Re: [PATCH 5/5] ASoC: codecs: Add WCD939x Codec driver

[Neil Armstrong <neil.armstrong@xxxxxxxxxx>]

On 25/11/2023 13:07, Konrad Dybcio wrote:
> On 23.11.2023 15:49, Neil Armstrong wrote:

<snip>

> > +
> > +static int wcd939x_io_init(struct snd_soc_component *component)
> > +{
> > +	snd_soc_component_write_field(component, WCD939X_ANA_BIAS,
> > +				      WCD939X_BIAS_ANALOG_BIAS_EN, 1);
> All of these values are BIT()s or 2-4 ORed BIT()s, can you check what they
> mean?
>
> Same for almost all other snd_soc_component_ write/modify functions

Ack

> [...]
>
> > +static int wcd939x_tx_swr_ctrl(struct snd_soc_dapm_widget *w,
> > +			       struct snd_kcontrol *kcontrol, int event)
> > +{
> > +	struct snd_soc_component *component = snd_soc_dapm_to_component(w->dapm);
> > +	struct wcd939x_priv *wcd939x = snd_soc_component_get_drvdata(component);
> > +	int bank;
> > +	int rate;
> > +
> > +	bank = (wcd939x_swr_get_current_bank(wcd939x->sdw_priv[AIF1_CAP]->sdev)) ? 0 : 1;
> > +	bank = bank ? 0 : 1;
> this is just the return value of wcd939x_swr_get_current_bank NOTed twice

Yep, I also found it wild, will simplify

> > +
> > +	switch (event) {
> > +	case SND_SOC_DAPM_PRE_PMU:
> > +		if (strnstr(w->name, "ADC", sizeof("ADC"))) {
> > +			int i = 0, mode = 0;
> > +
> > +			if (test_bit(WCD_ADC1, &wcd939x->status_mask))
> > +				mode |= tx_mode_bit[wcd939x->tx_mode[WCD_ADC1]];
> > +			if (test_bit(WCD_ADC2, &wcd939x->status_mask))
> > +				mode |= tx_mode_bit[wcd939x->tx_mode[WCD_ADC2]];
> > +			if (test_bit(WCD_ADC3, &wcd939x->status_mask))
> > +				mode |= tx_mode_bit[wcd939x->tx_mode[WCD_ADC3]];
> > +			if (test_bit(WCD_ADC4, &wcd939x->status_mask))
> > +				mode |= tx_mode_bit[wcd939x->tx_mode[WCD_ADC4]];
> > +
> > +			if (mode != 0) {
> > +				for (i = 0; i < ADC_MODE_ULP2; i++) {
> > +					if (mode & (1 << i)) {
> BIT(i)

Ack

> > +						i++;
> > +						break;
> > +					}
> the immediate break makes it look like you're just looking for the first set
> bit, 1-indexed, and there's functions for that

Indeed, let's use that

> > +				}
> > +			}
> > +			rate = wcd939x_get_clk_rate(i);
> > +			wcd939x_set_swr_clk_rate(component, rate, bank);
> > +			/* Copy clk settings to active bank */
> > +			wcd939x_set_swr_clk_rate(component, rate, !bank);
> This comment is very confusing

It is, not sure if I can clarify it more

> [...]
>
> > +static void wcd939x_mbhc_get_result_params(struct snd_soc_component *component,
> > +					   s16 *d1_a, u16 noff,
> > +						int32_t *zdet)
> very weird alignment

Will fix

> > +{
> > +	struct wcd939x_priv *wcd939x = snd_soc_component_get_drvdata(component);
> > +	int i;
> > +	int val, val1;
> > +	s16 c1;
> > +	s32 x1, d1;
> > +	s32 denom;
> > +	static const int mincode_param[] = {
> static inside a function means something else than static global
>
> > +		3277, 1639, 820, 410, 205, 103, 52, 26
> > +	};
> reverse-Christmas-tree

Will fix

> > +
> > +	regmap_update_bits(wcd939x->regmap, WCD939X_ANA_MBHC_ZDET, 0x20, 0x20);
> > +	for (i = 0; i < WCD939X_ZDET_NUM_MEASUREMENTS; i++) {
> > +		regmap_read(wcd939x->regmap, WCD939X_ANA_MBHC_RESULT_2, &val);
> > +		if (val & 0x80)
> BIT(7), please name it
Will fix

> > +			break;
> > +	}
> > +	val = val << 0x8;
> << 8 makes more sense

Will fix

> > +	regmap_read(wcd939x->regmap, WCD939X_ANA_MBHC_RESULT_1, &val1);
> > +	val |= val1;
> > +	regmap_update_bits(wcd939x->regmap, WCD939X_ANA_MBHC_ZDET, 0x20, 0x00);
> BIT()s
Will fix

> > +	x1 = WCD939X_MBHC_GET_X1(val);
> > +	c1 = WCD939X_MBHC_GET_C1(val);
> > +	/* If ramp is not complete, give additional 5ms */
> > +	if (c1 < 2 && x1)
> > +		usleep_range(5000, 5050);
> udelay, see timers-howto.txt
Will fix

> [...]
>
> > +static void wcd939x_wcd_mbhc_qfuse_cal(struct snd_soc_component *component,
> > +				       s32 *z_val, int flag_l_r)
> > +{
> > +	s16 q1;
> > +	int q1_cal;
> > +
> > +	q1 = snd_soc_component_read(component, WCD939X_DIGITAL_EFUSE_REG_21 + flag_l_r);
> > +	if (q1 & 0x80)
> BIT
> > +		q1_cal = (10000 - ((q1 & 0x7F) * 10));
> GENMASK
Will fix

> > +	else
> > +		q1_cal = (10000 + (q1 * 10));
> > +	if (q1_cal > 0)
> > +		*z_val = ((*z_val) * 10000) / q1_cal;
> > +}
> > +
> > +static void wcd939x_wcd_mbhc_calc_impedance(struct snd_soc_component *component,
> > +					    u32 *zl, uint32_t *zr)
> > +{
> > +	struct wcd939x_priv *wcd939x = dev_get_drvdata(component->dev);
> > +	s16 reg0, reg1, reg2, reg3, reg4;
> > +	s32 z1l, z1r, z1ls;
> > +	int z_mono, z_diff1, z_diff2;
> > +	bool is_fsm_disable = false;
> > +	struct wcd939x_mbhc_zdet_param zdet_param[] = {
> const
Will fix

> > +		{4, 0, 4, 0x08, 0x14, 0x18}, /* < 32ohm */
> > +		{4, 0, 6, 0x18, 0x60, 0x78}, /* 32ohm < Z < 400ohm */
> > +		{1, 4, 5, 0x18, 0x7C, 0x90}, /* 400ohm < Z < 1200ohm */
> > +		{1, 6, 7, 0x18, 0x7C, 0x90}, /* >1200ohm */
> > +	};
> > +	struct wcd939x_mbhc_zdet_param *zdet_param_ptr = NULL;
> > +	s16 d1_a[][4] = {
> const
Will fix

> > +		{0, 30, 90, 30},
> > +		{0, 30, 30, 6},
> > +		{0, 30, 30, 5},
> > +		{0, 30, 30, 5},
> > +	};
> > +	s16 *d1 = NULL;
> ????
Will fix

> [...]
>
> > +
> > +	rc = of_property_read_u32(np, "qcom,micbias1-microvolt",  &prop_val);
> similarly to the swr port map, this should be commonized for WCDxxxx

Yes, but where ?

> Konrad

Thanks,
Neil