Hi Niklas,
On 26/02/2020 11:24, Niklas Söderlund wrote:
> Hi Kieran,
>
> Thanks for your feedback.
>
> On 2020-02-26 09:05:22 +0000, Kieran Bingham wrote:
>> Hi Niklas,
>>
>> On 12/02/2020 22:49, Niklas Söderlund wrote:
>>> The desired behavior of the driver is to generate an interrupt and call
>>
>> s/behavior/behaviour/ but that's me being English, so you can ignore
>> that ... (at your peril ... :-D )
>>
>
> I have a hard time as it is trying to spell the "other" version of the
> language I been exposed to since early years, throwing more 'u' at it
> will only add to my confusion ;-)
>
>>> thermal_zone_device_update() as soon as the temperature have changed
>>> more then one degree.
>>>
>>> When the set_trips operation was implemented it was believed that the
>>> trip window set by the framework would move around the current
>>> temperature and the hysteresis value described in devicetree. The
>>
>> Should the hysteresis value described in devicetree be a part of the
>> +-MCELCIUS(1) calculations? or is it determined that a one degree window
>> each side is sufficient to contain such hysteresis of the readings?
>
> The hardware testers wanted a fixed 1 degree window.
But my question is what is the value described in devicetree.
Should it be involved in the calculation somewhere?
I.e. if the devicetree specifies that the hysteresis of the hardware is
(guessing) say "0.2 degree", does that mean that the 'window' must be at
least that value.
And picking on that purely as an example as I don't know the value or
meaning of the one in DT, but therefore - if there was hardware that
then had a 2 degree hysteresis, this fixed-1-degree window would have to
be:
+ low = temperature - min(MCELSIUS(1), MCELSIUS(hysteresis_val);
+ high = temperature + min(MCELSIUS(1), MCELSIUS(hysteresis_val));
Or perhaps I'm just mis-understanding the reference of the hysteresis
value anyway.
I see there is still a 'trips' section in the thermal documentation example.
Should that be removed? or is it handled by core?
>>> behavior of the framework is however to set a window based on the trip
>>> points described in devicetree.
>>>
>>> Remove the set_trips operation which was not used correctly and update
>>> the temperatures that triggers interrupts directly from the interrupt
>>> handler.
>>>
>>> Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@xxxxxxxxxxxx>
>>
>> Sounds good to me.
>>
>> Reviewed-by: Kieran Bingham <kieran.bingham+renesas@xxxxxxxxxxxxxxxx>
>
> Thanks! Unfortunately the patch is already merged so it will be hard to
> add your tag.
No problem, I saw it on linux-renesas-soc and was curious ;-)
--
Kieran
>
>>
>>> ---
>>> drivers/thermal/rcar_gen3_thermal.c | 24 ++++++++++++------------
>>> 1 file changed, 12 insertions(+), 12 deletions(-)
>>>
>>> diff --git a/drivers/thermal/rcar_gen3_thermal.c b/drivers/thermal/rcar_gen3_thermal.c
>>> index 72877bdc072daaed..55d1736f532cdb33 100644
>>> --- a/drivers/thermal/rcar_gen3_thermal.c
>>> +++ b/drivers/thermal/rcar_gen3_thermal.c
>>> @@ -81,8 +81,6 @@ struct rcar_gen3_thermal_tsc {
>>> void __iomem *base;
>>> struct thermal_zone_device *zone;
>>> struct equation_coefs coef;
>>> - int low;
>>> - int high;
>>> int tj_t;
>>> int id; /* thermal channel id */
>>> };
>>> @@ -204,12 +202,14 @@ static int rcar_gen3_thermal_mcelsius_to_temp(struct rcar_gen3_thermal_tsc *tsc,
>>> return INT_FIXPT(val);
>>> }
>>>
>>> -static int rcar_gen3_thermal_set_trips(void *devdata, int low, int high)
>>> +static int rcar_gen3_thermal_update_range(struct rcar_gen3_thermal_tsc *tsc)
>>> {
>>> - struct rcar_gen3_thermal_tsc *tsc = devdata;
>>> + int temperature, low, high;
>>>
>>> - low = clamp_val(low, -40000, 120000);
>>> - high = clamp_val(high, -40000, 120000);
>>> + rcar_gen3_thermal_get_temp(tsc, &temperature);
>>> +
>>> + low = temperature - MCELSIUS(1);
>>> + high = temperature + MCELSIUS(1);
>>>
>>> rcar_gen3_thermal_write(tsc, REG_GEN3_IRQTEMP1,
>>> rcar_gen3_thermal_mcelsius_to_temp(tsc, low));
>>> @@ -217,15 +217,11 @@ static int rcar_gen3_thermal_set_trips(void *devdata, int low, int high)
>>> rcar_gen3_thermal_write(tsc, REG_GEN3_IRQTEMP2,
>>> rcar_gen3_thermal_mcelsius_to_temp(tsc, high));
>>>
>>> - tsc->low = low;
>>> - tsc->high = high;
>>> -
>>> return 0;
>>> }
>>>
>>> static const struct thermal_zone_of_device_ops rcar_gen3_tz_of_ops = {
>>> .get_temp = rcar_gen3_thermal_get_temp,
>>> - .set_trips = rcar_gen3_thermal_set_trips,
>>> };
>>>
>>> static void rcar_thermal_irq_set(struct rcar_gen3_thermal_priv *priv, bool on)
>>> @@ -246,9 +242,11 @@ static irqreturn_t rcar_gen3_thermal_irq(int irq, void *data)
>>> for (i = 0; i < priv->num_tscs; i++) {
>>> status = rcar_gen3_thermal_read(priv->tscs[i], REG_GEN3_IRQSTR);
>>> rcar_gen3_thermal_write(priv->tscs[i], REG_GEN3_IRQSTR, 0);
>>> - if (status)
>>> + if (status) {
>>> + rcar_gen3_thermal_update_range(priv->tscs[i]);
>>> thermal_zone_device_update(priv->tscs[i]->zone,
>>> THERMAL_EVENT_UNSPECIFIED);
>>> + }
>>> }
>>>
>>> return IRQ_HANDLED;
>>> @@ -454,6 +452,8 @@ static int rcar_gen3_thermal_probe(struct platform_device *pdev)
>>> if (ret < 0)
>>> goto error_unregister;
>>>
>>> + rcar_gen3_thermal_update_range(tsc);
>>> +
>>> dev_info(dev, "TSC%d: Loaded %d trip points\n", i, ret);
>>> }
>>>
>>> @@ -492,7 +492,7 @@ static int __maybe_unused rcar_gen3_thermal_resume(struct device *dev)
>>> struct rcar_gen3_thermal_tsc *tsc = priv->tscs[i];
>>>
>>> priv->thermal_init(tsc);
>>> - rcar_gen3_thermal_set_trips(tsc, tsc->low, tsc->high);
>>> + rcar_gen3_thermal_update_range(tsc);
>>> }
>>>
>>> rcar_thermal_irq_set(priv, true);
>>>
>>
>
