diff mbox

[2/2] ASoC: samsung: Add machine driver for odroidx2

Message ID 1400759708-25831-2-git-send-email-s.nawrocki@samsung.com (mailing list archive)
State New, archived
Headers show

Commit Message

Sylwester Nawrocki/Kernel \(PLT\) /SRPOL/Staff Engineer/Samsung Electronics May 22, 2014, 11:55 a.m. UTC 
From: Chen Zhen <zhen1.chen@samsung.com>

This machine driver primary defines the audio path, the supported
data formats and sample frequency.
The codec works in I2S master mode, in order to achieve required
exact audio sampling frequencies without a need to adjust the EPLL
frequency to each sample rate at runtime.

Signed-off-by: Chen Zhen <zhen1.chen@samsung.com>
[s.nawrocki@samsung.com: coding style changes, fixed of_match table name]
Signed-off-by: Sylwester Nawrocki <s.nawrocki@samsung.com>
---
The only bits missing for v3.16 for audio on the Odroid X2/U3 to work
now seem to be only my clock defaults setting patches [1] and the related
dts updates.

[1] http://www.spinics.net/lists/devicetree/msg34718.html

 sound/soc/samsung/Kconfig             |    8 ++
 sound/soc/samsung/Makefile            |    2 +
 sound/soc/samsung/odroidx2_max98090.c |  150 +++++++++++++++++++++++++++++++++
 3 files changed, 160 insertions(+)
 create mode 100644 sound/soc/samsung/odroidx2_max98090.c

--
1.7.9.5

Comments

Mark Brown May 22, 2014, 6:53 p.m. UTC | #1 
On Thu, May 22, 2014 at 01:55:08PM +0200, Sylwester Nawrocki wrote:
> From: Chen Zhen <zhen1.chen@samsung.com>
> 
> This machine driver primary defines the audio path, the supported
> data formats and sample frequency.

I guess it can't yet, it would have been good to explain that in the
changelog for the bindings (and include the binding in this patch...).

> +	ret = snd_soc_dai_set_fmt(codec_dai, SND_SOC_DAIFMT_I2S
> +					| SND_SOC_DAIFMT_NB_NF
> +					| SND_SOC_DAIFMT_CBM_CFM);
> +	if (ret < 0)
> +		return ret;
> +
> +	ret = snd_soc_dai_set_fmt(cpu_dai, SND_SOC_DAIFMT_I2S
> +					| SND_SOC_DAIFMT_NB_NF
> +					| SND_SOC_DAIFMT_CBM_CFM);
> +	if (ret < 0)
> +		return ret;

These are constant, set these in the dai_link.

> +
> +	ret = snd_soc_dai_set_sysclk(codec_dai, 3, MAX98090_MCLK_FREQ,
> +				     SND_SOC_CLOCK_IN);
> +	if (ret < 0) {
> +		dev_err(codec_dai->dev,
> +			"Unable to switch to FLL1: %d\n", ret);
> +		return ret;
> +	}

So's this - also what is 3?  Set it on init.

> +	ret = snd_soc_dai_set_sysclk(cpu_dai, SAMSUNG_I2S_OPCLK,
> +					0, MOD_OPCLK_PCLK);

Similarly for the rest of the function.

> +	return snd_soc_register_card(card);

devm_snd_soc_register_card().
Sylwester Nawrocki/Kernel \(PLT\) /SRPOL/Staff Engineer/Samsung Electronics May 23, 2014, 8:44 a.m. UTC | #2 
On 22/05/14 20:53, Mark Brown wrote:
> On Thu, May 22, 2014 at 01:55:08PM +0200, Sylwester Nawrocki wrote:
>> From: Chen Zhen <zhen1.chen@samsung.com>
>>
>> This machine driver primary defines the audio path, the supported
>> data formats and sample frequency.
> 
> I guess it can't yet, it would have been good to explain that in the
> changelog for the bindings (and include the binding in this patch...).

Thanks for the review, I'm still rather inexperienced in the ASoC stuff,
I've obviously overlooked there is actually no routing handling. I'd rather
add those bits.
As for the DT binding documentation patch, I thought we're supposed to keep
documentation in separate patches, so processes of separating it from the
kernel is easier ?

>> +	ret = snd_soc_dai_set_fmt(codec_dai, SND_SOC_DAIFMT_I2S
>> +					| SND_SOC_DAIFMT_NB_NF
>> +					| SND_SOC_DAIFMT_CBM_CFM);
>> +	if (ret < 0)
>> +		return ret;
>> +
>> +	ret = snd_soc_dai_set_fmt(cpu_dai, SND_SOC_DAIFMT_I2S
>> +					| SND_SOC_DAIFMT_NB_NF
>> +					| SND_SOC_DAIFMT_CBM_CFM);
>> +	if (ret < 0)
>> +		return ret;
> 
> These are constant, set these in the dai_link.
> 
>> +
>> +	ret = snd_soc_dai_set_sysclk(codec_dai, 3, MAX98090_MCLK_FREQ,
>> +				     SND_SOC_CLOCK_IN);
>> +	if (ret < 0) {
>> +		dev_err(codec_dai->dev,
>> +			"Unable to switch to FLL1: %d\n", ret);
>> +		return ret;
>> +	}
> 
> So's this - also what is 3?  Set it on init.

OK, will move it. I'm not sure why 3, it seems to be ignored by the codec
anyway. Chen, can you confirm the second argument is actually a don't care
value ?

>> +	ret = snd_soc_dai_set_sysclk(cpu_dai, SAMSUNG_I2S_OPCLK,
>> +					0, MOD_OPCLK_PCLK);
> 
> Similarly for the rest of the function.
> 
>> +	return snd_soc_register_card(card);
> 
> devm_snd_soc_register_card().

Thanks for the suggestions, I'll amend that.


--
Regards,
Sylwester
Mark Brown May 23, 2014, 11:05 a.m. UTC | #3 
On Fri, May 23, 2014 at 10:44:56AM +0200, Sylwester Nawrocki wrote:
> On 22/05/14 20:53, Mark Brown wrote:

> As for the DT binding documentation patch, I thought we're supposed to keep
> documentation in separate patches, so processes of separating it from the
> kernel is easier ?

I'm not sure what the current theory is - they should definitely be
separate if they're part of a multi-patch series but for a single patch
I don't remember what the current taste is.  If you're sending to the
same recipients anyway splitting isn't going to accomplish much, the
idea was to only send the DT binding to the DT maintainers and list and
not the code.

> >> +	ret = snd_soc_dai_set_sysclk(codec_dai, 3, MAX98090_MCLK_FREQ,
> >> +				     SND_SOC_CLOCK_IN);
> >> +	if (ret < 0) {
> >> +		dev_err(codec_dai->dev,
> >> +			"Unable to switch to FLL1: %d\n", ret);
> >> +		return ret;
> >> +	}

> > So's this - also what is 3?  Set it on init.

> OK, will move it. I'm not sure why 3, it seems to be ignored by the codec
> anyway. Chen, can you confirm the second argument is actually a don't care
> value ?

If it's don't care supply 0 instead.
Sylwester Nawrocki/Kernel \(PLT\) /SRPOL/Staff Engineer/Samsung Electronics July 4, 2014, 11:04 a.m. UTC | #4 
On 22/05/14 20:53, Mark Brown wrote:
>> +	ret = snd_soc_dai_set_fmt(codec_dai, SND_SOC_DAIFMT_I2S
>> > +					| SND_SOC_DAIFMT_NB_NF
>> > +					| SND_SOC_DAIFMT_CBM_CFM);
>> > +	if (ret < 0)
>> > +		return ret;
>> > +
>> > +	ret = snd_soc_dai_set_fmt(cpu_dai, SND_SOC_DAIFMT_I2S
>> > +					| SND_SOC_DAIFMT_NB_NF
>> > +					| SND_SOC_DAIFMT_CBM_CFM);
>> > +	if (ret < 0)
>> > +		return ret;
>
> These are constant, set these in the dai_link.

set_fmt also sets master/slave mode of the I2S DAI, after I moved
this into the cpu_dai link data structure after suspend/resume cycle
the I2S IP block is not being properly re-configured. Should the
format setting be added in resume_post callback, or is there any
other preferred way ? Similarly the syclk settings are being lost
over suspend/resume cycle and nothing restores them.

--
Thanks,
Sylwester
Lars-Peter Clausen July 4, 2014, 11:10 a.m. UTC | #5 
On 07/04/2014 01:04 PM, Sylwester Nawrocki wrote:
> On 22/05/14 20:53, Mark Brown wrote:
>>> +	ret = snd_soc_dai_set_fmt(codec_dai, SND_SOC_DAIFMT_I2S
>>>> +					| SND_SOC_DAIFMT_NB_NF
>>>> +					| SND_SOC_DAIFMT_CBM_CFM);
>>>> +	if (ret < 0)
>>>> +		return ret;
>>>> +
>>>> +	ret = snd_soc_dai_set_fmt(cpu_dai, SND_SOC_DAIFMT_I2S
>>>> +					| SND_SOC_DAIFMT_NB_NF
>>>> +					| SND_SOC_DAIFMT_CBM_CFM);
>>>> +	if (ret < 0)
>>>> +		return ret;
>>
>> These are constant, set these in the dai_link.
>
> set_fmt also sets master/slave mode of the I2S DAI, after I moved
> this into the cpu_dai link data structure after suspend/resume cycle
> the I2S IP block is not being properly re-configured. Should the
> format setting be added in resume_post callback, or is there any
> other preferred way ? Similarly the syclk settings are being lost
> over suspend/resume cycle and nothing restores them.

The I2S driver should save and restore it's register during suspend/resume.

- Lars
Sylwester Nawrocki/Kernel \(PLT\) /SRPOL/Staff Engineer/Samsung Electronics July 4, 2014, 11:24 a.m. UTC | #6 
On 04/07/14 13:10, Lars-Peter Clausen wrote:
> On 07/04/2014 01:04 PM, Sylwester Nawrocki wrote:
>> On 22/05/14 20:53, Mark Brown wrote:
>>>> +	ret = snd_soc_dai_set_fmt(codec_dai, SND_SOC_DAIFMT_I2S
>>>>> +					| SND_SOC_DAIFMT_NB_NF
>>>>> +					| SND_SOC_DAIFMT_CBM_CFM);
>>>>> +	if (ret < 0)
>>>>> +		return ret;
>>>>> +
>>>>> +	ret = snd_soc_dai_set_fmt(cpu_dai, SND_SOC_DAIFMT_I2S
>>>>> +					| SND_SOC_DAIFMT_NB_NF
>>>>> +					| SND_SOC_DAIFMT_CBM_CFM);
>>>>> +	if (ret < 0)
>>>>> +		return ret;
>>>
>>> These are constant, set these in the dai_link.
>>
>> set_fmt also sets master/slave mode of the I2S DAI, after I moved
>> this into the cpu_dai link data structure after suspend/resume cycle
>> the I2S IP block is not being properly re-configured. Should the
>> format setting be added in resume_post callback, or is there any
>> other preferred way ? Similarly the syclk settings are being lost
>> over suspend/resume cycle and nothing restores them.
> 
> The I2S driver should save and restore it's register during suspend/resume.

OK, thanks. I checked the I2S driver does it, but only when dai->active
flags is set [1]. However, during system resume this flag is not set
(in situation where playback or capture wasn't active right before
system suspend). Is the dai->active test wrong in that driver then ?
I'm not sure if it could just be removed.

[1] http://lxr.free-electrons.com/source/sound/soc/samsung/i2s.c#L913

--
Regards,
Sylwester
Lars-Peter Clausen July 4, 2014, 11:28 a.m. UTC | #7 
On 07/04/2014 01:24 PM, Sylwester Nawrocki wrote:
> On 04/07/14 13:10, Lars-Peter Clausen wrote:
>> On 07/04/2014 01:04 PM, Sylwester Nawrocki wrote:
>>> On 22/05/14 20:53, Mark Brown wrote:
>>>>> +	ret = snd_soc_dai_set_fmt(codec_dai, SND_SOC_DAIFMT_I2S
>>>>>> +					| SND_SOC_DAIFMT_NB_NF
>>>>>> +					| SND_SOC_DAIFMT_CBM_CFM);
>>>>>> +	if (ret < 0)
>>>>>> +		return ret;
>>>>>> +
>>>>>> +	ret = snd_soc_dai_set_fmt(cpu_dai, SND_SOC_DAIFMT_I2S
>>>>>> +					| SND_SOC_DAIFMT_NB_NF
>>>>>> +					| SND_SOC_DAIFMT_CBM_CFM);
>>>>>> +	if (ret < 0)
>>>>>> +		return ret;
>>>>
>>>> These are constant, set these in the dai_link.
>>>
>>> set_fmt also sets master/slave mode of the I2S DAI, after I moved
>>> this into the cpu_dai link data structure after suspend/resume cycle
>>> the I2S IP block is not being properly re-configured. Should the
>>> format setting be added in resume_post callback, or is there any
>>> other preferred way ? Similarly the syclk settings are being lost
>>> over suspend/resume cycle and nothing restores them.
>>
>> The I2S driver should save and restore it's register during suspend/resume.
>
> OK, thanks. I checked the I2S driver does it, but only when dai->active
> flags is set [1]. However, during system resume this flag is not set
> (in situation where playback or capture wasn't active right before
> system suspend). Is the dai->active test wrong in that driver then ?
> I'm not sure if it could just be removed.

Yes, just remove the check.

- Lars
diff mbox

Patch

diff --git a/sound/soc/samsung/Kconfig b/sound/soc/samsung/Kconfig
index f2e2891..aa5ce7c 100644
--- a/sound/soc/samsung/Kconfig
+++ b/sound/soc/samsung/Kconfig
@@ -231,3 +231,11 @@  config SND_SOC_LITTLEMILL
 	select SND_SAMSUNG_I2S
 	select MFD_WM8994
 	select SND_SOC_WM8994
+
+config SND_SOC_ODROIDX2
+	tristate "Audio support for Odroid-X2 and Odroid-U3"
+	depends on SND_SOC_SAMSUNG
+	select SND_SOC_MAX98090
+	select SND_SAMSUNG_I2S
+	help
+	  Say Y here to enable audio support for the Odroid-X2/U3.
diff --git a/sound/soc/samsung/Makefile b/sound/soc/samsung/Makefile
index 86715d8..db89489 100644
--- a/sound/soc/samsung/Makefile
+++ b/sound/soc/samsung/Makefile
@@ -45,6 +45,7 @@  snd-soc-tobermory-objs := tobermory.o
 snd-soc-lowland-objs := lowland.o
 snd-soc-littlemill-objs := littlemill.o
 snd-soc-bells-objs := bells.o
+snd-soc-odroidx2-max98090-objs := odroidx2_max98090.o

 obj-$(CONFIG_SND_SOC_SAMSUNG_JIVE_WM8750) += snd-soc-jive-wm8750.o
 obj-$(CONFIG_SND_SOC_SAMSUNG_NEO1973_WM8753) += snd-soc-neo1973-wm8753.o
@@ -69,3 +70,4 @@  obj-$(CONFIG_SND_SOC_TOBERMORY) += snd-soc-tobermory.o
 obj-$(CONFIG_SND_SOC_LOWLAND) += snd-soc-lowland.o
 obj-$(CONFIG_SND_SOC_LITTLEMILL) += snd-soc-littlemill.o
 obj-$(CONFIG_SND_SOC_BELLS) += snd-soc-bells.o
+obj-$(CONFIG_SND_SOC_ODROIDX2) += snd-soc-odroidx2-max98090.o
diff --git a/sound/soc/samsung/odroidx2_max98090.c b/sound/soc/samsung/odroidx2_max98090.c
new file mode 100644
index 0000000..c3065a9
--- /dev/null
+++ b/sound/soc/samsung/odroidx2_max98090.c
@@ -0,0 +1,150 @@ 
+/*
+ * Copyright (C) 2014 Samsung Electronics Co., Ltd.
+ *
+ * This program is free software; you can redistribute  it and/or modify it
+ * under  the terms of  the GNU General  Public License as published by the
+ * Free Software Foundation;  either version 2 of the  License, or (at your
+ * option) any later version.
+ */
+
+#include <linux/of.h>
+#include <linux/module.h>
+#include <sound/soc.h>
+#include <sound/pcm_params.h>
+#include "i2s.h"
+#include "i2s-regs.h"
+
+/* The I2S CDCLK output clock frequency to the MAX98090 codec. */
+#define MAX98090_MCLK_FREQ 19200000
+
+static int odroidx2_hw_params(struct snd_pcm_substream *substream,
+			      struct snd_pcm_hw_params *params)
+{
+	struct snd_soc_pcm_runtime *rtd = substream->private_data;
+	struct snd_soc_dai *cpu_dai = rtd->cpu_dai;
+	struct snd_soc_dai *codec_dai = rtd->codec_dai;
+	int ret;
+
+	ret = snd_soc_dai_set_fmt(codec_dai, SND_SOC_DAIFMT_I2S
+					| SND_SOC_DAIFMT_NB_NF
+					| SND_SOC_DAIFMT_CBM_CFM);
+	if (ret < 0)
+		return ret;
+
+	ret = snd_soc_dai_set_fmt(cpu_dai, SND_SOC_DAIFMT_I2S
+					| SND_SOC_DAIFMT_NB_NF
+					| SND_SOC_DAIFMT_CBM_CFM);
+	if (ret < 0)
+		return ret;
+
+	ret = snd_soc_dai_set_sysclk(codec_dai, 3, MAX98090_MCLK_FREQ,
+				     SND_SOC_CLOCK_IN);
+	if (ret < 0) {
+		dev_err(codec_dai->dev,
+			"Unable to switch to FLL1: %d\n", ret);
+		return ret;
+	}
+
+	ret = snd_soc_dai_set_sysclk(cpu_dai, SAMSUNG_I2S_OPCLK,
+					0, MOD_OPCLK_PCLK);
+	if (ret < 0) {
+		dev_err(cpu_dai->dev,
+			"Unable to set i2s opclk: 0x%x\n", ret);
+		return ret;
+	}
+
+	/* Set the cpu DAI configuration in order to use CDCLK */
+	ret = snd_soc_dai_set_sysclk(cpu_dai, SAMSUNG_I2S_CDCLK,
+					0, SND_SOC_CLOCK_OUT);
+	if (ret < 0)
+		return ret;
+
+	dev_dbg(codec_dai->dev,
+		"HiFi DAI %s params: channel %d, rate %d as I2S slave\n",
+		((substream->stream == SNDRV_PCM_STREAM_PLAYBACK) ?
+		 "playback" : "capture"),  params_channels(params),
+		params_rate(params));
+
+	return 0;
+}
+
+static struct snd_soc_ops odroidx2_ops = {
+	.hw_params	= odroidx2_hw_params,
+};
+
+static struct snd_soc_dai_link odroidx2_dai = {
+	.name		= "MAX98090",
+	.stream_name	= "MAX98090 PCM",
+	.codec_dai_name	= "HiFi",
+	.ops		= &odroidx2_ops,
+};
+
+static struct snd_soc_card odroidx2 = {
+	.name		= "odroidx2",
+	.owner		= THIS_MODULE,
+	.dai_link	= &odroidx2_dai,
+	.num_links	= 1,
+};
+
+static int odroidx2_audio_probe(struct platform_device *pdev)
+{
+	struct device_node *np = pdev->dev.of_node;
+	struct snd_soc_card *card = &odroidx2;
+
+	if (!np)
+		return -ENODEV;
+
+	card->dev = &pdev->dev;
+
+	odroidx2_dai.codec_name = NULL;
+	odroidx2_dai.codec_of_node = of_parse_phandle(np,
+						"samsung,audio-codec", 0);
+	if (!odroidx2_dai.codec_of_node) {
+		dev_err(&pdev->dev,
+			"missing or invalid samsung,audio-codec property\n");
+		return -EINVAL;
+	}
+
+	odroidx2_dai.cpu_name = NULL;
+	odroidx2_dai.cpu_of_node = of_parse_phandle(np,
+						"samsung,i2s-controller", 0);
+	if (!odroidx2_dai.cpu_of_node) {
+		dev_err(&pdev->dev,
+			"missing or invalid samsung,i2s-controller property\n");
+		return -EINVAL;
+	}
+
+	odroidx2_dai.platform_of_node = odroidx2_dai.cpu_of_node;
+
+	return snd_soc_register_card(card);
+}
+
+static int odroidx2_audio_remove(struct platform_device *pdev)
+{
+	struct snd_soc_card *card = platform_get_drvdata(pdev);
+
+	snd_soc_unregister_card(card);
+
+	return 0;
+}
+
+static const struct of_device_id odroidx2_audio_of_match[] = {
+	{ .compatible = "samsung,odroidx2-audio" },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, odroidx2_audio_of_match);
+
+static struct platform_driver odroidx2_audio_driver = {
+	.driver = {
+		.name		= "odroidx2-audio",
+		.owner		= THIS_MODULE,
+		.of_match_table	= odroidx2_audio_of_match,
+	},
+	.probe	= odroidx2_audio_probe,
+	.remove	= odroidx2_audio_remove,
+};
+module_platform_driver(odroidx2_audio_driver);
+
+MODULE_AUTHOR("zhen1.chen@samsung.com");
+MODULE_DESCRIPTION("ALSA SoC Odroidx2 Audio Support");
+MODULE_LICENSE("GPL v2");