2012-11-05 03:40:53

by Drunkard Zhang

[permalink] [raw]
Subject: Auto reboot when CPU at full load with X86_ACPI_CPUFREQ

I'm using Intel Xeon X5570 x2 with Asus Z8PE-D18, and experiencing
auto reboot when CPU full loaded for minutes, like building kernel
with "make -j17". After a lot of bisecting of config file, I found the
option leads to the reboot: X86_ACPI_CPUFREQ, both configed
X86_ACPI_CPUFREQ as a module or built in will lead to reboot.

Config file finally bisected appended, config-3.7.0-rc3+-bad is the
one leads to reboot, config-3.7.0-rc3+-ok works OK. Hardware info also
appended.

I think it is a bug, anything I can do? When the bug triggered, the
screen blanked immediately, any advice for me to debug? Happy to match
to code :-)

Tested kernel version: 3.1.x, 3.3.x, 3.5.x, 3.6.x, 3.7*, they are all affected.


Attachments:
config-3.7.0-rc3+-bad (70.49 kB)
config-3.7.0-rc3+-ok (70.46 kB)
hardware-Asus_Z8PE-D18-Xeon5570 (131.24 kB)
Download all attachments

2012-11-05 05:18:35

by Drunkard Zhang

[permalink] [raw]
Subject: Re: Auto reboot when CPU at full load with X86_ACPI_CPUFREQ

> Config file finally bisected appended, config-3.7.0-rc3+-bad is the
> one leads to reboot, config-3.7.0-rc3+-ok works OK. Hardware info also
> appended.
>
Sorry, I forgot one thing, this bug is CPU specific, with Xeon E5606
or E5620 it's all fine, just triggered with Xeon X5570, or maybe all
Xeon X serial.

2012-11-05 16:13:59

by Drunkard Zhang

[permalink] [raw]
Subject: Auto reboot when CPU at full load with X86_ACPI_CPUFREQ

Resending to right lists :)

I'm using Intel Xeon X5570 x2 with Asus Z8PE-D18, and experiencing
auto reboot when CPU full loaded for minutes, like building kernel
with "make -j17". After a lot of bisecting of config file, I found the
option leads to the reboot: X86_ACPI_CPUFREQ, both configed
X86_ACPI_CPUFREQ as a module or built in will lead to reboot.

Config file finally bisected appended, config-3.7.0-rc3+-bad is the
one leads to reboot, config-3.7.0-rc3+-ok works OK. Hardware info also
appended.

I think it is a bug, anything I can do? When the bug triggered, the
screen blanked immediately, any advice for me to debug? Happy to match
to code :-)

This bug is CPU specific, with Xeon E5606 or E5620 it's all fine, just
triggered with Xeon X5570, or maybe all Xeon X serial.

Tested kernel version: 3.1.x, 3.3.x, 3.5.x, 3.6.x, 3.7*, they are all affected.


Attachments:
config-3.7.0-rc3+-bad (70.49 kB)
config-3.7.0-rc3+-ok (70.46 kB)
hardware-Asus_Z8PE-D18-Xeon5570 (131.24 kB)
Download all attachments

2012-11-11 21:01:22

by Pavel Machek

[permalink] [raw]
Subject: Re: Auto reboot when CPU at full load with X86_ACPI_CPUFREQ

Hi!

> I'm using Intel Xeon X5570 x2 with Asus Z8PE-D18, and experiencing
> auto reboot when CPU full loaded for minutes, like building kernel
> with "make -j17". After a lot of bisecting of config file, I found the
> option leads to the reboot: X86_ACPI_CPUFREQ, both configed
> X86_ACPI_CPUFREQ as a module or built in will lead to reboot.
>
> Config file finally bisected appended, config-3.7.0-rc3+-bad is the
> one leads to reboot, config-3.7.0-rc3+-ok works OK. Hardware info also
> appended.
>
> I think it is a bug, anything I can do? When the bug triggered, the
> screen blanked immediately, any advice for me to debug? Happy to match
> to code :-)
>
> This bug is CPU specific, with Xeon E5606 or E5620 it's all fine, just
> triggered with Xeon X5570, or maybe all Xeon X serial.
>
> Tested kernel version: 3.1.x, 3.3.x, 3.5.x, 3.6.x, 3.7*, they are all affected.

What does temperature do during those runs?
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

2012-11-12 02:41:33

by Drunkard Zhang

[permalink] [raw]
Subject: Re: Auto reboot when CPU at full load with X86_ACPI_CPUFREQ

2012/11/12 Pavel Machek <[email protected]>:
> Hi!
>
>> I'm using Intel Xeon X5570 x2 with Asus Z8PE-D18, and experiencing
>> auto reboot when CPU full loaded for minutes, like building kernel
>> with "make -j17". After a lot of bisecting of config file, I found the
>> option leads to the reboot: X86_ACPI_CPUFREQ, both configed
>> X86_ACPI_CPUFREQ as a module or built in will lead to reboot.
>>
>> Config file finally bisected appended, config-3.7.0-rc3+-bad is the
>> one leads to reboot, config-3.7.0-rc3+-ok works OK. Hardware info also
>> appended.
>>
>> I think it is a bug, anything I can do? When the bug triggered, the
>> screen blanked immediately, any advice for me to debug? Happy to match
>> to code :-)
>>
>> This bug is CPU specific, with Xeon E5606 or E5620 it's all fine, just
>> triggered with Xeon X5570, or maybe all Xeon X serial.
>>
>> Tested kernel version: 3.1.x, 3.3.x, 3.5.x, 3.6.x, 3.7*, they are all affected.
>
> What does temperature do during those runs?
> Pavel

CPU temperature didn't over 60 by degrees. I'm sure it's not hardware
faults, because without X86_ACPI_CPUFREQ it works fine, built in or
load as module it will reboot. I also configured some debug options to
catch something, but the reboot is too fast, dam. Any sugests?