[PATCH 1/5] thermal/core: Emit a warning if the thermal zone is updated without ops

[PATCH 1/5] thermal/core: Emit a warning if the thermal zone is updated without ops

[Daniel Lezcano]

The actual code is silently ignoring a thermal zone update when a
driver is requesting it without a get_temp ops set.

That looks not correct, as the caller should not have called this
function if the thermal zone is unable to read the temperature.

That makes the code less robust as the check won't detect the driver
is inconsistently using the thermal API and that does not help to
improve the framework as these circumvolutions hide the problem at the
source.

In order to detect the situation when it happens, let's add a warning
when the update is requested without the get_temp() ops set.

Any warning emitted will have to be fixed at the source of the
problem: the caller must not call thermal_zone_device_update if there
is not get_temp callback set.

Cc: Thara Gopinath <thara.gopinath@linaro.org>
Cc: Amit Kucheria <amitk@kernel.org>
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
---
 drivers/thermal/thermal_core.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c
index dee40ff41177..e6771e5aeedb 100644
--- a/drivers/thermal/thermal_core.c
+++ b/drivers/thermal/thermal_core.c
@@ -548,7 +548,8 @@ void thermal_zone_device_update(struct thermal_zone_device *tz,
 	if (atomic_read(&in_suspend))
 		return;
 
-	if (!tz->ops->get_temp)
+	if (WARN_ONCE(!tz->ops->get_temp, "'%s' must not be called without "
+		      "'get_temp' ops set\n", __func__))
 		return;
 
 	update_temperature(tz);
-- 
2.17.1

[PATCH 2/5] thermal/core: Add critical and hot ops

[Daniel Lezcano]

Currently there is no way to the sensors to directly call an ops in
interrupt mode without calling thermal_zone_device_update assuming all
the trip points are defined.

A sensor may want to do something special if a trip point is hot or
critical.

This patch adds the critical and hot ops to the thermal zone device,
so a sensor can directly invoke them or let the thermal framework to
call the sensor specific ones.

Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
---
 drivers/thermal/thermal_core.c | 43 +++++++++++++++++++++-------------
 include/linux/thermal.h        |  3 +++
 2 files changed, 30 insertions(+), 16 deletions(-)

diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c
index e6771e5aeedb..cee0b31b5cd7 100644
--- a/drivers/thermal/thermal_core.c
+++ b/drivers/thermal/thermal_core.c
@@ -375,6 +375,25 @@ static void thermal_emergency_poweroff(void)
 			      msecs_to_jiffies(poweroff_delay_ms));
 }
 
+void thermal_zone_device_critical(struct thermal_zone_device *tz)
+{
+	dev_emerg(&tz->device, "%s: critical temperature reached, "
+		  "shutting down\n", tz->type);
+
+	mutex_lock(&poweroff_lock);
+	if (!power_off_triggered) {
+		/*
+		 * Queue a backup emergency shutdown in the event of
+		 * orderly_poweroff failure
+		 */
+		thermal_emergency_poweroff();
+		orderly_poweroff(true);
+		power_off_triggered = true;
+	}
+	mutex_unlock(&poweroff_lock);
+}
+EXPORT_SYMBOL(thermal_zone_device_critical);
+
 static void handle_critical_trips(struct thermal_zone_device *tz,
 				  int trip, enum thermal_trip_type trip_type)
 {
@@ -391,22 +410,10 @@ static void handle_critical_trips(struct thermal_zone_device *tz,
 	if (tz->ops->notify)
 		tz->ops->notify(tz, trip, trip_type);
 
-	if (trip_type == THERMAL_TRIP_CRITICAL) {
-		dev_emerg(&tz->device,
-			  "critical temperature reached (%d C), shutting down\n",
-			  tz->temperature / 1000);
-		mutex_lock(&poweroff_lock);
-		if (!power_off_triggered) {
-			/*
-			 * Queue a backup emergency shutdown in the event of
-			 * orderly_poweroff failure
-			 */
-			thermal_emergency_poweroff();
-			orderly_poweroff(true);
-			power_off_triggered = true;
-		}
-		mutex_unlock(&poweroff_lock);
-	}
+	if (trip_type == THERMAL_TRIP_HOT && tz->ops->hot)
+		tz->ops->hot(tz);
+	else if (trip_type == THERMAL_TRIP_CRITICAL)
+		tz->ops->critical(tz);
 }
 
 static void handle_thermal_trip(struct thermal_zone_device *tz, int trip)
@@ -1331,6 +1338,10 @@ thermal_zone_device_register(const char *type, int trips, int mask,
 
 	tz->id = id;
 	strlcpy(tz->type, type, sizeof(tz->type));
+
+	if (!ops->critical)
+		ops->critical = thermal_zone_device_critical;
+
 	tz->ops = ops;
 	tz->tzp = tzp;
 	tz->device.class = &thermal_class;
diff --git a/include/linux/thermal.h b/include/linux/thermal.h
index f23a388ded15..125c8a4d52e6 100644
--- a/include/linux/thermal.h
+++ b/include/linux/thermal.h
@@ -79,6 +79,8 @@ struct thermal_zone_device_ops {
 			  enum thermal_trend *);
 	int (*notify) (struct thermal_zone_device *, int,
 		       enum thermal_trip_type);
+	void (*hot)(struct thermal_zone_device *);
+	void (*critical)(struct thermal_zone_device *);
 };
 
 struct thermal_cooling_device_ops {
@@ -399,6 +401,7 @@ void thermal_cdev_update(struct thermal_cooling_device *);
 void thermal_notify_framework(struct thermal_zone_device *, int);
 int thermal_zone_device_enable(struct thermal_zone_device *tz);
 int thermal_zone_device_disable(struct thermal_zone_device *tz);
+void thermal_zone_device_critical(struct thermal_zone_device *tz);
 #else
 static inline struct thermal_zone_device *thermal_zone_device_register(
 	const char *type, int trips, int mask, void *devdata,
-- 
2.17.1

[PATCH 3/5] thermal/drivers/acpi: Use hot and critical ops

[Daniel Lezcano]

The acpi driver wants to do a netlink notification in case of a hot or
critical trip point. Implement the corresponding ops to be used for
the thermal zone and use them instead of the notify ops.

Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
---
 drivers/acpi/thermal.c | 30 ++++++++++++++----------------
 1 file changed, 14 insertions(+), 16 deletions(-)

diff --git a/drivers/acpi/thermal.c b/drivers/acpi/thermal.c
index 12c0ece746f0..b5e4bc9e3282 100644
--- a/drivers/acpi/thermal.c
+++ b/drivers/acpi/thermal.c
@@ -677,27 +677,24 @@ static int thermal_get_trend(struct thermal_zone_device *thermal,
 	return 0;
 }
 
-
-static int thermal_notify(struct thermal_zone_device *thermal, int trip,
-			   enum thermal_trip_type trip_type)
+static void acpi_thermal_zone_device_hot(struct thermal_zone_device *thermal)
 {
-	u8 type = 0;
 	struct acpi_thermal *tz = thermal->devdata;
 
-	if (trip_type == THERMAL_TRIP_CRITICAL)
-		type = ACPI_THERMAL_NOTIFY_CRITICAL;
-	else if (trip_type == THERMAL_TRIP_HOT)
-		type = ACPI_THERMAL_NOTIFY_HOT;
-	else
-		return 0;
-
 	acpi_bus_generate_netlink_event(tz->device->pnp.device_class,
-					dev_name(&tz->device->dev), type, 1);
+					dev_name(&tz->device->dev),
+					ACPI_THERMAL_NOTIFY_HOT, 1);
+}
 
-	if (trip_type == THERMAL_TRIP_CRITICAL && nocrt)
-		return 1;
+static void acpi_thermal_zone_device_critical(struct thermal_zone_device *thermal)
+{
+	struct acpi_thermal *tz = thermal->devdata;
 
-	return 0;
+	acpi_bus_generate_netlink_event(tz->device->pnp.device_class,
+					dev_name(&tz->device->dev),
+					ACPI_THERMAL_NOTIFY_CRITICAL, 1);
+
+	thermal_zone_device_critical(thermal);
 }
 
 static int acpi_thermal_cooling_device_cb(struct thermal_zone_device *thermal,
@@ -812,7 +809,8 @@ static struct thermal_zone_device_ops acpi_thermal_zone_ops = {
 	.get_trip_temp = thermal_get_trip_temp,
 	.get_crit_temp = thermal_get_crit_temp,
 	.get_trend = thermal_get_trend,
-	.notify = thermal_notify,
+	.hot = acpi_thermal_zone_device_hot,
+	.critical = acpi_thermal_zone_device_critical,
 };
 
 static int acpi_thermal_register_thermal_zone(struct acpi_thermal *tz)
-- 
2.17.1

[PATCH 4/5] thermal/drivers/rcar: Remove notification usage

[Daniel Lezcano]

The ops is only showing a trace telling a critical trip point is
crossed. The same information is given by the thermal framework.

This is redundant, remove the code.

Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
---
 drivers/thermal/rcar_thermal.c | 19 -------------------
 1 file changed, 19 deletions(-)

diff --git a/drivers/thermal/rcar_thermal.c b/drivers/thermal/rcar_thermal.c
index 6ae757d66f46..b49f04daaf47 100644
--- a/drivers/thermal/rcar_thermal.c
+++ b/drivers/thermal/rcar_thermal.c
@@ -323,24 +323,6 @@ static int rcar_thermal_get_trip_temp(struct thermal_zone_device *zone,
 	return 0;
 }
 
-static int rcar_thermal_notify(struct thermal_zone_device *zone,
-			       int trip, enum thermal_trip_type type)
-{
-	struct rcar_thermal_priv *priv = rcar_zone_to_priv(zone);
-	struct device *dev = rcar_priv_to_dev(priv);
-
-	switch (type) {
-	case THERMAL_TRIP_CRITICAL:
-		/* FIXME */
-		dev_warn(dev, "Thermal reached to critical temperature\n");
-		break;
-	default:
-		break;
-	}
-
-	return 0;
-}
-
 static const struct thermal_zone_of_device_ops rcar_thermal_zone_of_ops = {
 	.get_temp	= rcar_thermal_of_get_temp,
 };
@@ -349,7 +331,6 @@ static struct thermal_zone_device_ops rcar_thermal_zone_ops = {
 	.get_temp	= rcar_thermal_get_temp,
 	.get_trip_type	= rcar_thermal_get_trip_type,
 	.get_trip_temp	= rcar_thermal_get_trip_temp,
-	.notify		= rcar_thermal_notify,
 };
 
 /*
-- 
2.17.1

[PATCH 5/5] thermal/core: Remove notify ops

[Daniel Lezcano]

With the remove of the notify user in a previous patch, the ops is no
longer needed, remove it.

Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
---
 drivers/thermal/thermal_core.c | 3 ---
 include/linux/thermal.h        | 2 --
 2 files changed, 5 deletions(-)

diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c
index cee0b31b5cd7..d7481fdf4e4c 100644
--- a/drivers/thermal/thermal_core.c
+++ b/drivers/thermal/thermal_core.c
@@ -407,9 +407,6 @@ static void handle_critical_trips(struct thermal_zone_device *tz,
 
 	trace_thermal_zone_trip(tz, trip, trip_type);
 
-	if (tz->ops->notify)
-		tz->ops->notify(tz, trip, trip_type);
-
 	if (trip_type == THERMAL_TRIP_HOT && tz->ops->hot)
 		tz->ops->hot(tz);
 	else if (trip_type == THERMAL_TRIP_CRITICAL)
diff --git a/include/linux/thermal.h b/include/linux/thermal.h
index 125c8a4d52e6..7e051b4cf715 100644
--- a/include/linux/thermal.h
+++ b/include/linux/thermal.h
@@ -77,8 +77,6 @@ struct thermal_zone_device_ops {
 	int (*set_emul_temp) (struct thermal_zone_device *, int);
 	int (*get_trend) (struct thermal_zone_device *, int,
 			  enum thermal_trend *);
-	int (*notify) (struct thermal_zone_device *, int,
-		       enum thermal_trip_type);
 	void (*hot)(struct thermal_zone_device *);
 	void (*critical)(struct thermal_zone_device *);
 };
-- 
2.17.1

Re: [PATCH 2/5] thermal/core: Add critical and hot ops

[Lukasz Luba]

On 12/10/20 12:15 PM, Daniel Lezcano wrote:
> Currently there is no way to the sensors to directly call an ops in
> interrupt mode without calling thermal_zone_device_update assuming all
> the trip points are defined.
> 
> A sensor may want to do something special if a trip point is hot or
> critical.
> 
> This patch adds the critical and hot ops to the thermal zone device,
> so a sensor can directly invoke them or let the thermal framework to
> call the sensor specific ones.
> 
> Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
> Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
> ---
>   drivers/thermal/thermal_core.c | 43 +++++++++++++++++++++-------------
>   include/linux/thermal.h        |  3 +++
>   2 files changed, 30 insertions(+), 16 deletions(-)
> 
> diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c
> index e6771e5aeedb..cee0b31b5cd7 100644
> --- a/drivers/thermal/thermal_core.c
> +++ b/drivers/thermal/thermal_core.c
> @@ -375,6 +375,25 @@ static void thermal_emergency_poweroff(void)
>   			      msecs_to_jiffies(poweroff_delay_ms));
>   }
>   
> +void thermal_zone_device_critical(struct thermal_zone_device *tz)
> +{
> +	dev_emerg(&tz->device, "%s: critical temperature reached, "
> +		  "shutting down\n", tz->type);
> +
> +	mutex_lock(&poweroff_lock);
> +	if (!power_off_triggered) {
> +		/*
> +		 * Queue a backup emergency shutdown in the event of
> +		 * orderly_poweroff failure
> +		 */
> +		thermal_emergency_poweroff();
> +		orderly_poweroff(true);
> +		power_off_triggered = true;
> +	}
> +	mutex_unlock(&poweroff_lock);
> +}
> +EXPORT_SYMBOL(thermal_zone_device_critical);
> +
>   static void handle_critical_trips(struct thermal_zone_device *tz,
>   				  int trip, enum thermal_trip_type trip_type)
>   {
> @@ -391,22 +410,10 @@ static void handle_critical_trips(struct thermal_zone_device *tz,
>   	if (tz->ops->notify)
>   		tz->ops->notify(tz, trip, trip_type);
>   
> -	if (trip_type == THERMAL_TRIP_CRITICAL) {
> -		dev_emerg(&tz->device,
> -			  "critical temperature reached (%d C), shutting down\n",
> -			  tz->temperature / 1000);
> -		mutex_lock(&poweroff_lock);
> -		if (!power_off_triggered) {
> -			/*
> -			 * Queue a backup emergency shutdown in the event of
> -			 * orderly_poweroff failure
> -			 */
> -			thermal_emergency_poweroff();
> -			orderly_poweroff(true);
> -			power_off_triggered = true;
> -		}
> -		mutex_unlock(&poweroff_lock);
> -	}
> +	if (trip_type == THERMAL_TRIP_HOT && tz->ops->hot)
> +		tz->ops->hot(tz);
> +	else if (trip_type == THERMAL_TRIP_CRITICAL)
> +		tz->ops->critical(tz);

I can see that in the patch 3/5 there driver .critical() callback
calls framework thermal_zone_device_critical() at the end.
I wonder if we could always call this framework function.

Something like a helper function always called:
void _handle_critical( *tz)
{
	if (tz->ops->critical)
		tz->ops->critical(tz);

	thermal_zone_device_critical(tz);
}

and then called:

	else if (trip_type == THERMAL_TRIP_CRITICAL)
		_handle_critical(tz);

>   }
>   
>   static void handle_thermal_trip(struct thermal_zone_device *tz, int trip)
> @@ -1331,6 +1338,10 @@ thermal_zone_device_register(const char *type, int trips, int mask,
>   
>   	tz->id = id;
>   	strlcpy(tz->type, type, sizeof(tz->type));
> +
> +	if (!ops->critical)
> +		ops->critical = thermal_zone_device_critical;

Then we could drop this default assignment.

> +
>   	tz->ops = ops;
>   	tz->tzp = tzp;
>   	tz->device.class = &thermal_class;
> diff --git a/include/linux/thermal.h b/include/linux/thermal.h
> index f23a388ded15..125c8a4d52e6 100644
> --- a/include/linux/thermal.h
> +++ b/include/linux/thermal.h
> @@ -79,6 +79,8 @@ struct thermal_zone_device_ops {
>   			  enum thermal_trend *);
>   	int (*notify) (struct thermal_zone_device *, int,
>   		       enum thermal_trip_type);
> +	void (*hot)(struct thermal_zone_device *);
> +	void (*critical)(struct thermal_zone_device *);
>   };
>   
>   struct thermal_cooling_device_ops {
> @@ -399,6 +401,7 @@ void thermal_cdev_update(struct thermal_cooling_device *);
>   void thermal_notify_framework(struct thermal_zone_device *, int);
>   int thermal_zone_device_enable(struct thermal_zone_device *tz);
>   int thermal_zone_device_disable(struct thermal_zone_device *tz);
> +void thermal_zone_device_critical(struct thermal_zone_device *tz);
>   #else
>   static inline struct thermal_zone_device *thermal_zone_device_register(
>   	const char *type, int trips, int mask, void *devdata,
> 

I am just concerned about drivers which provide own .critical() callback
but forgot to call thermal_zone_device_critical() at the end and
framework could skip it.

Or we can make sure during the review that it's not an issue (and ignore
out of tree drivers)?

Regards,
Lukasz

Re: [PATCH 2/5] thermal/core: Add critical and hot ops

[Daniel Lezcano]

On 10/12/2020 13:44, Lukasz Luba wrote:
> 
> 
> On 12/10/20 12:15 PM, Daniel Lezcano wrote:
>> Currently there is no way to the sensors to directly call an ops in
>> interrupt mode without calling thermal_zone_device_update assuming all
>> the trip points are defined.
>>
>> A sensor may want to do something special if a trip point is hot or
>> critical.
>>
>> This patch adds the critical and hot ops to the thermal zone device,
>> so a sensor can directly invoke them or let the thermal framework to
>> call the sensor specific ones.
>>
>> Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
>> Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
>> ---
>>   drivers/thermal/thermal_core.c | 43 +++++++++++++++++++++-------------
>>   include/linux/thermal.h        |  3 +++
>>   2 files changed, 30 insertions(+), 16 deletions(-)
>>
>> diff --git a/drivers/thermal/thermal_core.c
>> b/drivers/thermal/thermal_core.c
>> index e6771e5aeedb..cee0b31b5cd7 100644
>> --- a/drivers/thermal/thermal_core.c
>> +++ b/drivers/thermal/thermal_core.c
>> @@ -375,6 +375,25 @@ static void thermal_emergency_poweroff(void)
>>                     msecs_to_jiffies(poweroff_delay_ms));
>>   }
>>   +void thermal_zone_device_critical(struct thermal_zone_device *tz)
>> +{
>> +    dev_emerg(&tz->device, "%s: critical temperature reached, "
>> +          "shutting down\n", tz->type);
>> +
>> +    mutex_lock(&poweroff_lock);
>> +    if (!power_off_triggered) {
>> +        /*
>> +         * Queue a backup emergency shutdown in the event of
>> +         * orderly_poweroff failure
>> +         */
>> +        thermal_emergency_poweroff();
>> +        orderly_poweroff(true);
>> +        power_off_triggered = true;
>> +    }
>> +    mutex_unlock(&poweroff_lock);
>> +}
>> +EXPORT_SYMBOL(thermal_zone_device_critical);
>> +
>>   static void handle_critical_trips(struct thermal_zone_device *tz,
>>                     int trip, enum thermal_trip_type trip_type)
>>   {
>> @@ -391,22 +410,10 @@ static void handle_critical_trips(struct
>> thermal_zone_device *tz,
>>       if (tz->ops->notify)
>>           tz->ops->notify(tz, trip, trip_type);
>>   -    if (trip_type == THERMAL_TRIP_CRITICAL) {
>> -        dev_emerg(&tz->device,
>> -              "critical temperature reached (%d C), shutting down\n",
>> -              tz->temperature / 1000);
>> -        mutex_lock(&poweroff_lock);
>> -        if (!power_off_triggered) {
>> -            /*
>> -             * Queue a backup emergency shutdown in the event of
>> -             * orderly_poweroff failure
>> -             */
>> -            thermal_emergency_poweroff();
>> -            orderly_poweroff(true);
>> -            power_off_triggered = true;
>> -        }
>> -        mutex_unlock(&poweroff_lock);
>> -    }
>> +    if (trip_type == THERMAL_TRIP_HOT && tz->ops->hot)
>> +        tz->ops->hot(tz);
>> +    else if (trip_type == THERMAL_TRIP_CRITICAL)
>> +        tz->ops->critical(tz);
> 
> I can see that in the patch 3/5 there driver .critical() callback
> calls framework thermal_zone_device_critical() at the end.
> I wonder if we could always call this framework function.

It is actually done on purpose, we want to let the driver to handle the
critical routine which may not end up with an emergency shutdown.

[ ... ]

>>   #else
>>   static inline struct thermal_zone_device *thermal_zone_device_register(
>>       const char *type, int trips, int mask, void *devdata,
>>
> 
> I am just concerned about drivers which provide own .critical() callback
> but forgot to call thermal_zone_device_critical() at the end and
> framework could skip it.
> 
> Or we can make sure during the review that it's not an issue (and ignore
> out of tree drivers)?

Yes, the framework guarantees if the critical trip point is crossed we
call the emergency shutdown by default. If the driver choose to override
it, it takes responsibility of the change.



-- 
<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

Re: [PATCH 2/5] thermal/core: Add critical and hot ops

[Lukasz Luba]

On 12/10/20 1:37 PM, Daniel Lezcano wrote:
> On 10/12/2020 13:44, Lukasz Luba wrote:
>>
>>
>> On 12/10/20 12:15 PM, Daniel Lezcano wrote:
>>> Currently there is no way to the sensors to directly call an ops in
>>> interrupt mode without calling thermal_zone_device_update assuming all
>>> the trip points are defined.
>>>
>>> A sensor may want to do something special if a trip point is hot or
>>> critical.
>>>
>>> This patch adds the critical and hot ops to the thermal zone device,
>>> so a sensor can directly invoke them or let the thermal framework to
>>> call the sensor specific ones.
>>>
>>> Tested-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
>>> Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
>>> ---
>>>    drivers/thermal/thermal_core.c | 43 +++++++++++++++++++++-------------
>>>    include/linux/thermal.h        |  3 +++
>>>    2 files changed, 30 insertions(+), 16 deletions(-)
>>>
>>> diff --git a/drivers/thermal/thermal_core.c
>>> b/drivers/thermal/thermal_core.c
>>> index e6771e5aeedb..cee0b31b5cd7 100644
>>> --- a/drivers/thermal/thermal_core.c
>>> +++ b/drivers/thermal/thermal_core.c
>>> @@ -375,6 +375,25 @@ static void thermal_emergency_poweroff(void)
>>>                      msecs_to_jiffies(poweroff_delay_ms));
>>>    }
>>>    +void thermal_zone_device_critical(struct thermal_zone_device *tz)
>>> +{
>>> +    dev_emerg(&tz->device, "%s: critical temperature reached, "
>>> +          "shutting down\n", tz->type);
>>> +
>>> +    mutex_lock(&poweroff_lock);
>>> +    if (!power_off_triggered) {
>>> +        /*
>>> +         * Queue a backup emergency shutdown in the event of
>>> +         * orderly_poweroff failure
>>> +         */
>>> +        thermal_emergency_poweroff();
>>> +        orderly_poweroff(true);
>>> +        power_off_triggered = true;
>>> +    }
>>> +    mutex_unlock(&poweroff_lock);
>>> +}
>>> +EXPORT_SYMBOL(thermal_zone_device_critical);
>>> +
>>>    static void handle_critical_trips(struct thermal_zone_device *tz,
>>>                      int trip, enum thermal_trip_type trip_type)
>>>    {
>>> @@ -391,22 +410,10 @@ static void handle_critical_trips(struct
>>> thermal_zone_device *tz,
>>>        if (tz->ops->notify)
>>>            tz->ops->notify(tz, trip, trip_type);
>>>    -    if (trip_type == THERMAL_TRIP_CRITICAL) {
>>> -        dev_emerg(&tz->device,
>>> -              "critical temperature reached (%d C), shutting down\n",
>>> -              tz->temperature / 1000);
>>> -        mutex_lock(&poweroff_lock);
>>> -        if (!power_off_triggered) {
>>> -            /*
>>> -             * Queue a backup emergency shutdown in the event of
>>> -             * orderly_poweroff failure
>>> -             */
>>> -            thermal_emergency_poweroff();
>>> -            orderly_poweroff(true);
>>> -            power_off_triggered = true;
>>> -        }
>>> -        mutex_unlock(&poweroff_lock);
>>> -    }
>>> +    if (trip_type == THERMAL_TRIP_HOT && tz->ops->hot)
>>> +        tz->ops->hot(tz);
>>> +    else if (trip_type == THERMAL_TRIP_CRITICAL)
>>> +        tz->ops->critical(tz);
>>
>> I can see that in the patch 3/5 there driver .critical() callback
>> calls framework thermal_zone_device_critical() at the end.
>> I wonder if we could always call this framework function.
> 
> It is actually done on purpose, we want to let the driver to handle the
> critical routine which may not end up with an emergency shutdown.

I see.

> 
> [ ... ]
> 
>>>    #else
>>>    static inline struct thermal_zone_device *thermal_zone_device_register(
>>>        const char *type, int trips, int mask, void *devdata,
>>>
>>
>> I am just concerned about drivers which provide own .critical() callback
>> but forgot to call thermal_zone_device_critical() at the end and
>> framework could skip it.
>>
>> Or we can make sure during the review that it's not an issue (and ignore
>> out of tree drivers)?
> 
> Yes, the framework guarantees if the critical trip point is crossed we
> call the emergency shutdown by default. If the driver choose to override
> it, it takes responsibility of the change.
> 

Fair enough. Thus, the patch LGTM

Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>

Regards,
Lukasz

Re: [PATCH 5/5] thermal/core: Remove notify ops

[Lukasz Luba]

On 12/10/20 12:15 PM, Daniel Lezcano wrote:
> With the remove of the notify user in a previous patch, the ops is no
> longer needed, remove it.
> 
> Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
> ---
>   drivers/thermal/thermal_core.c | 3 ---
>   include/linux/thermal.h        | 2 --
>   2 files changed, 5 deletions(-)
> 
> diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c
> index cee0b31b5cd7..d7481fdf4e4c 100644
> --- a/drivers/thermal/thermal_core.c
> +++ b/drivers/thermal/thermal_core.c
> @@ -407,9 +407,6 @@ static void handle_critical_trips(struct thermal_zone_device *tz,
>   
>   	trace_thermal_zone_trip(tz, trip, trip_type);
>   
> -	if (tz->ops->notify)
> -		tz->ops->notify(tz, trip, trip_type);
> -
>   	if (trip_type == THERMAL_TRIP_HOT && tz->ops->hot)
>   		tz->ops->hot(tz);
>   	else if (trip_type == THERMAL_TRIP_CRITICAL)
> diff --git a/include/linux/thermal.h b/include/linux/thermal.h
> index 125c8a4d52e6..7e051b4cf715 100644
> --- a/include/linux/thermal.h
> +++ b/include/linux/thermal.h
> @@ -77,8 +77,6 @@ struct thermal_zone_device_ops {
>   	int (*set_emul_temp) (struct thermal_zone_device *, int);
>   	int (*get_trend) (struct thermal_zone_device *, int,
>   			  enum thermal_trend *);
> -	int (*notify) (struct thermal_zone_device *, int,
> -		       enum thermal_trip_type);
>   	void (*hot)(struct thermal_zone_device *);
>   	void (*critical)(struct thermal_zone_device *);
>   };
> 

I couldn't find other users apart from those in patch 3/5 and 4/5.
I will leave to someone else to review those patches.
This patch looks good

Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>

Re: [PATCH 4/5] thermal/drivers/rcar: Remove notification usage

[Niklas Söderlund]

Hi Daniel,

Thanks for your work.

On 2020-12-10 13:15:13 +0100, Daniel Lezcano wrote:
> The ops is only showing a trace telling a critical trip point is
> crossed. The same information is given by the thermal framework.
> 
> This is redundant, remove the code.
> 
> Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>

Reviewed-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>

> ---
>  drivers/thermal/rcar_thermal.c | 19 -------------------
>  1 file changed, 19 deletions(-)
> 
> diff --git a/drivers/thermal/rcar_thermal.c b/drivers/thermal/rcar_thermal.c
> index 6ae757d66f46..b49f04daaf47 100644
> --- a/drivers/thermal/rcar_thermal.c
> +++ b/drivers/thermal/rcar_thermal.c
> @@ -323,24 +323,6 @@ static int rcar_thermal_get_trip_temp(struct thermal_zone_device *zone,
>  	return 0;
>  }
>  
> -static int rcar_thermal_notify(struct thermal_zone_device *zone,
> -			       int trip, enum thermal_trip_type type)
> -{
> -	struct rcar_thermal_priv *priv = rcar_zone_to_priv(zone);
> -	struct device *dev = rcar_priv_to_dev(priv);
> -
> -	switch (type) {
> -	case THERMAL_TRIP_CRITICAL:
> -		/* FIXME */
> -		dev_warn(dev, "Thermal reached to critical temperature\n");
> -		break;
> -	default:
> -		break;
> -	}
> -
> -	return 0;
> -}
> -
>  static const struct thermal_zone_of_device_ops rcar_thermal_zone_of_ops = {
>  	.get_temp	= rcar_thermal_of_get_temp,
>  };
> @@ -349,7 +331,6 @@ static struct thermal_zone_device_ops rcar_thermal_zone_ops = {
>  	.get_temp	= rcar_thermal_get_temp,
>  	.get_trip_type	= rcar_thermal_get_trip_type,
>  	.get_trip_temp	= rcar_thermal_get_trip_temp,
> -	.notify		= rcar_thermal_notify,
>  };
>  
>  /*
> -- 
> 2.17.1
> 

-- 
Regards,
Niklas Söderlund

[thermal: thermal/next] thermal/core: Emit a warning if the thermal zone is updated without ops

[thermal-bot for Daniel Lezcano]

The following commit has been merged into the thermal/next branch of thermal:

Commit-ID:     433178e75834dc35f1ae79b56ec2cf396f2c6f3c
Gitweb:        https://git.kernel.org/pub/scm/linux/kernel/git/thermal/linux.git//433178e75834dc35f1ae79b56ec2cf396f2c6f3c
Author:        Daniel Lezcano <daniel.lezcano@linaro.org>
AuthorDate:    Thu, 10 Dec 2020 13:15:10 +01:00
Committer:     Daniel Lezcano <daniel.lezcano@linaro.org>
CommitterDate: Fri, 11 Dec 2020 14:11:13 +01:00

thermal/core: Emit a warning if the thermal zone is updated without ops

The actual code is silently ignoring a thermal zone update when a
driver is requesting it without a get_temp ops set.

That looks not correct, as the caller should not have called this
function if the thermal zone is unable to read the temperature.

That makes the code less robust as the check won't detect the driver
is inconsistently using the thermal API and that does not help to
improve the framework as these circumvolutions hide the problem at the
source.

In order to detect the situation when it happens, let's add a warning
when the update is requested without the get_temp() ops set.

Any warning emitted will have to be fixed at the source of the
problem: the caller must not call thermal_zone_device_update if there
is not get_temp callback set.

Cc: Thara Gopinath <thara.gopinath@linaro.org>
Cc: Amit Kucheria <amitk@kernel.org>
Cc: linux-pm@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Reviewed-by: Lukasz Luba <lukasz.luba@arm.com>
Link: https://lore.kernel.org/r/20201210121514.25760-1-daniel.lezcano@linaro.org
---
 drivers/thermal/thermal_core.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/drivers/thermal/thermal_core.c b/drivers/thermal/thermal_core.c
index 90e38cc..64de098 100644
--- a/drivers/thermal/thermal_core.c
+++ b/drivers/thermal/thermal_core.c
@@ -553,7 +553,8 @@ void thermal_zone_device_update(struct thermal_zone_device *tz,
 	if (atomic_read(&in_suspend))
 		return;
 
-	if (!tz->ops->get_temp)
+	if (WARN_ONCE(!tz->ops->get_temp, "'%s' must not be called without "
+		      "'get_temp' ops set\n", __func__))
 		return;
 
 	update_temperature(tz);

Re: [PATCH 3/5] thermal/drivers/acpi: Use hot and critical ops

[Daniel Lezcano]

On 10/12/2020 13:15, Daniel Lezcano wrote:
> The acpi driver wants to do a netlink notification in case of a hot or
> critical trip point. Implement the corresponding ops to be used for
> the thermal zone and use them instead of the notify ops.
> 
> Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>

Is there any comment on this patch ?

> ---
>  drivers/acpi/thermal.c | 30 ++++++++++++++----------------
>  1 file changed, 14 insertions(+), 16 deletions(-)
> 
> diff --git a/drivers/acpi/thermal.c b/drivers/acpi/thermal.c
> index 12c0ece746f0..b5e4bc9e3282 100644
> --- a/drivers/acpi/thermal.c
> +++ b/drivers/acpi/thermal.c
> @@ -677,27 +677,24 @@ static int thermal_get_trend(struct thermal_zone_device *thermal,
>  	return 0;
>  }
>  
> -
> -static int thermal_notify(struct thermal_zone_device *thermal, int trip,
> -			   enum thermal_trip_type trip_type)
> +static void acpi_thermal_zone_device_hot(struct thermal_zone_device *thermal)
>  {
> -	u8 type = 0;
>  	struct acpi_thermal *tz = thermal->devdata;
>  
> -	if (trip_type == THERMAL_TRIP_CRITICAL)
> -		type = ACPI_THERMAL_NOTIFY_CRITICAL;
> -	else if (trip_type == THERMAL_TRIP_HOT)
> -		type = ACPI_THERMAL_NOTIFY_HOT;
> -	else
> -		return 0;
> -
>  	acpi_bus_generate_netlink_event(tz->device->pnp.device_class,
> -					dev_name(&tz->device->dev), type, 1);
> +					dev_name(&tz->device->dev),
> +					ACPI_THERMAL_NOTIFY_HOT, 1);
> +}
>  
> -	if (trip_type == THERMAL_TRIP_CRITICAL && nocrt)
> -		return 1;
> +static void acpi_thermal_zone_device_critical(struct thermal_zone_device *thermal)
> +{
> +	struct acpi_thermal *tz = thermal->devdata;
>  
> -	return 0;
> +	acpi_bus_generate_netlink_event(tz->device->pnp.device_class,
> +					dev_name(&tz->device->dev),
> +					ACPI_THERMAL_NOTIFY_CRITICAL, 1);
> +
> +	thermal_zone_device_critical(thermal);
>  }
>  
>  static int acpi_thermal_cooling_device_cb(struct thermal_zone_device *thermal,
> @@ -812,7 +809,8 @@ static struct thermal_zone_device_ops acpi_thermal_zone_ops = {
>  	.get_trip_temp = thermal_get_trip_temp,
>  	.get_crit_temp = thermal_get_crit_temp,
>  	.get_trend = thermal_get_trend,
> -	.notify = thermal_notify,
> +	.hot = acpi_thermal_zone_device_hot,
> +	.critical = acpi_thermal_zone_device_critical,
>  };
>  
>  static int acpi_thermal_register_thermal_zone(struct acpi_thermal *tz)
> 


-- 
<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

Re: [PATCH 3/5] thermal/drivers/acpi: Use hot and critical ops

[Srinivas Pandruvada]

On Thu, 2020-12-17 at 07:28 +0100, Daniel Lezcano wrote:
> On 10/12/2020 13:15, Daniel Lezcano wrote:
> > The acpi driver wants to do a netlink notification in case of a hot
> > or
> > critical trip point. Implement the corresponding ops to be used for
> > the thermal zone and use them instead of the notify ops.
> > 
> > Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
> 
> Is there any comment on this patch ?

Looks good to me.

Thanks,
Srinivas

> 
> > ---
> >  drivers/acpi/thermal.c | 30 ++++++++++++++----------------
> >  1 file changed, 14 insertions(+), 16 deletions(-)
> > 
> > diff --git a/drivers/acpi/thermal.c b/drivers/acpi/thermal.c
> > index 12c0ece746f0..b5e4bc9e3282 100644
> > --- a/drivers/acpi/thermal.c
> > +++ b/drivers/acpi/thermal.c
> > @@ -677,27 +677,24 @@ static int thermal_get_trend(struct
> > thermal_zone_device *thermal,
> >         return 0;
> >  }
> >  
> > -
> > -static int thermal_notify(struct thermal_zone_device *thermal, int
> > trip,
> > -                          enum thermal_trip_type trip_type)
> > +static void acpi_thermal_zone_device_hot(struct
> > thermal_zone_device *thermal)
> >  {
> > -       u8 type = 0;
> >         struct acpi_thermal *tz = thermal->devdata;
> >  
> > -       if (trip_type == THERMAL_TRIP_CRITICAL)
> > -               type = ACPI_THERMAL_NOTIFY_CRITICAL;
> > -       else if (trip_type == THERMAL_TRIP_HOT)
> > -               type = ACPI_THERMAL_NOTIFY_HOT;
> > -       else
> > -               return 0;
> > -
> >         acpi_bus_generate_netlink_event(tz->device-
> > >pnp.device_class,
> > -                                       dev_name(&tz->device->dev),
> > type, 1);
> > +                                       dev_name(&tz->device->dev),
> > +                                       ACPI_THERMAL_NOTIFY_HOT,
> > 1);
> > +}
> >  
> > -       if (trip_type == THERMAL_TRIP_CRITICAL && nocrt)
> > -               return 1;
> > +static void acpi_thermal_zone_device_critical(struct
> > thermal_zone_device *thermal)
> > +{
> > +       struct acpi_thermal *tz = thermal->devdata;
> >  
> > -       return 0;
> > +       acpi_bus_generate_netlink_event(tz->device-
> > >pnp.device_class,
> > +                                       dev_name(&tz->device->dev),
> > +                                       ACPI_THERMAL_NOTIFY_CRITICA
> > L, 1);
> > +
> > +       thermal_zone_device_critical(thermal);
> >  }
> >  
> >  static int acpi_thermal_cooling_device_cb(struct
> > thermal_zone_device *thermal,
> > @@ -812,7 +809,8 @@ static struct thermal_zone_device_ops
> > acpi_thermal_zone_ops = {
> >         .get_trip_temp = thermal_get_trip_temp,
> >         .get_crit_temp = thermal_get_crit_temp,
> >         .get_trend = thermal_get_trend,
> > -       .notify = thermal_notify,
> > +       .hot = acpi_thermal_zone_device_hot,
> > +       .critical = acpi_thermal_zone_device_critical,
> >  };
> >  
> >  static int acpi_thermal_register_thermal_zone(struct acpi_thermal
> > *tz)
> > 
> 
> 

Re: [PATCH 3/5] thermal/drivers/acpi: Use hot and critical ops

[Daniel Lezcano]

On 17/12/2020 12:38, Srinivas Pandruvada wrote:
> On Thu, 2020-12-17 at 07:28 +0100, Daniel Lezcano wrote:
>> On 10/12/2020 13:15, Daniel Lezcano wrote:
>>> The acpi driver wants to do a netlink notification in case of a hot
>>> or
>>> critical trip point. Implement the corresponding ops to be used for
>>> the thermal zone and use them instead of the notify ops.
>>>
>>> Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
>>
>> Is there any comment on this patch ?
> 
> Looks good to me.

Thanks for reviewing


-- 
<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