no locking so i used its dmesg output as non-problematic.

I also noted an error message at boot so i write it down here.

PCI BIOS Bug: MCFG area at e0000000 is not E820-reserved
PCI not using MMCONFIG
ACPI: error attaching device data 
ACPI: error attaching device data
Comment 8 ykzhao 2008-08-13 18:18:42 UTC
Hi, Cenk
   Sorry that you misuse the boot option. 
   From the log it seems that the processor driver is compiled as modules. Will you please compile it as built-in kernel and try the boot option of "processor.max_cstate=1"?
   Will you please try the boot option of "idle=poll"?
   thanks.
Comment 9 GuidoI 2008-09-11 05:40:04 UTC
(In reply to comment #0)
> Latest working kernel version:2.6.23
> Earliest failing kernel version:2.6.25
> Distribution:Arch Linux, Pardus, OpenSuSe 11.0, gOS etc..
> Hardware Environment:MSI VR330 Laptop (AMD Sempron 3600+ 2gb 667mhz kingston
> DDR2 RAM, Nvidia Go6100, Realtek ALC883 Sound card, Ralink rt73 wireless
> adapter)

Hi,
I have the same laptop, with exact same configuration except CPU is AMD Sempron 3500+ with 1gb ram.

In my case, the lockups occur even with 2.6.22, so (at least for me) it doesn't seem like a regression on 2.6.25.

I've just found a spanish thread where a user comments he has run (on this laptop) Ubuntu 7.04 without problems, but that on 7.10 he had to set acpi=off boot option.
http://www.psicofxp.com/forums/gnu-linux.50/737610-instalar-fedora-en-notebook-msi-drivers.html

So I'm right now installing kernel 2.6.20 , (the Feisty-stock one) to test, and will report here the results.

I'm really interested in getting to the root of this issue, since I'm pretty happy with kernel performance/compatibility on this laptop... except for the acpi problem.

> Software Environment:
> Problem Description:Computer hard locks at any time. Many different
> distrubitons made this. Not only the X server, when i boot from terminal it's
> locked in terminal also. When i boot the distro with "acpi=off" parameter
> there's no locking.

I'm confirming these informations. Lock-ups have ocurred anywhere from an hour after booting, up to a couple of days uptime. Even with X server not running. acpi=off solves the problem. Tried disabling the powernow_k8 module (cpu scaling) without positive results.

Still haven't tried compiling the processor driver as built-in, i understand you refer to this module: kernel/drivers/acpi/processor.ko

Thanks a lot both of you!
Comment 10 GuidoI 2008-09-11 05:43:47 UTC
Created attachment 17726 [details]
Ubuntu Feisty 2.6.20-17 dmesg output

I've installed and booted the 2.6.20 kernel supposedly reported to be stable with acpi on.
I'm right now running on that kernel to post this message. Will leave the computer on a couple of days and report what happens...
Comment 11 GuidoI 2008-09-18 23:47:57 UTC
(In reply to comment #10)
> Created an attachment (id=17726) [details]
> Ubuntu Feisty 2.6.20-17 dmesg output
> 
> I've installed and booted the 2.6.20 kernel supposedly reported to be stable
> with acpi on.
> I'm right now running on that kernel to post this message. Will leave the
> computer on a couple of days and report what happens...
> 
So far, so good: Not even one lock-up, I'm really gladly suprised :)
I've scored +48 hs uptimes, with ACPI on, something that never happened since I bought this laptop and installed linux.

So, now more than encouraged, i'm going to rebuild the 2.6.22 kernel with the processor module built-in into the kernel, and try the options you suggested (ykzhao)

Cheers!
Comment 12 GuidoI 2008-09-20 00:12:18 UTC
Created attachment 17888 [details]
Ubuntu Hardy 2.6.22 with acpi processor built-in

I've recompiled the kernel 2.6.22 (Exactly the same tree and config that caused the lockups) changing CONFIG_ACPI_PROCESSOR=y  (the locking-up one was a module)
I then booted with the mentioned options (both at the same time):
processor.max_cstate=1 idle=poll

I'm running right now on that configuration, with an uptime of 9 hours. Hope the dmesg is worth something.

Cheers!

ps. I'm again really really grateful for (in?)directly helping me solve this issue. Thanks!
Comment 13 ykzhao 2008-09-23 02:32:57 UTC
Thanks for the test.
   If both options of "processor.max_csate=1" and "idle=poll" are added, the processor.max_cstate will be useless. 
   The CPU on this laptop is AMD. And it seems that there exists some problems about the C1E.
   Will you please try the following boot option on the latest kernel(2.6.27-rc6/rc7)and see whether the problem still exists?
   a. idle=poll
   b. processor.max_cstate=1
   c. nolapic_timer
   d. nohz=off (disable tickless feature).
   thanks.
Comment 14 GuidoI 2008-09-23 05:31:59 UTC
Hi ykzhao!
Spoke too soon ;)
I finally experienced a lockup when using both options together, after about a day of uptime.
Then, i tried only putting "idle=poll" and i had another lockup after a couple of hours.
I then put only processor.max_cstate=1 , and once again after a day or so of uptime, it locked up while playing music.
The nolapic_timer option, i've tried it before (only that with the processor.ko as a module, not with processor compiled in). without luck, of course.
I will compile latest kernel and see what happens with those options.
I'm a bit surprised that an -older- kernel works... i'd never imagined that, in fact i was always installing (to try) latest kernel from ubuntu repositories in a passive hope of it being fixed :)

Thanks a lot.
Comment 15 Zhang Rui 2008-11-16 23:14:41 UTC
will you please try the latest kernel?
there are a couple of timer/clock fixes shipped in recently.
Comment 16 Cenk Limni 2008-11-17 02:50:44 UTC
I installed Ubuntu 8.10 Intrepid with kernel 2.6.27-7 1 or 2 weeks ago and there's no lock up anymore. The problem has gone.
Comment 17 GuidoI 2008-11-17 12:46:16 UTC
(In reply to comment #16)
> I installed Ubuntu 8.10 Intrepid with kernel 2.6.27-7 1 or 2 weeks ago and
> there's no lock up anymore. The problem has gone.
> 

That's great, I procrastinated too much the test of the vanilla 2.6.27 kernel, and I wondered if the new Ubuntu would have a sufficiently updated kernel - you've just answered me!
Thanks a lot,

Guido
Comment 18 Zhang Rui 2008-11-17 17:15:52 UTC
so the problem has already been fixed in the latest kernel.
close this bug.

Note You need to log in before you can comment on or make changes to this bug.