Skip to contentSkip to conversation

ASoC: Tegra machine ASoC driver for boards using ALC5332 codec

PATCH Superseded Leon Romanovsky 2 participants 7 messages Dec 12, 2011 – Dec 13, 2011

Subsystems (3)

v2 of 3 Superseded Latest is v3 →

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 ASoC: Tegra machine ASoC driver for boards using ALC5332 codec

At this stage only Toshiba AC100/Dynabook supported.

--
Changes since v1:
* Use module_platform_driver.
* Move controls, dapm_widgets and dapm_routes to snd_soc_card structure.
* Set dai_fmt in snd_soc_dai_link structure instead probe function.
* Remove check board in probe function
* Fix warning about unused codec_dai.
* Use devm_kzalloc to manage memory allocation.

---
 sound/soc/tegra/Kconfig         |    9 ++
 sound/soc/tegra/Makefile        |    2 +
 sound/soc/tegra/tegra_alc5632.c |  233 +++++++++++++++++++++++++++++++++++++++
 3 files changed, 244 insertions(+), 0 deletions(-)
 create mode 100644 sound/soc/tegra/tegra_alc5632.c
sound/soc/tegra/Kconfig +9 -0
index c6af1fd..ce1b773 100644
@@ -47,3 +47,12 @@ config SND_SOC_TEGRA_TRIMSLICE
 	help
 	  Say Y or M here if you want to add support for SoC audio on the
 	  TrimSlice platform.
+
+config SND_SOC_TEGRA_ALC5632
+       tristate "SoC Audio support for Tegra boards using an ALC5632 codec"
+       depends on SND_SOC_TEGRA && I2C
+       select SND_SOC_TEGRA_I2S
+       select SND_SOC_ALC5632
+       help
+         Say Y or M here if you want to add support for SoC audio on the
+         Toshiba AC100 netbook.
sound/soc/tegra/Makefile +2 -0
index 4d943b3..8e584b8 100644
@@ -14,6 +14,8 @@ obj-$(CONFIG_SND_SOC_TEGRA_SPDIF) += snd-soc-tegra-spdif.o
 # Tegra machine Support
 snd-soc-tegra-wm8903-objs := tegra_wm8903.o
 snd-soc-tegra-trimslice-objs := trimslice.o
+snd-soc-tegra-alc5632-objs := tegra_alc5632.o
 
 obj-$(CONFIG_SND_SOC_TEGRA_WM8903) += snd-soc-tegra-wm8903.o
 obj-$(CONFIG_SND_SOC_TEGRA_TRIMSLICE) += snd-soc-tegra-trimslice.o
+obj-$(CONFIG_SND_SOC_TEGRA_ALC5632) += snd-soc-tegra-alc5632.o
sound/soc/tegra/tegra_alc5632.c new file +233 -0

233 lines not on this page · raw message

Signature · 1 line
1.7.3.4
#1 lore raw

Leon Romanovsky wrote at Monday, December 12, 2011 12:46 PM:

Quoted 4 lines

At this stage only Toshiba AC100/Dynabook supported.

sound/soc/tegra/tegra_alc5632.c new file +2 -0
+static int tegra_alc5632_asoc_hw_params(struct snd_pcm_substream *substream,
+					struct snd_pcm_hw_params *params)

...

Quoted 4 lines
sound/soc/tegra/tegra_alc5632.c new file +4 -0
+	srate = params_rate(params);
+	switch (srate) {
+	case 64000:
+	case 88200:

sound/soc/tegra/tegra_alc5632.c line 65

Quoted 10 lines
sound/soc/tegra/tegra_alc5632.c new file +10 -0
+	case 96000:
+		mclk = 128 * srate;
+		break;
+	default:
+		mclk = 512 * srate;
+		break;
+	}
+	/* FIXME: Codec only requires >= 3MHz if OSR==0 */
+	while (mclk < 6000000)
+		mclk *= 2;

I think you should replace all that with:

mclk = 512 * srate;

The reason being:

* The codec only supports up to 48KHz as far as I can tell, thus making the special-case in the switch statement never fire, and irrelevant.

* The codec wants 512Fs, not min(6MHz, 512Fs) as far as I can tell, so the extra while loop at the end is going to set the wrong rate for 8KHz and 11.025KHz.

Quoted 5 lines
sound/soc/tegra/tegra_alc5632.c new file +5 -0
+static struct snd_soc_jack_gpio tegra_alc5632_hs_jack_gpio = {
+	.name = "Headset Detect",
+	.report = SND_JACK_HEADSET,
+	.debounce_time = 150,
+};

...

Quoted 2 lines
(quoted diff) +2 -0
+static int tegra_alc5632_asoc_init(struct snd_soc_pcm_runtime *rtd)
+{

...

sound/soc/tegra/tegra_alc5632.c line 141

Quoted 2 lines
sound/soc/tegra/tegra_alc5632.c new file +2 -0
+	snd_soc_jack_add_gpios(&tegra_alc5632_hs_jack,
+			1, &tegra_alc5632_hs_jack_gpio);

Since tegra_alc5632_hs_jack_gpio is never set, doesn't this end up using GPIO 0 as the headset detect, which most likely is something completely unrelated to audio.

Other than those two things, this looks fine to me.

Signature · 1 line
nvpublic
#2 lore raw
Quoted 25 lines · Stephen Warren

Leon Romanovsky wrote at Monday, December 12, 2011 12:46 PM:

Quoted 4 lines

At this stage only Toshiba AC100/Dynabook supported.

+static int tegra_alc5632_asoc_hw_params(struct snd_pcm_substream *substream,
+                                     struct snd_pcm_hw_params *params)

...

Quoted 6 lines

+     srate = params_rate(params);
+     switch (srate) {
+     case 64000:
+     case 88200:
+     case 96000:
+             mclk = 128 * srate;

Quoted 8 lines

+             break;
+     default:
+             mclk = 512 * srate;
+             break;
+     }
+     /* FIXME: Codec only requires >= 3MHz if OSR==0 */
+     while (mclk < 6000000)
+             mclk *= 2;

I think you should replace all that with:

mclk = 512 * srate;

Quoted 8 lines

The reason being:

* The codec only supports up to 48KHz as far as I can tell, thus making the special-case in the switch statement never fire, and irrelevant.

* The codec wants 512Fs, not min(6MHz, 512Fs) as far as I can tell, so the extra while loop at the end is going to set the wrong rate for 8KHz and 11.025KHz.

You are right, I'll check it at the spec.

Quoted 9 lines
Quoted 5 lines
sound/soc/tegra/tegra_alc5632.c new file +5 -0
+static struct snd_soc_jack_gpio tegra_alc5632_hs_jack_gpio = {
+     .name = "Headset Detect",
+     .report = SND_JACK_HEADSET,
+     .debounce_time = 150,
+};

...

Quoted 2 lines
(quoted diff) +2 -0
+static int tegra_alc5632_asoc_init(struct snd_soc_pcm_runtime *rtd)
+{

...

Quoted 6 lines
Quoted 2 lines

+     snd_soc_jack_add_gpios(&tegra_alc5632_hs_jack,
+                     1, &tegra_alc5632_hs_jack_gpio);

Since tegra_alc5632_hs_jack_gpio is never set, doesn't this end up using GPIO 0 as the headset detect, which most likely is something completely unrelated to audio.

This is initial support of board file, at the coming patch set, I'll update board-paz00-pinmux.c [1] with audio related pins and I'll add platform device support.
I think It is easy for review in this way.

[1] http://git.kernel.org/?p=linux/kernel/git/olof/tegra.git;a=blob;f=arch/arm/mach-tegra/board-paz00-pinmux.c;h=126892cfddce1ce222b0251f1bac972cb16b6db1;hb=refs/heads/for-next

Quoted 6 lines

Other than those two things, this looks fine to me.

--
nvpublic

Signature · 6 lines
Leon Romanovsky | Independent Linux Consultant
        www.leon.nu | [email protected]
_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#3 lore raw

Leon Romanovsky wrote at Monday, December 12, 2011 11:36 PM:

Quoted 3 lines · Stephen Warren
Quoted 2 lines

Leon Romanovsky wrote at Monday, December 12, 2011 12:46 PM:

Quoted 1 line

At this stage only Toshiba AC100/Dynabook supported.

...

Quoted 12 lines
Quoted 12 lines
Quoted 5 lines
sound/soc/tegra/tegra_alc5632.c new file +5 -0
+static struct snd_soc_jack_gpio tegra_alc5632_hs_jack_gpio = {
+     .name = "Headset Detect",
+     .report = SND_JACK_HEADSET,
+     .debounce_time = 150,
+};

...

Quoted 2 lines
(quoted diff) +2 -0
+static int tegra_alc5632_asoc_init(struct snd_soc_pcm_runtime *rtd)
+{

...

Quoted 2 lines

+     snd_soc_jack_add_gpios(&tegra_alc5632_hs_jack,
+                     1, &tegra_alc5632_hs_jack_gpio);

Quoted 8 lines
Quoted 3 lines

Since tegra_alc5632_hs_jack_gpio is never set, doesn't this end up using GPIO 0 as the headset detect, which most likely is something completely unrelated to audio.

This is initial support of board file, at the coming patch set, I'll update board-paz00-pinmux.c [1] with audio related pins and I'll add platform device support.
I think It is easy for review in this way.

I'm fine adding the feature later. But, I think you shouldn't call snd_soc_jack_add_gpios() until the platform data is there, or you'll end up enabling the feature already, but using the wrong GPIO.

Signature · 6 lines
nvpublic

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#4 lore raw
Quoted 22 lines · Stephen Warren

Leon Romanovsky wrote at Monday, December 12, 2011 11:36 PM:

Quoted 3 lines · Stephen Warren
Quoted 2 lines

Leon Romanovsky wrote at Monday, December 12, 2011 12:46 PM:

Quoted 1 line

At this stage only Toshiba AC100/Dynabook supported.

...

Quoted 16 lines
Quoted 15 lines
Quoted 5 lines
sound/soc/tegra/tegra_alc5632.c new file +5 -0
+static struct snd_soc_jack_gpio tegra_alc5632_hs_jack_gpio = {
+     .name = "Headset Detect",
+     .report = SND_JACK_HEADSET,
+     .debounce_time = 150,
+};

...

Quoted 2 lines
(quoted diff) +2 -0
+static int tegra_alc5632_asoc_init(struct snd_soc_pcm_runtime *rtd)
+{

...

Quoted 2 lines

+     snd_soc_jack_add_gpios(&tegra_alc5632_hs_jack,
+                     1, &tegra_alc5632_hs_jack_gpio);

Since tegra_alc5632_hs_jack_gpio is never set, doesn't this end up using GPIO 0 as the headset detect, which most likely is something completely unrelated to audio.

Quoted 8 lines
Quoted 4 lines

This is initial support of board file, at the coming patch set, I'll update board-paz00-pinmux.c [1] with audio related pins and I'll add platform device support.
I think It is easy for review in this way.

I'm fine adding the feature later. But, I think you shouldn't call snd_soc_jack_add_gpios() until the platform data is there, or you'll end up enabling the feature already, but using the wrong GPIO.

Does it matter ? Anyway, this file is useless without proper audio support, and currently only paz00 will be using this file.

--
Leon Romanovsky | Independent Linux Consultant
        www.leon.nu | [email protected]
_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

#5 lore raw

Leon Romanovsky wrote at Tuesday, December 13, 2011 11:27 AM:

Quoted 28 lines · Stephen Warren
Quoted 27 lines

Leon Romanovsky wrote at Monday, December 12, 2011 11:36 PM:

Quoted 3 lines · Stephen Warren
Quoted 2 lines

Leon Romanovsky wrote at Monday, December 12, 2011 12:46 PM:

Quoted 1 line

At this stage only Toshiba AC100/Dynabook supported.

...

Quoted 20 lines
Quoted 9 lines
Quoted 5 lines
sound/soc/tegra/tegra_alc5632.c new file +5 -0
+static struct snd_soc_jack_gpio tegra_alc5632_hs_jack_gpio = {
+     .name = "Headset Detect",
+     .report = SND_JACK_HEADSET,
+     .debounce_time = 150,
+};

...

Quoted 2 lines
(quoted diff) +2 -0
+static int tegra_alc5632_asoc_init(struct snd_soc_pcm_runtime *rtd)
+{

...

Quoted 6 lines
Quoted 2 lines

+     snd_soc_jack_add_gpios(&tegra_alc5632_hs_jack,
+                     1, &tegra_alc5632_hs_jack_gpio);

Since tegra_alc5632_hs_jack_gpio is never set, doesn't this end up using GPIO 0 as the headset detect, which most likely is something completely unrelated to audio.

This is initial support of board file, at the coming patch set, I'll update board-paz00-pinmux.c [1] with audio related pins and I'll add platform device support.
I think It is easy for review in this way.

Quoted 6 lines
Quoted 3 lines

I'm fine adding the feature later. But, I think you shouldn't call snd_soc_jack_add_gpios() until the platform data is there, or you'll end up enabling the feature already, but using the wrong GPIO.

Does it matter ? Anyway, this file is useless without proper audio support, and currently only paz00 will be using this file.

Yes, I think so.

The issue is that by calling snd_soc_jack_add_gpios(), ASoC will internally call gpio_request() and request_any_context_irq() for GPIO 0, which:

a) Will prevent it from being used as a GPIO by any other code.

b) Will configure the GPIO HW to make this pin an input, such that if the pin is enabled as a GPIO, might stop Tegra driving the pin if it was previously configured as an output.

c) If the GPIO is attached to some random piece of HW that toggles the line, this could cause an interrupt storm, or at least bogus jack detect events.

Just removing that one call and the unused GPIO structure would solve all this; it can easily be added back when adding the platform data support.

Signature · 6 lines
nvpublic

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel
#6 lore raw
Quoted 47 lines · Stephen Warren

Leon Romanovsky wrote at Tuesday, December 13, 2011 11:27 AM:

Quoted 28 lines · Stephen Warren
Quoted 27 lines

Leon Romanovsky wrote at Monday, December 12, 2011 11:36 PM:

Quoted 3 lines · Stephen Warren
Quoted 2 lines

Leon Romanovsky wrote at Monday, December 12, 2011 12:46 PM:

Quoted 1 line

At this stage only Toshiba AC100/Dynabook supported.

...

Quoted 20 lines
Quoted 9 lines
Quoted 5 lines
sound/soc/tegra/tegra_alc5632.c new file +5 -0
+static struct snd_soc_jack_gpio tegra_alc5632_hs_jack_gpio = {
+     .name = "Headset Detect",
+     .report = SND_JACK_HEADSET,
+     .debounce_time = 150,
+};

...

Quoted 2 lines
(quoted diff) +2 -0
+static int tegra_alc5632_asoc_init(struct snd_soc_pcm_runtime *rtd)
+{

...

Quoted 6 lines
Quoted 2 lines

+     snd_soc_jack_add_gpios(&tegra_alc5632_hs_jack,
+                     1, &tegra_alc5632_hs_jack_gpio);

Since tegra_alc5632_hs_jack_gpio is never set, doesn't this end up using GPIO 0 as the headset detect, which most likely is something completely unrelated to audio.

This is initial support of board file, at the coming patch set, I'll update board-paz00-pinmux.c [1] with audio related pins and I'll add platform device support.
I think It is easy for review in this way.

Quoted 6 lines
Quoted 3 lines

I'm fine adding the feature later. But, I think you shouldn't call snd_soc_jack_add_gpios() until the platform data is there, or you'll end up enabling the feature already, but using the wrong GPIO.

Does it matter ? Anyway, this file is useless without proper audio support, and currently only paz00 will be using this file.

Yes, I think so.

The issue is that by calling snd_soc_jack_add_gpios(), ASoC will internally call gpio_request() and request_any_context_irq() for GPIO 0, which:

a) Will prevent it from being used as a GPIO by any other code.

b) Will configure the GPIO HW to make this pin an input, such that if the pin is enabled as a GPIO, might stop Tegra driving the pin if it was previously configured as an output.

Quoted 7 lines

c) If the GPIO is attached to some random piece of HW that toggles the line, this could cause an interrupt storm, or at least bogus jack detect events.

Just removing that one call and the unused GPIO structure would solve all this; it can easily be added back when adding the platform data support.

Thanks for explanation, I'll try to resend the patch till the weekends.

Quoted 8 lines

--
nvpublic

_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

Signature · 6 lines
Leon Romanovsky | Independent Linux Consultant
        www.leon.nu | [email protected]
_______________________________________________
Alsa-devel mailing list
[email protected]
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel