Hello again, hope you are still reading my texts...
I digged through the code and narrowed down the issue I wanted to fix.
It appears to be related to the `bool thermal_active` dpm struct
member and this piece of code:
if (rdev->asic->dpm.force_performance_level) {
if (rdev->pm.dpm.thermal_active) {
enum radeon_dpm_forced_level level = rdev->pm.dpm.forced_level;
/* force low perf level for thermal */
radeon_dpm_force_performance_level(rdev,
RADEON_DPM_FORCED_LEVEL_LOW);
/* save the user's level */
rdev->pm.dpm.forced_level = level;
} else {
/* otherwise, user selected level */
radeon_dpm_force_performance_level(rdev, rdev->pm.dpm.forced_level);
}
}
I did a double check here - at boot `thermal_active` is `false` and
thus, performance level is properly initiated. But at resume from
suspend `thermal_active` is true and performance level is strictly
bound to low profile.
Besides you cannot change it via echo 1 > /sys/.../force_dpm_level,
again thanks to `thermal_active` checked there.
Could you explain meaning of this small boolean to me? I'd like to
make a small neat patch fixing this, but I'm scared of doing it in
wrong way.
Sorry if I'm being too persistent.
Thanks,
Oleg
2014-07-22 20:05 GMT+04:00 Alex Deucher <alexdeucher@xxxxxxxxx>:
> On Tue, Jul 22, 2014 at 8:39 AM, Адонай Элохим <algonkvel@xxxxxxxxx> wrote:
>> Hello all!
>>
>> I have some spare time and knowledge in C to try to fix some bugs I am
>> seeing on my machine.
>> So I've checked out and compiled all git trees that I may need and now I'm
>> beginning to read articles.
>>
>> And this is the point from where I don't know where to go. I want to fix
>> particular bug #79806 [1].
>> For me there are many places where this bug can hide - mesa? dri? radeon
>> kernel module? and I just don't know whether should I start reading articles
>> about mesa hacking or about dri architecture or about kernel module
>> development.
>>
>> Now I think the best thing for me is to start looking through radeon kernel
>> module code (I've got ingenious idea that power management resides there)
>> and read more about its architecture. Is this right? I mean, I just want to
>> find out, is this a right place to start looking at for this bug?
>
> The power management is handled in the kernel driver. See radeon_pm.c
> and the relevant *_dpm.c files depending on what asic you have.
>
> Alex
>
>>
>> P.S. Sorry for my English in case it's not good, I'm learning it now
>>
>> P.P.S. And thanks for your hard work!
>>
>> -------------------------------------------
>> [1] https://bugs.freedesktop.org/show_bug.cgi?id=79806
>>
>> _______________________________________________
>> dri-devel mailing list
>> dri-devel@xxxxxxxxxxxxxxxxxxxxx
>> http://lists.freedesktop.org/mailman/listinfo/dri-devel
>>
_______________________________________________
dri-devel mailing list
dri-devel@xxxxxxxxxxxxxxxxxxxxx
http://lists.freedesktop.org/mailman/listinfo/dri-devel
