On Mon, Feb 17, 2020 at 11:57 AM Samuel Holland samuel@sholland.org wrote:
Hello,
On 2/16/20 9:48 PM, Chen-Yu Tsai wrote:
Hi,
On Mon, Feb 17, 2020 at 10:18 AM Samuel Holland samuel@sholland.org wrote:
This matches the hardware more accurately, and is necessary for including the (stereo) headphone mute switch in the DAPM graph.
Signed-off-by: Samuel Holland samuel@sholland.org
sound/soc/sunxi/sun50i-codec-analog.c | 28 +++++++++++++++++++-------- 1 file changed, 20 insertions(+), 8 deletions(-)
diff --git a/sound/soc/sunxi/sun50i-codec-analog.c b/sound/soc/sunxi/sun50i-codec-analog.c index 17165f1ddb63..f98851067f97 100644 --- a/sound/soc/sunxi/sun50i-codec-analog.c +++ b/sound/soc/sunxi/sun50i-codec-analog.c @@ -311,9 +311,15 @@ static const struct snd_soc_dapm_widget sun50i_a64_codec_widgets[] = { */
SND_SOC_DAPM_REGULATOR_SUPPLY("cpvdd", 0, 0),
SND_SOC_DAPM_MUX("Headphone Source Playback Route",
SND_SOC_DAPM_MUX("Left Headphone Source", SND_SOC_NOPM, 0, 0, sun50i_codec_hp_src),
SND_SOC_DAPM_OUT_DRV("Headphone Amp", SUN50I_ADDA_HP_CTRL,
SND_SOC_DAPM_MUX("Right Headphone Source",
Please don't remove the widget suffix. The suffix was chosen to work with alsa-lib's control parsing code. The term "Playback Route" is appropriate because it is playback only, and it is a routing switch, not a source or sink.
Removing the suffix here doesn't affect the control name as seen by userspace, because the control is shared between multiple widgets (Left and Right).
You're right.
Also, what's wrong with just having a single "stereo" widget instead of two "mono" widgets? I added stereo (2-channel) support to DAPM quite some time ago. You just have to have two routes in and out.
If you have any completed path through a widget, the widget is turned on. A stereo mute switch is logically two separate paths. So if I break one path by muting one channel, that lets me turn off everything before and after in the path (e.g. turning off the right side of the DAC lets DAPM turn off the right mixer, the right side of the output amp, even the right side of the AIF or the ADC if that was the only input. That only works if there are separate Left/Right widgets.
Looks like that's the case indeed. Might be worth revisiting the core DAPM code later on if I can find the time.
Since the widgets and routes changed are internal to the codec, there won't be any issue if we rework this stuff later on. So for now,
Reviewed-by: Chen-Yu Tsai wens@csie.org