processor module locks up hyperthreading enabled machine

processor module locks up hyperthreading enabled machine

[Reinhard Nissl]

[-- Attachment #1: Type: text/plain, Size: 461 bytes --]

Hi,

I have this problem here on my system since starting to use kernel 
2.6.13's release candidats. Attached you'll find the console output of 
SuSE kernels 2.6.13 (OpenSuSE 10.0) and 2.6.13.2 (OpenSuSE 10.1). The 
latter is "identical" to the output of a self compiled kernel 
2.6.14-rc4, with the .config taken from 2.6.13.2's /proc/config.gz.

How can I further assist in solving this issue?

Bye.
-- 
Dipl.-Inform. (FH) Reinhard Nissl
mailto:rnissl@gmx.de

[-- Attachment #2: kernel-2.6.13.2.txt.gz --]
[-- Type: application/x-gzip, Size: 3997 bytes --]

[-- Attachment #3: kernel-2.6.13.txt.gz --]
[-- Type: application/x-gzip, Size: 6294 bytes --]

RE: processor module locks up hyperthreading enabled machine

[Pallipadi, Venkatesh]

1) Does this problem go way if you boot with maxcpus=1?
2) Does the problem go away if you not load the acpi processor module?

Looks somewhat similar to
http://bugzilla.kernel.org/show_bug.cgi?id=5331

Thanks,
-Venki

>-----Original Message-----
>From: linux-kernel-owner@vger.kernel.org 
>[mailto:linux-kernel-owner@vger.kernel.org] On Behalf Of Reinhard Nissl
>Sent: Wednesday, October 12, 2005 2:30 PM
>To: linux-kernel@vger.kernel.org
>Subject: processor module locks up hyperthreading enabled machine
>
>Hi,
>
>I have this problem here on my system since starting to use kernel 
>2.6.13's release candidats. Attached you'll find the console output of 
>SuSE kernels 2.6.13 (OpenSuSE 10.0) and 2.6.13.2 (OpenSuSE 10.1). The 
>latter is "identical" to the output of a self compiled kernel 
>2.6.14-rc4, with the .config taken from 2.6.13.2's /proc/config.gz.
>
>How can I further assist in solving this issue?
>
>Bye.
>-- 
>Dipl.-Inform. (FH) Reinhard Nissl
>mailto:rnissl@gmx.de
>

Re: processor module locks up hyperthreading enabled machine

[Reinhard Nissl]

Hi,

I'm sorry for the delay. Here are the results:

Pallipadi, Venkatesh wrote:

> 1) Does this problem go way if you boot with maxcpus=1?

No.

The modules are loaded via initrd and SuSE prints the name of the module 
before loading it. So I get on console

Loading processor
ACPI: CPU0 (power states:C1[C1])
ACPI: CPU0 (power states:C1[C1])
Loading thermal
Loading fan
Waiting for device /dev/sda1 to appear: ok
Attempting manual resume
Loading jdb
Loading ext
Waiting for device /dev/sda2 to appear: ok
rootfs: major=8 minor=2 devn=2050

Then the system freezes.

Without the maxcpus argument the freeze happens already in front of the 
above "Loading thermal" line.

> 2) Does the problem go away if you not load the acpi processor module?

Yes.

When I remove processor (and thermal as it has a dependency on 
processor) to be loaded by initrd then the freeze happens at some time 
later when acpid gets started, as SuSE loads these modules at that time 
again in case they were not loaded already.

When I remove loading the processor (and thermal) module at this 
location too, then the system is running ok so far besides the 
restrictions which result from not loading these modules.

What should I try next?

Bye.
-- 
Dipl.-Inform. (FH) Reinhard Nissl
mailto:rnissl@gmx.de

RE: processor module locks up hyperthreading enabled machine

[Pallipadi, Venkatesh]

 

>-----Original Message-----
>From: Reinhard Nissl [mailto:rnissl@gmx.de] 
>Sent: Tuesday, November 01, 2005 2:47 PM
>To: Pallipadi, Venkatesh
>Cc: linux-kernel@vger.kernel.org; Brown, Len
>Subject: Re: processor module locks up hyperthreading enabled machine
>
>Hi,
>
>I'm sorry for the delay. Here are the results:


Hmmmmm.... Please try the latest patch here
http://bugme.osdl.org/show_bug.cgi?id=5452


Thanks,
Venki



>
>Pallipadi, Venkatesh wrote:
>
>> 1) Does this problem go way if you boot with maxcpus=1?
>
>No.
>
>The modules are loaded via initrd and SuSE prints the name of 
>the module 
>before loading it. So I get on console
>
>Loading processor
>ACPI: CPU0 (power states:C1[C1])
>ACPI: CPU0 (power states:C1[C1])
>Loading thermal
>Loading fan
>Waiting for device /dev/sda1 to appear: ok
>Attempting manual resume
>Loading jdb
>Loading ext
>Waiting for device /dev/sda2 to appear: ok
>rootfs: major=8 minor=2 devn=2050
>
>Then the system freezes.
>
>Without the maxcpus argument the freeze happens already in 
>front of the 
>above "Loading thermal" line.
>
>> 2) Does the problem go away if you not load the acpi 
>processor module?
>
>Yes.
>
>When I remove processor (and thermal as it has a dependency on 
>processor) to be loaded by initrd then the freeze happens at some time 
>later when acpid gets started, as SuSE loads these modules at 
>that time 
>again in case they were not loaded already.
>
>When I remove loading the processor (and thermal) module at this 
>location too, then the system is running ok so far besides the 
>restrictions which result from not loading these modules.
>
>What should I try next?
>
>Bye.
>-- 
>Dipl.-Inform. (FH) Reinhard Nissl
>mailto:rnissl@gmx.de
>

Re: processor module locks up hyperthreading enabled machine

[Reinhard Nissl]

Hi,

Pallipadi, Venkatesh wrote:

> Hmmmmm.... Please try the latest patch here
> http://bugme.osdl.org/show_bug.cgi?id=5452

Your patch fixed the freeze. Tested with kernel 2.6.14.

I should have mentioned, that my system is equiped with a 
Fujitsu-Siemens Mainboard D1562-A2, most recent BIOS V4.06 R1.09.

I'll try to get this bug fixed in the BIOS, but chances are not that 
good as the board is already "quite old".

Would you please be so kind and drop a few lines about the drawbacks of 
this BIOS bug? What kind of functionality is lost by "ignoring" the 
second CPU at this particular code section?

Thanks for your support!

Bye.
-- 
Dipl.-Inform. (FH) Reinhard Nissl
mailto:rnissl@gmx.de

RE: processor module locks up hyperthreading enabled machine

[Pallipadi, Venkatesh]

 

>-----Original Message-----
>From: Reinhard Nissl [mailto:rnissl@gmx.de] 
>Sent: Wednesday, November 02, 2005 11:30 AM
>To: Pallipadi, Venkatesh
>Cc: linux-kernel@vger.kernel.org; Brown, Len
>Subject: Re: processor module locks up hyperthreading enabled machine
>
>Hi,
>
>Pallipadi, Venkatesh wrote:
>
>> Hmmmmm.... Please try the latest patch here
>> http://bugme.osdl.org/show_bug.cgi?id=5452
>
>Your patch fixed the freeze. Tested with kernel 2.6.14.
>
>I should have mentioned, that my system is equiped with a 
>Fujitsu-Siemens Mainboard D1562-A2, most recent BIOS V4.06 R1.09.
>
>I'll try to get this bug fixed in the BIOS, but chances are not that 
>good as the board is already "quite old".
>
>Would you please be so kind and drop a few lines about the 
>drawbacks of 
>this BIOS bug? What kind of functionality is lost by "ignoring" the 
>second CPU at this particular code section?
>

It will basically disable ACPI Processor Power Management features. 
P-state (speedstep or dynamic change of frequency to save power) and
C-states beyond C1 and T-states. 
>From dmesg, I know that on your system C2 and beyond is not supported.
T-states are kind of unimportantish. Only other difference will be
P-state. Not sure whether your processor supports speedstep and dynamic
change of cpufreq. If it doesn't, then you won't loose anything by this
workaround.

Thanks,
Venki