Kernel 2.4.23 passive cooling

Kernel 2.4.23 passive cooling

[Nils Faerber]

Hi!
Passive cooling still seems a little strange, at least to me on my Asus
L3800C notebook.
What happens:
 - active cooling is perfectly fine and I succesfully use
speedstepdaemon to automatically set my CPU to lower performance in idle
mode.
 - passive cooling works until it reaches the first hot state. After
that the fan(s) power up for a very short time and the CPU performance
is set to maximum (!) again. The performance setting can afterwards not
be reset to low performance anymore; it will be reset to high
automatically!
 - Changing back to active cooling restores the machine into a normal
"cool" state but still leaves the performance unchangeable. Only fix is
to reboot (I have ACPI fix in kernel).

So basically for me passive cooling is not usable. I think this is a bug
in the hysteresis handling in the thermal zone management.

Is there any chance that this will be fixed?
Or is this a BIOS bug?

CU
  nils faerber

-- 
kernel concepts          Tel: +49-271-771091-12
Dreisbachstr. 24         Fax: +49-271-771091-19
D-57250 Netphen          D1 : +49-170-2729106
--



-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

Re: Kernel 2.4.23 passive cooling

[Karol Kozimor]

Thus wrote Nils Faerber:
> Hi!
> Passive cooling still seems a little strange, at least to me on my Asus
> L3800C notebook.
[...]
>  - passive cooling works until it reaches the first hot state. After
> that the fan(s) power up for a very short time and the CPU performance
> is set to maximum (!) again. The performance setting can afterwards not
> be reset to low performance anymore; it will be reset to high
> automatically!

Use the limit interface instead of directly manipulating the
CPU0/performance -- this *may* help.

> Is there any chance that this will be fixed?
> Or is this a BIOS bug?

I'm not sure, but I haven't really had enough time to trace the appropriate 
AML codepath -- I can obviously also experience the problem. Oddly enough,
the first time I booted ACPI 20031002 (just in order to see if improved
index field access will fix the passive cooling problem), the processor
went up to about 83 C without the fan kicking in (I can't remember if
throttling was on at that time, but I guess it should have been). I then
switched to active (the fans started spinning like mad), and back to
passive, which apparently didn't stop the fans.

What I suspect is that there are some semi-obscure things going on in the
AML and it would take me at least two or three cups of coffee to trace it.
I'll see what I can do.

Note: most of the thermal management is done by the AML and not by the OS,
as th OS doesn't know about most tresholds.

Best regards,

-- 
Karol 'sziwan' Kozimor
sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

Re: Kernel 2.4.23 passive cooling

[Karol Kozimor]

Thus wrote Karol Kozimor:
> What I suspect is that there are some semi-obscure things going on in the
> AML and it would take me at least two or three cups of coffee to trace it.
> I'll see what I can do.

Oh, and it would *very much* help me if I knew how to make _all_ thermal 
events visible and reported (be it through /proc/acpi/event or at least
through a debug switch), and not only the 0x81 ones (hint hint!).
Best regards,

-- 
Karol 'sziwan' Kozimor
sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

RE: Kernel 2.4.23 passive cooling

[Yu, Luming]

Try
#echo -n 0x04000000 > /proc/acpi/debug_layer
#echo -n 0xffffffff > /proc/acpi/debug_level

Thanks,
Luming

-----Original Message-----
From: acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org [mailto:acpi-devel-admin-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org]On Behalf Of Karol Kozimor
Sent: 2003?12?2? 20:43
To: Nils Faerber; acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
Subject: Re: [ACPI] Kernel 2.4.23 passive cooling


Thus wrote Karol Kozimor:
> What I suspect is that there are some semi-obscure things going on in the
> AML and it would take me at least two or three cups of coffee to trace it.
> I'll see what I can do.

Oh, and it would *very much* help me if I knew how to make _all_ thermal 
events visible and reported (be it through /proc/acpi/event or at least
through a debug switch), and not only the 0x81 ones (hint hint!).
Best regards,

-- 
Karol 'sziwan' Kozimor
sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/
_______________________________________________
Acpi-devel mailing list
Acpi-devel-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
https://lists.sourceforge.net/lists/listinfo/acpi-devel


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

Re: Kernel 2.4.23 passive cooling

[Karol Kozimor]

Thus wrote Yu, Luming:
> Try
> #echo -n 0x04000000 > /proc/acpi/debug_layer
> #echo -n 0xffffffff > /proc/acpi/debug_level

I produced a log (http://hell.org.pl/~sziwan/thermal.log) using those
switches. A bit of commentary:
18:57:59 -- I switch to passive cooling
18:59:07 -- Back to active
18:59:19 -- Forth to passive
[no thermal events till 19:25:12, the readings are trigerred manually]
After that, the temperature rocks up and down, as fan switch in at 3480 dK. 
This seems to be a DSDT bug after all, since the debug output doesn't seem
suspicious and it even fails to notice the fans (which is normal, since the
method to reevaluate the fans is internally called from within the GPE
context). See http://bugme.osdl.org/show_bug.cgi?id=1185 for the DSDT and
additional info.
Best regards,

-- 
Karol 'sziwan' Kozimor
sziwan-DETuoxkZsSqrDJvtcaxF/A@public.gmane.org


-------------------------------------------------------
This SF.net email is sponsored by: Perforce Software.
Perforce is the Fast Software Configuration Management System offering
advanced branching capabilities and atomic changes on 50+ platforms.
Free Eval! http://www.perforce.com/perforce/loadprog.html

RE: Kernel 2.4.23 passive cooling

[Yu, Luming]

> > #echo -n 0x04000000 > /proc/acpi/debug_layer
> > #echo -n 0xffffffff > /proc/acpi/debug_level
> 
> I produced a log (http://hell.org.pl/~sziwan/thermal.log) using those
> switches. A bit of commentary:
> 18:57:59 -- I switch to passive cooling
> 18:59:07 -- Back to active
> 18:59:19 -- Forth to passive
> [no thermal events till 19:25:12, the readings are trigerred manually]

It's not strange , because when any GP events is detected, 
the method associated with this GPE  (_LXxx or _Exx) will be put at 
the end of queue, which will be asynch executed later.

> After that, the temperature rocks up and down, as fan switch 
> in at 3480 dK. 
> This seems to be a DSDT bug after all, since the debug output 
> doesn't seem

If that message is from execution of method _Lxx or _Exx, 
then it's not strange too.




-------------------------------------------------------
This SF.net email is sponsored by: Perforce Software.
Perforce is the Fast Software Configuration Management System offering
advanced branching capabilities and atomic changes on 50+ platforms.
Free Eval! http://www.perforce.com/perforce/loadprog.html