Unexpected shutdown - perhaps ACPI related?

Unexpected shutdown - perhaps ACPI related?

[Brian Schau]

[-- Attachment #1: Type: text/plain, Size: 656 bytes --]

Hello,

I am seeing strange behaviour on my laptop.   I've been running
diverse distributions (OpenSUSE, Ubuntu and others) - my
current being gOS 3.0 gadgets.
No matter the distribution I see unexpected shutdowns!   That is - one
second I work, the other I don't.    The laptop shuts
itself off without any prior notice.

If I disable ACPI the unexpected shutdown disappears (and I loose a
bunch of other good stuff :-(

I don't know how to diagnose this further so I am asking for help here.

For starters I've attached the output from lsusb and lspci.

I am running kernel 2.6.24-19.

How to proceed?   I am willing to test ... :-)

Best regards,
Brian

[-- Attachment #2: lspci.txt.gz --]
[-- Type: application/x-gzip, Size: 2523 bytes --]

[-- Attachment #3: lsusb.txt.gz --]
[-- Type: application/x-gzip, Size: 1559 bytes --]

Re: Unexpected shutdown - perhaps ACPI related?

[Thomas Renninger]

On Wednesday 08 October 2008 12:02:09 Brian Schau wrote:
> Hello,
>
> I am seeing strange behaviour on my laptop.   I've been running
> diverse distributions (OpenSUSE, Ubuntu and others) - my
> current being gOS 3.0 gadgets.
> No matter the distribution I see unexpected shutdowns!   That is - one
> second I work, the other I don't.    The laptop shuts
> itself off without any prior notice.
>
> If I disable ACPI the unexpected shutdown disappears (and I loose a
> bunch of other good stuff :-(
>
> I don't know how to diagnose this further so I am asking for help here.
>
> For starters I've attached the output from lsusb and lspci.
>
> I am running kernel 2.6.24-19.
>
> How to proceed?   I am willing to test ... :-)

Are this critical thermal shutdowns?
grep -i critical /var/log/messages
should reveal that.
If yes:
  - watch your fan/temperature states:
    /proc/acpi/{thermal_zone,fan}
  - Does cpufreq work?
  - Maybe
    echo 15 >/proc/acpi/thermal_zone/*/polling_frequency
    helps then?

     Thomas

Re: Unexpected shutdown - perhaps ACPI related?

[Len Brown]

On Wed, 8 Oct 2008, Thomas Renninger wrote:

> On Wednesday 08 October 2008 12:02:09 Brian Schau wrote:
> > Hello,
> >
> > I am seeing strange behaviour on my laptop.   I've been running
> > diverse distributions (OpenSUSE, Ubuntu and others) - my
> > current being gOS 3.0 gadgets.
> > No matter the distribution I see unexpected shutdowns!   That is - one
> > second I work, the other I don't.    The laptop shuts
> > itself off without any prior notice.

Is this a hardware poweroff, or a software poweroff?
Hardware poweroff would be sudden light's out -- often
you can hear the disk make a funny sound.
This is what would happen if you have no mattery and
you yank out the A/C.

Software poweroff would be something that invoked shutdown
command, and thus you'd see it in the logs at Thomas' suggested.

Thomas' guess is a good one.  See if this is thermal related.
load your cpu with a copy of "cat /dev/zero > /dev/null"
for each core.  Listen for fans to speed up,
watch the temperature files 
and see if they change.
see at what temperature the shutdown occurs.

For that matter, what temperature is it set to shutdown at? --
please respond with

grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*

thanks,
-Len

> > If I disable ACPI the unexpected shutdown disappears (and I loose a
> > bunch of other good stuff :-(
> >
> > I don't know how to diagnose this further so I am asking for help here.
> >
> > For starters I've attached the output from lsusb and lspci.
> >
> > I am running kernel 2.6.24-19.
> >
> > How to proceed?   I am willing to test ... :-)
> 
> Are this critical thermal shutdowns?
> grep -i critical /var/log/messages
> should reveal that.
> If yes:
>   - watch your fan/temperature states:
>     /proc/acpi/{thermal_zone,fan}
>   - Does cpufreq work?
>   - Maybe
>     echo 15 >/proc/acpi/thermal_zone/*/polling_frequency
>     helps then?
> 
>      Thomas
> --
> To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 

Re: Unexpected shutdown - perhaps ACPI related?

[Brian Schau]

Hi,


> Thomas' guess is a good one.  See if this is thermal related.
> load your cpu with a copy of "cat /dev/zero > /dev/null
> for each core.  Listen for fans to speed up,

I did so for 10 minutes and the temperatur never rose beyond 72 deg C.  I also
did some other work during that time.
I couldn't get it to shutdown ...

BTW, the system is a dual core 2GHz system.

> grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*

root@stacy:~# grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*
/proc/acpi/thermal_zone/THRM/cooling_mode:<setting not supported>
/proc/acpi/thermal_zone/THRM/polling_frequency:<polling disabled>
/proc/acpi/thermal_zone/THRM/state:state:                   ok
/proc/acpi/thermal_zone/THRM/temperature:temperature:             73 C
/proc/acpi/thermal_zone/THRM/trip_points:critical (S5):           104 C
grep: /proc/acpi/fan/*/*: No such file or directory

wrt.to the last line (grep error) - the 'fan' modules is loaded but
there are not
files present in the /proc/acpi/fan directory.

Also, when I see the shutdowns it occurs as you described above - the hardware
shutdown.   I've grepped my logfiles but there are no indication of
ACPI anomalities.

Best regards,
Brian

Re: Unexpected shutdown - perhaps ACPI related?

[Thomas Renninger]

On Thursday 09 October 2008 13:40:36 Brian Schau wrote:
> Hi,
>
> > Thomas' guess is a good one.  See if this is thermal related.
> > load your cpu with a copy of "cat /dev/zero > /dev/null
> > for each core.  Listen for fans to speed up,
>
> I did so for 10 minutes and the temperatur never rose beyond 72 deg C.
> I also did some other work during that time.
> I couldn't get it to shutdown ...
>
> BTW, the system is a dual core 2GHz system.
>
> > grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*
>
> root@stacy:~# grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*
> /proc/acpi/thermal_zone/THRM/cooling_mode:<setting not supported>
> /proc/acpi/thermal_zone/THRM/polling_frequency:<polling disabled>
> /proc/acpi/thermal_zone/THRM/state:state:                   ok
> /proc/acpi/thermal_zone/THRM/temperature:temperature:             73 C
> /proc/acpi/thermal_zone/THRM/trip_points:critical (S5):           104 C
> grep: /proc/acpi/fan/*/*: No such file or directory
73 C is beyond 72 C :)
104 sounds as if the HW would switch off before, could still be a thermal 
issue. Hmm, but the fans seem to be controlled by HW...

> wrt.to the last line (grep error) - the 'fan' modules is loaded but
> there are not
> files present in the /proc/acpi/fan directory.
>
> Also, when I see the shutdowns it occurs as you described above - the
> hardware shutdown.
Means, the power just gets switched off?
You should try to reproduce this easier somehow. Heavy battery/thermal 
reading, switching cpufreq up and down or something?
How often does that happen?
If it happens often you could try to exclude/not load ACPI drivers, cpufreq 
drivers. Switch off C-states. These are good candidates.

Does this only happen with X?
Maybe using another X driver (vesa fb) solves it?

Did this machine ever worked fine with older kernels/distributions?

     Thomas

Re: Unexpected shutdown - perhaps ACPI related?

[Len Brown]

On Thu, 9 Oct 2008, Thomas Renninger wrote:

> On Thursday 09 October 2008 13:40:36 Brian Schau wrote:

> > BTW, the system is a dual core 2GHz system.
> >
> > > grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*
> >
> > root@stacy:~# grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*
> > /proc/acpi/thermal_zone/THRM/cooling_mode:<setting not supported>
> > /proc/acpi/thermal_zone/THRM/polling_frequency:<polling disabled>
> > /proc/acpi/thermal_zone/THRM/state:state:                   ok
> > /proc/acpi/thermal_zone/THRM/temperature:temperature:             73 C
> > /proc/acpi/thermal_zone/THRM/trip_points:critical (S5):           104 C
> > grep: /proc/acpi/fan/*/*: No such file or directory

> 73 C is beyond 72 C :)
> 104 sounds as if the HW would switch off before, could still be a thermal 
> issue. Hmm, but the fans seem to be controlled by HW...
> 
> > wrt.to the last line (grep error) - the 'fan' modules is loaded but
> > there are not
> > files present in the /proc/acpi/fan directory.
> >
> > Also, when I see the shutdowns it occurs as you described above - the
> > hardware shutdown.

> Means, the power just gets switched off?
> You should try to reproduce this easier somehow. Heavy battery/thermal 
> reading, switching cpufreq up and down or something?
> How often does that happen?
> If it happens often you could try to exclude/not load ACPI drivers, cpufreq 
> drivers. Switch off C-states. These are good candidates.
> 
> Does this only happen with X?
> Maybe using another X driver (vesa fb) solves it?
> 
> Did this machine ever worked fine with older kernels/distributions?

Also, try running memtest overnight.

cheers,
-Len

Re: Unexpected shutdown - perhaps ACPI related?

[Brian Schau]

>> root@stacy:~# grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*
>> /proc/acpi/thermal_zone/THRM/cooling_mode:<setting not supported>
>> /proc/acpi/thermal_zone/THRM/polling_frequency:<polling disabled>
>> /proc/acpi/thermal_zone/THRM/state:state:                   ok
>> /proc/acpi/thermal_zone/THRM/temperature:temperature:             73 C
>> /proc/acpi/thermal_zone/THRM/trip_points:critical (S5):           104 C
>> grep: /proc/acpi/fan/*/*: No such file or directory
> 73 C is beyond 72 C :)

Sorry, my bad :-)

> 104 sounds as if the HW would switch off before, could still be a thermal
> issue. Hmm, but the fans seem to be controlled by HW...

I am pretty sure it somehow is hardware related.   And as the subject suggests
I am not even sure it is ACPI related (I don't know jack about ACPI so I cannot
identify if it is involved :-)
However, it seems that my laptop is shutting itself off when ACPI is enabled.
(Though, it hasn't done that for almost 2 days now - it's like when
you've been coughing
all night and then when you go to the doctors office you can't cough.
The minute you
step outside the coughing starts again.   Maybe I should stay on the
linux-acpi list :-)

>> Also, when I see the shutdowns it occurs as you described above - the
>> hardware shutdown.
> Means, the power just gets switched off?

Yes.   Just like that.

> You should try to reproduce this easier somehow. Heavy battery/thermal
> reading, switching cpufreq up and down or something?

I will try to create a script which scales the CPU and reads the battery ..

> How often does that happen?

Seldom enough to be easily diagnosed but often enough to be a pain in the a**!

> If it happens often you could try to exclude/not load ACPI drivers, cpufreq
> drivers. Switch off C-states. These are good candidates.

To have a stable system I boot with "apci=off noacpi" will that also
kill the cpufreq
drivers?     Turning off the acpi means I cannot have the machine automatically
turn off when I request a shutdown.   A minor thing I can live with ...
How do I switch off C-states?   (And what does it mean?  :-)

> Does this only happen with X?

So far, yes.  But I am only using X.   If I need to do console work
(which I often
do) I use xterm or gnome-termina.


> Did this machine ever worked fine with older kernels/distributions?

Yes.  I've been an avid slackware user and I never saw these problems.   OTOH
back then I rolled my own kernel and as far as I recall it did have
parts of the ACPI
interface included (primarily the battery interface) ...

/brian

Re: Unexpected shutdown - perhaps ACPI related?

[Zhao Yakui]

On Thu, 2008-10-09 at 13:40 +0200, Brian Schau wrote:

> BTW, the system is a dual core 2GHz system.
> 
> > grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*
> 
> root@stacy:~# grep . /proc/acpi/thermal*/*/* /proc/acpi/fan/*/*
> /proc/acpi/thermal_zone/THRM/cooling_mode:<setting not supported>
> /proc/acpi/thermal_zone/THRM/polling_frequency:<polling disabled>
> /proc/acpi/thermal_zone/THRM/state:state:                   ok
> /proc/acpi/thermal_zone/THRM/temperature:temperature:             73 C
> /proc/acpi/thermal_zone/THRM/trip_points:critical (S5):           104 C
> grep: /proc/acpi/fan/*/*: No such file or directory
> 
> wrt.to the last line (grep error) - the 'fan' modules is loaded but
> there are not
> files present in the /proc/acpi/fan directory.
The critical trip point (S5) is 104, which is above the current thermal
temperature. Maybe the unexpected shutdown is not related with the ACPI
thermal driver. 

Will you please attach the output of acpidump?
thanks.
> 
> Also, when I see the shutdowns it occurs as you described above - the hardware
> shutdown.   I've grepped my logfiles but there are no indication of
> ACPI anomalities.
> 
> Best regards,
> Brian
> --
> To unsubscribe from this list: send the line "unsubscribe linux-acpi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html