Skip to contentSkip to conversation

2.6.12-rc6 missing commit(s) in cpufreq?

PATCH Voluspa 3 participants 5 messages Jun 17, 2005

lore
Share & tools
lore

Voluspa

author
#0 lore raw

According to:
http://www.kernel.org/pub/linux/kernel/v2.6/testing/ChangeLog-2.6.12-rc6

There should be a:

commit 1206aaac285904e3e3995eecbf4129b6555a8973

Author: Dave Jones <[email protected]>
Date:   Tue May 31 19:03:48 2005 -0700

[CPUFREQ] Allow ondemand stepping to be changed by user.

And when I look at:
http://www.kernel.org/git/?p=linux/kernel/git/davej/cpufreq.git;a=commit;h=1206aaac285904e3e3995eecbf4129b6555a8973

There are changes in the diff like:

cpufreq/cpufreq_ondemand.c +0 -0
@@ -79,6 +79,7 @@ struct dbs_tuners {

unsigned int up_threshold;
unsigned int down_threshold;
unsigned int ignore_nice;
+ unsigned int freq_step;
};

Problem is that neither a clean 2.6.11 patched with patch-2.6.12-rc6 nor a full linux-2.6.12-rc6.tar.bz (I just downloaded it) contain that commit.

The first example from above looks like:

struct dbs_tuners {
        unsigned int            sampling_rate;
        unsigned int            sampling_down_factor;
        unsigned int            up_threshold;
        unsigned int            ignore_nice;
};

Even more strange is the other discrepancies in that list, suggesting other missed commits. Live directory from 2.6.12-rc :

root:sleipner:/sys/devices/system/cpu/cpu0/cpufreq/ondemand# ls
total 0
0 ignore_nice           0 sampling_rate      0 sampling_rate_min
0 sampling_down_factor  0 sampling_rate_max  0 up_threshold

Perhaps this is why I can't get the conservative governor to work at all (it just sits at the freq at which it was loaded, never going up/down no matter the load).

Mvh
Mats Johannesson
--

#1 lore raw
Quoted 25 lines · On Fri, 17 Jun 2005 12:21:55 +0200 Voluspa <[email protected]> wrote:

commit 1206aaac285904e3e3995eecbf4129b6555a8973

Author: Dave Jones <[email protected]>
Date:   Tue May 31 19:03:48 2005 -0700

[CPUFREQ] Allow ondemand stepping to be changed by user.

And when I look at:
http://www.kernel.org/git/?p=linux/kernel/git/davej/cpufreq.git;a=commit;h=1206aaac285904e3e3995eecbf4129b6555a8973

There are changes in the diff like:

cpufreq/cpufreq_ondemand.c +1 -0
@@ -79,6 +79,7 @@ struct dbs_tuners {
 unsigned int up_threshold;
 unsigned int down_threshold;
 unsigned int ignore_nice;
+ unsigned int freq_step;
 };
 

Problem is that neither a clean 2.6.11 patched with patch-2.6.12-rc6 nor a full linux-2.6.12-rc6.tar.bz (I just downloaded it) contain that commit.

Quoted 8 lines

The first example from above looks like:

struct dbs_tuners {
        unsigned int            sampling_rate;
        unsigned int            sampling_down_factor;
        unsigned int            up_threshold;
        unsigned int            ignore_nice;
};

Commit c29f1403098135bdef75b190a5037db514701031 removes that element of struct dbs_tuners again.

Signature · 3 lines
Cheers,
Stephen Rothwell                    [email protected]
http://www.canb.auug.org.au/~sfr/
author
#2 lore raw
Quoted 1 line · On Fri, 17 Jun 2005 20:35:49 +1000 Stephen Rothwell wrote:

On Fri, 17 Jun 2005 12:21:55 +0200 Voluspa <[email protected]> wrote:

[...]

Quoted 2 lines

Commit c29f1403098135bdef75b190a5037db514701031 removes that element of struct dbs_tuners again.

Yeah, I just saw that while going through the full git history. Argh, I'm a git. Better go on a vacation now. Bye.

Mvh
Mats Johannesson
--

#3 lore raw
In reply to Voluspa
Quoted 5 lines · Voluspa wrote/

According to:
http://www.kernel.org/pub/linux/kernel/v2.6/testing/ChangeLog-2.6.12-rc6

There should be a:

Quoted 9 lines

commit 1206aaac285904e3e3995eecbf4129b6555a8973

Author: Dave Jones <[email protected]>
Date:   Tue May 31 19:03:48 2005 -0700

[CPUFREQ] Allow ondemand stepping to be changed by user.

And when I look at:
http://www.kernel.org/git/?p=linux/kernel/git/davej/cpufreq.git;a=commit;h=1206aaac285904e3e3995eecbf4129b6555a8973

:

Quoted 4 lines

Problem is that neither a clean 2.6.11 patched with patch-2.6.12-rc6 nor a full linux-2.6.12-rc6.tar.bz (I just downloaded it) contain that commit.

Yes, it is in linux-2.6.12-rc6, but also "[CPUFREQ] ondemand governor automatic downscaling" which back off this change because it introduce a new algorithm which compute automatically this step to the best value.

The cpufreq tree was going during three months separatetly and then all the commits went inside the Linus's tree at once. It might look strange then, but nothing to worry :-)

:

Quoted 7 lines

Even more strange is the other discrepancies in that list, suggesting other missed commits. Live directory from 2.6.12-rc :

root:sleipner:/sys/devices/system/cpu/cpu0/cpufreq/ondemand# ls
total 0
0 ignore_nice           0 sampling_rate      0 sampling_rate_min
0 sampling_down_factor  0 sampling_rate_max  0 up_threshold

That's what you should see, everything is ok :-)

Quoted 4 lines

Perhaps this is why I can't get the conservative governor to work at all (it just sits at the freq at which it was loaded, never going up/down no matter the load).

I think it's completetly unrelated. Probably it's better to discuss about this bug on the cpufreq mailing list: [email protected]

Eric

author
#4 lore raw
Quoted 1 line · On Fri, 17 Jun 2005 12:41:00 +0200 Eric Piel wrote:

That's what you should see, everything is ok :-)

Thank's for the nice reply (in face of my uninformed spam).

Quoted 5 lines
Quoted 3 lines

Perhaps this is why I can't get the conservative governor to work at all (it just sits at the freq at which it was loaded, never going up/down no matter the load).

I think it's completetly unrelated. Probably it's better to discuss about this bug on the cpufreq mailing list: [email protected]

Pointer was welcome. I'll pop over when the vacation (I sure am in need of) is over.

Mvh
Mats Johannesson
--