Re: [PATCH v8 5/5] PM / DEVFREQ: add basic governors

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, Aug 24, 2011 at 1:22 AM, MyungJoo Ham <myungjoo.ham@xxxxxxxxxxx> wrote:
> Four CPUFREQ-like governors are provided as examples.
>
> powersave: use the lowest frequency possible. The user (device) should
> set the polling_ms as 0 because polling is useless for this governor.
>
> performance: use the highest freqeuncy possible. The user (device)
> should set the polling_ms as 0 because polling is useless for this
> governor.
>
> userspace: use the user specified frequency stored at
> devfreq.user_set_freq. With sysfs support in the following patch, a user
> may set the value with the sysfs interface.
>
> simple_ondemand: simplified version of CPUFREQ's ONDEMAND governor.
>
> When a user updates OPP entries (enable/disable/add), OPP framework
> automatically notifies DEVFREQ to update operating frequency
> accordingly. Thus, DEVFREQ users (device drivers) do not need to update
> DEVFREQ manually with OPP entry updates or set polling_ms for powersave
> , performance, userspace, or any other "static" governors.
>
> Note that these are given only as basic examples for governors and any
> devices with DEVFREQ may implement their own governors with the drivers
> and use them.
>
> Signed-off-by: MyungJoo Ham <myungjoo.ham@xxxxxxxxxxx>
> Signed-off-by: Kyungmin Park <kyungmin.park@xxxxxxxxxxx>
>
> ---
> Changed from v7
> - Userspace uses its own sysfs interface.
>
> Changed from v5
> - Seperated governor files from devfreq.c
> - Allow simple ondemand to be tuned for each device
> ---
>  Documentation/ABI/testing/sysfs-devices-power |    9 ++
>  drivers/devfreq/Kconfig                       |   36 ++++++++
>  drivers/devfreq/Makefile                      |    4 +
>  drivers/devfreq/governor_performance.c        |   24 +++++
>  drivers/devfreq/governor_powersave.c          |   24 +++++
>  drivers/devfreq/governor_simpleondemand.c     |   88 ++++++++++++++++++
>  drivers/devfreq/governor_userspace.c          |  119 +++++++++++++++++++++++++
>  include/linux/devfreq.h                       |   41 +++++++++
>  8 files changed, 345 insertions(+), 0 deletions(-)
>  create mode 100644 drivers/devfreq/governor_performance.c
>  create mode 100644 drivers/devfreq/governor_powersave.c
>  create mode 100644 drivers/devfreq/governor_simpleondemand.c
>  create mode 100644 drivers/devfreq/governor_userspace.c
>
> diff --git a/Documentation/ABI/testing/sysfs-devices-power b/Documentation/ABI/testing/sysfs-devices-power
> index 57f4591..c7f6977 100644
> --- a/Documentation/ABI/testing/sysfs-devices-power
> +++ b/Documentation/ABI/testing/sysfs-devices-power
> @@ -202,3 +202,12 @@ Description:
>                shows the requested polling interval of the corresponding
>                device. The values are represented in ms. If the value is less
>                than 1 jiffy, it is considered to be 0, which means no polling.
> +
> +What:          /sys/devices/.../power/devfreq_userspace_set_freq

How about just .../devfreq_set_freq?  I think the userspace bit is
implied and the name is very long.

> +Date:          August 2011
> +Contact:       MyungJoo Ham <myungjoo.ham@xxxxxxxxxxx>
> +Description:
> +               The /sys/devices/.../power/devfreq_userspace_set_freq sets
> +               and shows the user specified frequency in kHz. This sysfs
> +               entry is created and managed by userspace DEVFREQ governor.
> +               If other governors are used, it won't be supported.
> diff --git a/drivers/devfreq/Kconfig b/drivers/devfreq/Kconfig
> index 1fb42de..643b055 100644
> --- a/drivers/devfreq/Kconfig
> +++ b/drivers/devfreq/Kconfig
> @@ -34,6 +34,42 @@ menuconfig PM_DEVFREQ
>
>  if PM_DEVFREQ
>
> +comment "DEVFREQ Governors"
> +
> +config DEVFREQ_GOV_SIMPLE_ONDEMAND
> +       bool "Simple Ondemand"
> +       help
> +         Chooses frequency based on the recent load on the device. Works
> +         similar as ONDEMAND governor of CPUFREQ does. A device with
> +         Simple-Ondemand should be able to provide busy/total counter
> +         values that imply the usage rate. A device may provide tuned
> +         values to the governor with data field at devfreq_add_device().
> +
> +config DEVFREQ_GOV_PERFORMANCE
> +       bool "Performance"
> +       help
> +         Sets the frequency at the maximum available frequency.
> +         This governor always returns UINT_MAX as frequency so that
> +         the DEVFREQ framework returns the highest frequency available
> +         at any time.
> +
> +config DEVFREQ_GOV_POWERSAVE
> +       bool "Powersave"
> +       help
> +         Sets the frequency at the minimum available frequency.
> +         This governor always returns 0 as frequency so that
> +         the DEVFREQ framework returns the lowest frequency available
> +         at any time.
> +
> +config DEVFREQ_GOV_USERSPACE
> +       bool "Userspace"
> +       help
> +         Sets the frequency at the user specified one.
> +         This governor returns the user configured frequency if there
> +         has been an input to /sys/devices/.../power/devfreq_set_freq.
> +         Otherwise, the governor does not change the frequnecy
> +         given at the initialization.
> +
>  comment "DEVFREQ Drivers"
>
>  endif # PM_DEVFREQ
> diff --git a/drivers/devfreq/Makefile b/drivers/devfreq/Makefile
> index 168934a..4564a89 100644
> --- a/drivers/devfreq/Makefile
> +++ b/drivers/devfreq/Makefile
> @@ -1 +1,5 @@
>  obj-$(CONFIG_PM_DEVFREQ)       += devfreq.o
> +obj-$(CONFIG_DEVFREQ_GOV_SIMPLE_ONDEMAND)      += governor_simpleondemand.o
> +obj-$(CONFIG_DEVFREQ_GOV_PERFORMANCE)  += governor_performance.o
> +obj-$(CONFIG_DEVFREQ_GOV_POWERSAVE)    += governor_powersave.o
> +obj-$(CONFIG_DEVFREQ_GOV_USERSPACE)    += governor_userspace.o
> diff --git a/drivers/devfreq/governor_performance.c b/drivers/devfreq/governor_performance.c
> new file mode 100644
> index 0000000..c47eff8
> --- /dev/null
> +++ b/drivers/devfreq/governor_performance.c
> @@ -0,0 +1,24 @@
> +/*
> + *  linux/drivers/devfreq/governor_performance.c
> + *
> + *  Copyright (C) 2011 Samsung Electronics
> + *     MyungJoo Ham <myungjoo.ham@xxxxxxxxxxx>
> + *
> + * 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.
> + */
> +
> +#include <linux/devfreq.h>
> +
> +static int devfreq_performance_func(struct devfreq *df,
> +                                   unsigned long *freq)
> +{
> +       *freq = UINT_MAX; /* devfreq_do will run "floor" */
> +       return 0;
> +}
> +
> +struct devfreq_governor devfreq_performance = {
> +       .name = "performance",
> +       .get_target_freq = devfreq_performance_func,
> +};
> diff --git a/drivers/devfreq/governor_powersave.c b/drivers/devfreq/governor_powersave.c
> new file mode 100644
> index 0000000..4f128d8
> --- /dev/null
> +++ b/drivers/devfreq/governor_powersave.c
> @@ -0,0 +1,24 @@
> +/*
> + *  linux/drivers/devfreq/governor_powersave.c
> + *
> + *  Copyright (C) 2011 Samsung Electronics
> + *     MyungJoo Ham <myungjoo.ham@xxxxxxxxxxx>
> + *
> + * 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.
> + */
> +
> +#include <linux/devfreq.h>
> +
> +static int devfreq_powersave_func(struct devfreq *df,
> +                                 unsigned long *freq)
> +{
> +       *freq = 0; /* devfreq_do will run "ceiling" to 0 */
> +       return 0;
> +}
> +
> +struct devfreq_governor devfreq_powersave = {
> +       .name = "powersave",
> +       .get_target_freq = devfreq_powersave_func,
> +};
> diff --git a/drivers/devfreq/governor_simpleondemand.c b/drivers/devfreq/governor_simpleondemand.c
> new file mode 100644
> index 0000000..18fe8be
> --- /dev/null
> +++ b/drivers/devfreq/governor_simpleondemand.c
> @@ -0,0 +1,88 @@
> +/*
> + *  linux/drivers/devfreq/governor_simpleondemand.c
> + *
> + *  Copyright (C) 2011 Samsung Electronics
> + *     MyungJoo Ham <myungjoo.ham@xxxxxxxxxxx>
> + *
> + * 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.
> + */
> +
> +#include <linux/errno.h>
> +#include <linux/devfreq.h>
> +#include <linux/math64.h>
> +
> +/* Default constants for DevFreq-Simple-Ondemand (DFSO) */
> +#define DFSO_UPTHRESHOLD       (90)
> +#define DFSO_DOWNDIFFERENCTIAL (5)
> +static int devfreq_simple_ondemand_func(struct devfreq *df,
> +                                       unsigned long *freq)
> +{
> +       struct devfreq_dev_status stat;
> +       int err = df->profile->get_dev_status(df->dev, &stat);
> +       unsigned long long a, b;
> +       unsigned int dfso_upthreshold = DFSO_UPTHRESHOLD;
> +       unsigned int dfso_downdifferential = DFSO_DOWNDIFFERENCTIAL;
> +       struct devfreq_simple_ondemand_data *data = df->data;
> +
> +       if (err)
> +               return err;
> +
> +       if (data) {
> +               if (data->upthreshold)
> +                       dfso_upthreshold = data->upthreshold;
> +               if (data->downdifferential)
> +                       dfso_downdifferential = data->downdifferential;
> +       }
> +       if (dfso_upthreshold > 100 ||
> +           dfso_upthreshold < dfso_downdifferential)
> +               return -EINVAL;
> +
> +       /* Assume MAX if it is going to be divided by zero */
> +       if (stat.total_time == 0) {
> +               *freq = UINT_MAX;
> +               return 0;
> +       }
> +
> +       /* Prevent overflow */
> +       if (stat.busy_time >= (1 << 24) || stat.total_time >= (1 << 24)) {
> +               stat.busy_time >>= 7;
> +               stat.total_time >>= 7;
> +       }
> +
> +       /* Set MAX if it's busy enough */
> +       if (stat.busy_time * 100 >
> +           stat.total_time * dfso_upthreshold) {
> +               *freq = UINT_MAX;
> +               return 0;
> +       }
> +
> +       /* Set MAX if we do not know the initial frequency */
> +       if (stat.current_frequency == 0) {
> +               *freq = UINT_MAX;
> +               return 0;
> +       }
> +
> +       /* Keep the current frequency */
> +       if (stat.busy_time * 100 >
> +           stat.total_time * (dfso_upthreshold - dfso_downdifferential)) {
> +               *freq = stat.current_frequency;
> +               return 0;
> +       }
> +
> +       /* Set the desired frequency based on the load */
> +       a = stat.busy_time;
> +       a *= stat.current_frequency;
> +       b = div_u64(a, stat.total_time);
> +       b *= 100;
> +       b = div_u64(b, (dfso_upthreshold - dfso_downdifferential / 2));
> +       *freq = (unsigned long) b;
> +
> +       return 0;
> +}
> +
> +struct devfreq_governor devfreq_simple_ondemand = {
> +       .name = "simple_ondemand",
> +       .get_target_freq = devfreq_simple_ondemand_func,
> +};
> diff --git a/drivers/devfreq/governor_userspace.c b/drivers/devfreq/governor_userspace.c
> new file mode 100644
> index 0000000..53a4574
> --- /dev/null
> +++ b/drivers/devfreq/governor_userspace.c
> @@ -0,0 +1,119 @@
> +/*
> + *  linux/drivers/devfreq/governor_simpleondemand.c
> + *
> + *  Copyright (C) 2011 Samsung Electronics
> + *     MyungJoo Ham <myungjoo.ham@xxxxxxxxxxx>
> + *
> + * 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.
> + */
> +
> +#include <linux/slab.h>
> +#include <linux/device.h>
> +#include <linux/devfreq.h>
> +#include <linux/pm.h>
> +#include "governor.h"
> +
> +struct userspace_data {
> +       unsigned long user_frequency;
> +       bool valid;
> +};
> +
> +static int devfreq_userspace_func(struct devfreq *df, unsigned long *freq)
> +{
> +       struct userspace_data *data = df->data;
> +
> +       if (!data->valid)
> +               *freq = df->previous_freq; /* No user freq specified yet */
> +       else
> +               *freq = data->user_frequency;
> +       return 0;
> +}
> +
> +static ssize_t store_freq(struct device *dev, struct device_attribute *attr,
> +                         const char *buf, size_t count)
> +{
> +       struct devfreq *devfreq = get_devfreq(dev);
> +       struct userspace_data *data;
> +       unsigned long wanted;
> +       int err = 0;
> +
> +       if (IS_ERR(devfreq)) {
> +               err = PTR_ERR(devfreq);
> +               goto out;
> +       }
> +       data = devfreq->data;
> +
> +       sscanf(buf, "%lu", &wanted);
> +       data->user_frequency = wanted;
> +       data->valid = true;
> +       err = update_devfreq(devfreq);
> +       if (err == 0)
> +               err = count;
> +out:
> +       return err;
> +}
> +
> +static ssize_t show_freq(struct device *dev, struct device_attribute *attr,
> +                        char *buf)
> +{
> +       struct devfreq *devfreq = get_devfreq(dev);
> +       struct userspace_data *data;
> +       int err = 0;
> +
> +       if (IS_ERR(devfreq)) {
> +               err = PTR_ERR(devfreq);
> +               goto out;
> +       }
> +       data = devfreq->data;
> +
> +       if (data->valid)
> +               err = sprintf(buf, "%lu\n", data->user_frequency);
> +       else
> +               err = sprintf(buf, "undefined\n");
> +out:
> +       return err;
> +}

Shouldn't accesses to devfreq->data be protected by a mutex?

Regards,
Mike

> +
> +static DEVICE_ATTR(devfreq_userspace_set_freq, 0644, show_freq, store_freq);
> +static struct attribute *dev_entries[] = {
> +       &dev_attr_devfreq_userspace_set_freq.attr,
> +       NULL,
> +};
> +static struct attribute_group dev_attr_group = {
> +       .name   = power_group_name,
> +       .attrs  = dev_entries,
> +};
> +
> +static int userspace_init(struct devfreq *devfreq)
> +{
> +       int err = 0;
> +       struct userspace_data *data = kzalloc(sizeof(struct userspace_data),
> +                                             GFP_KERNEL);
> +
> +       if (!data) {
> +               err = -ENOMEM;
> +               goto out;
> +       }
> +       data->valid = false;
> +       devfreq->data = data;
> +
> +       sysfs_merge_group(&devfreq->dev->kobj, &dev_attr_group);
> +out:
> +       return err;
> +}
> +
> +static void userspace_exit(struct devfreq *devfreq)
> +{
> +       sysfs_unmerge_group(&devfreq->dev->kobj, &dev_attr_group);
> +       kfree(devfreq->data);
> +       devfreq->data = NULL;
> +}
> +
> +struct devfreq_governor devfreq_userspace = {
> +       .name = "userspace",
> +       .get_target_freq = devfreq_userspace_func,
> +       .init = userspace_init,
> +       .exit = userspace_exit,
> +};
> diff --git a/include/linux/devfreq.h b/include/linux/devfreq.h
> index fdc6916..cbafcdf 100644
> --- a/include/linux/devfreq.h
> +++ b/include/linux/devfreq.h
> @@ -13,6 +13,7 @@
>  #ifndef __LINUX_DEVFREQ_H__
>  #define __LINUX_DEVFREQ_H__
>
> +#include <linux/opp.h>
>  #include <linux/notifier.h>
>
>  #define DEVFREQ_NAME_LEN 16
> @@ -65,6 +66,8 @@ struct devfreq_governor {
>  *                     "devfreq_monitor" executions to reevaluate
>  *                     frequency/voltage of the device. Set by
>  *                     profile's polling_ms interval.
> + * @user_set_freq      User specified adequete frequency value (thru sysfs
> + *             interface). Governors may and may not use this value.
>  * @data       Private data of the governor. The devfreq framework does not
>  *             touch this.
>  *
> @@ -82,6 +85,7 @@ struct devfreq {
>        unsigned long previous_freq;
>        unsigned int next_polling;
>
> +       unsigned long user_set_freq; /* governors may ignore this. */
>        void *data; /* private data for governors */
>  };
>
> @@ -91,6 +95,37 @@ extern int devfreq_add_device(struct device *dev,
>                           struct devfreq_governor *governor,
>                           void *data);
>  extern int devfreq_remove_device(struct device *dev);
> +
> +#ifdef CONFIG_DEVFREQ_GOV_POWERSAVE
> +extern struct devfreq_governor devfreq_powersave;
> +#endif
> +#ifdef CONFIG_DEVFREQ_GOV_PERFORMANCE
> +extern struct devfreq_governor devfreq_performance;
> +#endif
> +#ifdef CONFIG_DEVFREQ_GOV_USERSPACE
> +extern struct devfreq_governor devfreq_userspace;
> +#endif
> +#ifdef CONFIG_DEVFREQ_GOV_SIMPLE_ONDEMAND
> +extern struct devfreq_governor devfreq_simple_ondemand;
> +/**
> + * struct devfreq_simple_ondemand_data - void *data fed to struct devfreq
> + *     and devfreq_add_device
> + * @ upthreshold       If the load is over this value, the frequency jumps.
> + *                     Specify 0 to use the default. Valid value = 0 to 100.
> + * @ downdifferential  If the load is under upthreshold - downdifferential,
> + *                     the governor may consider slowing the frequency down.
> + *                     Specify 0 to use the default. Valid value = 0 to 100.
> + *                     downdifferential < upthreshold must hold.
> + *
> + * If the fed devfreq_simple_ondemand_data pointer is NULL to the governor,
> + * the governor uses the default values.
> + */
> +struct devfreq_simple_ondemand_data {
> +       unsigned int upthreshold;
> +       unsigned int downdifferential;
> +};
> +#endif
> +
>  #else /* !CONFIG_PM_DEVFREQ */
>  static int devfreq_add_device(struct device *dev,
>                           struct devfreq_dev_profile *profile,
> @@ -104,6 +139,12 @@ static int devfreq_remove_device(struct device *dev)
>  {
>        return 0;
>  }
> +
> +#define devfreq_powersave      NULL
> +#define devfreq_performance    NULL
> +#define devfreq_userspace      NULL
> +#define devfreq_simple_ondemand        NULL
> +
>  #endif /* CONFIG_PM_DEVFREQ */
>
>  #endif /* __LINUX_DEVFREQ_H__ */
> --
> 1.7.4.1
>
>
_______________________________________________
linux-pm mailing list
linux-pm@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/linux-pm



[Index of Archives]     [Linux ACPI]     [Netdev]     [Ethernet Bridging]     [Linux Wireless]     [CPU Freq]     [Kernel Newbies]     [Fedora Kernel]     [Security]     [Linux for Hams]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux RAID]     [Linux Admin]     [Samba]

  Powered by Linux