Skip to content

Commit d7bf0a1

Browse files
Andre-ARMdlezcano
authored andcommitted
thermal/drivers/sun8i: Explain unknown H6 register value
So far we were ORing in some "unknown" value into the THS control register on the Allwinner H6. This part of the register is not explained in the H6 manual, but the H616 manual details those bits, and on closer inspection the THS IP blocks in both SoCs seem very close: - The BSP code for both SoCs writes the same values into THS_CTRL. - The reset values of at least the first three registers are the same. Replace the "unknown" value with its proper meaning: "acquire time", most probably the sample part of the sample & hold circuit of the ADC, according to its explanation in the H616 manual. No functional change, just a macro rename and adjustment. Signed-off-by: Andre Przywara <[email protected]> Reviewed-by: Jernej Skrabec <[email protected]> Acked-by: Vasily Khoruzhick <[email protected]> Signed-off-by: Daniel Lezcano <[email protected]> Link: https://lore.kernel.org/r/[email protected]
1 parent d1dc7ee commit d7bf0a1

File tree

1 file changed

+16
-13
lines changed

1 file changed

+16
-13
lines changed

drivers/thermal/sun8i_thermal.c

Lines changed: 16 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -50,7 +50,8 @@
5050
#define SUN8I_THS_CTRL2_T_ACQ1(x) ((GENMASK(15, 0) & (x)) << 16)
5151
#define SUN8I_THS_DATA_IRQ_STS(x) BIT(x + 8)
5252

53-
#define SUN50I_THS_CTRL0_T_ACQ(x) ((GENMASK(15, 0) & (x)) << 16)
53+
#define SUN50I_THS_CTRL0_T_ACQ(x) (GENMASK(15, 0) & ((x) - 1))
54+
#define SUN50I_THS_CTRL0_T_SAMPLE_PER(x) ((GENMASK(15, 0) & ((x) - 1)) << 16)
5455
#define SUN50I_THS_FILTER_EN BIT(2)
5556
#define SUN50I_THS_FILTER_TYPE(x) (GENMASK(1, 0) & (x))
5657
#define SUN50I_H6_THS_PC_TEMP_PERIOD(x) ((GENMASK(19, 0) & (x)) << 12)
@@ -410,25 +411,27 @@ static int sun8i_h3_thermal_init(struct ths_device *tmdev)
410411
return 0;
411412
}
412413

413-
/*
414-
* Without this undocumented value, the returned temperatures would
415-
* be higher than real ones by about 20C.
416-
*/
417-
#define SUN50I_H6_CTRL0_UNK 0x0000002f
418-
419414
static int sun50i_h6_thermal_init(struct ths_device *tmdev)
420415
{
421416
int val;
422417

423418
/*
424-
* T_acq = 20us
425-
* clkin = 24MHz
426-
*
427-
* x = T_acq * clkin - 1
428-
* = 479
419+
* The manual recommends an overall sample frequency of 50 KHz (20us,
420+
* 480 cycles at 24 MHz), which provides plenty of time for both the
421+
* acquisition time (>24 cycles) and the actual conversion time
422+
* (>14 cycles).
423+
* The lower half of the CTRL register holds the "acquire time", in
424+
* clock cycles, which the manual recommends to be 2us:
425+
* 24MHz * 2us = 48 cycles.
426+
* The high half of THS_CTRL encodes the sample frequency, in clock
427+
* cycles: 24MHz * 20us = 480 cycles.
428+
* This is explained in the H616 manual, but apparently wrongly
429+
* described in the H6 manual, although the BSP code does the same
430+
* for both SoCs.
429431
*/
430432
regmap_write(tmdev->regmap, SUN50I_THS_CTRL0,
431-
SUN50I_H6_CTRL0_UNK | SUN50I_THS_CTRL0_T_ACQ(479));
433+
SUN50I_THS_CTRL0_T_ACQ(48) |
434+
SUN50I_THS_CTRL0_T_SAMPLE_PER(480));
432435
/* average over 4 samples */
433436
regmap_write(tmdev->regmap, SUN50I_H6_THS_MFC,
434437
SUN50I_THS_FILTER_EN |

0 commit comments

Comments
 (0)