Re: regression between -rc4 and -rc5

Re: regression between -rc4 and -rc5

[Olaf Conradi]

On 12/9/05, Karasyov, Konstantin A <konstantin.a.karasyov-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org> wrote:
> Hi,
>
> It looks like your trip points are getting reset on thermal
> notifications.
> Your DSDT implements hysteresis for passive and active cooling (i.e.
> when the temperature raises above the trip point, the trip point value
> is decreased to avoid too frequent trip point crossing), so hardware
> sends the notification to the system to update the trip points, when the
> trip point is being crossed.
>
> So, it's not a bug, it is a firmware implementation, but your DSDT could
> be slightly modified to set higher trip points values:

I've started looking into the DSDT to fix various methods, so far just
tried to figure out what all methods do.

In the mean time I upgraded to kernel 2.6.15-rc4 and most problems
disappeared. Still using the original DSDT. During heavy usage cpu
temperature rises to 76-78, and never reaches 80.

Yesterday I upgraded to kernel 2.6.15-rc5, and during compilation of
ndiswrapper, while browsing the internet, it shutdown. If I compile
ndiswrapper and don't do anything else, the cpu temp touches 80
briefly, but does not shutdown. Under -rc4 this does not happen.

 -Olaf


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_idv37&alloc_id\x16865&op=click

Re: regression between -rc4 and -rc5

[Thomas Renninger]

Olaf Conradi wrote:
> 
> In the mean time I upgraded to kernel 2.6.15-rc4 and most problems
> disappeared. Still using the original DSDT. During heavy usage cpu
> temperature rises to 76-78, and never reaches 80.
> 
> Yesterday I upgraded to kernel 2.6.15-rc5, and during compilation of
> ndiswrapper, while browsing the internet, it shutdown. If I compile
> ndiswrapper and don't do anything else, the cpu temp touches 80
> briefly, but does not shutdown. Under -rc4 this does not happen.
> 
What you see is probably a bug that slipped in 2.6.15-rc5:
It totally disables the use of cpufreq for passive cooling.
I already pointed Len to it and as this is really sever I am sure it
will be added before 2.6.15 is out if possible.
Also see Dirk Mueller's mail (12/21/2005):
Subject: make cpu_has_cpufreq() work
or the bug:
http://bugzilla.kernel.org/show_bug.cgi?id=3410

   Thomas


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click

Re: regression between -rc4 and -rc5

[Olaf Conradi]

Hi, thanks for your reply.

On 12/22/05, Thomas Renninger <trenn-l3A5Bk7waGM@public.gmane.org> wrote:
> What you see is probably a bug that slipped in 2.6.15-rc5:
> It totally disables the use of cpufreq for passive cooling.
> I already pointed Len to it and as this is really sever I am sure it
> will be added before 2.6.15 is out if possible.
> Also see Dirk Mueller's mail (12/21/2005):
> Subject: make cpu_has_cpufreq() work
> or the bug:
> http://bugzilla.kernel.org/show_bug.cgi?id=3410

My processor does not support speedstep for frequency scaling, it only
supports throttling.

The bug you pointed to does not seem to be entirely cpufreq specific,
so maybe it is relevant. Am not sure.

How can I diagnose the problem? I can put the system under load by
compiling something, but how can I best monitor acpi?

Cheers and a merry xmas,
 -Olaf


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_idv37&alloc_id\x16865&op=click