Merge tag 'pm-4.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management updates from Rafael Wysocki:
 "Traditionally, cpufreq is the area with the greatest number of
  changes, but there are fewer of them than last time. There also is
  some activity in the generic power domains and the devfreq frameworks,
  a couple of system suspend and hibernation fixes and some assorted
  changes in other places.

  One new feature is the cpufreq change to allow the scheduler to pass
  hints to the governors' utilization update callbacks and some code
  rework based on that. Another one is the support for domain removal in
  the generic power domains framework. Also it is now possible to use
  hibernation with PAGE_POISONING_ZERO enabled and devfreq supports the
  RockChip DFI controller and the rk3399 DMC.

  The rest of the changes is mostly fixes and cleanups in a number of
  places.

  Specifics:

   - Add a mechanism for passing hints from the scheduler to cpufreq
     governors via their utilization update callbacks and use it to
     introduce "IOwait boosting" into the schedutil governor and
     intel_pstate that will make them boost performance if the enqueued
     task was previously waiting on I/O (Rafael Wysocki).

   - Fix a schedutil governor problem that causes it to overestimate
     utilization if SMT is in use (Steve Muckle).

   - Update defconfigs trying to use the schedutil governor as a module
     which is not possible any more (Javier Martinez Canillas).

   - Update the intel_pstate's pstate_sample tracepoint to take "IOwait
     boosting" into account (Srinivas Pandruvada).

   - Fix a problem in the cpufreq core causing it to mishandle the
     initialization of CPUs registered after the cpufreq driver (Viresh
     Kumar, Rafael Wysocki).

   - Make the cpufreq-dt driver support per-policy governor tunables,
     clean it up and update its Kconfig description (Viresh Kumar).

   - Add support for more ARM platforms to the cpufreq-dt driver
     (Chanwoo Choi, Dave Gerlach, Geert Uytterhoeven).

   - Make the cpufreq CPPC driver report frequencies in KHz to avoid
     user space compatiblility issues (Al Stone, Hoan Tran).

   - Clean up a few cpufreq drivers (st, kirkwood, SCPI) a bit (Colin
     Ian King, Markus Elfring).

   - Constify some local structures in the intel_pstate driver (Julia
     Lawall).

   - Add a Documentation/cpu-freq/ entry to MAINTAINERS (Jean Delvare).

   - Add support for PM domain removal to the generic power domains
     (genpd) framework, add new DT helper functions to it and make it
     always enable debugfs support if available (Jon Hunter, Tomeu
     Vizoso).

   - Clean up the generic power domains (genpd) framework and make it
     avoid measuring power-on and power-off latencies during system-wide
     PM transitions (Ulf Hansson).

   - Add support for the RockChip DFI controller and the rk3399 DMC to
     the devfreq framework (Lin Huang, Axel Lin, Arnd Bergmann).

   - Add COMPILE_TEST to the devfreq framework (Krzysztof Kozlowski,
     Stephen Rothwell).

   - Fix a minor issue in the exynos-ppmu devfreq driver and fix up
     devfreq Kconfig indentation style (Wei Yongjun, Jisheng Zhang).

   - Fix the system suspend interface to make suspend-to-idle work if
     platform suspend operations have not been registered (Sudeep
     Holla).

   - Make it possible to use hibernation with PAGE_POISONING_ZERO
     enabled (Anisse Astier).

   - Increas the default timeout of the system suspend/resume watchdog
     and make it depend on EXPERT (Chen Yu).

   - Make the operating performance points (OPP) framework avoid using
     OPPs that aren't supported by the platform and fix a build warning
     in it (Dave Gerlach, Arnd Bergmann).

   - Fix the ARM cpuidle driver's return value (Christophe Jaillet).

   - Make the SmartReflex AVS (Adaptive Voltage Scaling) driver use more
     common logging style (Joe Perches)"

* tag 'pm-4.9-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (58 commits)
  PM / OPP: Don't support OPP if it provides supported-hw but platform does not
  cpufreq: st: add missing \n to end of dev_err message
  cpufreq: kirkwood: add missing \n to end of dev_err messages
  PM / Domains: Rename pm_genpd_sync_poweron|poweroff()
  PM / Domains: Don't measure latency of ->power_on|off() during system PM
  PM / Domains: Remove redundant system PM callbacks
  PM / Domains: Simplify detaching a device from its genpd
  PM / devfreq: rk3399_dmc: Remove explictly regulator_put call in .remove
  PM / devfreq: rockchip: add PM_DEVFREQ_EVENT dependency
  PM / OPP: avoid maybe-uninitialized warning
  PM / Domains: Allow holes in genpd_data.domains array
  cpufreq: CPPC: Avoid overflow when calculating desired_perf
  cpufreq: ti: Use generic platdev driver
  cpufreq: intel_pstate: Add io_boost trace
  partial revert of "PM / devfreq: Add COMPILE_TEST for build coverage"
  cpufreq: intel_pstate: Use IOWAIT flag in Atom algorithm
  cpufreq: schedutil: Add iowait boosting
  cpufreq / sched: SCHED_CPUFREQ_IOWAIT flag to indicate iowait condition
  PM / Domains: Add support for removing nested PM domains by provider
  PM / Domains: Add support for removing PM domains
  ...

1 files changed, 49 insertions, 4 deletions

diff --git a/drivers/cpufreq/cppc_cpufreq.c b/drivers/cpufreq/cppc_cpufreq.c
index 8882b8e2ecd0..99db4227ae38 100644
--- a/drivers/cpufreq/cppc_cpufreq.c
+++ b/drivers/cpufreq/cppc_cpufreq.c
@@ -19,10 +19,19 @@
 #include <linux/delay.h>
 #include <linux/cpu.h>
 #include <linux/cpufreq.h>
+#include <linux/dmi.h>
 #include <linux/vmalloc.h>
 
+#include <asm/unaligned.h>
+
 #include <acpi/cppc_acpi.h>
 
+/* Minimum struct length needed for the DMI processor entry we want */
+#define DMI_ENTRY_PROCESSOR_MIN_LENGTH	48
+
+/* Offest in the DMI processor structure for the max frequency */
+#define DMI_PROCESSOR_MAX_SPEED  0x14
+
 /*
  * These structs contain information parsed from per CPU
  * ACPI _CPC structures.
@@ -32,6 +41,39 @@
  */
 static struct cpudata **all_cpu_data;
 
+/* Capture the max KHz from DMI */
+static u64 cppc_dmi_max_khz;
+
+/* Callback function used to retrieve the max frequency from DMI */
+static void cppc_find_dmi_mhz(const struct dmi_header *dm, void *private)
+{
+	const u8 *dmi_data = (const u8 *)dm;
+	u16 *mhz = (u16 *)private;
+
+	if (dm->type == DMI_ENTRY_PROCESSOR &&
+	    dm->length >= DMI_ENTRY_PROCESSOR_MIN_LENGTH) {
+		u16 val = (u16)get_unaligned((const u16 *)
+				(dmi_data + DMI_PROCESSOR_MAX_SPEED));
+		*mhz = val > *mhz ? val : *mhz;
+	}
+}
+
+/* Look up the max frequency in DMI */
+static u64 cppc_get_dmi_max_khz(void)
+{
+	u16 mhz = 0;
+
+	dmi_walk(cppc_find_dmi_mhz, &mhz);
+
+	/*
+	 * Real stupid fallback value, just in case there is no
+	 * actual value set.
+	 */
+	mhz = mhz ? mhz : 1;
+
+	return (1000 * mhz);
+}
+
 static int cppc_cpufreq_set_target(struct cpufreq_policy *policy,
 		unsigned int target_freq,
 		unsigned int relation)
@@ -42,7 +84,7 @@ static int cppc_cpufreq_set_target(struct cpufreq_policy *policy,
 
 	cpu = all_cpu_data[policy->cpu];
 
-	cpu->perf_ctrls.desired_perf = target_freq;
+	cpu->perf_ctrls.desired_perf = (u64)target_freq * policy->max / cppc_dmi_max_khz;
 	freqs.old = policy->cur;
 	freqs.new = target_freq;
 
@@ -94,8 +136,10 @@ static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
 		return ret;
 	}
 
-	policy->min = cpu->perf_caps.lowest_perf;
-	policy->max = cpu->perf_caps.highest_perf;
+	cppc_dmi_max_khz = cppc_get_dmi_max_khz();
+
+	policy->min = cpu->perf_caps.lowest_perf * cppc_dmi_max_khz / cpu->perf_caps.highest_perf;
+	policy->max = cppc_dmi_max_khz;
 	policy->cpuinfo.min_freq = policy->min;
 	policy->cpuinfo.max_freq = policy->max;
 	policy->shared_type = cpu->shared_type;
@@ -112,7 +156,8 @@ static int cppc_cpufreq_cpu_init(struct cpufreq_policy *policy)
 	cpu->cur_policy = policy;
 
 	/* Set policy->cur to max now. The governors will adjust later. */
-	policy->cur = cpu->perf_ctrls.desired_perf = cpu->perf_caps.highest_perf;
+	policy->cur = cppc_dmi_max_khz;
+	cpu->perf_ctrls.desired_perf = cpu->perf_caps.highest_perf;
 
 	ret = cppc_set_perf(cpu_num, &cpu->perf_ctrls);
 	if (ret)