diff mbox

[v3,2/3] thermal: Add devfreq cooling

Message ID 1438096672-10867-3-git-send-email-javi.merino@arm.com (mailing list archive)
State Not Applicable, archived
Headers show

Commit Message

Javi Merino July 28, 2015, 3:17 p.m. UTC
From: Ørjan Eide <orjan.eide@arm.com>

Add a generic thermal cooling device for devfreq, that is similar to
cpu_cooling.

The device must use devfreq.  In order to use the power extension of the
cooling device, it must have registered its OPPs using the OPP library.

Cc: Zhang Rui <rui.zhang@intel.com>
Cc: Eduardo Valentin <edubezval@gmail.com>
Signed-off-by: Javi Merino <javi.merino@arm.com>
Signed-off-by: Ørjan Eide <orjan.eide@arm.com>
---
Changes since v2:
  - Use the opp library to enable/disable opps instead of exporting
    devfreq_set_{max,min}
  - Use idr API to get a unique name for each devfreq cooling device
    registered.

 drivers/thermal/Kconfig           |  10 +
 drivers/thermal/Makefile          |   3 +
 drivers/thermal/devfreq_cooling.c | 466 ++++++++++++++++++++++++++++++++++++++
 include/linux/devfreq_cooling.h   |  93 ++++++++
 4 files changed, 572 insertions(+)
 create mode 100644 drivers/thermal/devfreq_cooling.c
 create mode 100644 include/linux/devfreq_cooling.h
diff mbox

Patch

diff --git a/drivers/thermal/Kconfig b/drivers/thermal/Kconfig
index 118938ee8552..da339ff7c3c9 100644
--- a/drivers/thermal/Kconfig
+++ b/drivers/thermal/Kconfig
@@ -147,6 +147,16 @@  config CLOCK_THERMAL
 	  device that is configured to use this cooling mechanism will be
 	  controlled to reduce clock frequency whenever temperature is high.
 
+config DEVFREQ_THERMAL
+	bool "Generic device cooling support"
+	depends on PM_DEVFREQ
+	help
+	  This implements the generic devfreq cooling mechanism through
+	  frequency reduction for devices using devfreq.
+
+	  This will throttle the device by limiting the maximum allowed DVFS
+	  frequency corresponding to the cooling level.
+
 	  If you want this support, you should say Y here.
 
 config THERMAL_EMULATION
diff --git a/drivers/thermal/Makefile b/drivers/thermal/Makefile
index 535dfee1496f..45f26978ff74 100644
--- a/drivers/thermal/Makefile
+++ b/drivers/thermal/Makefile
@@ -22,6 +22,9 @@  thermal_sys-$(CONFIG_CPU_THERMAL)	+= cpu_cooling.o
 # clock cooling
 thermal_sys-$(CONFIG_CLOCK_THERMAL)	+= clock_cooling.o
 
+# devfreq cooling
+thermal_sys-$(CONFIG_DEVFREQ_THERMAL) += devfreq_cooling.o
+
 # platform thermal drivers
 obj-$(CONFIG_QCOM_SPMI_TEMP_ALARM)	+= qcom-spmi-temp-alarm.o
 obj-$(CONFIG_SPEAR_THERMAL)	+= spear_thermal.o
diff --git a/drivers/thermal/devfreq_cooling.c b/drivers/thermal/devfreq_cooling.c
new file mode 100644
index 000000000000..81e357fce751
--- /dev/null
+++ b/drivers/thermal/devfreq_cooling.c
@@ -0,0 +1,466 @@ 
+/*
+ * devfreq_cooling: Thermal cooling device implementation for devices using
+ *                  devfreq
+ *
+ * Copyright (C) 2014-2015 ARM Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ */
+
+#include <linux/devfreq.h>
+#include <linux/devfreq_cooling.h>
+#include <linux/export.h>
+#include <linux/slab.h>
+#include <linux/pm_opp.h>
+#include <linux/thermal.h>
+
+static DEFINE_MUTEX(devfreq_lock);
+static DEFINE_IDR(devfreq_idr);
+
+/**
+ * get_idr - function to get a unique id.
+ * @idr: struct idr * handle used to create a id.
+ * @id: int * value generated by this function.
+ *
+ * This function will populate @id with an unique
+ * id, using the idr API.
+ *
+ * Return: 0 on success, an error code on failure.
+ */
+static int get_idr(struct idr *idr, int *id)
+{
+	int ret;
+
+	mutex_lock(&devfreq_lock);
+	ret = idr_alloc(idr, NULL, 0, 0, GFP_KERNEL);
+	mutex_unlock(&devfreq_lock);
+	if (unlikely(ret < 0))
+		return ret;
+	*id = ret;
+
+	return 0;
+}
+
+/**
+ * release_idr - function to free the unique id.
+ * @idr: struct idr * handle used for creating the id.
+ * @id: int value representing the unique id.
+ */
+static void release_idr(struct idr *idr, int id)
+{
+	mutex_lock(&devfreq_lock);
+	idr_remove(idr, id);
+	mutex_unlock(&devfreq_lock);
+}
+
+static int enable_disable_opps(struct devfreq *df, int max_enabled_state)
+{
+	int i;
+	struct device *dev = df->dev.parent;
+
+	for (i = 0; i < df->profile->max_state; i++) {
+		struct dev_pm_opp *opp;
+		unsigned int freq = df->profile->freq_table[i] * 1000;
+		bool want_enable = i <= max_enabled_state ? true : false;
+
+		rcu_read_lock();
+		opp = dev_pm_opp_find_freq_exact(dev, freq, !want_enable);
+		rcu_read_unlock();
+
+		if (PTR_ERR(opp) == -ERANGE)
+			continue;
+		else if (IS_ERR(opp))
+			return PTR_ERR(opp);
+
+		if (want_enable)
+			dev_pm_opp_enable(dev, freq);
+		else
+			dev_pm_opp_disable(dev, freq);
+	}
+
+	return 0;
+}
+
+static int devfreq_cooling_get_max_state(struct thermal_cooling_device *cdev,
+					 unsigned long *state)
+{
+	struct devfreq_cooling_device *dfc = cdev->devdata;
+	struct devfreq *df = dfc->devfreq;
+
+	*state = df->profile->max_state - 1;
+
+	return 0;
+}
+
+static int devfreq_cooling_get_cur_state(struct thermal_cooling_device *cdev,
+					 unsigned long *state)
+{
+	struct devfreq_cooling_device *dfc = cdev->devdata;
+
+	*state = dfc->cooling_state;
+
+	return 0;
+}
+
+static int devfreq_cooling_set_cur_state(struct thermal_cooling_device *cdev,
+					 unsigned long state)
+{
+	struct devfreq_cooling_device *dfc = cdev->devdata;
+	struct devfreq *df = dfc->devfreq;
+	struct device *dev = df->dev.parent;
+	unsigned long index;
+	unsigned long max_state = df->profile->max_state;
+	int ret;
+
+	if (state == dfc->cooling_state)
+		return 0;
+
+	dev_dbg(dev, "Setting cooling state %lu\n", state);
+
+	if (state >= max_state)
+		return -EINVAL;
+
+	index = (max_state - 1) - state;
+
+	ret = enable_disable_opps(df, index);
+	if (ret)
+		return ret;
+
+	dfc->cooling_state = state;
+
+	return 0;
+}
+
+static unsigned long
+freq_get_state(struct devfreq *df, unsigned long freq)
+{
+	unsigned long state = THERMAL_CSTATE_INVALID;
+	int i;
+
+	for (i = 0; i < df->profile->max_state; i++) {
+		if (df->profile->freq_table[i] == freq)
+			state = (df->profile->max_state - 1) - i;
+	}
+
+	return state;
+}
+
+static unsigned long
+state_get_freq(struct devfreq *df, unsigned long state)
+{
+	unsigned long index = (df->profile->max_state - 1) - state;
+
+	if (index >= df->profile->max_state)
+		return 0;
+
+	return df->profile->freq_table[index];
+}
+
+static unsigned long
+get_static_power(struct devfreq_cooling_device *dfc, unsigned long freq)
+{
+	struct devfreq *df = dfc->devfreq;
+	struct device *dev = df->dev.parent;
+	unsigned long voltage;
+	struct dev_pm_opp *opp;
+
+	rcu_read_lock();
+	opp = dev_pm_opp_find_freq_exact(dev, freq * 1000, true);
+	voltage = dev_pm_opp_get_voltage(opp) / 1000; /* mV */
+	rcu_read_unlock();
+
+	if (voltage == 0) {
+		dev_warn_ratelimited(dev,
+				     "Failed to get voltage for frequency %lu: %ld\n",
+				     freq, IS_ERR(opp) ? PTR_ERR(opp) : 0);
+		return 0;
+	}
+
+	return dfc->power_ops->get_static_power(voltage);
+}
+
+static int devfreq_cooling_get_requested_power(struct thermal_cooling_device *cdev,
+					       struct thermal_zone_device *tz,
+					       u32 *power)
+{
+	struct devfreq_cooling_device *dfc = cdev->devdata;
+	struct devfreq *df = dfc->devfreq;
+	struct devfreq_dev_status *status = &df->last_status;
+	unsigned long state;
+	unsigned long freq = status->current_frequency;
+	u32 dyn_power, static_power;
+
+	/* Get dynamic power for state */
+	state = freq_get_state(df, freq);
+	if (state == THERMAL_CSTATE_INVALID)
+		return -EAGAIN;
+
+	dyn_power = dfc->power_table[state];
+
+	/* Scale dynamic power for utilization */
+	dyn_power = (dyn_power * status->busy_time) / status->total_time;
+
+	/* Get static power */
+	static_power = get_static_power(dfc, freq);
+
+	*power = dyn_power + static_power;
+
+	return 0;
+}
+
+static int devfreq_cooling_state2power(struct thermal_cooling_device *cdev,
+				       struct thermal_zone_device *tz,
+				       unsigned long state,
+				       u32 *power)
+{
+	struct devfreq_cooling_device *dfc = cdev->devdata;
+	struct devfreq *df = dfc->devfreq;
+	unsigned long freq;
+	u32 static_power;
+
+	freq = state_get_freq(df, state);
+	static_power = get_static_power(dfc, freq);
+
+	*power = dfc->power_table[state] + static_power;
+	return 0;
+}
+
+static int devfreq_cooling_power2state(struct thermal_cooling_device *cdev,
+				       struct thermal_zone_device *tz,
+				       u32 power, unsigned long *state)
+{
+	struct devfreq_cooling_device *dfc = cdev->devdata;
+	struct devfreq *df = dfc->devfreq;
+	struct devfreq_dev_status *status = &df->last_status;
+	unsigned long freq = status->current_frequency;
+	unsigned long busy_time;
+	s32 dyn_power;
+	u32 static_power;
+	int i;
+
+	static_power = get_static_power(dfc, freq);
+
+	dyn_power = power - static_power;
+	dyn_power = dyn_power > 0 ? dyn_power : 0;
+
+	/* Scale dynamic power for utilization */
+	busy_time = status->busy_time ?: 1;
+	dyn_power = (dyn_power * status->total_time) / busy_time;
+
+	/*
+	 * Find the first cooling state that is within the power
+	 * budget for dynamic power.
+	 */
+	for (i = 0; i < df->profile->max_state - 1; i++)
+		if (dyn_power >= dfc->power_table[i])
+			break;
+
+	*state = i;
+	return 0;
+}
+
+static struct thermal_cooling_device_ops devfreq_cooling_ops = {
+	.get_max_state = devfreq_cooling_get_max_state,
+	.get_cur_state = devfreq_cooling_get_cur_state,
+	.set_cur_state = devfreq_cooling_set_cur_state,
+};
+
+/**
+ * devfreq_cooling_gen_power_table(): - Generate power table.
+ * @dfc: Pointer to devfreq cooling device.
+ *
+ * Generate a table with the device's maximum power usage at each cooling
+ * state (OPP). The static and dynamic powerm using the appropriate voltage and
+ * frequency for the state, is acquired from the struct devfreq_cooling_ops, and
+ * summed to make the maximum power draw.
+ *
+ * The table is malloced, and a pointer put in dfc->power_table. This must be
+ * freed when unregistering the devfreq cooling device.
+ *
+ * Return: 0 on success, negative error code on failure.
+ */
+static int devfreq_cooling_gen_power_table(struct devfreq_cooling_device *dfc)
+{
+	struct devfreq *df = dfc->devfreq;
+	struct device *dev = df->dev.parent;
+	int num_opps;
+	struct devfreq_cooling_ops *callbacks = dfc->power_ops;
+	unsigned long freq;
+	u32 *table;
+	int i;
+
+	if (!IS_ENABLED(CONFIG_PM_OPP)) {
+		dev_warn(dev, "Can't use power extensions of the devfreq cooling device without CONFIG_PM_OPP\n");
+		return -EINVAL;
+	}
+
+	rcu_read_lock();
+	num_opps = dev_pm_opp_get_opp_count(dev);
+
+	if (num_opps != dfc->devfreq->profile->max_state) {
+		rcu_read_unlock();
+		return -ERANGE;
+	}
+
+	table = devm_kcalloc(dev, num_opps, sizeof(*table), GFP_KERNEL);
+	if (!table) {
+		rcu_read_unlock();
+		return -ENOMEM;
+	}
+
+	for (i = 0, freq = ULONG_MAX; i < num_opps; i++, freq--) {
+		unsigned long power_dyn, voltage;
+		struct dev_pm_opp *opp;
+
+		opp = dev_pm_opp_find_freq_floor(dev, &freq);
+		if (IS_ERR(opp))
+			break;
+
+		voltage = dev_pm_opp_get_voltage(opp) / 1000; /* mV */
+
+		power_dyn = callbacks->get_dynamic_power(freq, voltage);
+
+		dev_info(dev, "Dynamic power table: %lu MHz @ %lu mV: %lu = %lu mW\n",
+			 freq / 1000000, voltage, power_dyn, power_dyn);
+
+		table[i] = power_dyn;
+	}
+	rcu_read_unlock();
+
+	if (i != num_opps) {
+		devm_kfree(dev, table);
+		return -EFAULT;
+	}
+
+	dfc->power_table = table;
+
+	return 0;
+}
+
+/**
+ * of_devfreq_cooling_register_power(): - Register devfreq cooling device,
+ *                                      with OF and power information.
+ * @np: Pointer to OF device_node.
+ * @df: Pointer to devfreq device.
+ * @ops: Pointer to power ops.
+ */
+struct devfreq_cooling_device *
+of_devfreq_cooling_register_power(struct device_node *np, struct devfreq *df,
+				  struct devfreq_cooling_ops *ops)
+{
+	struct thermal_cooling_device *cdev;
+	struct devfreq_cooling_device *dfc;
+	struct device *dev = df->dev.parent;
+	char dev_name[THERMAL_NAME_LENGTH];
+	int err;
+
+	/* freq_table is required to look map state index to frequency. */
+	if (!df->profile->max_state || !df->profile->freq_table)
+		return ERR_PTR(-EINVAL);
+
+	dfc = devm_kzalloc(dev, sizeof(*dfc), GFP_KERNEL);
+	if (!dfc)
+		return ERR_PTR(-ENOMEM);
+
+	dfc->devfreq = df;
+
+	if (ops) {
+		if (!ops->get_static_power || !ops->get_dynamic_power) {
+			err = -EINVAL;
+			goto free_dfc;
+		}
+		dfc->power_ops = ops;
+
+		err = devfreq_cooling_gen_power_table(dfc);
+		if (err)
+			goto free_dfc;
+
+		devfreq_cooling_ops.get_requested_power =
+			devfreq_cooling_get_requested_power;
+		devfreq_cooling_ops.state2power = devfreq_cooling_state2power;
+		devfreq_cooling_ops.power2state = devfreq_cooling_power2state;
+	}
+
+	err = get_idr(&devfreq_idr, &dfc->id);
+	if (err)
+		goto free_power_table;
+
+	snprintf(dev_name, sizeof(dev_name), "devfreq-%d", dfc->id);
+
+	cdev = thermal_of_cooling_device_register(np, dev_name, dfc,
+						  &devfreq_cooling_ops);
+	if (IS_ERR(cdev)) {
+		err = PTR_ERR(cdev);
+		dev_err(df->dev.parent,
+			"Failed to register devfreq cooling device (%d)\n",
+			err);
+		goto release_idr;
+	}
+
+	dfc->cdev = cdev;
+
+	return dfc;
+
+release_idr:
+	release_idr(&devfreq_idr, dfc->id);
+free_power_table:
+	if (dfc->power_table)
+		devm_kfree(dev, dfc->power_table);
+free_dfc:
+	devm_kfree(dev, dfc);
+
+	return ERR_PTR(err);
+}
+EXPORT_SYMBOL(of_devfreq_cooling_register_power);
+
+/**
+ * of_devfreq_cooling_register(): - Register devfreq cooling device,
+ *                                with OF information.
+ * @np: Pointer to OF device_node.
+ * @df: Pointer to devfreq device.
+ */
+struct devfreq_cooling_device *
+of_devfreq_cooling_register(struct device_node *np, struct devfreq *df)
+{
+	return of_devfreq_cooling_register_power(np, df, NULL);
+}
+EXPORT_SYMBOL(of_devfreq_cooling_register);
+
+/**
+ * devfreq_cooling_register(): - Register devfreq cooling device.
+ * @df: Pointer to devfreq device.
+ */
+struct devfreq_cooling_device *devfreq_cooling_register(struct devfreq *df)
+{
+	return of_devfreq_cooling_register(NULL, df);
+}
+EXPORT_SYMBOL(devfreq_cooling_register);
+
+/**
+ * devfreq_cooling_unregister(): - Unregister devfreq cooling device.
+ * @dfc: Pointer to devfreq cooling device to unregister.
+ */
+void devfreq_cooling_unregister(struct devfreq_cooling_device *dfc)
+{
+	struct device *dev;
+
+	if (!dfc)
+		return;
+
+	dev = dfc->devfreq->dev.parent;
+
+	thermal_cooling_device_unregister(dfc->cdev);
+
+	release_idr(&devfreq_idr, dfc->id);
+	if (dfc->power_table)
+		devm_kfree(dev, dfc->power_table);
+	devm_kfree(dev, dfc);
+}
+EXPORT_SYMBOL(devfreq_cooling_unregister);
diff --git a/include/linux/devfreq_cooling.h b/include/linux/devfreq_cooling.h
new file mode 100644
index 000000000000..8e7346ebbeb2
--- /dev/null
+++ b/include/linux/devfreq_cooling.h
@@ -0,0 +1,93 @@ 
+/*
+ * devfreq_cooling: Thermal cooling device implementation for devices using
+ *                  devfreq
+ *
+ * Copyright (C) 2014 ARM Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * This program is distributed "as is" WITHOUT ANY WARRANTY of any
+ * kind, whether express or implied; without even the implied warranty
+ * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ */
+
+#ifndef __DEVFREQ_COOLING_H__
+#define __DEVFREQ_COOLING_H__
+
+#include <linux/devfreq.h>
+#include <linux/thermal.h>
+
+#ifdef CONFIG_DEVFREQ_THERMAL
+
+/**
+ * struct devfreq_cooling_ops - Devfreq cooling power ops
+ * @get_static_power: Take voltage, in mV, and return the static power in mW.
+ * @get_dynamic_power: Take voltage, in mV, and frequency, in HZ, and return
+ * the dynamic power draw in mW.
+ */
+struct devfreq_cooling_ops {
+	unsigned long (*get_static_power)(unsigned long voltage);
+	unsigned long (*get_dynamic_power)(unsigned long freq,
+					   unsigned long voltage);
+};
+
+/**
+ * struct devfreq_cooling_device - Devfreq cooling device
+ * @id:            unique integer value corresponding to each
+ *                 devfreq_cooling_device registered.
+ * @cdev:          Pointer to associated thermal cooling device.
+ * @devfreq:       Pointer to associated devfreq device.
+ * @cooling_state: Current cooling state.
+ * @power_table:   Pointer to table with maximum power draw for each cooling
+ *                 state. State is the index into the table, and the power is
+ *                 stated in mW.
+ * @power_ops:     Pointer to power operations, used to generate @power_table.
+ */
+struct devfreq_cooling_device {
+	int id;
+	struct thermal_cooling_device *cdev;
+	struct devfreq *devfreq;
+	unsigned long cooling_state;
+	u32 *power_table;
+	struct devfreq_cooling_ops *power_ops;
+};
+
+struct devfreq_cooling_device *
+of_devfreq_cooling_register_power(struct device_node *np, struct devfreq *df,
+				  struct devfreq_cooling_ops *ops);
+struct devfreq_cooling_device *
+of_devfreq_cooling_register(struct device_node *np, struct devfreq *df);
+struct devfreq_cooling_device *devfreq_cooling_register(struct devfreq *df);
+void devfreq_cooling_unregister(struct devfreq_cooling_device *dfc);
+
+#else /* !CONFIG_DEVFREQ_THERMAL */
+
+struct devfreq_cooling_device *
+of_devfreq_cooling_register_power(struct device_node *np, struct devfreq *df,
+				  struct devfreq_cooling_ops *ops)
+{
+	return ERR_PTR(-EINVAL);
+}
+
+static inline struct devfreq_cooling_device *
+of_devfreq_cooling_register(struct device_node *np, struct devfreq *df)
+{
+	return ERR_PTR(-EINVAL);
+}
+
+static inline struct devfreq_cooling_device *
+devfreq_cooling_register(struct devfreq *df)
+{
+	return ERR_PTR(-EINVAL);
+}
+
+static inline void
+devfreq_cooling_unregister(struct devfreq_cooling_device *dfc)
+{
+}
+
+#endif /* CONFIG_DEVFREQ_THERMAL */
+#endif /* __DEVFREQ_COOLING_H__ */