Merge tag 'pm+acpi-3.19-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull more ACPI and power management updates from Rafael Wysocki:
 "These are regression fixes (leds-gpio, ACPI backlight driver,
  operating performance points library, ACPI device enumeration
  messages, cpupower tool), other bug fixes (ACPI EC driver, ACPI device
  PM), some cleanups in the operating performance points (OPP)
  framework, continuation of CONFIG_PM_RUNTIME elimination, a couple of
  minor intel_pstate driver changes, a new MAINTAINERS entry for it and
  an ACPI fan driver change needed for better support of thermal
  management in user space.

  Specifics:

   - Fix a regression in leds-gpio introduced by a recent commit that
     inadvertently changed the name of one of the properties used by the
     driver (Fabio Estevam).

   - Fix a regression in the ACPI backlight driver introduced by a
     recent fix that missed one special case that had to be taken into
     account (Aaron Lu).

   - Drop the level of some new kernel messages from the ACPI core
     introduced by a recent commit to KERN_DEBUG which they should have
     used from the start and drop some other unuseful KERN_ERR messages
     printed by ACPI (Rafael J Wysocki).

   - Revert an incorrect commit modifying the cpupower tool (Prarit
     Bhargava).

   - Fix two regressions introduced by recent commits in the OPP library
     and clean up some existing minor issues in that code (Viresh
     Kumar).

   - Continue to replace CONFIG_PM_RUNTIME with CONFIG_PM throughout the
     tree (or drop it where that can be done) in order to make it
     possible to eliminate CONFIG_PM_RUNTIME (Rafael J Wysocki, Ulf
     Hansson, Ludovic Desroches).

     There will be one more "CONFIG_PM_RUNTIME removal" batch after this
     one, because some new uses of it have been introduced during the
     current merge window, but that should be sufficient to finally get
     rid of it.

   - Make the ACPI EC driver more robust against race conditions related
     to GPE handler installation failures (Lv Zheng).

   - Prevent the ACPI device PM core code from attempting to disable
     GPEs that it has not enabled which confuses ACPICA and makes it
     report errors unnecessarily (Rafael J Wysocki).

   - Add a "force" command line switch to the intel_pstate driver to
     make it possible to override the blacklisting of some systems in
     that driver if needed (Ethan Zhao).

   - Improve intel_pstate code documentation and add a MAINTAINERS entry
     for it (Kristen Carlson Accardi).

   - Make the ACPI fan driver create cooling device interfaces witn
     names that reflect the IDs of the ACPI device objects they are
     associated with, except for "generic" ACPI fans (PNP ID "PNP0C0B").

     That's necessary for user space thermal management tools to be able
     to connect the fans with the parts of the system they are supposed
     to be cooling properly.  From Srinivas Pandruvada"

* tag 'pm+acpi-3.19-rc1-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (32 commits)
  MAINTAINERS: add entry for intel_pstate
  ACPI / video: update the skip case for acpi_video_device_in_dod()
  power / PM: Eliminate CONFIG_PM_RUNTIME
  NFC / PM: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  SCSI / PM: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  ACPI / EC: Fix unexpected ec_remove_handlers() invocations
  Revert "tools: cpupower: fix return checks for sysfs_get_idlestate_count()"
  tracing / PM: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  x86 / PM: Replace CONFIG_PM_RUNTIME in io_apic.c
  PM: Remove the SET_PM_RUNTIME_PM_OPS() macro
  mmc: atmel-mci: use SET_RUNTIME_PM_OPS() macro
  PM / Kconfig: Replace PM_RUNTIME with PM in dependencies
  ARM / PM: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  sound / PM: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  phy / PM: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  video / PM: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  tty / PM: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  spi: Replace CONFIG_PM_RUNTIME with CONFIG_PM
  ACPI / PM: Do not disable wakeup GPEs that have not been enabled
  ACPI / utils: Drop error messages from acpi_evaluate_reference()
  ...

1 files changed, 36 insertions, 2 deletions

diff --git a/drivers/cpufreq/intel_pstate.c b/drivers/cpufreq/intel_pstate.c
index 1405b39..742eefb 100644
--- a/drivers/cpufreq/intel_pstate.c
+++ b/drivers/cpufreq/intel_pstate.c
@@ -199,7 +199,14 @@ static signed int pid_calc(struct _pid *pid, int32_t busy)
 
 	pid->integral += fp_error;
 
-	/* limit the integral term */
+	/*
+	 * We limit the integral here so that it will never
+	 * get higher than 30.  This prevents it from becoming
+	 * too large an input over long periods of time and allows
+	 * it to get factored out sooner.
+	 *
+	 * The value of 30 was chosen through experimentation.
+	 */
 	integral_limit = int_tofp(30);
 	if (pid->integral > integral_limit)
 		pid->integral = integral_limit;
@@ -616,6 +623,11 @@ static void intel_pstate_get_min_max(struct cpudata *cpu, int *min, int *max)
 	if (limits.no_turbo || limits.turbo_disabled)
 		max_perf = cpu->pstate.max_pstate;
 
+	/*
+	 * performance can be limited by user through sysfs, by cpufreq
+	 * policy, or by cpu specific default values determined through
+	 * experimentation.
+	 */
 	max_perf_adj = fp_toint(mul_fp(int_tofp(max_perf), limits.max_perf));
 	*max = clamp_t(int, max_perf_adj,
 			cpu->pstate.min_pstate, cpu->pstate.turbo_pstate);
@@ -717,11 +729,29 @@ static inline int32_t intel_pstate_get_scaled_busy(struct cpudata *cpu)
 	u32 duration_us;
 	u32 sample_time;
 
+	/*
+	 * core_busy is the ratio of actual performance to max
+	 * max_pstate is the max non turbo pstate available
+	 * current_pstate was the pstate that was requested during
+	 * 	the last sample period.
+	 *
+	 * We normalize core_busy, which was our actual percent
+	 * performance to what we requested during the last sample
+	 * period. The result will be a percentage of busy at a
+	 * specified pstate.
+	 */
 	core_busy = cpu->sample.core_pct_busy;
 	max_pstate = int_tofp(cpu->pstate.max_pstate);
 	current_pstate = int_tofp(cpu->pstate.current_pstate);
 	core_busy = mul_fp(core_busy, div_fp(max_pstate, current_pstate));
 
+	/*
+	 * Since we have a deferred timer, it will not fire unless
+	 * we are in C0.  So, determine if the actual elapsed time
+	 * is significantly greater (3x) than our sample interval.  If it
+	 * is, then we were idle for a long enough period of time
+	 * to adjust our busyness.
+	 */
 	sample_time = pid_params.sample_rate_ms  * USEC_PER_MSEC;
 	duration_us = (u32) ktime_us_delta(cpu->sample.time,
 					   cpu->last_sample_time);
@@ -948,6 +978,7 @@ static struct cpufreq_driver intel_pstate_driver = {
 
 static int __initdata no_load;
 static int __initdata no_hwp;
+static unsigned int force_load;
 
 static int intel_pstate_msrs_not_valid(void)
 {
@@ -1094,7 +1125,8 @@ static bool intel_pstate_platform_pwr_mgmt_exists(void)
 			case PSS:
 				return intel_pstate_no_acpi_pss();
 			case PPC:
-				return intel_pstate_has_acpi_ppc();
+				return intel_pstate_has_acpi_ppc() &&
+					(!force_load);
 			}
 	}
 
@@ -1175,6 +1207,8 @@ static int __init intel_pstate_setup(char *str)
 		no_load = 1;
 	if (!strcmp(str, "no_hwp"))
 		no_hwp = 1;
+	if (!strcmp(str, "force"))
+		force_load = 1;
 	return 0;
 }
 early_param("intel_pstate", intel_pstate_setup);