Thermal framework today supports monitoring for rising temperatures and
subsequently initiating cooling action in case of a trip point being
crossed. There are scenarios where a SoC needs some warming action to be
activated if the temperature falls below a cetain allowable limit.
Since warming action can be considered mirror opposite of cooling action,
most of the thermal framework can be re-used to achieve this.
To enable thermal framework to monitor falling temperature, a new parameter
is added to the thermal trip point binding in the device tree to indicate
the direction(rising/falling) of temperature monitoring. Thermal DT
driver is extended to capture this information from the device tree
entries and to reflect it in the thermal framework as a new enum
variable in the thermal trip point structure.
As an initial attempt, step-wise governor is extended to support
bi-directional monitoring of temprature if a trip point is hit, depending
on the newly introduced enum variable. Finally thermal sysfs entries are
extended to indicate the trip point monitor direction.
Patch series introducing various resources that are used as warming devices
on Qualcomm sdm845:
https://lkml.org/lkml/2019/7/29/749 (already merged)
https://lkml.org/lkml/2019/9/10/727 (under review)
Thara Gopinath (4):
dt-bindings: thermal: Introduce monitor-falling binding to thermal
trip point description
thermal: Thermal core and sysfs changes needed to support
bi-directional monitoring of trip points.
thermal: of-thermal: Extend thermal dt driver to support
bi-directional monitoring of a thermal trip point.
thermal: step_wise: Extend thermal step-wise governor to monitor
falling temperature.
.../devicetree/bindings/thermal/thermal.txt | 8 +++
drivers/thermal/of-thermal.c | 22 ++++++++
drivers/thermal/step_wise.c | 59 +++++++++++++++------
drivers/thermal/thermal_sysfs.c | 60 ++++++++++++++++++++--
include/linux/thermal.h | 10 ++++
include/uapi/linux/thermal.h | 2 +-
6 files changed, 141 insertions(+), 20 deletions(-)
--
2.1.4
From the step wise governor point of view, the policy decisions
that has to taken on a thermal trip point that is defined to be monitored
for falling temprature is the mirror opposite of the decisions it has
to take on a trip point that is monitored for rising temperature.
Signed-off-by: Thara Gopinath <[email protected]>
---
drivers/thermal/step_wise.c | 59 +++++++++++++++++++++++++++++++++------------
1 file changed, 44 insertions(+), 15 deletions(-)
diff --git a/drivers/thermal/step_wise.c b/drivers/thermal/step_wise.c
index 6e051cb..aa8e0a0 100644
--- a/drivers/thermal/step_wise.c
+++ b/drivers/thermal/step_wise.c
@@ -35,7 +35,8 @@
* deactivate the thermal instance
*/
static unsigned long get_target_state(struct thermal_instance *instance,
- enum thermal_trend trend, bool throttle)
+ enum thermal_trend trend, bool throttle,
+ enum thermal_trip_monitor_type type)
{
struct thermal_cooling_device *cdev = instance->cdev;
unsigned long cur_state;
@@ -65,11 +66,21 @@ static unsigned long get_target_state(struct thermal_instance *instance,
switch (trend) {
case THERMAL_TREND_RAISING:
- if (throttle) {
- next_target = cur_state < instance->upper ?
- (cur_state + 1) : instance->upper;
- if (next_target < instance->lower)
- next_target = instance->lower;
+ if (type == THERMAL_TRIP_MONITOR_FALLING) {
+ if (cur_state <= instance->lower) {
+ if (!throttle)
+ next_target = THERMAL_NO_TARGET;
+ } else {
+ if (!throttle)
+ next_target = cur_state - 1;
+ }
+ } else {
+ if (throttle) {
+ next_target = cur_state < instance->upper ?
+ (cur_state + 1) : instance->upper;
+ if (next_target < instance->lower)
+ next_target = instance->lower;
+ }
}
break;
case THERMAL_TREND_RAISE_FULL:
@@ -77,14 +88,23 @@ static unsigned long get_target_state(struct thermal_instance *instance,
next_target = instance->upper;
break;
case THERMAL_TREND_DROPPING:
- if (cur_state <= instance->lower) {
- if (!throttle)
- next_target = THERMAL_NO_TARGET;
+ if (type == THERMAL_TRIP_MONITOR_FALLING) {
+ if (throttle) {
+ next_target = cur_state < instance->upper ?
+ (cur_state + 1) : instance->upper;
+ if (next_target < instance->lower)
+ next_target = instance->lower;
+ }
} else {
- if (!throttle) {
- next_target = cur_state - 1;
- if (next_target > instance->upper)
- next_target = instance->upper;
+ if (cur_state <= instance->lower) {
+ if (!throttle)
+ next_target = THERMAL_NO_TARGET;
+ } else {
+ if (!throttle) {
+ next_target = cur_state - 1;
+ if (next_target > instance->upper)
+ next_target = instance->upper;
+ }
}
}
break;
@@ -117,6 +137,8 @@ static void thermal_zone_trip_update(struct thermal_zone_device *tz, int trip)
{
int trip_temp;
enum thermal_trip_type trip_type;
+ enum thermal_trip_monitor_type monitor_type =
+ THERMAL_TRIP_MONITOR_RISING;
enum thermal_trend trend;
struct thermal_instance *instance;
bool throttle = false;
@@ -130,9 +152,15 @@ static void thermal_zone_trip_update(struct thermal_zone_device *tz, int trip)
tz->ops->get_trip_type(tz, trip, &trip_type);
}
+ if (tz->ops->get_trip_monitor_type)
+ tz->ops->get_trip_monitor_type(tz, trip, &monitor_type);
+
trend = get_tz_trend(tz, trip);
- if (tz->temperature >= trip_temp) {
+ if (((monitor_type == THERMAL_TRIP_MONITOR_RISING) &&
+ (tz->temperature >= trip_temp)) ||
+ ((monitor_type == THERMAL_TRIP_MONITOR_FALLING) &&
+ (tz->temperature <= trip_temp))) {
throttle = true;
trace_thermal_zone_trip(tz, trip, trip_type);
}
@@ -147,7 +175,8 @@ static void thermal_zone_trip_update(struct thermal_zone_device *tz, int trip)
continue;
old_target = instance->target;
- instance->target = get_target_state(instance, trend, throttle);
+ instance->target = get_target_state(instance, trend, throttle,
+ monitor_type);
dev_dbg(&instance->cdev->device, "old_target=%d, target=%d\n",
old_target, (int)instance->target);
--
2.1.4
Introduce a new binding parameter to thermal trip point description
to indicate whether the temperature level specified by the trip point
is monitored for a rise or fall in temperature.
Signed-off-by: Thara Gopinath <[email protected]>
---
Documentation/devicetree/bindings/thermal/thermal.txt | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/Documentation/devicetree/bindings/thermal/thermal.txt b/Documentation/devicetree/bindings/thermal/thermal.txt
index ca14ba9..849a2a9 100644
--- a/Documentation/devicetree/bindings/thermal/thermal.txt
+++ b/Documentation/devicetree/bindings/thermal/thermal.txt
@@ -90,6 +90,14 @@ Required properties:
"critical": Hardware not reliable.
Type: string
+Optional property:
+- monitor-falling: Indicate whether the system action is kick
+ Type: boolean started when the temperature falls below or rises
+ above the trip temperature level indicated in
+ "temperature".If true, the trip point is monitored
+ for falling temperature else the trip point is
+ monitored for rising temperature.
+
* Cooling device maps
The cooling device maps node is a node to describe how cooling devices
--
2.1.4
On Wed, Sep 18, 2019 at 10:18:20PM -0400, Thara Gopinath wrote:
> Introduce a new binding parameter to thermal trip point description
> to indicate whether the temperature level specified by the trip point
> is monitored for a rise or fall in temperature.
What if it is both?
When do you need this? Seems like you'd always want to monitor both
directions to undo any action done on rising temp. Unless you want a
hysteresis, but this doesn't seem like the best way to implement that.
>
> Signed-off-by: Thara Gopinath <[email protected]>
> ---
> Documentation/devicetree/bindings/thermal/thermal.txt | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/Documentation/devicetree/bindings/thermal/thermal.txt b/Documentation/devicetree/bindings/thermal/thermal.txt
> index ca14ba9..849a2a9 100644
> --- a/Documentation/devicetree/bindings/thermal/thermal.txt
> +++ b/Documentation/devicetree/bindings/thermal/thermal.txt
> @@ -90,6 +90,14 @@ Required properties:
> "critical": Hardware not reliable.
> Type: string
>
> +Optional property:
> +- monitor-falling: Indicate whether the system action is kick
> + Type: boolean started when the temperature falls below or rises
> + above the trip temperature level indicated in
> + "temperature".If true, the trip point is monitored
> + for falling temperature else the trip point is
> + monitored for rising temperature.
> +
> * Cooling device maps
>
> The cooling device maps node is a node to describe how cooling devices
> --
> 2.1.4
>
Hi Rob,
Thanks for the review.
On 10/01/2019 06:09 PM, Rob Herring wrote:
> On Wed, Sep 18, 2019 at 10:18:20PM -0400, Thara Gopinath wrote:
>> Introduce a new binding parameter to thermal trip point description
>> to indicate whether the temperature level specified by the trip point
>> is monitored for a rise or fall in temperature.
>
> What if it is both?
>
> When do you need this? Seems like you'd always want to monitor both
> directions to undo any action done on rising temp. Unless you want a
> hysteresis, but this doesn't seem like the best way to implement that.
>
The thermal framework is designed in such a manner that I cannot think
of a use case for both.
The framework takes care of removing the warming/cooling action when the
trip point is crossed in the opposite direction. It only needs an
indication on when to start implementing the
action.
For eg. When the temperature crosses/increases above 90 degree, the
framework will start the cooling action and will continue monitoring
till the temperature falls below 90 and the cooling action is removed.
Vice versa when the temperature decreases below say 5 degree, the
framework should initiate the warming action and monitor till the
temperature rises above and remove the warming action.
So the trip point is really an indication of the temperature crossing a
threshold in the specified direction.
Now this parameter is needed to indicate whether the thermal framework
has to start implementing the warming/cooling action when the
temperature rises above the trip point or falls below the trip point.
Till now the framework was always assuming that the cooling action had
to be implemented when temperature rises above the trip point.
--
Warm Regards
Thara
On 09/18/2019 10:18 PM, Thara Gopinath wrote:
> Thermal framework today supports monitoring for rising temperatures and
> subsequently initiating cooling action in case of a trip point being
> crossed. There are scenarios where a SoC needs some warming action to be
> activated if the temperature falls below a cetain allowable limit.
> Since warming action can be considered mirror opposite of cooling action,
> most of the thermal framework can be re-used to achieve this.
>
> To enable thermal framework to monitor falling temperature, a new parameter
> is added to the thermal trip point binding in the device tree to indicate
> the direction(rising/falling) of temperature monitoring. Thermal DT
> driver is extended to capture this information from the device tree
> entries and to reflect it in the thermal framework as a new enum
> variable in the thermal trip point structure.
> As an initial attempt, step-wise governor is extended to support
> bi-directional monitoring of temprature if a trip point is hit, depending
> on the newly introduced enum variable. Finally thermal sysfs entries are
> extended to indicate the trip point monitor direction.
>
> Patch series introducing various resources that are used as warming devices
> on Qualcomm sdm845:
> https://lkml.org/lkml/2019/7/29/749 (already merged)
>
> https://lkml.org/lkml/2019/9/10/727 (under review)
Gentle reminder for reviews!
>
>
> Thara Gopinath (4):
> dt-bindings: thermal: Introduce monitor-falling binding to thermal
> trip point description
> thermal: Thermal core and sysfs changes needed to support
> bi-directional monitoring of trip points.
> thermal: of-thermal: Extend thermal dt driver to support
> bi-directional monitoring of a thermal trip point.
> thermal: step_wise: Extend thermal step-wise governor to monitor
> falling temperature.
>
> .../devicetree/bindings/thermal/thermal.txt | 8 +++
> drivers/thermal/of-thermal.c | 22 ++++++++
> drivers/thermal/step_wise.c | 59 +++++++++++++++------
> drivers/thermal/thermal_sysfs.c | 60 ++++++++++++++++++++--
> include/linux/thermal.h | 10 ++++
> include/uapi/linux/thermal.h | 2 +-
> 6 files changed, 141 insertions(+), 20 deletions(-)
>
--
Warm Regards
Thara
On 9/18/2019 8:18 PM, Thara Gopinath wrote:
>>From the step wise governor point of view, the policy decisions
> that has to taken on a thermal trip point that is defined to be monitored
> for falling temprature is the mirror opposite of the decisions it has
> to take on a trip point that is monitored for rising temperature.
>
> Signed-off-by: Thara Gopinath <[email protected]>
> ---
> drivers/thermal/step_wise.c | 59 +++++++++++++++++++++++++++++++++------------
> 1 file changed, 44 insertions(+), 15 deletions(-)
>
> diff --git a/drivers/thermal/step_wise.c b/drivers/thermal/step_wise.c
> index 6e051cb..aa8e0a0 100644
> --- a/drivers/thermal/step_wise.c
> +++ b/drivers/thermal/step_wise.c
> @@ -35,7 +35,8 @@
> * deactivate the thermal instance
> */
> static unsigned long get_target_state(struct thermal_instance *instance,
> - enum thermal_trend trend, bool throttle)
> + enum thermal_trend trend, bool throttle,
> + enum thermal_trip_monitor_type type)
> {
> struct thermal_cooling_device *cdev = instance->cdev;
> unsigned long cur_state;
> @@ -65,11 +66,21 @@ static unsigned long get_target_state(struct thermal_instance *instance,
>
> switch (trend) {
> case THERMAL_TREND_RAISING:
> - if (throttle) {
> - next_target = cur_state < instance->upper ?
> - (cur_state + 1) : instance->upper;
> - if (next_target < instance->lower)
> - next_target = instance->lower;
> + if (type == THERMAL_TRIP_MONITOR_FALLING) {
> + if (cur_state <= instance->lower) {
> + if (!throttle)
> + next_target = THERMAL_NO_TARGET;
> + } else {
> + if (!throttle)
> + next_target = cur_state - 1;
> + }
> + } else {
> + if (throttle) {
> + next_target = cur_state < instance->upper ?
> + (cur_state + 1) : instance->upper;
> + if (next_target < instance->lower)
> + next_target = instance->lower;
> + }
> }
> break;
> case THERMAL_TREND_RAISE_FULL:
> @@ -77,14 +88,23 @@ static unsigned long get_target_state(struct thermal_instance *instance,
> next_target = instance->upper;
> break;
> case THERMAL_TREND_DROPPING:
> - if (cur_state <= instance->lower) {
> - if (!throttle)
> - next_target = THERMAL_NO_TARGET;
> + if (type == THERMAL_TRIP_MONITOR_FALLING) {
> + if (throttle) {
> + next_target = cur_state < instance->upper ?
> + (cur_state + 1) : instance->upper;
> + if (next_target < instance->lower)
> + next_target = instance->lower;
> + }
> } else {
> - if (!throttle) {
> - next_target = cur_state - 1;
> - if (next_target > instance->upper)
> - next_target = instance->upper;
> + if (cur_state <= instance->lower) {
> + if (!throttle)
> + next_target = THERMAL_NO_TARGET;
> + } else {
> + if (!throttle) {
> + next_target = cur_state - 1;
> + if (next_target > instance->upper)
> + next_target = instance->upper;
> + }
> }
> }
> break;
> @@ -117,6 +137,8 @@ static void thermal_zone_trip_update(struct thermal_zone_device *tz, int trip)
> {
> int trip_temp;
> enum thermal_trip_type trip_type;
> + enum thermal_trip_monitor_type monitor_type =
> + THERMAL_TRIP_MONITOR_RISING;
> enum thermal_trend trend;
> struct thermal_instance *instance;
> bool throttle = false;
> @@ -130,9 +152,15 @@ static void thermal_zone_trip_update(struct thermal_zone_device *tz, int trip)
> tz->ops->get_trip_type(tz, trip, &trip_type);
> }
>
> + if (tz->ops->get_trip_monitor_type)
> + tz->ops->get_trip_monitor_type(tz, trip, &monitor_type);
> +
> trend = get_tz_trend(tz, trip);
>
> - if (tz->temperature >= trip_temp) {
> + if (((monitor_type == THERMAL_TRIP_MONITOR_RISING) &&
> + (tz->temperature >= trip_temp)) ||
> + ((monitor_type == THERMAL_TRIP_MONITOR_FALLING) &&
> + (tz->temperature <= trip_temp))) {
Governors monitoring warming devices need to have support for
hysteresis. Assume a case where the device is in idle when the
temperature goes below threshold and we trigger a mitigation. Even a
minimal workload or even the processing of the threshold by the governor
could warm the device and put the temperature above the threshold and we
will have to remove any mitigation. To avoid this ping-pong, its best to
add a hysteresis support.
> throttle = true;
> trace_thermal_zone_trip(tz, trip, trip_type);
> }
> @@ -147,7 +175,8 @@ static void thermal_zone_trip_update(struct thermal_zone_device *tz, int trip)
> continue;
>
> old_target = instance->target;
> - instance->target = get_target_state(instance, trend, throttle);
> + instance->target = get_target_state(instance, trend, throttle,
> + monitor_type);
> dev_dbg(&instance->cdev->device, "old_target=%d, target=%d\n",
> old_target, (int)instance->target);
>
>
On Thu, Sep 19, 2019 at 7:48 AM Thara Gopinath
<[email protected]> wrote:
>
> Introduce a new binding parameter to thermal trip point description
> to indicate whether the temperature level specified by the trip point
> is monitored for a rise or fall in temperature.
>
> Signed-off-by: Thara Gopinath <[email protected]>
> ---
> Documentation/devicetree/bindings/thermal/thermal.txt | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/Documentation/devicetree/bindings/thermal/thermal.txt b/Documentation/devicetree/bindings/thermal/thermal.txt
> index ca14ba9..849a2a9 100644
> --- a/Documentation/devicetree/bindings/thermal/thermal.txt
> +++ b/Documentation/devicetree/bindings/thermal/thermal.txt
> @@ -90,6 +90,14 @@ Required properties:
> "critical": Hardware not reliable.
> Type: string
>
> +Optional property:
> +- monitor-falling: Indicate whether the system action is kick
Stray space after :
> + Type: boolean started when the temperature falls below or rises
Unnecessary tab after boolean (I'll fix up the rest of the file in the
yaml conversion)
I suggest not making this boolean. Just use the property as a flag by
itself to denote a falling trip point. No need to deal with true/false
values.
Similarly, the sysfs file would show up only in case of a trip that
sets this flag and just contain a 1, for example.
> + above the trip temperature level indicated in
> + "temperature".If true, the trip point is monitored
Add space after full stop.
> + for falling temperature else the trip point is
> + monitored for rising temperature.
> +
> * Cooling device maps
>
> The cooling device maps node is a node to describe how cooling devices
> --
> 2.1.4
>