pm: fix pm configuration

[Hyungsin]

This PR fixes PM configuration.

PM functions in periph_common should not be executed if FEATURE_PERIPH_PM is used. PM_LAYERED will be used instead.

[haukepetersen]

seems to me that this PR is defeating its purpose, those functions are defined weak as to actually be overridden by other implementation if present... Could you elaborate a little more on you actual problem?

[vincent-d]

those functions are defined weak as to actually be overridden by other implementation if present

While working on #7351 I saw that it was not working, but I don't know why.

[vincent-d]

Actually, this is the same as #6709

[Hyungsin]

@haukepetersen, yeah the problem is that it does not work as expected.
That weak function is actually called when using pm_layer! FYI, my platform is SAMR21.
This problem has been reported multiple times, such as #6709 and #6655, as @vincent-d mentioned.

The fundamental solution would be making this function overridden. But for now, using #ifdef seems to be the simplest way of getting around this problem, at least temporarily.

[jnohlgard]

While working on Kinetis pm I noticed that adding USEMODULE += pm_layered has no effect unless pm_block/pm_unblock is used at least once in the drivers (periph). The linker doesn't discard the weak pm_set_lowest unless there is something else required from the file containing the strong pm_set_lowest.

[vincent-d]

The linker doesn't discard the weak pm_set_lowest unless there is something else required from the file containing the strong pm_set_lowest.

The strong symbol is also used is the object file (.a) of pm_layered is linked before periph_common. But our current build system pass them alphabetically, so periph_common is used before pm_layered.

[Hyungsin]

So, is there any problem if this PR is applied?
I think it is possible to use this PR before having a complete solution, if it has nothing wrong.

[Hyungsin]

@haukepetersen, @kaspar030, any feedback?

[kaspar030]

This fix is included in #7241.

[Hyungsin]

@kaspar030, Cool, waiting for your job finished!

[roberthartung]

This will be superseded by #7597

[kaspar030]

Closed in favor of #7726.