Bug 29522 - Unable to use any kind of frequency scaling
Summary: Unable to use any kind of frequency scaling
Status: CLOSED INSUFFICIENT_DATA
Alias: None
Product: Power Management
Classification: Unclassified
Component: cpufreq (show other bugs)
Hardware: All Linux
: P1 normal
Assignee: Rafael J. Wysocki
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2011-02-20 07:51 UTC by Rogério Brito
Modified: 2012-05-24 07:53 UTC (History)
2 users (show)

See Also:
Kernel Version: 2.6.38-rc5
Subsystem:
Regression: No
Bisected commit-id:


Attachments

Description Rogério Brito 2011-02-20 07:51:38 UTC
Hi.

I have a Pentium D 805 with a Gigabyte motherboard that is only able to reduce its speed with p4-clockmod and with the userspace governor (with the help of the powernowd daemon controlling it), not with the ondemand/conservative (since these modules say---correctly---that the latency is too high for them).

Since I see that p4-clockmod is frowned upon, I would like to get this fixed.

Is there anything that I could help with so that this could get fixed?


Thanks,

Rogério Brito.


P.S.: Right now, I am compiling a new kernel (from Linus git tree, up-to-date) with the following change, just to see its behaviour:

--- a/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c
+++ b/arch/x86/kernel/cpu/cpufreq/p4-clockmod.c
@@ -244,7 +244,7 @@ static int cpufreq_p4_cpu_init(struct cpufreq_policy *policy)
 
        /* the transition latency is set to be 1 higher than the maximum
         * transition latency of the ondemand governor */
-       policy->cpuinfo.transition_latency = 10000001;
+       policy->cpuinfo.transition_latency = 100000;
        policy->cur = stock_freq;
 
        return cpufreq_frequency_table_cpuinfo(policy, &p4clockmod_table[0]);
Comment 1 Rogério Brito 2011-02-20 10:48:12 UTC
Hi there.

Just as an extra data point, with the patch that I sent in the earlier package, I can indeed (as expected) set lower frequencies with ondemand (that is, *it* sets the frequency dynamically).

One benefit of this is that, when my computer is left idle, it does not heat as much as it used to, but I fear that there is no voltage being decreased, and, therefore, no power economy.

Is that right? How should one implement it "the right way"? I just know some C, but I don't know how program physical devices. I can, OTOH, test as many patches as you'd like me to.


Thanks in advance for any help, Rogério Brito.
Comment 2 Rogério Brito 2011-02-20 10:50:29 UTC
Oh, and I can provide as many details from this computer as is feasible. It is currently my main workstation and development box.

I may not have access to it during vacations and travels, though.


Thanks again,
Rogério Brito.
Comment 3 Zhang Rui 2012-01-18 03:15:02 UTC
It's great that the kernel bugzilla is back.

Can you please verify if the problem still exists in the latest upstream kernel?
Comment 4 Zhang Rui 2012-05-24 07:53:13 UTC
bug closed as there is no response from the bug reporter.
please feel free to reopen it if the problem still exists in the latest upstream kernel.

Note You need to log in before you can comment on or make changes to this bug.