powercap/drivers/dtpm: Add dtpm devfreq with energy model support

Message ID 20210319162836.9364-1-daniel.lezcano@linaro.org
State New
Headers show
Series
  • powercap/drivers/dtpm: Add dtpm devfreq with energy model support
Related show

Commit Message

Daniel Lezcano March 19, 2021, 4:28 p.m.
Currently the dtpm supports the CPUs via cpufreq and the energy
model. This change provides the same for the device which supports
devfreq.

Each device supporting devfreq and having an energy model can register
themselves in the list of supported devices.

The concept is the same as the cpufreq dtpm support: the QoS is used
to aggregate the requests and the energy model gives the value of the
instantaneous power consumption ponderated by the load of the device.

Cc: Chanwoo Choi <cwchoi00@gmail.com>
Cc: Lukasz Luba <lukasz.luba@arm.com>
Cc: Kyungmin Park <kyungmin.park@samsung.com>
Cc: MyungJoo Ham <myungjoo.ham@samsung.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>

---
 drivers/powercap/Kconfig        |   7 ++
 drivers/powercap/Makefile       |   1 +
 drivers/powercap/dtpm_devfreq.c | 198 ++++++++++++++++++++++++++++++++
 include/linux/dtpm.h            |  13 +++
 4 files changed, 219 insertions(+)
 create mode 100644 drivers/powercap/dtpm_devfreq.c

-- 
2.17.1

Comments

Lukasz Luba March 23, 2021, 3:56 p.m. | #1
Hi Daniel,

On 3/19/21 4:28 PM, Daniel Lezcano wrote:
> Currently the dtpm supports the CPUs via cpufreq and the energy

> model. This change provides the same for the device which supports

> devfreq.

> 

> Each device supporting devfreq and having an energy model can register

> themselves in the list of supported devices.

> 

> The concept is the same as the cpufreq dtpm support: the QoS is used

> to aggregate the requests and the energy model gives the value of the

> instantaneous power consumption ponderated by the load of the device.

> 



I've just started the review, but I have a blocking question:

Why there is no unregister function (like 'dtmp_unregister_devfreq')?
Do you consider any devfreq drivers to be modules?

The code looks like an API that it's going to be called directly in
e.g. GPU driver in it's probe function. In that case probably the
module unloading should call dtmp unregister.

Could you explain this to me please? So I can continue the review.

Regards,
Lukasz
Daniel Lezcano March 23, 2021, 3:59 p.m. | #2
On 23/03/2021 16:56, Lukasz Luba wrote:
> Hi Daniel,

> 

> On 3/19/21 4:28 PM, Daniel Lezcano wrote:

>> Currently the dtpm supports the CPUs via cpufreq and the energy

>> model. This change provides the same for the device which supports

>> devfreq.

>>

>> Each device supporting devfreq and having an energy model can register

>> themselves in the list of supported devices.

>>

>> The concept is the same as the cpufreq dtpm support: the QoS is used

>> to aggregate the requests and the energy model gives the value of the

>> instantaneous power consumption ponderated by the load of the device.

>>

> 

> 

> I've just started the review, but I have a blocking question:

> 

> Why there is no unregister function (like 'dtmp_unregister_devfreq')?

> Do you consider any devfreq drivers to be modules?

> 

> The code looks like an API that it's going to be called directly in

> e.g. GPU driver in it's probe function. In that case probably the

> module unloading should call dtmp unregister.

> 

> Could you explain this to me please? So I can continue the review.


Just forgot the unregister function :)


-- 
<http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog
Daniel Lezcano March 23, 2021, 4 p.m. | #3
On 23/03/2021 16:56, Lukasz Luba wrote:
> Hi Daniel,

> 

> On 3/19/21 4:28 PM, Daniel Lezcano wrote:

>> Currently the dtpm supports the CPUs via cpufreq and the energy

>> model. This change provides the same for the device which supports

>> devfreq.

>>

>> Each device supporting devfreq and having an energy model can register

>> themselves in the list of supported devices.

>>

>> The concept is the same as the cpufreq dtpm support: the QoS is used

>> to aggregate the requests and the energy model gives the value of the

>> instantaneous power consumption ponderated by the load of the device.

>>

> 

> 

> I've just started the review, but I have a blocking question:

> 

> Why there is no unregister function (like 'dtmp_unregister_devfreq')?

> Do you consider any devfreq drivers to be modules?

> 

> The code looks like an API that it's going to be called directly in

> e.g. GPU driver in it's probe function. In that case probably the

> module unloading should call dtmp unregister.

> 

> Could you explain this to me please? So I can continue the review.


BTW, thanks for taking the time to review the patch.


-- 
<http://www.linaro.org/> Linaro.org │ Open source software for ARM SoCs

Follow Linaro:  <http://www.facebook.com/pages/Linaro> Facebook |
<http://twitter.com/#!/linaroorg> Twitter |
<http://www.linaro.org/linaro-blog/> Blog

Patch

diff --git a/drivers/powercap/Kconfig b/drivers/powercap/Kconfig
index 599b41e4e0a7..acdb047d8f1b 100644
--- a/drivers/powercap/Kconfig
+++ b/drivers/powercap/Kconfig
@@ -64,4 +64,11 @@  config DTPM_CPU
 	help
 	  This enables support for CPU power limitation based on
 	  energy model.
+
+config DTPM_DEVFREQ
+	bool "Add device power capping based on the energy model"
+	depends on DTPM && ENERGY_MODEL
+	help
+	  This enables support for device power limitation based on
+	  energy model.
 endif
diff --git a/drivers/powercap/Makefile b/drivers/powercap/Makefile
index 519cabc624c3..e47f4fd68fb9 100644
--- a/drivers/powercap/Makefile
+++ b/drivers/powercap/Makefile
@@ -2,6 +2,7 @@ 
 obj-$(CONFIG_DTPM) += dtpm.o
 obj-$(CONFIG_DTPM_CONFIGFS) += dtpm_configfs.o
 obj-$(CONFIG_DTPM_CPU) += dtpm_cpu.o
+obj-$(CONFIG_DTPM_DEVFREQ) += dtpm_devfreq.o
 obj-$(CONFIG_POWERCAP)	+= powercap_sys.o
 obj-$(CONFIG_INTEL_RAPL_CORE) += intel_rapl_common.o
 obj-$(CONFIG_INTEL_RAPL) += intel_rapl_msr.o
diff --git a/drivers/powercap/dtpm_devfreq.c b/drivers/powercap/dtpm_devfreq.c
new file mode 100644
index 000000000000..0f259238a45d
--- /dev/null
+++ b/drivers/powercap/dtpm_devfreq.c
@@ -0,0 +1,198 @@ 
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * Copyright 2021 Linaro Limited
+ *
+ * Author: Daniel Lezcano <daniel.lezcano@linaro.org>
+ *
+ * The devfreq device combined with the energy model and the load can
+ * give an estimation of the power consumption as well as limiting the
+ * power.
+ *
+ */
+#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+
+#include <linux/cpumask.h>
+#include <linux/devfreq.h>
+#include <linux/dtpm.h>
+#include <linux/energy_model.h>
+#include <linux/pm_qos.h>
+#include <linux/slab.h>
+#include <linux/units.h>
+
+#ifndef HZ_PER_KHZ
+#define HZ_PER_KHZ 1000UL
+#endif
+
+struct dtpm_devfreq {
+	struct dtpm dtpm;
+	struct dev_pm_qos_request qos_req;
+	struct devfreq *devfreq;
+};
+
+struct dtpm_devfreq *to_dtpm_devfreq(struct dtpm *dtpm)
+{
+	return container_of(dtpm, struct dtpm_devfreq, dtpm);
+}
+
+static int update_pd_power_uw(struct dtpm *dtpm)
+{
+	struct dtpm_devfreq *dtpm_devfreq = to_dtpm_devfreq(dtpm);
+	struct devfreq *devfreq = dtpm_devfreq->devfreq;
+	struct device *dev = devfreq->dev.parent;
+	struct em_perf_domain *pd = em_pd_get(dev);
+
+	dtpm->power_min = pd->table[0].power;
+	dtpm->power_min *= MICROWATT_PER_MILLIWATT;
+
+	dtpm->power_max = pd->table[pd->nr_perf_states - 1].power;
+	dtpm->power_max *= MICROWATT_PER_MILLIWATT;
+
+	return 0;
+}
+
+static u64 set_pd_power_limit(struct dtpm *dtpm, u64 power_limit)
+{
+	struct dtpm_devfreq *dtpm_devfreq = to_dtpm_devfreq(dtpm);
+	struct devfreq *devfreq = dtpm_devfreq->devfreq;
+	struct device *dev = devfreq->dev.parent;
+	struct em_perf_domain *pd = em_pd_get(dev);
+	unsigned long freq;
+	u64 power;
+	int i;
+
+	for (i = 0; i < pd->nr_perf_states; i++) {
+
+		power = pd->table[i].power * MICROWATT_PER_MILLIWATT;
+		if (power > power_limit)
+			break;
+	}
+
+	freq = pd->table[i - 1].frequency;
+
+	dev_pm_qos_update_request(&dtpm_devfreq->qos_req, freq);
+
+	power_limit = pd->table[i - 1].power * MICROWATT_PER_MILLIWATT;
+
+	return power_limit;
+}
+
+static void _normalize_load(struct devfreq_dev_status *status)
+{
+	if (status->total_time > 0xfffff) {
+		status->total_time >>= 10;
+		status->busy_time >>= 10;
+	}
+
+	status->busy_time <<= 10;
+	status->busy_time /= status->total_time ? : 1;
+
+	status->busy_time = status->busy_time ? : 1;
+	status->total_time = 1024;
+}
+
+static u64 get_pd_power_uw(struct dtpm *dtpm)
+{
+	struct dtpm_devfreq *dtpm_devfreq = to_dtpm_devfreq(dtpm);
+	struct devfreq *devfreq = dtpm_devfreq->devfreq;
+	struct device *dev = devfreq->dev.parent;
+	struct em_perf_domain *pd = em_pd_get(dev);
+	struct devfreq_dev_status status;
+	unsigned long freq;
+	u64 power;
+	int i;
+
+	mutex_lock(&devfreq->lock);
+	status = devfreq->last_status;
+	mutex_unlock(&devfreq->lock);
+
+	freq = DIV_ROUND_UP(status.current_frequency, HZ_PER_KHZ);
+	_normalize_load(&status);
+
+	for (i = 0; i < pd->nr_perf_states; i++) {
+
+		if (pd->table[i].frequency < freq)
+			continue;
+
+		power = pd->table[i].power * MICROWATT_PER_MILLIWATT;
+		power *= status.busy_time;
+		power >>= 10;
+
+		return power;
+	}
+
+	return 0;
+}
+
+static void pd_release(struct dtpm *dtpm)
+{
+	struct dtpm_devfreq *dtpm_devfreq = to_dtpm_devfreq(dtpm);
+
+	if (dev_pm_qos_request_active(&dtpm_devfreq->qos_req))
+		dev_pm_qos_remove_request(&dtpm_devfreq->qos_req);
+
+	kfree(dtpm_devfreq);
+}
+
+static struct dtpm_ops dtpm_ops = {
+	.set_power_uw = set_pd_power_limit,
+	.get_power_uw = get_pd_power_uw,
+	.update_power_uw = update_pd_power_uw,
+	.release = pd_release,
+};
+
+int dtpm_register_devfreq(struct devfreq *devfreq)
+{
+	struct device *dev = devfreq->dev.parent;
+	struct dtpm_devfreq *dtpm_devfreq;
+	struct em_perf_domain *pd;
+	char *name;
+	int ret = -ENOMEM;
+
+	pd = em_pd_get(dev);
+	if (!pd) {
+		ret = dev_pm_opp_of_register_em(dev, NULL);
+		if (ret) {
+			pr_err("No energy model available for '%s'\n", name);
+			return -EINVAL;
+		}
+	}
+
+	dtpm_devfreq = kzalloc(sizeof(*dtpm_devfreq), GFP_KERNEL);
+	if (!dtpm_devfreq)
+		return -ENOMEM;
+
+	dtpm_init(&dtpm_devfreq->dtpm, &dtpm_ops);
+
+	dtpm_devfreq->devfreq = devfreq;
+
+	name = kasprintf(GFP_KERNEL, "%s-devfreq", dev_name(dev));
+	if (!name)
+		goto out_dtpm_devfreq;
+
+	ret = dtpm_register(name, &dtpm_devfreq->dtpm);
+	if (ret) {
+		pr_err("Failed to register '%s': %d\n", name, ret);
+		goto out_kfree_name;
+	}
+
+	ret = dev_pm_qos_add_request(dev, &dtpm_devfreq->qos_req,
+				     DEV_PM_QOS_MAX_FREQUENCY,
+				     PM_QOS_MAX_FREQUENCY_DEFAULT_VALUE);
+	if (ret) {
+		pr_err("Failed to add QoS request: %d\n", ret);
+		goto out_dtpm_unregister;
+	}
+
+	kfree(name);
+
+	return 0;
+
+out_dtpm_unregister:
+	dtpm_unregister(name);
+out_kfree_name:
+	kfree(name);
+out_dtpm_devfreq:
+	kfree(dtpm_devfreq);
+
+	return ret;
+}
diff --git a/include/linux/dtpm.h b/include/linux/dtpm.h
index 6b9761aa5ea7..5df4f99c3553 100644
--- a/include/linux/dtpm.h
+++ b/include/linux/dtpm.h
@@ -80,4 +80,17 @@  void dtpm_unregister(const char *name);
 struct dtpm *dtpm_get(const char *name);
 
 void dtpm_put(struct dtpm *dtpm);
+
+#ifdef CONFIG_DTPM_DEVFREQ
+
+struct devfreq;
+
+int dtpm_register_devfreq(struct devfreq *devfreq);
+#else
+static inline int dtpm_register_devfreq(struct devfreq *devfreq)
+{
+	return 0;
+}
+#endif
+
 #endif