Re: [GIT PULL] thermal for v5.7-rc1

[Linus Torvalds <torvalds@linux-foundation.org>]

On Tue, Apr 7, 2020 at 4:26 AM Daniel Lezcano <daniel.lezcano@linaro.org> wrote:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/thermal/linux.git
> tags/thermal-v5.7-rc1

Ho humm.

This caused a conflict between commit

  f12e4f66ab6a ("thermal/cpu-cooling: Update thermal pressure in case
of a maximum frequency capping")

that came in through the scheduler updates from Ingo Molnar, and commit

  ff44f672d741 ("thermal/drivers/cpufreq_cooling: Fix return of
cpufreq_set_cur_state")

from the thermal tree.

The conflict wasn't complicated, but the reason I mention it is that I
resolved it in a way that neither of those commits had done.

In particular, the thermal tree did

  ret = freq_qos_update_request(..)
  return ret < 0 ? ret : 0;

where that whole "return negative or zero" logic is new (it used to
return positive values, the fix was to return zero instead).

The scheduler tree did

  ret = freq_qos_update_request(..)
  if (ret > 0) {.. do thermal pressure thing ..}
  return ret;

which obviously still returns that positive value.

My resolution to the conflict was to not take that return with a
conditional operation, but instead just add a

  ret = 0;

to inside that thermal pressure if-statement, and avoid returning a
non-zero positive value that way.

I just wanted both sides to be aware of my non-traditional merge
resolution, and take a look.

               Linus