Power management updates for 7.0-rc4

- Make the idle loop skip the cpuidle governor .reflect() callback
    after it has skipped the .select() one (Rafael Wysocki)
 
  - Fix swapped power/energy unit labels in cpupower (Kaushlendra Kumar)
 
  - Add support for setting EPP via systemd service and intel_pstate
    turbo boost support to cpupower (Jan Kiszka, Zhang Rui)
 -----BEGIN PGP SIGNATURE-----
 
 iQFGBAABCAAwFiEEcM8Aw/RY0dgsiRUR7l+9nS/U47UFAmmzAVcSHHJqd0Byand5
 c29ja2kubmV0AAoJEO5fvZ0v1OO1NFEH/1wByyw7EUSbiCkHKtPGFUELcEseg9vT
 OwfWX3wdeXrAcadc5ZW8leYheZIX3NsJjmesSKyzGuHHPnZt++/eDA+PxcXyK/5F
 PqUKT1x3heCkoXlqUc21AsH+Yq9B9kj5fQ+D4yN//NL8J91CGlvqYN9HxRKYFena
 z7p0sN2wZ0ka0mu6b0yKmT4FTTHFVKunB1ReY/t6BDMp0T30LTRHHLHhBkgaQBz6
 fhlC/MyAldyGDg2mt2RsmB7HGMuePl69qTAArAwsC7wbwCcBhnUNx1Lxf5XtPQqv
 oZhBrBvpfzZYObk85qbZEnXn0kxhpTbliJX1gHOoAiq5iBL9TxMs9HM=
 =h9bj
 -----END PGP SIGNATURE-----

Merge tag 'pm-7.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:

 - Make the idle loop skip the cpuidle governor .reflect() callback
   after it has skipped the .select() one (Rafael Wysocki)

 - Fix swapped power/energy unit labels in cpupower (Kaushlendra Kumar)

 - Add support for setting EPP via systemd service and intel_pstate
   turbo boost support to cpupower (Jan Kiszka, Zhang Rui)

* tag 'pm-7.0-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  sched: idle: Make skipping governor callbacks more consistent
  cpupower: Add intel_pstate turbo boost support for Intel platforms
  cpupower: Add support for setting EPP via systemd service
  cpupower: fix swapped power/energy unit labels

drivers/cpuidle/cpuidle.c

@@ -359,16 +359,6 @@ noinstr int cpuidle_enter_state(struct cpuidle_device *dev,
 int cpuidle_select(struct cpuidle_driver *drv, struct cpuidle_device *dev,
 		   bool *stop_tick)
 {
-	/*
-	 * If there is only a single idle state (or none), there is nothing
-	 * meaningful for the governor to choose. Skip the governor and
-	 * always use state 0 with the tick running.
-	 */
-	if (drv->state_count <= 1) {
-		*stop_tick = false;
-		return 0;
-	}
-
 	return cpuidle_curr_governor->select(drv, dev, stop_tick);
 }
 

kernel/sched/idle.c

@@ -221,7 +221,7 @@ static void cpuidle_idle_call(void)
 
 		next_state = cpuidle_find_deepest_state(drv, dev, max_latency_ns);
 		call_cpuidle(drv, dev, next_state);
-	} else {
+	} else if (drv->state_count > 1) {
 		bool stop_tick = true;
 
 		/*
@@ -239,6 +239,15 @@ static void cpuidle_idle_call(void)
 		 * Give the governor an opportunity to reflect on the outcome
 		 */
 		cpuidle_reflect(dev, entered_state);
+	} else {
+		tick_nohz_idle_retain_tick();
+
+		/*
+		 * If there is only a single idle state (or none), there is
+		 * nothing meaningful for the governor to choose.  Skip the
+		 * governor and always use state 0.
+		 */
+		call_cpuidle(drv, dev, 0);
 	}
 
 exit_idle:

tools/power/cpupower/cpupower-service.conf

@@ -30,3 +30,8 @@
 # its policy for the relative importance of performance versus energy savings to
 # the processor. See man CPUPOWER-SET(1) for additional details
 #PERF_BIAS=
+
+# Set the Energy Performance Preference
+# Available options can be read from
+# /sys/devices/system/cpu/cpufreq/policy0/energy_performance_available_preferences
+#EPP=

tools/power/cpupower/cpupower.sh

@@ -23,4 +23,10 @@ then
     cpupower set -b "$PERF_BIAS" > /dev/null || ESTATUS=1
 fi
 
+# apply Energy Performance Preference
+if test -n "$EPP"
+then
+    cpupower set -e "$EPP" > /dev/null || ESTATUS=1
+fi
+
 exit $ESTATUS

tools/power/cpupower/utils/cpupower-set.c

@@ -124,7 +124,11 @@ int cmd_set(int argc, char **argv)
 	}
 
 	if (params.turbo_boost) {
-		ret = cpupower_set_turbo_boost(turbo_boost);
+		if (cpupower_cpu_info.vendor == X86_VENDOR_INTEL)
+			ret = cpupower_set_intel_turbo_boost(turbo_boost);
+		else
+			ret = cpupower_set_generic_turbo_boost(turbo_boost);
+
 		if (ret)
 			fprintf(stderr, "Error setting turbo-boost\n");
 	}

tools/power/cpupower/utils/helpers/helpers.h

@@ -104,7 +104,7 @@ extern struct cpupower_cpu_info cpupower_cpu_info;
 /* cpuid and cpuinfo helpers  **************************/
 
 int cpufreq_has_generic_boost_support(bool *active);
-int cpupower_set_turbo_boost(int turbo_boost);
+int cpupower_set_generic_turbo_boost(int turbo_boost);
 
 /* X86 ONLY ****************************************/
 #if defined(__i386__) || defined(__x86_64__)
@@ -143,6 +143,7 @@ extern int decode_pstates(unsigned int cpu, int boost_states,
 
 int cpufreq_has_x86_boost_support(unsigned int cpu, int *support,
 				  int *active, int *states);
+int cpupower_set_intel_turbo_boost(int turbo_boost);
 
 /* AMD P-State stuff **************************/
 bool cpupower_amd_pstate_enabled(void);
@@ -189,6 +190,8 @@ static inline int cpupower_set_amd_pstate_mode(char *mode)
 static inline int cpufreq_has_x86_boost_support(unsigned int cpu, int *support,
 						int *active, int *states)
 { return -1; }
+static inline int cpupower_set_intel_turbo_boost(int turbo_boost)
+{ return -1; }
 
 static inline bool cpupower_amd_pstate_enabled(void)
 { return false; }

tools/power/cpupower/utils/helpers/misc.c

@@ -19,6 +19,9 @@ int cpufreq_has_x86_boost_support(unsigned int cpu, int *support, int *active,
 {
 	int ret;
 	unsigned long long val;
+	char linebuf[MAX_LINE_LEN];
+	char path[SYSFS_PATH_MAX];
+	char *endp;
 
 	*support = *active = *states = 0;
 
@@ -42,8 +45,42 @@ int cpufreq_has_x86_boost_support(unsigned int cpu, int *support, int *active,
 		}
 	} else if (cpupower_cpu_info.caps & CPUPOWER_CAP_AMD_PSTATE) {
 		amd_pstate_boost_init(cpu, support, active);
-	} else if (cpupower_cpu_info.caps & CPUPOWER_CAP_INTEL_IDA)
+	} else if (cpupower_cpu_info.caps & CPUPOWER_CAP_INTEL_IDA) {
 		*support = *active = 1;
+
+		snprintf(path, sizeof(path), PATH_TO_CPU "intel_pstate/no_turbo");
+
+		if (!is_valid_path(path))
+			return 0;
+
+		if (cpupower_read_sysfs(path, linebuf, MAX_LINE_LEN) == 0)
+			return -1;
+
+		val = strtol(linebuf, &endp, 0);
+		if (endp == linebuf || errno == ERANGE)
+			return -1;
+
+		*active = !val;
+	}
+	return 0;
+}
+
+int cpupower_set_intel_turbo_boost(int turbo_boost)
+{
+	char path[SYSFS_PATH_MAX];
+	char linebuf[2] = {};
+
+	snprintf(path, sizeof(path), PATH_TO_CPU "intel_pstate/no_turbo");
+
+	/* Fallback to generic solution when intel_pstate driver not running */
+	if (!is_valid_path(path))
+		return cpupower_set_generic_turbo_boost(turbo_boost);
+
+	snprintf(linebuf, sizeof(linebuf), "%d", !turbo_boost);
+
+	if (cpupower_write_sysfs(path, linebuf, 2) <= 0)
+		return -1;
+
 	return 0;
 }
 
@@ -274,7 +311,7 @@ void print_speed(unsigned long speed, int no_rounding)
 	}
 }
 
-int cpupower_set_turbo_boost(int turbo_boost)
+int cpupower_set_generic_turbo_boost(int turbo_boost)
 {
 	char path[SYSFS_PATH_MAX];
 	char linebuf[2] = {};

tools/power/cpupower/utils/powercap-info.c

@@ -38,11 +38,11 @@ static int powercap_print_one_zone(struct powercap_zone *zone)
 	printf(" (%s)\n", mode ? "enabled" : "disabled");
 
 	if (zone->has_power_uw)
-		printf(_("%sPower can be monitored in micro Jules\n"),
+		printf(_("%sPower can be monitored in micro Watts\n"),
 		       pr_prefix);
 
 	if (zone->has_energy_uj)
-		printf(_("%sPower can be monitored in micro Watts\n"),
+		printf(_("%sPower can be monitored in micro Jules\n"),
 		       pr_prefix);
 
 	printf("\n");