Re: [PATCH v1 5/6] ASoC: codecs: msm8916-wcd-analog: add MBHC support

From: Srinivas Kandagatla
Date: Fri Jul 28 2017 - 10:12:17 EST




On 28/07/17 14:51, Mark Brown wrote:
On Wed, Jul 26, 2017 at 02:35:11AM +0200, srinivas.kandagatla@xxxxxxxxxx wrote:

+ if (wcd->hphl_jack_type_normally_open)
+ plug_type = CDC_A_HPHL_PLUG_TYPE_NO;
+
+ if (wcd->gnd_jack_type_normally_open)
+ plug_type |= CDC_A_GND_PLUG_TYPE_NO;

It'd be clearer to use |= consistently, this is just confusing as it
makes it look like there's a path where things aren't initialized.

Yep, I will fix this in next version.

+ btn_result = snd_soc_read(codec, CDC_A_MBHC_RESULT_1) &
+ CDC_A_MBHC_RESULT_1_BTN_RESULT_MASK;
+
+ if (!btn_result)
+ priv->mbhc_btn0_pressed = false;
+
+ snd_jack_report(priv->jack->jack, 0);

This checks to see if the read worked, uses it to decide if it should
clear the button pressed flag but unconditionally reports that the
button is not pressed. It also discards the read error if there was one
without reporting that. This needs cleanup.
Okay, will clean this part in next version.

+ btn_result = snd_soc_read(codec, CDC_A_MBHC_RESULT_1) &
+ CDC_A_MBHC_RESULT_1_BTN_RESULT_MASK;
+
+ switch (btn_result) {
+ case 0xf:
+ snd_jack_report(priv->jack->jack, SND_JACK_BTN_4);
+ break;
+ case 0x7:
+ snd_jack_report(priv->jack->jack, SND_JACK_BTN_3);
+ break;
+ case 0x3:
+ snd_jack_report(priv->jack->jack, SND_JACK_BTN_2);
+ break;
+ case 0x1:
+ snd_jack_report(priv->jack->jack, SND_JACK_BTN_1);
+ break;
+ case 0:
+ priv->mbhc_btn0_pressed = true;
+ snd_jack_report(priv->jack->jack, SND_JACK_BTN_0);
+ break;
+ }

Here we silently ignore any value we get back from the chip that we
didn't expect and all read errors.

I would not expect anyother values if the vrefs are programmed correctly, however i will add a error message in default case to notify this to user.


+struct msm8916_wcd_mbhc_data {
+ /* Voltage threshold when internal current source of 100uA is used */
+ int vref_btn_cs[MBHC_MAX_BUTTONS];
+ /* Voltage threshold when microphone bias is ON */
+ int vref_btn_micb[MBHC_MAX_BUTTONS];
+};

I'd expect to see some mechanism for configuring the threasholds via DT,
not all vendors use the same specs for resistive buttons.

Yep makes sense, I will do this in next version.