From 266c13d767be61a17d8e6f2310b9b7c46278273b Mon Sep 17 00:00:00 2001
From: Viresh Kumar
Date: Tue, 2 Jul 2013 16:36:28 +0530
Subject: cpufreq: Fix serialization of frequency transitions

Commit 7c30ed ("cpufreq: make sure frequency transitions are serialized")
interacts poorly with systems that have a single core freqency for all
cores.  On such systems we have a single policy for all cores with
several CPUs.  When we do a frequency transition the governor calls the
pre and post change notifiers which causes cpufreq_notify_transition()
per CPU.  Since the policy is the same for all of them all CPUs after
the first and the warnings added are generated by checking a per-policy
flag the warnings will be triggered for all cores after the first.

Fix this by allowing notifier to be called for n times. Where n is the number of
cpus in policy->cpus.

Reported-and-tested-by: Mark Brown <broonie@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/cpufreq/cpufreq.c | 7 ++++---
 include/linux/cpufreq.h   | 2 +-
 2 files changed, 5 insertions(+), 4 deletions(-)

diff --git a/drivers/cpufreq/cpufreq.c b/drivers/cpufreq/cpufreq.c
index 6a015ada5285..0937b8d6c2a4 100644
--- a/drivers/cpufreq/cpufreq.c
+++ b/drivers/cpufreq/cpufreq.c
@@ -312,11 +312,12 @@ static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
 	switch (state) {
 
 	case CPUFREQ_PRECHANGE:
-		if (WARN(policy->transition_ongoing,
+		if (WARN(policy->transition_ongoing ==
+					cpumask_weight(policy->cpus),
 				"In middle of another frequency transition\n"))
 			return;
 
-		policy->transition_ongoing = true;
+		policy->transition_ongoing++;
 
 		/* detect if the driver reported a value as "old frequency"
 		 * which is not equal to what the cpufreq core thinks is
@@ -341,7 +342,7 @@ static void __cpufreq_notify_transition(struct cpufreq_policy *policy,
 				"No frequency transition in progress\n"))
 			return;
 
-		policy->transition_ongoing = false;
+		policy->transition_ongoing--;
 
 		adjust_jiffies(CPUFREQ_POSTCHANGE, freqs);
 		pr_debug("FREQ: %lu - CPU: %lu", (unsigned long)freqs->new,
diff --git a/include/linux/cpufreq.h b/include/linux/cpufreq.h
index 4d7390bc1727..90d5a15120d5 100644
--- a/include/linux/cpufreq.h
+++ b/include/linux/cpufreq.h
@@ -119,7 +119,7 @@ struct cpufreq_policy {
 
 	struct kobject		kobj;
 	struct completion	kobj_unregister;
-	bool			transition_ongoing; /* Tracks transition status */
+	int			transition_ongoing; /* Tracks transition status */
 };
 
 #define CPUFREQ_ADJUST			(0)
-- 
cgit v1.2.3


From 91bdad0b6237c25a7bf8fd4604d0cc64a2005a23 Mon Sep 17 00:00:00 2001
From: Rafael J. Wysocki
Date: Thu, 4 Jul 2013 13:22:11 +0200
Subject: ACPI / PM: Fix corner case in acpi_bus_update_power()

The role of acpi_bus_update_power() is to update the given ACPI
device object's power.state field to reflect the current physical
state of the device (as inferred from the configuration of power
resources and _PSC, if available).  For this purpose it calls
acpi_device_set_power() that should update the power resources'
reference counters and set power.state as appropriate.  However,
that doesn't work if the "new" state is D1, D2 or D3hot and the
the current value of power.state means D3cold, because in that
case acpi_device_set_power() will refuse to transition the device
from D3cold to non-D0.

To address this problem, make acpi_bus_update_power() call
acpi_power_transition() directly to update the power resources'
reference counters and only use acpi_device_set_power() to put
the device into D0 if the current physical state of it cannot
be determined.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: 3.9+ <stable@vger.kernel.org>
---
 drivers/acpi/device_pm.c | 23 ++++++++++++++++++-----
 1 file changed, 18 insertions(+), 5 deletions(-)

diff --git a/drivers/acpi/device_pm.c b/drivers/acpi/device_pm.c
index e9e8bb24785b..4ab807dc8518 100644
--- a/drivers/acpi/device_pm.c
+++ b/drivers/acpi/device_pm.c
@@ -324,14 +324,27 @@ int acpi_bus_update_power(acpi_handle handle, int *state_p)
 	if (result)
 		return result;
 
-	if (state == ACPI_STATE_UNKNOWN)
+	if (state == ACPI_STATE_UNKNOWN) {
 		state = ACPI_STATE_D0;
-
-	result = acpi_device_set_power(device, state);
-	if (!result && state_p)
+		result = acpi_device_set_power(device, state);
+		if (result)
+			return result;
+	} else {
+		if (device->power.flags.power_resources) {
+			/*
+			 * We don't need to really switch the state, bu we need
+			 * to update the power resources' reference counters.
+			 */
+			result = acpi_power_transition(device, state);
+			if (result)
+				return result;
+		}
+		device->power.state = state;
+	}
+	if (state_p)
 		*state_p = state;
 
-	return result;
+	return 0;
 }
 EXPORT_SYMBOL_GPL(acpi_bus_update_power);
 
-- 
cgit v1.2.3


From 2ce65fe89153b6091393541de7e211d505436ff7 Mon Sep 17 00:00:00 2001
From: Rafael J. Wysocki
Date: Thu, 4 Jul 2013 13:25:04 +0200
Subject: ACPI / dock: Actually define acpi_dock_init() as void

Commit 94add0f (ACPI / dock: Initialize ACPI dock subsystem upfront)
changed the header of acpi_dock_init() in internal.h so that it is
supposed to be a void function now, but it forgot to update its
actual definition in dock.c according to which it still is supposed
to return int.

Although that didn't cause any visible breakage or even a compiler
warning to be thrown, which is odd enough, fix it.

Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Cc: 3.10+ <stable@vger.kernel.org>
---
 drivers/acpi/dock.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/drivers/acpi/dock.c b/drivers/acpi/dock.c
index 14de9f46972e..826560753389 100644
--- a/drivers/acpi/dock.c
+++ b/drivers/acpi/dock.c
@@ -1064,10 +1064,10 @@ find_dock_and_bay(acpi_handle handle, u32 lvl, void *context, void **rv)
 	return AE_OK;
 }
 
-int __init acpi_dock_init(void)
+void __init acpi_dock_init(void)
 {
 	if (acpi_disabled)
-		return 0;
+		return;
 
 	/* look for dock stations and bays */
 	acpi_walk_namespace(ACPI_TYPE_DEVICE, ACPI_ROOT_OBJECT,
@@ -1075,11 +1075,10 @@ int __init acpi_dock_init(void)
 
 	if (!dock_station_count) {
 		pr_info(PREFIX "No dock devices found.\n");
-		return 0;
+		return;
 	}
 
 	register_acpi_bus_notifier(&dock_acpi_notifier);
 	pr_info(PREFIX "%s: %d docks/bays found\n",
 		ACPI_DOCK_DRIVER_DESCRIPTION, dock_station_count);
-	return 0;
 }
-- 
cgit v1.2.3


From b67cf7c44c7e6b485f1c255ac3c1fe98cc99b677 Mon Sep 17 00:00:00 2001
From: Haicheng Li
Date: Thu, 4 Jul 2013 12:07:11 +0800
Subject: ACPI / scan: remove unused LIST_HEAD(acpi_device_list)

The acpi_device_list list is not used, so removed it.

[rjw: Changelog]
Signed-off-by: Haicheng Li <haicheng.li@linux.intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/acpi/scan.c | 1 -
 1 file changed, 1 deletion(-)

diff --git a/drivers/acpi/scan.c b/drivers/acpi/scan.c
index dfe76f17cfc4..10985573aaa7 100644
--- a/drivers/acpi/scan.c
+++ b/drivers/acpi/scan.c
@@ -35,7 +35,6 @@ bool acpi_force_hot_remove;
 
 static const char *dummy_hid = "device";
 
-static LIST_HEAD(acpi_device_list);
 static LIST_HEAD(acpi_bus_id_list);
 static DEFINE_MUTEX(acpi_scan_lock);
 static LIST_HEAD(acpi_scan_handlers_list);
-- 
cgit v1.2.3


From 85eb98274fd81c44e721c275ced54661ee9d5b05 Mon Sep 17 00:00:00 2001
From: Naresh Bhat
Date: Mon, 1 Jul 2013 19:51:00 +0530
Subject: ACPI / fan: Initialize acpi_state variable

Make the following compiler warning go away:

 CC      drivers/acpi/fan.o
drivers/acpi/fan.c: In function ‘fan_get_cur_state’:
drivers/acpi/fan.c:96:9: warning: ‘acpi_state’ may be used uninitialized in this function [-Wuninitialized]

by initializing the local variable acpi_state in fan_get_cur_state().

[rjw: Changelog]
Signed-off-by: Naresh Bhat <naresh.bhat@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/acpi/fan.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/acpi/fan.c b/drivers/acpi/fan.c
index 8d1c0105e113..5b02a0aa540c 100644
--- a/drivers/acpi/fan.c
+++ b/drivers/acpi/fan.c
@@ -84,7 +84,7 @@ static int fan_get_cur_state(struct thermal_cooling_device *cdev, unsigned long
 {
 	struct acpi_device *device = cdev->devdata;
 	int result;
-	int acpi_state;
+	int acpi_state = ACPI_STATE_D0;
 
 	if (!device)
 		return -EINVAL;
-- 
cgit v1.2.3


From f605181abd95a109031a23c67a824eb8e5dcfe67 Mon Sep 17 00:00:00 2001
From: Thomas Renninger
Date: Fri, 28 Jun 2013 15:34:29 +0200
Subject: cpupower: Make idlestate usage unsigned

Use unsigned int as the data type for some variables related to CPU
idle states which allows the code to be simplified slightly.

[rjw: Changelog]
Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 tools/power/cpupower/utils/cpuidle-info.c  | 18 ++++--------------
 tools/power/cpupower/utils/helpers/sysfs.c |  2 +-
 tools/power/cpupower/utils/helpers/sysfs.h |  2 +-
 3 files changed, 6 insertions(+), 16 deletions(-)

diff --git a/tools/power/cpupower/utils/cpuidle-info.c b/tools/power/cpupower/utils/cpuidle-info.c
index 8145af5f93a6..edd5dba3e249 100644
--- a/tools/power/cpupower/utils/cpuidle-info.c
+++ b/tools/power/cpupower/utils/cpuidle-info.c
@@ -22,7 +22,7 @@
 
 static void cpuidle_cpu_output(unsigned int cpu, int verbose)
 {
-	int idlestates, idlestate;
+	unsigned int idlestates, idlestate;
 	char *tmp;
 
 	printf(_ ("Analyzing CPU %d:\n"), cpu);
@@ -31,10 +31,8 @@ static void cpuidle_cpu_output(unsigned int cpu, int verbose)
 	if (idlestates == 0) {
 		printf(_("CPU %u: No idle states\n"), cpu);
 		return;
-	} else if (idlestates <= 0) {
-		printf(_("CPU %u: Can't read idle state info\n"), cpu);
-		return;
 	}
+
 	printf(_("Number of idle states: %d\n"), idlestates);
 	printf(_("Available idle states:"));
 	for (idlestate = 0; idlestate < idlestates; idlestate++) {
@@ -98,21 +96,13 @@ static void cpuidle_general_output(void)
 static void proc_cpuidle_cpu_output(unsigned int cpu)
 {
 	long max_allowed_cstate = 2000000000;
-	int cstates, cstate;
+	unsigned int cstate, cstates;
 
 	cstates = sysfs_get_idlestate_count(cpu);
 	if (cstates == 0) {
-		/*
-		 * Go on and print same useless info as you'd see with
-		 * cat /proc/acpi/processor/../power
-		 *	printf(_("CPU %u: No C-states available\n"), cpu);
-		 *	return;
-		 */
-	} else if (cstates <= 0) {
-		printf(_("CPU %u: Can't read C-state info\n"), cpu);
+		printf(_("CPU %u: No C-states info\n"), cpu);
 		return;
 	}
-	/* printf("Cstates: %d\n", cstates); */
 
 	printf(_("active state:            C0\n"));
 	printf(_("max_cstate:              C%u\n"), cstates-1);
diff --git a/tools/power/cpupower/utils/helpers/sysfs.c b/tools/power/cpupower/utils/helpers/sysfs.c
index 38ab91629463..891f6710d026 100644
--- a/tools/power/cpupower/utils/helpers/sysfs.c
+++ b/tools/power/cpupower/utils/helpers/sysfs.c
@@ -238,7 +238,7 @@ char *sysfs_get_idlestate_desc(unsigned int cpu, unsigned int idlestate)
  * Negativ in error case
  * Zero if cpuidle does not export any C-states
  */
-int sysfs_get_idlestate_count(unsigned int cpu)
+unsigned int sysfs_get_idlestate_count(unsigned int cpu)
 {
 	char file[SYSFS_PATH_MAX];
 	struct stat statbuf;
diff --git a/tools/power/cpupower/utils/helpers/sysfs.h b/tools/power/cpupower/utils/helpers/sysfs.h
index 8cb797bbceb0..0401a97a4cab 100644
--- a/tools/power/cpupower/utils/helpers/sysfs.h
+++ b/tools/power/cpupower/utils/helpers/sysfs.h
@@ -19,7 +19,7 @@ extern char *sysfs_get_idlestate_name(unsigned int cpu,
 				unsigned int idlestate);
 extern char *sysfs_get_idlestate_desc(unsigned int cpu,
 				unsigned int idlestate);
-extern int sysfs_get_idlestate_count(unsigned int cpu);
+extern unsigned int sysfs_get_idlestate_count(unsigned int cpu);
 
 extern char *sysfs_get_cpuidle_governor(void);
 extern char *sysfs_get_cpuidle_driver(void);
-- 
cgit v1.2.3


From 0924c369bc5492cf181a066fc2d459aa18ffa5ac Mon Sep 17 00:00:00 2001
From: Thomas Renninger
Date: Fri, 28 Jun 2013 15:34:30 +0200
Subject: cpupower: Implement disabling of cstate interface

Latest kernel allows to disable C-states via:
/sys/devices/system/cpu/cpuX/cpuidle/stateY/disable

This patch provides lower level sysfs access functions to make use of
this interface.  A later patch will implement the higher level stuff.

Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 tools/power/cpupower/utils/helpers/sysfs.c | 116 ++++++++++++++++++++++++++++-
 tools/power/cpupower/utils/helpers/sysfs.h |   8 ++
 2 files changed, 123 insertions(+), 1 deletion(-)

diff --git a/tools/power/cpupower/utils/helpers/sysfs.c b/tools/power/cpupower/utils/helpers/sysfs.c
index 891f6710d026..5cdc600e8152 100644
--- a/tools/power/cpupower/utils/helpers/sysfs.c
+++ b/tools/power/cpupower/utils/helpers/sysfs.c
@@ -87,8 +87,35 @@ int sysfs_is_cpu_online(unsigned int cpu)
 	return value;
 }
 
+/* CPUidle idlestate specific /sys/devices/system/cpu/cpuX/cpuidle/ access */
+
+
 /* CPUidle idlestate specific /sys/devices/system/cpu/cpuX/cpuidle/ access */
 
+/*
+ * helper function to check whether a file under "../cpuX/cpuidle/stateX/" dir
+ * exists.
+ * For example the functionality to disable c-states was introduced in later
+ * kernel versions, this function can be used to explicitly check for this
+ * feature.
+ *
+ * returns 1 if the file exists, 0 otherwise.
+ */
+unsigned int sysfs_idlestate_file_exists(unsigned int cpu,
+					 unsigned int idlestate,
+					 const char *fname)
+{
+	char path[SYSFS_PATH_MAX];
+	struct stat statbuf;
+
+
+	snprintf(path, sizeof(path), PATH_TO_CPU "cpu%u/cpuidle/state%u/%s",
+		 cpu, idlestate, fname);
+	if (stat(path, &statbuf) != 0)
+		return 0;
+	return 1;
+}
+
 /*
  * helper function to read file from /sys into given buffer
  * fname is a relative path under "cpuX/cpuidle/stateX/" dir
@@ -121,6 +148,40 @@ unsigned int sysfs_idlestate_read_file(unsigned int cpu, unsigned int idlestate,
 	return (unsigned int) numread;
 }
 
+/* 
+ * helper function to write a new value to a /sys file
+ * fname is a relative path under "../cpuX/cpuidle/cstateY/" dir
+ *
+ * Returns the number of bytes written or 0 on error
+ */
+static
+unsigned int sysfs_idlestate_write_file(unsigned int cpu,
+					unsigned int idlestate,
+					const char *fname,
+					const char *value, size_t len)
+{
+	char path[SYSFS_PATH_MAX];
+	int fd;
+	ssize_t numwrite;
+
+	snprintf(path, sizeof(path), PATH_TO_CPU "cpu%u/cpuidle/state%u/%s",
+		 cpu, idlestate, fname);
+
+	fd = open(path, O_WRONLY);
+	if (fd == -1)
+		return 0;
+
+	numwrite = write(fd, value, len);
+	if (numwrite < 1) {
+		close(fd);
+		return 0;
+	}
+
+	close(fd);
+
+	return (unsigned int) numwrite;
+}
+
 /* read access to files which contain one numeric value */
 
 enum idlestate_value {
@@ -128,6 +189,7 @@ enum idlestate_value {
 	IDLESTATE_POWER,
 	IDLESTATE_LATENCY,
 	IDLESTATE_TIME,
+	IDLESTATE_DISABLE,
 	MAX_IDLESTATE_VALUE_FILES
 };
 
@@ -136,6 +198,7 @@ static const char *idlestate_value_files[MAX_IDLESTATE_VALUE_FILES] = {
 	[IDLESTATE_POWER] = "power",
 	[IDLESTATE_LATENCY] = "latency",
 	[IDLESTATE_TIME]  = "time",
+	[IDLESTATE_DISABLE]  = "disable",
 };
 
 static unsigned long long sysfs_idlestate_get_one_value(unsigned int cpu,
@@ -205,8 +268,59 @@ static char *sysfs_idlestate_get_one_string(unsigned int cpu,
 	return result;
 }
 
+/*
+ * Returns:
+ *    1  if disabled
+ *    0  if enabled
+ *    -1 if idlestate is not available
+ *    -2 if disabling is not supported by the kernel
+ */
+int sysfs_is_idlestate_disabled(unsigned int cpu,
+				unsigned int idlestate)
+{
+	if (sysfs_get_idlestate_count(cpu) < idlestate)
+		return -1;
+
+	if (!sysfs_idlestate_file_exists(cpu, idlestate,
+				 idlestate_value_files[IDLESTATE_DISABLE]))
+		return -2;
+	return sysfs_idlestate_get_one_value(cpu, idlestate, IDLESTATE_DISABLE);
+}
+
+/*
+ * Pass 1 as last argument to disable or 0 to enable the state
+ * Returns:
+ *    0  on success
+ *    negative values on error, for example:
+ *      -1 if idlestate is not available
+ *      -2 if disabling is not supported by the kernel
+ *      -3 No write access to disable/enable C-states
+ */
+int sysfs_idlestate_disable(unsigned int cpu,
+			    unsigned int idlestate,
+			    unsigned int disable)
+{
+	char value[SYSFS_PATH_MAX];
+	int bytes_written;
+
+	if (sysfs_get_idlestate_count(cpu) < idlestate)
+		return -1;
+
+	if (!sysfs_idlestate_file_exists(cpu, idlestate,
+				 idlestate_value_files[IDLESTATE_DISABLE]))
+		return -2;
+
+	snprintf(value, SYSFS_PATH_MAX, "%u", disable);
+
+	bytes_written = sysfs_idlestate_write_file(cpu, idlestate, "disable",
+						   value, sizeof(disable));
+	if (bytes_written)
+		return 0;
+	return -3;
+}
+
 unsigned long sysfs_get_idlestate_latency(unsigned int cpu,
-					unsigned int idlestate)
+					  unsigned int idlestate)
 {
 	return sysfs_idlestate_get_one_value(cpu, idlestate, IDLESTATE_LATENCY);
 }
diff --git a/tools/power/cpupower/utils/helpers/sysfs.h b/tools/power/cpupower/utils/helpers/sysfs.h
index 0401a97a4cab..d28f11fedbda 100644
--- a/tools/power/cpupower/utils/helpers/sysfs.h
+++ b/tools/power/cpupower/utils/helpers/sysfs.h
@@ -7,8 +7,16 @@
 
 extern unsigned int sysfs_read_file(const char *path, char *buf, size_t buflen);
 
+extern unsigned int sysfs_idlestate_file_exists(unsigned int cpu,
+						unsigned int idlestate,
+						const char *fname);
+
 extern int sysfs_is_cpu_online(unsigned int cpu);
 
+extern int sysfs_is_idlestate_disabled(unsigned int cpu,
+				       unsigned int idlestate);
+extern int sysfs_idlestate_disable(unsigned int cpu, unsigned int idlestate,
+				   unsigned int disable);
 extern unsigned long sysfs_get_idlestate_latency(unsigned int cpu,
 						unsigned int idlestate);
 extern unsigned long sysfs_get_idlestate_usage(unsigned int cpu,
-- 
cgit v1.2.3


From c4f3610eba69321b9cf35779cd67e68b5138cc16 Mon Sep 17 00:00:00 2001
From: Thomas Renninger
Date: Fri, 28 Jun 2013 15:34:31 +0200
Subject: cpupower: Introduce idle-set subcommand and C-state
 enabling/disabling

Example:

cpupower idle-set -d 3

will disable C-state 3 on all processors (set commands are active on
all CPUs by default), same as:

cpupower -c all idle-set -d 3

Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 tools/power/cpupower/Makefile               |   3 +-
 tools/power/cpupower/man/cpupower-monitor.1 |  10 ++-
 tools/power/cpupower/utils/builtin.h        |   1 +
 tools/power/cpupower/utils/cpuidle-info.c   |   6 +-
 tools/power/cpupower/utils/cpuidle-set.c    | 118 ++++++++++++++++++++++++++++
 tools/power/cpupower/utils/cpupower.c       |  13 +--
 6 files changed, 142 insertions(+), 9 deletions(-)
 create mode 100644 tools/power/cpupower/utils/cpuidle-set.c

diff --git a/tools/power/cpupower/Makefile b/tools/power/cpupower/Makefile
index d875a74a3bdf..ce17f30f04ba 100644
--- a/tools/power/cpupower/Makefile
+++ b/tools/power/cpupower/Makefile
@@ -131,7 +131,8 @@ UTIL_OBJS =  utils/helpers/amd.o utils/helpers/topology.o utils/helpers/msr.o \
 	utils/idle_monitor/amd_fam14h_idle.o utils/idle_monitor/cpuidle_sysfs.o \
 	utils/idle_monitor/mperf_monitor.o utils/idle_monitor/cpupower-monitor.o \
 	utils/cpupower.o utils/cpufreq-info.o utils/cpufreq-set.o \
-	utils/cpupower-set.o utils/cpupower-info.o utils/cpuidle-info.o
+	utils/cpupower-set.o utils/cpupower-info.o utils/cpuidle-info.o \
+	utils/cpuidle-set.o
 
 UTIL_SRC := $(UTIL_OBJS:.o=.c)
 
diff --git a/tools/power/cpupower/man/cpupower-monitor.1 b/tools/power/cpupower/man/cpupower-monitor.1
index e01c35d13b6e..914cbb9d9cd0 100644
--- a/tools/power/cpupower/man/cpupower-monitor.1
+++ b/tools/power/cpupower/man/cpupower-monitor.1
@@ -110,13 +110,21 @@ May work poorly on Linux-2.6.20 through 2.6.29, as the \fBacpi-cpufreq \fP
 kernel frequency driver periodically cleared aperf/mperf registers in those
 kernels.
 
-.SS "Nehalem" "SandyBridge"
+.SS "Nehalem" "SandyBridge" "HaswellExtended"
 Intel Core and Package sleep state counters.
 Threads (hyperthreaded cores) may not be able to enter deeper core states if
 its sibling is utilized.
 Deepest package sleep states may in reality show up as machine/platform wide
 sleep states and can only be entered if all cores are idle. Look up Intel
 manuals (some are provided in the References section) for further details.
+The monitors are named after the CPU family where the sleep state capabilities
+got introduced and may not match exactly the CPU name of the platform.
+For example an IvyBridge processor has sleep state capabilities which got
+introduced in Nehalem and SandyBridge processor families.
+Thus on an IvyBridge processor one will get Nehalem and SandyBridge sleep
+state monitors.
+HaswellExtended extra package sleep state capabilities are available only in a
+specific Haswell (family 0x45) and probably also other future processors.
 
 .SS "Fam_12h" "Fam_14h"
 AMD laptop and desktop processor (family 12h and 14h) sleep state counters.
diff --git a/tools/power/cpupower/utils/builtin.h b/tools/power/cpupower/utils/builtin.h
index c10496fbe3c6..2284c8ea4e2a 100644
--- a/tools/power/cpupower/utils/builtin.h
+++ b/tools/power/cpupower/utils/builtin.h
@@ -5,6 +5,7 @@ extern int cmd_set(int argc, const char **argv);
 extern int cmd_info(int argc, const char **argv);
 extern int cmd_freq_set(int argc, const char **argv);
 extern int cmd_freq_info(int argc, const char **argv);
+extern int cmd_idle_set(int argc, const char **argv);
 extern int cmd_idle_info(int argc, const char **argv);
 extern int cmd_monitor(int argc, const char **argv);
 
diff --git a/tools/power/cpupower/utils/cpuidle-info.c b/tools/power/cpupower/utils/cpuidle-info.c
index edd5dba3e249..75e66de7e7a7 100644
--- a/tools/power/cpupower/utils/cpuidle-info.c
+++ b/tools/power/cpupower/utils/cpuidle-info.c
@@ -48,10 +48,14 @@ static void cpuidle_cpu_output(unsigned int cpu, int verbose)
 		return;
 
 	for (idlestate = 0; idlestate < idlestates; idlestate++) {
+		int disabled = sysfs_is_idlestate_disabled(cpu, idlestate);
+		/* Disabled interface not supported on older kernels */
+		if (disabled < 0)
+			disabled = 0;
 		tmp = sysfs_get_idlestate_name(cpu, idlestate);
 		if (!tmp)
 			continue;
-		printf("%s:\n", tmp);
+		printf("%s%s:\n", tmp, (disabled) ? " (DISABLED) " : "");
 		free(tmp);
 
 		tmp = sysfs_get_idlestate_desc(cpu, idlestate);
diff --git a/tools/power/cpupower/utils/cpuidle-set.c b/tools/power/cpupower/utils/cpuidle-set.c
new file mode 100644
index 000000000000..c78141c5dfac
--- /dev/null
+++ b/tools/power/cpupower/utils/cpuidle-set.c
@@ -0,0 +1,118 @@
+#include <unistd.h>
+#include <stdio.h>
+#include <errno.h>
+#include <stdlib.h>
+#include <limits.h>
+#include <string.h>
+#include <ctype.h>
+
+#include <getopt.h>
+
+#include "cpufreq.h"
+#include "helpers/helpers.h"
+#include "helpers/sysfs.h"
+
+static struct option info_opts[] = {
+	{ .name = "disable",	.has_arg = required_argument,	.flag = NULL,	.val = 'd'},
+	{ .name = "enable",	.has_arg = required_argument,	.flag = NULL,	.val = 'e'},
+	{ },
+};
+
+
+int cmd_idle_set(int argc, char **argv)
+{
+	extern char *optarg;
+	extern int optind, opterr, optopt;
+	int ret = 0, cont = 1, param = 0, idlestate = 0;
+	unsigned int cpu = 0;
+
+	do {
+		ret = getopt_long(argc, argv, "d:e:", info_opts, NULL);
+		if (ret == -1)
+			break;
+		switch (ret) {
+		case '?':
+			param = '?';
+			cont = 0;
+			break;
+		case 'd':
+			if (param) {
+				param = -1;
+				cont = 0;
+				break;
+			}
+			param = ret;
+			idlestate = atoi(optarg);
+			break;
+		case 'e':
+			if (param) {
+				param = -1;
+				cont = 0;
+				break;
+			}
+			param = ret;
+			idlestate = atoi(optarg);
+			break;
+		case -1:
+			cont = 0;
+			break;
+		}
+	} while (cont);
+
+	switch (param) {
+	case -1:
+		printf(_("You can't specify more than one "
+			 "output-specific argument\n"));
+		exit(EXIT_FAILURE);
+	case '?':
+		printf(_("invalid or unknown argument\n"));
+		exit(EXIT_FAILURE);
+	}
+
+	/* Default is: set all CPUs */
+	if (bitmask_isallclear(cpus_chosen))
+		bitmask_setall(cpus_chosen);
+
+	for (cpu = bitmask_first(cpus_chosen);
+	     cpu <= bitmask_last(cpus_chosen); cpu++) {
+
+		if (!bitmask_isbitset(cpus_chosen, cpu))
+			continue;
+
+		switch (param) {
+
+		case 'd':
+			ret = sysfs_idlestate_disable(cpu, idlestate, 1);
+			if (ret == 0)
+		printf(_("Idlestate %u disabled on CPU %u\n"),  idlestate, cpu);
+			else if (ret == -1)
+		printf(_("Idlestate %u not available on CPU %u\n"),
+		       idlestate, cpu);
+			else if (ret == -2)
+		printf(_("Idlestate disabling not supported by kernel\n"));
+			else
+		printf(_("Idlestate %u not disabled on CPU %u\n"),
+		       idlestate, cpu);
+			break;
+		case 'e':
+			ret = sysfs_idlestate_disable(cpu, idlestate, 0);
+			if (ret == 0)
+		printf(_("Idlestate %u enabled on CPU %u\n"),  idlestate, cpu);
+			else if (ret == -1)
+		printf(_("Idlestate %u not available on CPU %u\n"),
+		       idlestate, cpu);
+			else if (ret == -2)
+		printf(_("Idlestate enabling not supported by kernel\n"));
+			else
+		printf(_("Idlestate %u not enabled on CPU %u\n"),
+		       idlestate, cpu);
+			break;
+		default:
+			/* Not reachable with proper args checking */
+			printf(_("Invalid or unknown argument\n"));
+			exit(EXIT_FAILURE);
+			break;
+		}
+	}
+	return EXIT_SUCCESS;
+}
diff --git a/tools/power/cpupower/utils/cpupower.c b/tools/power/cpupower/utils/cpupower.c
index 52bee591c1c5..7efc570ffbaa 100644
--- a/tools/power/cpupower/utils/cpupower.c
+++ b/tools/power/cpupower/utils/cpupower.c
@@ -17,12 +17,6 @@
 #include "helpers/helpers.h"
 #include "helpers/bitmask.h"
 
-struct cmd_struct {
-	const char *cmd;
-	int (*main)(int, const char **);
-	int needs_root;
-};
-
 #define ARRAY_SIZE(x) (sizeof(x)/sizeof(x[0]))
 
 static int cmd_help(int argc, const char **argv);
@@ -43,10 +37,17 @@ int be_verbose;
 
 static void print_help(void);
 
+struct cmd_struct {
+	const char *cmd;
+	int (*main)(int, const char **);
+	int needs_root;
+};
+
 static struct cmd_struct commands[] = {
 	{ "frequency-info",	cmd_freq_info,	0	},
 	{ "frequency-set",	cmd_freq_set,	1	},
 	{ "idle-info",		cmd_idle_info,	0	},
+	{ "idle-set",		cmd_idle_set,	1	},
 	{ "set",		cmd_set,	1	},
 	{ "info",		cmd_info,	0	},
 	{ "monitor",		cmd_monitor,	0	},
-- 
cgit v1.2.3


From 2aa1ca75c48f1da2c9bdc81ec4403eaf89a410d5 Mon Sep 17 00:00:00 2001
From: Thomas Renninger
Date: Fri, 28 Jun 2013 15:34:32 +0200
Subject: cpupower: Haswell also supports the C-states introduced with
 SandyBridge

Add Haswell model numbers to snb_register() as it also supports the
C-states introduced in SandyBridge processors.

[rjw: Changelog]
Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 tools/power/cpupower/utils/idle_monitor/snb_idle.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/tools/power/cpupower/utils/idle_monitor/snb_idle.c b/tools/power/cpupower/utils/idle_monitor/snb_idle.c
index a99b43b97d6d..efc8a69c9aba 100644
--- a/tools/power/cpupower/utils/idle_monitor/snb_idle.c
+++ b/tools/power/cpupower/utils/idle_monitor/snb_idle.c
@@ -155,6 +155,10 @@ static struct cpuidle_monitor *snb_register(void)
 	case 0x2D: /* SNB Xeon */
 	case 0x3A: /* IVB */
 	case 0x3E: /* IVB Xeon */
+	case 0x3C: /* HSW */
+	case 0x3F: /* HSW */
+	case 0x45: /* HSW */
+	case 0x46: /* HSW */
 		break;
 	default:
 		return NULL;
-- 
cgit v1.2.3


From 7ee767b69b6885dd81bafaf1881c5028033a6177 Mon Sep 17 00:00:00 2001
From: Thomas Renninger
Date: Fri, 28 Jun 2013 15:34:33 +0200
Subject: cpupower: Add Haswell family 0x45 specific idle monitor to show
 PC8,9,10 states

This specific processor supports 3 new package sleep states.
Provide a monitor, so that the user can see their usage.

Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 tools/power/cpupower/Makefile                      |   1 +
 .../cpupower/utils/idle_monitor/hsw_ext_idle.c     | 196 +++++++++++++++++++++
 .../cpupower/utils/idle_monitor/idle_monitors.def  |   1 +
 3 files changed, 198 insertions(+)
 create mode 100644 tools/power/cpupower/utils/idle_monitor/hsw_ext_idle.c

diff --git a/tools/power/cpupower/Makefile b/tools/power/cpupower/Makefile
index ce17f30f04ba..cbfec92af327 100644
--- a/tools/power/cpupower/Makefile
+++ b/tools/power/cpupower/Makefile
@@ -128,6 +128,7 @@ UTIL_OBJS =  utils/helpers/amd.o utils/helpers/topology.o utils/helpers/msr.o \
 	utils/helpers/sysfs.o utils/helpers/misc.o utils/helpers/cpuid.o \
 	utils/helpers/pci.o utils/helpers/bitmask.o \
 	utils/idle_monitor/nhm_idle.o utils/idle_monitor/snb_idle.o \
+	utils/idle_monitor/hsw_ext_idle.o \
 	utils/idle_monitor/amd_fam14h_idle.o utils/idle_monitor/cpuidle_sysfs.o \
 	utils/idle_monitor/mperf_monitor.o utils/idle_monitor/cpupower-monitor.o \
 	utils/cpupower.o utils/cpufreq-info.o utils/cpufreq-set.o \
diff --git a/tools/power/cpupower/utils/idle_monitor/hsw_ext_idle.c b/tools/power/cpupower/utils/idle_monitor/hsw_ext_idle.c
new file mode 100644
index 000000000000..ebeaba6571a3
--- /dev/null
+++ b/tools/power/cpupower/utils/idle_monitor/hsw_ext_idle.c
@@ -0,0 +1,196 @@
+/*
+ *  (C) 2010,2011       Thomas Renninger <trenn@suse.de>, Novell Inc.
+ *
+ *  Licensed under the terms of the GNU GPL License version 2.
+ *
+ *  Based on SandyBridge monitor. Implements the new package C-states
+ *  (PC8, PC9, PC10) coming with a specific Haswell (family 0x45) CPU.
+ */
+
+#if defined(__i386__) || defined(__x86_64__)
+
+#include <stdio.h>
+#include <stdint.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "helpers/helpers.h"
+#include "idle_monitor/cpupower-monitor.h"
+
+#define MSR_PKG_C8_RESIDENCY           0x00000630
+#define MSR_PKG_C9_RESIDENCY           0x00000631
+#define MSR_PKG_C10_RESIDENCY          0x00000632
+
+#define MSR_TSC	0x10
+
+enum intel_hsw_ext_id { PC8 = 0, PC9, PC10, HSW_EXT_CSTATE_COUNT,
+			TSC = 0xFFFF };
+
+static int hsw_ext_get_count_percent(unsigned int self_id, double *percent,
+				 unsigned int cpu);
+
+static cstate_t hsw_ext_cstates[HSW_EXT_CSTATE_COUNT] = {
+	{
+		.name			= "PC8",
+		.desc			= N_("Processor Package C8"),
+		.id			= PC8,
+		.range			= RANGE_PACKAGE,
+		.get_count_percent	= hsw_ext_get_count_percent,
+	},
+	{
+		.name			= "PC9",
+		.desc			= N_("Processor Package C9"),
+		.desc			= N_("Processor Package C2"),
+		.id			= PC9,
+		.range			= RANGE_PACKAGE,
+		.get_count_percent	= hsw_ext_get_count_percent,
+	},
+	{
+		.name			= "PC10",
+		.desc			= N_("Processor Package C10"),
+		.id			= PC10,
+		.range			= RANGE_PACKAGE,
+		.get_count_percent	= hsw_ext_get_count_percent,
+	},
+};
+
+static unsigned long long tsc_at_measure_start;
+static unsigned long long tsc_at_measure_end;
+static unsigned long long *previous_count[HSW_EXT_CSTATE_COUNT];
+static unsigned long long *current_count[HSW_EXT_CSTATE_COUNT];
+/* valid flag for all CPUs. If a MSR read failed it will be zero */
+static int *is_valid;
+
+static int hsw_ext_get_count(enum intel_hsw_ext_id id, unsigned long long *val,
+			unsigned int cpu)
+{
+	int msr;
+
+	switch (id) {
+	case PC8:
+		msr = MSR_PKG_C8_RESIDENCY;
+		break;
+	case PC9:
+		msr = MSR_PKG_C9_RESIDENCY;
+		break;
+	case PC10:
+		msr = MSR_PKG_C10_RESIDENCY;
+		break;
+	case TSC:
+		msr = MSR_TSC;
+		break;
+	default:
+		return -1;
+	};
+	if (read_msr(cpu, msr, val))
+		return -1;
+	return 0;
+}
+
+static int hsw_ext_get_count_percent(unsigned int id, double *percent,
+				 unsigned int cpu)
+{
+	*percent = 0.0;
+
+	if (!is_valid[cpu])
+		return -1;
+
+	*percent = (100.0 *
+		(current_count[id][cpu] - previous_count[id][cpu])) /
+		(tsc_at_measure_end - tsc_at_measure_start);
+
+	dprint("%s: previous: %llu - current: %llu - (%u)\n",
+		hsw_ext_cstates[id].name, previous_count[id][cpu],
+		current_count[id][cpu], cpu);
+
+	dprint("%s: tsc_diff: %llu - count_diff: %llu - percent: %2.f (%u)\n",
+	       hsw_ext_cstates[id].name,
+	       (unsigned long long) tsc_at_measure_end - tsc_at_measure_start,
+	       current_count[id][cpu] - previous_count[id][cpu],
+	       *percent, cpu);
+
+	return 0;
+}
+
+static int hsw_ext_start(void)
+{
+	int num, cpu;
+	unsigned long long val;
+
+	for (num = 0; num < HSW_EXT_CSTATE_COUNT; num++) {
+		for (cpu = 0; cpu < cpu_count; cpu++) {
+			hsw_ext_get_count(num, &val, cpu);
+			previous_count[num][cpu] = val;
+		}
+	}
+	hsw_ext_get_count(TSC, &tsc_at_measure_start, 0);
+	return 0;
+}
+
+static int hsw_ext_stop(void)
+{
+	unsigned long long val;
+	int num, cpu;
+
+	hsw_ext_get_count(TSC, &tsc_at_measure_end, 0);
+
+	for (num = 0; num < HSW_EXT_CSTATE_COUNT; num++) {
+		for (cpu = 0; cpu < cpu_count; cpu++) {
+			is_valid[cpu] = !hsw_ext_get_count(num, &val, cpu);
+			current_count[num][cpu] = val;
+		}
+	}
+	return 0;
+}
+
+struct cpuidle_monitor intel_hsw_ext_monitor;
+
+static struct cpuidle_monitor *hsw_ext_register(void)
+{
+	int num;
+
+	if (cpupower_cpu_info.vendor != X86_VENDOR_INTEL
+	    || cpupower_cpu_info.family != 6)
+		return NULL;
+
+	switch (cpupower_cpu_info.model) {
+	case 0x45: /* HSW */
+		break;
+	default:
+		return NULL;
+	}
+
+	is_valid = calloc(cpu_count, sizeof(int));
+	for (num = 0; num < HSW_EXT_CSTATE_COUNT; num++) {
+		previous_count[num] = calloc(cpu_count,
+					sizeof(unsigned long long));
+		current_count[num]  = calloc(cpu_count,
+					sizeof(unsigned long long));
+	}
+	intel_hsw_ext_monitor.name_len = strlen(intel_hsw_ext_monitor.name);
+	return &intel_hsw_ext_monitor;
+}
+
+void hsw_ext_unregister(void)
+{
+	int num;
+	free(is_valid);
+	for (num = 0; num < HSW_EXT_CSTATE_COUNT; num++) {
+		free(previous_count[num]);
+		free(current_count[num]);
+	}
+}
+
+struct cpuidle_monitor intel_hsw_ext_monitor = {
+	.name			= "HaswellExtended",
+	.hw_states		= hsw_ext_cstates,
+	.hw_states_num		= HSW_EXT_CSTATE_COUNT,
+	.start			= hsw_ext_start,
+	.stop			= hsw_ext_stop,
+	.do_register		= hsw_ext_register,
+	.unregister		= hsw_ext_unregister,
+	.needs_root		= 1,
+	.overflow_s		= 922000000 /* 922337203 seconds TSC overflow
+					       at 20GHz */
+};
+#endif /* defined(__i386__) || defined(__x86_64__) */
diff --git a/tools/power/cpupower/utils/idle_monitor/idle_monitors.def b/tools/power/cpupower/utils/idle_monitor/idle_monitors.def
index e3f8d9b2b18f..0d6ba4dbb9c7 100644
--- a/tools/power/cpupower/utils/idle_monitor/idle_monitors.def
+++ b/tools/power/cpupower/utils/idle_monitor/idle_monitors.def
@@ -2,6 +2,7 @@
 DEF(amd_fam14h)
 DEF(intel_nhm)
 DEF(intel_snb)
+DEF(intel_hsw_ext)
 DEF(mperf)
 #endif
 DEF(cpuidle_sysfs)
-- 
cgit v1.2.3


From 10a0b6176b9f8b026ce07acd8f755297653c443c Mon Sep 17 00:00:00 2001
From: Mika Westerberg
Date: Fri, 5 Jul 2013 12:15:56 +0300
Subject: ACPI / power: add missing newline to debug messages

There are few places in power.c where debug messages have no newline
at the end.  Reading such debug messages from dmesg is not fun, so
fix this by adding the missing newlines.

Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Cc: All <stable@vger.kernel.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
---
 drivers/acpi/power.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/acpi/power.c b/drivers/acpi/power.c
index 288bb270f8ed..5c28c894c0fc 100644
--- a/drivers/acpi/power.c
+++ b/drivers/acpi/power.c
@@ -279,7 +279,7 @@ static int acpi_power_on_unlocked(struct acpi_power_resource *resource)
 
 	if (resource->ref_count++) {
 		ACPI_DEBUG_PRINT((ACPI_DB_INFO,
-				  "Power resource [%s] already on",
+				  "Power resource [%s] already on\n",
 				  resource->name));
 	} else {
 		result = __acpi_power_on(resource);
@@ -325,7 +325,7 @@ static int acpi_power_off_unlocked(struct acpi_power_resource *resource)
 
 	if (!resource->ref_count) {
 		ACPI_DEBUG_PRINT((ACPI_DB_INFO,
-				  "Power resource [%s] already off",
+				  "Power resource [%s] already off\n",
 				  resource->name));
 		return 0;
 	}
-- 
cgit v1.2.3