Hi Daniel,
Thanks for the review.
On Sat, 2021-04-24 at 16:14 +0200, Daniel Lezcano wrote:
> On 21/04/2021 22:04, Ezequiel Garcia wrote:
> > From: Finley Xiao <finley.xiao@xxxxxxxxxxxxxx>
> >
> > The RK3568 SoCs have two Temperature Sensors, channel 0 is for CPU,
> > channel 1 is for GPU.
> >
> > Signed-off-by: Finley Xiao <finley.xiao@xxxxxxxxxxxxxx>
> > Signed-off-by: Ezequiel Garcia <ezequiel@xxxxxxxxxxxxx>
> > ---
> > drivers/thermal/rockchip_thermal.c | 107 +++++++++++++++++++++++++++++
> > 1 file changed, 107 insertions(+)
> >
> > diff --git a/drivers/thermal/rockchip_thermal.c b/drivers/thermal/rockchip_thermal.c
> > index aa9e0e31ef98..8e26f2685003 100644
> > --- a/drivers/thermal/rockchip_thermal.c
> > +++ b/drivers/thermal/rockchip_thermal.c
> > @@ -211,7 +211,11 @@ struct rockchip_thermal_data {
> > #define TSADCV3_AUTO_PERIOD_TIME 1875 /* 2.5ms */
> > #define TSADCV3_AUTO_PERIOD_HT_TIME 1875 /* 2.5ms */
> >
> > +#define TSADCV5_AUTO_PERIOD_TIME 1622 /* 2.5ms */
> > +#define TSADCV5_AUTO_PERIOD_HT_TIME 1622 /* 2.5ms */
> > +
> > #define TSADCV2_USER_INTER_PD_SOC 0x340 /* 13 clocks */
> > +#define TSADCV5_USER_INTER_PD_SOC 0xfc0 /* 97us, at least 90us */
> >
> > #define GRF_SARADC_TESTBIT 0x0e644
> > #define GRF_TSADC_TESTBIT_L 0x0e648
> > @@ -219,6 +223,12 @@ struct rockchip_thermal_data {
> >
> > #define PX30_GRF_SOC_CON2 0x0408
> >
> > +#define RK3568_GRF_TSADC_CON 0x0600
> > +#define RK3568_GRF_TSADC_ANA_REG0 (0x10001 << 0)
> > +#define RK3568_GRF_TSADC_ANA_REG1 (0x10001 << 1)
> > +#define RK3568_GRF_TSADC_ANA_REG2 (0x10001 << 2)
> > +#define RK3568_GRF_TSADC_TSEN (0x10001 << 8)
> > +
> > #define GRF_SARADC_TESTBIT_ON (0x10001 << 2)
> > #define GRF_TSADC_TESTBIT_H_ON (0x10001 << 2)
> > #define GRF_TSADC_VCM_EN_L (0x10001 << 7)
> > @@ -474,6 +484,45 @@ static const struct tsadc_table rk3399_code_table[] = {
> > {TSADCV3_DATA_MASK, 125000},
> > };
> >
> > +static const struct tsadc_table rk3568_code_table[] = {
> > + {0, -40000},
> > + {1584, -40000},
> > + {1620, -35000},
> > + {1652, -30000},
> > + {1688, -25000},
> > + {1720, -20000},
> > + {1756, -15000},
> > + {1788, -10000},
> > + {1824, -5000},
> > + {1856, 0},
> > + {1892, 5000},
> > + {1924, 10000},
> > + {1956, 15000},
> > + {1992, 20000},
> > + {2024, 25000},
> > + {2060, 30000},
> > + {2092, 35000},
> > + {2128, 40000},
> > + {2160, 45000},
> > + {2196, 50000},
> > + {2228, 55000},
> > + {2264, 60000},
> > + {2300, 65000},
> > + {2332, 70000},
> > + {2368, 75000},
> > + {2400, 80000},
> > + {2436, 85000},
> > + {2468, 90000},
> > + {2500, 95000},
> > + {2536, 100000},
> > + {2572, 105000},
> > + {2604, 110000},
> > + {2636, 115000},
> > + {2672, 120000},
> > + {2704, 125000},
> > + {TSADCV2_DATA_MASK, 125000},
> > +};
> > +
> > static u32 rk_tsadcv2_temp_to_code(const struct chip_tsadc_table *table,
> > int temp)
> > {
> > @@ -701,6 +750,35 @@ static void rk_tsadcv4_initialize(struct regmap *grf, void __iomem *regs,
> > regmap_write(grf, PX30_GRF_SOC_CON2, GRF_CON_TSADC_CH_INV);
> > }
> >
> > +static void rk_tsadcv7_initialize(struct regmap *grf, void __iomem *regs,
> > + enum tshut_polarity tshut_polarity)
> > +{
> > + writel_relaxed(TSADCV5_USER_INTER_PD_SOC, regs + TSADCV2_USER_CON);
> > + writel_relaxed(TSADCV5_AUTO_PERIOD_TIME, regs + TSADCV2_AUTO_PERIOD);
> > + writel_relaxed(TSADCV2_HIGHT_INT_DEBOUNCE_COUNT,
> > + regs + TSADCV2_HIGHT_INT_DEBOUNCE);
> > + writel_relaxed(TSADCV5_AUTO_PERIOD_HT_TIME,
> > + regs + TSADCV2_AUTO_PERIOD_HT);
> > + writel_relaxed(TSADCV2_HIGHT_TSHUT_DEBOUNCE_COUNT,
> > + regs + TSADCV2_HIGHT_TSHUT_DEBOUNCE);
> > +
> > + if (tshut_polarity == TSHUT_HIGH_ACTIVE)
> > + writel_relaxed(0U | TSADCV2_AUTO_TSHUT_POLARITY_HIGH,
> > + regs + TSADCV2_AUTO_CON);
> > + else
> > + writel_relaxed(0U & ~TSADCV2_AUTO_TSHUT_POLARITY_HIGH,
> > + regs + TSADCV2_AUTO_CON);
> > +
> > + if (!IS_ERR(grf)) {
>
> That is strange to do this check with a parameter. Is the sensor
> functional if the regmap failed ?
>
Indeed, it seems it's optional. The same check is in rk_tsadcv3_initialize,
and also in rockchip_configure_from_dt():
/* The tsadc wont to handle the error in here since some SoCs didn't
* need this property.
*/
thermal->grf = syscon_regmap_lookup_by_phandle(np, "rockchip,grf");
if (IS_ERR(thermal->grf))
dev_warn(dev, "Missing rockchip,grf property\n");
But I completely agree it looks strange without a comment,
might be better to add one.
> > + regmap_write(grf, RK3568_GRF_TSADC_CON, RK3568_GRF_TSADC_TSEN);
> > + udelay(15);
> > + regmap_write(grf, RK3568_GRF_TSADC_CON, RK3568_GRF_TSADC_ANA_REG0);
> > + regmap_write(grf, RK3568_GRF_TSADC_CON, RK3568_GRF_TSADC_ANA_REG1);
> > + regmap_write(grf, RK3568_GRF_TSADC_CON, RK3568_GRF_TSADC_ANA_REG2);
> > + usleep_range(100, 200);
>
> Is it possible to put a comment on why these delays are necessary (if
> you have the info) ?
>
Sure, I'll add a comment there. It is described in RK3568 Part1 TRM,
section 18.5.2.
Thanks,
Ezequiel
