Currently, when a controller is configured to use filtered mode, thermal
readings are valid only about 30% of the time.

Upon testing, it was noticed that lowering any of the interval settings
resulted in an improved rate of valid data. The same was observed when
decreasing the number of samples for each sensor (which also results in
quicker measurements).

Retrying the read with a timeout longer than the time it takes to
resample (about 344us with these settings and 4 sensors) also improves
the rate.

Lower all timing settings to the minimum, configure the filtering to
single sample, and poll the measurement register for at least one period
to improve the data validity on filtered mode.  With these changes in
place, out of 100000 reads, a single one failed, ie 99.999% of the data
was valid.

Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
Tested-by: Chen-Yu Tsai <wenst@chromium.org>
Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>

---

Changes in v2:
- Added macros for the wait and timeout times (and in the process made the wait
  200us instead of 240us, just to simplify, gives the same result)
- Rebased on thermal/bleeding-edge (since the
  "thermal/drivers/mediatek/lvts_thermal: Fixes to the interrupt handling"
  series was merged)

 drivers/thermal/mediatek/lvts_thermal.c | 19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)

diff --git a/drivers/thermal/mediatek/lvts_thermal.c b/drivers/thermal/mediatek/lvts_thermal.c
index ea89b29ffe5f..1a3cb118ff32 100644
--- a/drivers/thermal/mediatek/lvts_thermal.c
+++ b/drivers/thermal/mediatek/lvts_thermal.c
@@ -58,11 +58,11 @@
 #define LVTS_PROTTC(__base)		(__base + 0x00CC)
 #define LVTS_CLKEN(__base)		(__base + 0x00E4)
 
-#define LVTS_PERIOD_UNIT			((118 * 1000) / (256 * 38))
-#define LVTS_GROUP_INTERVAL			1
-#define LVTS_FILTER_INTERVAL		1
-#define LVTS_SENSOR_INTERVAL		1
-#define LVTS_HW_FILTER				0x2
+#define LVTS_PERIOD_UNIT			0
+#define LVTS_GROUP_INTERVAL			0
+#define LVTS_FILTER_INTERVAL		0
+#define LVTS_SENSOR_INTERVAL		0
+#define LVTS_HW_FILTER				0x0
 #define LVTS_TSSEL_CONF				0x13121110
 #define LVTS_CALSCALE_CONF			0x300
 #define LVTS_MONINT_CONF			0x8300318C
@@ -86,6 +86,9 @@
 #define LVTS_MSR_IMMEDIATE_MODE		0
 #define LVTS_MSR_FILTERED_MODE		1
 
+#define LVTS_MSR_READ_TIMEOUT_US	400
+#define LVTS_MSR_READ_WAIT_US		(LVTS_MSR_READ_TIMEOUT_US / 2)
+
 #define LVTS_HW_SHUTDOWN_MT8195		105000
 
 #define LVTS_MINIMUM_THRESHOLD		20000
@@ -268,6 +271,7 @@ static int lvts_get_temp(struct thermal_zone_device *tz, int *temp)
 	struct lvts_sensor *lvts_sensor = thermal_zone_device_priv(tz);
 	void __iomem *msr = lvts_sensor->msr;
 	u32 value;
+	int rc;
 
 	/*
 	 * Measurement registers:
@@ -280,7 +284,8 @@ static int lvts_get_temp(struct thermal_zone_device *tz, int *temp)
 	 * 16	: Valid temperature
 	 * 15-0	: Raw temperature
 	 */
-	value = readl(msr);
+	rc = readl_poll_timeout(msr, value, value & BIT(16),
+				LVTS_MSR_READ_WAIT_US, LVTS_MSR_READ_TIMEOUT_US);
 
 	/*
 	 * As the thermal zone temperature will read before the
@@ -293,7 +298,7 @@ static int lvts_get_temp(struct thermal_zone_device *tz, int *temp)
 	 * functionning temperature and directly jump to a system
 	 * shutdown.
 	 */
-	if (!(value & BIT(16)))
+	if (rc)
 		return -EAGAIN;
 
 	*temp = lvts_raw_to_temp(value & 0xFFFF);
-- 
2.41.0

On 13/07/2023 17:42, Nícolas F. R. A. Prado wrote:
> Currently, when a controller is configured to use filtered mode, thermal
> readings are valid only about 30% of the time.
> 
> Upon testing, it was noticed that lowering any of the interval settings
> resulted in an improved rate of valid data. The same was observed when
> decreasing the number of samples for each sensor (which also results in
> quicker measurements).
> 
> Retrying the read with a timeout longer than the time it takes to
> resample (about 344us with these settings and 4 sensors) also improves
> the rate.
> 
> Lower all timing settings to the minimum, configure the filtering to
> single sample, and poll the measurement register for at least one period
> to improve the data validity on filtered mode.  With these changes in
> place, out of 100000 reads, a single one failed, ie 99.999% of the data
> was valid.
> 
> Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
> Tested-by: Chen-Yu Tsai <wenst@chromium.org>
> Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
> 
> ---

Applied, thanks

-- 
<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

Reviewed-by: Alexandre Mergnat <amergnat@baylibre.com>

On 13/07/2023 17:42, Nícolas F. R. A. Prado wrote:
> Currently, when a controller is configured to use filtered mode, thermal
> readings are valid only about 30% of the time.
> 
> Upon testing, it was noticed that lowering any of the interval settings
> resulted in an improved rate of valid data. The same was observed when
> decreasing the number of samples for each sensor (which also results in
> quicker measurements).
> 
> Retrying the read with a timeout longer than the time it takes to
> resample (about 344us with these settings and 4 sensors) also improves
> the rate.
> 
> Lower all timing settings to the minimum, configure the filtering to
> single sample, and poll the measurement register for at least one period
> to improve the data validity on filtered mode.  With these changes in
> place, out of 100000 reads, a single one failed, ie 99.999% of the data
> was valid.

-- 
Regards,
Alexandre

On Thu, Jul 13, 2023 at 11:42:37AM -0400, Nícolas F. R. A. Prado wrote:
> Currently, when a controller is configured to use filtered mode, thermal
> readings are valid only about 30% of the time.
> 
> Upon testing, it was noticed that lowering any of the interval settings
> resulted in an improved rate of valid data. The same was observed when
> decreasing the number of samples for each sensor (which also results in
> quicker measurements).
> 
> Retrying the read with a timeout longer than the time it takes to
> resample (about 344us with these settings and 4 sensors) also improves
> the rate.
> 
> Lower all timing settings to the minimum, configure the filtering to
> single sample, and poll the measurement register for at least one period
> to improve the data validity on filtered mode.  With these changes in
> place, out of 100000 reads, a single one failed, ie 99.999% of the data
> was valid.
> 
> Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
> Tested-by: Chen-Yu Tsai <wenst@chromium.org>
> Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>

Hi Daniel,

gentle ping on this patch.

Thanks,
Nícolas

Il 13/07/23 17:42, Nícolas F. R. A. Prado ha scritto:
> Currently, when a controller is configured to use filtered mode, thermal
> readings are valid only about 30% of the time.
> 
> Upon testing, it was noticed that lowering any of the interval settings
> resulted in an improved rate of valid data. The same was observed when
> decreasing the number of samples for each sensor (which also results in
> quicker measurements).
> 
> Retrying the read with a timeout longer than the time it takes to
> resample (about 344us with these settings and 4 sensors) also improves
> the rate.
> 
> Lower all timing settings to the minimum, configure the filtering to
> single sample, and poll the measurement register for at least one period
> to improve the data validity on filtered mode.  With these changes in
> place, out of 100000 reads, a single one failed, ie 99.999% of the data
> was valid.
> 
> Reviewed-by: Chen-Yu Tsai <wenst@chromium.org>
> Tested-by: Chen-Yu Tsai <wenst@chromium.org>
> Signed-off-by: Nícolas F. R. A. Prado <nfraprado@collabora.com>
> 

Good to go!

Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>