On Friday 26 February 2016 12:04:59 Li Yang wrote:

On Fri, Dec 18, 2015 at 4:32 PM, Arnd Bergmann [off-list ref] wrote:

quoted

On Tuesday 15 December 2015 00:58:26 Rafael J. Wysocki wrote:

quoted

On Thursday, November 26, 2015 05:21:11 PM Jia Hongtao wrote:

quoted

Register the qoriq cpufreq driver as a cooling device, based on the
thermal device tree framework. When temperature crosses the passive trip
point cpufreq is used to throttle CPUs.

Signed-off-by: Jia Hongtao <redacted>
Reviewed-by: Viresh Kumar <viresh.kumar@linaro.org>

Applied, thanks!

I got a randconfig build error today:

drivers/built-in.o: In function `qoriq_cpufreq_ready':
debugfs.c:(.text+0x1f4688): undefined reference to `of_cpufreq_cooling_register'

CONFIG_OF=y
CONFIG_QORIQ_CPUFREQ=y
CONFIG_THERMAL=m
CONFIG_THERMAL_OF=y

I think you need a 'depends on THERMAL' to prevent the driver from
being built-in when THERMAL=m.

Maybe this is not the best approach.  The cpufreq feature itself
should be working independently without thermal framework.  I think we
should make the qoriq_cpufreq_ready() defined as null function if
THERMAL is not defined.

It already does this when CONFIG_THERMAL is not defined, and my
patch doesn't change that. I'm not sure what you are asking for now.

Do you want to allow using the cpufreq driver as a built-in driver
even when the thermal code is in a module, and then silently skip
all thermal management as if it was turned off?

That would be this patch:

diff --git a/include/linux/cpu_cooling.h b/include/linux/cpu_cooling.h
index c156f5082758..a8d9241fc1bb 100644
--- a/include/linux/cpu_cooling.h
+++ b/include/linux/cpu_cooling.h

@@ -48,7 +48,7 @@ cpufreq_power_cooling_register(const struct cpumask *clip_cpus,
  * @np: a valid struct device_node to the cooling device device tree node.
  * @clip_cpus: cpumask of cpus where the frequency constraints will happen
  */
-#ifdef CONFIG_THERMAL_OF
+#if IS_REACHABLE(CONFIG_THERMAL_OF)
 struct thermal_cooling_device *
 of_cpufreq_cooling_register(struct device_node *np,
 			    const struct cpumask *clip_cpus);

but my feeling is that this would cause more surprises when users
find their thermal management is not active even though it was
enabled in Kconfig and the thermal module gets loaded.

	Arnd