All of lore.kernel.org
 help / color / mirror / Atom feed
From: Eduardo Valentin <edubezval@gmail.com>
To: Javi Merino <javi.merino@arm.com>
Cc: linux-pm@vger.kernel.org, linux-kernel@vger.kernel.org,
	rui.zang@intel.com, Zhang Rui <rui.zhang@intel.com>
Subject: Re: [PATCH v3 4/4] thermal: show the sub thermal zones in sysfs
Date: Wed, 2 Mar 2016 19:17:02 -0800	[thread overview]
Message-ID: <20160303031700.GB3379@localhost.localdomain> (raw)
In-Reply-To: <1448464186-26289-5-git-send-email-javi.merino@arm.com>

On Wed, Nov 25, 2015 at 03:09:46PM +0000, Javi Merino wrote:
> When a thermal zone is created that has sub thermal zones via
> devicetree or via thermal_zone_add_subtz() expose that relationship in
> sysfs.
> 
> Cc: Zhang Rui <rui.zhang@intel.com>
> Cc: Eduardo Valentin <edubezval@gmail.com>
> Signed-off-by: Javi Merino <javi.merino@arm.com>
> ---
>  Documentation/thermal/sysfs-api.txt |  33 ++++++++++
>  drivers/thermal/thermal_core.c      | 125 ++++++++++++++++++++++++++++++++++++
>  include/linux/thermal.h             |   3 +
>  3 files changed, 161 insertions(+)
> 
> diff --git a/Documentation/thermal/sysfs-api.txt b/Documentation/thermal/sysfs-api.txt
> index dda24a0bdeec..5636a531f0b1 100644
> --- a/Documentation/thermal/sysfs-api.txt
> +++ b/Documentation/thermal/sysfs-api.txt
> @@ -219,11 +219,16 @@ Thermal zone device sys I/F, created once it's registered:
>      |---temp:			Current temperature
>      |---mode:			Working mode of the thermal zone
>      |---policy:			Thermal governor used for this zone
> +    |---aggregation_function:	Function to use when aggregating the
> +				temperature of this zone's slave thermal zones
>      |---available_policies:	Available thermal governors for this zone
>      |---trip_point_[0-*]_temp:	Trip point temperature
>      |---trip_point_[0-*]_type:	Trip point type
>      |---trip_point_[0-*]_hyst:	Hysteresis value for this trip point
>      |---emul_temp:		Emulated temperature set node
> +    |---subtz[0-*]:		Slave thermal zones of this thermal zone
> +    |---subtz[0-*]_weight:	Weight of the slave thermal zone on
> +				this thermal zone
>      |---sustainable_power:      Sustainable dissipatable power
>      |---k_po:                   Proportional term during temperature overshoot
>      |---k_pu:                   Proportional term during temperature undershoot
> @@ -296,6 +301,19 @@ policy
>  	One of the various thermal governors used for a particular zone.
>  	RW, Required
>  
> +aggregation_function
> +	The aggregation function used to calculate the temperature of
> +	this thermal zone when it has sub thermal zones.  It can
> +	either be "max" or "weighted_average".  If it's "max", the
> +	temperature of this thermal zone is the maximum of the
> +	temperatures of the sub thermal zones.  If
> +	aggregation_function is "weighted_average", the temperature of
> +	this thermal zone is the weighted average of the temperatures
> +	of the sub thermal zones.  The weights are specified in
> +	subtz[0-*]_weights.  The aggregation_function sysfs entry
> +	doesn't exist for thermal zones without sub thermal zones.
> +	RW, Optional
> +
>  available_policies
>  	Available thermal governors which can be used for a particular zone.
>  	RO, Required
> @@ -359,6 +377,21 @@ emul_temp
>  	  because userland can easily disable the thermal policy by simply
>  	  flooding this sysfs node with low temperature values.
>  
> +subtz[0-*]
> +	sysfs link to the slave thermal zone device.
> +	RO, Optional
> +
> +subtz[0-*]_weight
> +	The influence of subtz[0-*] in this thermal zone.  This
> +	parameter is only used when using "weighted_average" as the
> +	aggregation_function.  When calculating the temperature of
> +	this thermal zone, the temperature of each of the thermal
> +	zone's slaves is multiplied by its weight.  The result is
> +	divided by the sum of all weights.  If all weights are 0,
> +	the temperature of this thermal zone is the average of the
> +	temperatures of the sub thermal zones.
> +	RW, Optional
> +
>  sustainable_power
>  	An estimate of the sustained power that can be dissipated by
>  	the thermal zone. Used by the power allocator governor. For
> diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c
> index 98bdb63b3b95..2bb1af0a9349 100644
> --- a/drivers/thermal/thermal_core.c
> +++ b/drivers/thermal/thermal_core.c
> @@ -63,14 +63,23 @@ static struct thermal_governor *def_governor;
>  
>  /**
>   * struct thermal_zone_link - a link between "master" and "slave" thermal zones
> + * @id:		link number in the master thermal zone
> + * @name:	filename in the master thermal zone's sysfs directory
>   * @weight:	weight of the @slave thermal zone in the master
>   *		calculation.  Used when the master thermal zone's
>   *		aggregation function is THERMAL_AGG_WEIGHTED_AVG
> + * @weight_attr_name:	filename of the weight attribute in the master's thermal
> + *			zone sysfs directory
> + * @weight_attr:	device attribute for the weight entry in sysfs
>   * @slave:	pointer to the slave thermal zone
>   * @node:	list node in the master thermal zone's slave_tzs list.
>   */
>  struct thermal_zone_link {
> +	int id;
> +	char name[THERMAL_NAME_LENGTH];
>  	int weight;
> +	char weight_attr_name[THERMAL_NAME_LENGTH];

Are you sure you need this?

> +	struct device_attribute weight_attr;
>  	struct thermal_zone_device *slave;
>  	struct list_head node;
>  };
> @@ -990,6 +999,45 @@ emul_temp_store(struct device *dev, struct device_attribute *attr,
>  static DEVICE_ATTR(emul_temp, S_IWUSR, NULL, emul_temp_store);
>  
>  static ssize_t
> +agg_func_show(struct device *dev, struct device_attribute *devattr, char *buf)
> +{
> +	struct thermal_zone_device *tz = to_thermal_zone(dev);
> +
> +	if (tz->slave_agg_function == THERMAL_AGG_MAX)
> +		return sprintf(buf, "max\n");
> +	else if (tz->slave_agg_function == THERMAL_AGG_WEIGHTED_AVG)
> +		return sprintf(buf, "weighted_average\n");
> +	else
> +		return sprintf(buf, "(invalid)\n");
> +}
> +
> +static ssize_t
> +agg_func_store(struct device *dev, struct device_attribute *attr,
> +	       const char *buf, size_t count)
> +{
> +	int ret = 0;
> +	struct thermal_zone_device *tz = to_thermal_zone(dev);
> +	char func_name[THERMAL_NAME_LENGTH];
> +
> +	snprintf(func_name, sizeof(func_name), "%s", buf);
> +
> +	mutex_lock(&tz->lock);
> +
> +	if (sysfs_streq(func_name, "max"))
> +		tz->slave_agg_function = THERMAL_AGG_MAX;
> +	else if (sysfs_streq(func_name, "weighted_average"))
> +		tz->slave_agg_function = THERMAL_AGG_WEIGHTED_AVG;
> +	else
> +		ret = -EINVAL;
> +
> +	mutex_unlock(&tz->lock);
> +
> +	return ret ? ret : count;
> +}
> +static DEVICE_ATTR(aggregation_function, S_IRUGO | S_IWUSR, agg_func_show,
> +		   agg_func_store);
> +
> +static ssize_t
>  sustainable_power_show(struct device *dev, struct device_attribute *devattr,
>  		       char *buf)
>  {
> @@ -1305,6 +1353,35 @@ thermal_cooling_device_weight_store(struct device *dev,
>  
>  	return count;
>  }
> +
> +static ssize_t
> +thermal_subtz_weight_show(struct device *dev, struct device_attribute *attr,
> +			  char *buf)
> +{
> +	struct thermal_zone_link *link;
> +
> +	link = container_of(attr, struct thermal_zone_link, weight_attr);
> +
> +	return sprintf(buf, "%d\n", link->weight);
> +}
> +
> +static ssize_t
> +thermal_subtz_weight_store(struct device *dev, struct device_attribute *attr,
> +			   const char *buf, size_t count)
> +{
> +	struct thermal_zone_link *link;
> +	int ret, weight;
> +
> +	ret = kstrtoint(buf, 0, &weight);
> +	if (ret)
> +		return ret;
> +
> +	link = container_of(attr, struct thermal_zone_link, weight_attr);
> +	link->weight = weight;
> +
> +	return count;
> +}
> +

Now thinking of this, I see how you could input:

(0.25 * T0 - 2C) + (0.75 * T1 + 1C) //where T0 is sensor of subtz0 and
T1 is sensor of subtz1

for example. Which is a typical case, from what I have been seeing. 

How about if we change weight to: (ak/bk)*Tk + ck ?

>  /* Device management */
>  
>  /**
> @@ -2135,6 +2212,47 @@ int thermal_zone_add_subtz(struct thermal_zone_device *tz,
>  	link->weight = weight;
>  	list_add_tail(&link->node, &tz->slave_tzs);
>  
> +	if (list_is_singular(&tz->slave_tzs)) {
> +		ret = device_create_file(&tz->device,
> +					 &dev_attr_aggregation_function);
> +		if (ret)
> +			pr_warn("Failed to create aggregation_function sysfs file: %d\n",
> +				ret);
> +			/*
> +			 * Fall through: we failed to create the sysfs file, but
> +			 * it's not fatal
> +			 */
> +	}
> +
> +	ret = get_idr(&tz->link_idr, NULL, &link->id);
> +	if (ret) {
> +		/*
> +		 * Even if we can't create the symlink in sysfs,
> +		 * we've linked the thermal zones, so return success
> +		 */
> +		ret = 0;
> +		goto unlock;
> +	}
> +
> +	snprintf(link->name, ARRAY_SIZE(link->name), "subtz%d", link->id);
> +	ret = sysfs_create_link(&tz->device.kobj, &subtz->device.kobj,
> +				link->name);
> +	if (ret)
> +		pr_warn("Failed to create symlink to sub thermal zone: %d\n",
> +			ret);
> +
> +	snprintf(link->weight_attr_name, THERMAL_NAME_LENGTH, "subtz%d_weight",
> +		 link->id);
> +	sysfs_attr_init(&link->weight_attr.attr);
> +	link->weight_attr.attr.name = link->weight_attr_name;
> +	link->weight_attr.attr.mode = S_IWUSR | S_IRUGO;
> +	link->weight_attr.show = thermal_subtz_weight_show;
> +	link->weight_attr.store = thermal_subtz_weight_store;
> +	ret = device_create_file(&tz->device, &link->weight_attr);
> +	if (ret)
> +		pr_warn("Failed to create sub thermal zone weight: %d\n",
> +			ret);
> +
>  	ret = 0;
>  
>  unlock:
> @@ -2166,6 +2284,10 @@ int thermal_zone_del_subtz(struct thermal_zone_device *tz,
>  
>  	list_for_each_entry(link, &tz->slave_tzs, node) {
>  		if (link->slave == subtz) {
> +			device_remove_file(&tz->device, &link->weight_attr);
> +			sysfs_remove_link(&tz->device.kobj, link->name);
> +			release_idr(&tz->link_idr, NULL, link->id);
> +
>  			list_del(&link->node);
>  			kfree(link);
>  
> @@ -2174,6 +2296,9 @@ int thermal_zone_del_subtz(struct thermal_zone_device *tz,
>  		}
>  	}
>  
> +	if (list_empty(&tz->slave_tzs))
> +		device_remove_file(&tz->device, &dev_attr_aggregation_function);
> +
>  	mutex_unlock(&tz->lock);
>  
>  	return ret;
> diff --git a/include/linux/thermal.h b/include/linux/thermal.h
> index dc3d2ab77ab6..f14884353230 100644
> --- a/include/linux/thermal.h
> +++ b/include/linux/thermal.h
> @@ -192,6 +192,8 @@ struct thermal_attr {
>   * @thermal_instances:	list of &struct thermal_instance of this thermal zone
>   * @idr:	&struct idr to generate unique id for this zone's cooling
>   *		devices
> + * @link_idr:	&struct idr to generate unique ids for this zone's links to
> + *		other thermal zones
>   * @lock:	lock to protect thermal_instances list
>   * @node:	node in thermal_tz_list (in thermal_core.c)
>   * @poll_queue:	delayed work for polling
> @@ -221,6 +223,7 @@ struct thermal_zone_device {
>  	void *governor_data;
>  	struct list_head thermal_instances;
>  	struct idr idr;
> +	struct idr link_idr;
>  	struct mutex lock;
>  	struct list_head node;
>  	struct delayed_work poll_queue;
> -- 
> 1.9.1
> 

  reply	other threads:[~2016-03-03  3:17 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-11-25 15:09 [PATCH v3 0/4] Hierarchical thermal zones Javi Merino
2015-11-25 15:09 ` [PATCH v3 1/4] thermal: Add support for hierarchical " Javi Merino
2016-03-03  3:12   ` Eduardo Valentin
2016-03-03  3:23     ` Eduardo Valentin
2015-11-25 15:09 ` [PATCH v3 2/4] devicetree: bindings: let thermal-sensor point to other " Javi Merino
2015-11-25 17:54   ` Mark Rutland
2015-11-25 18:41     ` Javi Merino
2015-11-25 18:41       ` Javi Merino
2016-01-04 14:17   ` Sascha Hauer
2016-03-03  3:21     ` Eduardo Valentin
2016-03-21 11:55       ` Javi Merino
2016-03-22 15:13         ` Eduardo Valentin
2016-03-22 15:13           ` Eduardo Valentin
2016-03-03  3:19   ` Eduardo Valentin
2015-11-25 15:09 ` [PATCH v3 3/4] thermal: of: parse stacked thermal zones from device tree Javi Merino
2015-11-25 15:09 ` [PATCH v3 4/4] thermal: show the sub thermal zones in sysfs Javi Merino
2016-03-03  3:17   ` Eduardo Valentin [this message]
2016-03-16 22:06   ` Eduardo Valentin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20160303031700.GB3379@localhost.localdomain \
    --to=edubezval@gmail.com \
    --cc=javi.merino@arm.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pm@vger.kernel.org \
    --cc=rui.zang@intel.com \
    --cc=rui.zhang@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.