overrun in case of record

overrun in case of record

[Rajeev kumar]

Hello Mark

I am testing my audio driver on spear platform.During recording I am 
getting two issues.
1. Overrun issue
2. pcm_read error

I am using arecord for capturing data.
#arecord -f dat > t
overrun!!! (at least 32.949 ms long)

Overrun issue: The issue is only reproducible if we try to record for 
more than 40 sec.The strange part is that Instead of overrun condition 
generated,
it will record complete data for 10sec.(This is verified with the number 
of bytes recorded).
So could you please let me know whether this is an error condition or 
just a warning.

pcm_read error: For this I have seen a bug is locked in the 
bugzila(*Bug 472469* <show_bug.cgi?id=472469> -arecord: input/output 
error ). and the status  is closed saying insufficient data.
Could you please let me know what is the meaning of this bug? Is it 
like, data is not available in the buffer and pcm_read from the user 
space fails? But why this situation will
arise?

Best Regards
Rajeev

Re: overrun in case of record

[Mark Brown]

On Tue, Sep 06, 2011 at 03:04:44PM +0530, Rajeev kumar wrote:

> I am using arecord for capturing data.
> #arecord -f dat > t
> overrun!!! (at least 32.949 ms long)
> 
> Overrun issue: The issue is only reproducible if we try to record
> for more than 40 sec.The strange part is that Instead of overrun
> condition generated,
> it will record complete data for 10sec.(This is verified with the
> number of bytes recorded).
> So could you please let me know whether this is an error condition
> or just a warning.

An overrun means that your driver is producing data faster than the
application is reading it and ran out of buffer space.  This may mean
that you're running at the wrong sample rate, or it may mean that your
system is just too slow to do recordings.

> pcm_read error: For this I have seen a bug is locked in the
> bugzila(*Bug 472469* <show_bug.cgi?id=472469> -arecord: input/output
> error ). and the status  is closed saying insufficient data.
> Could you please let me know what is the meaning of this bug? Is it
> like, data is not available in the buffer and pcm_read from the user
> space fails? But why this situation will
> arise?

You'll need to investigate the bug.

Re: overrun in case of record

[Rajeev kumar]

Hello Mark

On 9/6/2011 11:31 PM, Mark Brown wrote:
>> pcm_read error: For this I have seen a bug is locked in the
>> >  bugzila(*Bug 472469*<show_bug.cgi?id=472469>  -arecord: input/output
>> >  error ). and the status  is closed saying insufficient data.
>> >  Could you please let me know what is the meaning of this bug? Is it
>> >  like, data is not available in the buffer and pcm_read from the user
>> >  space fails? But why this situation will
>> >  arise?
> You'll need to investigate the bug.
> .
>
I analyzed the bug and found that if I use arecord  in non-blocking 
mode, it works fine.

#arecord -f dat  --nonblock >  t

Working in a non-blocking  mode means there is no data available in ring 
buffer and returning -EAGAIN from the alsa, so application is
wating for data to arrive in the ring buffer. Now it is not clear to me 
why this situation will arise?
I have checked the clock and all the clock are intact. The calculation 
for number of bytes received in a sec is correct.

Best Regards
Rajeev

Re: overrun in case of record

[Clemens Ladisch]

Rajeev kumar wrote:
> I analyzed the bug and found that if I use arecord  in non-blocking
> mode, it works fine.

In theory, with a correct driver, using non-blocking mode should not
make any difference.

All the symptoms indicate that the pointer callback and/or the calling
of snd_pcm_period_elapsed is somehow wrong.


Regards,
Clemens

Exporting playback and capture capability to user space

[Rajeev kumar]

Hello Mark

I have a single codec with playback and capture capability. We are using 
it for different i2s controller. Below are the details for that.

case 1: I2S controller-1: with playback and capture capability
case 2: I2S controller-2: with playback capability only
case 3: I2S controller-3: with capture capability only.

We are registering all the I2S controller with there respective  capability.

In command 'aplay -l' or 'cat /proc/asound/pcm', among playback devices, 
capture device is also listed.  This is true in case-1 but for case-2 
and case-3 also it is showing the same behavior. According to my 
understanding it should not show the list of all devices i.e. in case of 
controller having playback capability it should show only playback 
devices not capture devices and vice versa. Please correct me if I am wrong.

So  my questions are
1. How to separate playback and capture devices in case-2 and case-3. 
Are we need to register different codec for different i2s controller?

2. Is it codec capability only which is exported to user space ?


Best Regards
Rajeev

Re: Exporting playback and capture capability to user space

[Mark Brown]

On Thu, Nov 17, 2011 at 07:05:46PM +0530, Rajeev kumar wrote:

> In command 'aplay -l' or 'cat /proc/asound/pcm', among playback
> devices, capture device is also listed.  This is true in case-1 but

This is the intended behaviour, capture only PCMs are still PCMs.

Re: Exporting playback and capture capability to user space

[Rajeev kumar]

Hello Mark

Could you please help me by answering these questions?

Best Regards
Rajeev

On 11/17/2011 7:05 PM, Rajeev kumar wrote:
> Hello Mark
>
> I have a single codec with playback and capture capability. We are using
> it for different i2s controller. Below are the details for that.
>
> case 1: I2S controller-1: with playback and capture capability
> case 2: I2S controller-2: with playback capability only
> case 3: I2S controller-3: with capture capability only.
>
> We are registering all the I2S controller with there respective capability.
>
> In command 'aplay -l' or 'cat /proc/asound/pcm', among playback devices,
> capture device is also listed. This is true in case-1 but for case-2 and
> case-3 also it is showing the same behavior. According to my
> understanding it should not show the list of all devices i.e. in case of
> controller having playback capability it should show only playback
> devices not capture devices and vice versa. Please correct me if I am
> wrong.
>
> So my questions are
> 1. How to separate playback and capture devices in case-2 and case-3.
> Are we need to register different codec for different i2s controller?
>
> 2. Is it codec capability only which is exported to user space ?
>
>
> Best Regards
> Rajeev
>
>

Re: Exporting playback and capture capability to user space

[Vinod Koul]

On Mon, 2011-11-21 at 11:19 +0530, Rajeev kumar wrote:
> Hello Mark
> 
> Could you please help me by answering these questions?
> 
> Best Regards
> Rajeev
> 
Pls don't top-post!!
> On 11/17/2011 7:05 PM, Rajeev kumar wrote:
> > Hello Mark
> >
> > I have a single codec with playback and capture capability. We are using
> > it for different i2s controller. Below are the details for that.
> >
> > case 1: I2S controller-1: with playback and capture capability
> > case 2: I2S controller-2: with playback capability only
> > case 3: I2S controller-3: with capture capability only.
> >
> > We are registering all the I2S controller with there respective capability.
> >
> > In command 'aplay -l' or 'cat /proc/asound/pcm', among playback devices,
> > capture device is also listed. This is true in case-1 but for case-2 and
> > case-3 also it is showing the same behavior. According to my
> > understanding it should not show the list of all devices i.e. in case of
> > controller having playback capability it should show only playback
> > devices not capture devices and vice versa. Please correct me if I am
> > wrong.
> >
> > So my questions are
> > 1. How to separate playback and capture devices in case-2 and case-3.
> > Are we need to register different codec for different i2s controller?
How asoc builds this is based on dailinks you have provided. If for the
device 2 and device 3 the cpu and codec dais support both playback and
capture devices, it will go ahead and register these device.
The fact that for device 2 and device 3 capture is also registered
suggests me that your cpu dais are not correct.

Would help if you post your code, esp the codec, cpu dais and dialinks.

-- 
~Vinod

Re: Exporting playback and capture capability to user space

[Mark Brown]

On Mon, Nov 21, 2011 at 11:19:10AM +0530, Rajeev kumar wrote:

> Could you please help me by answering these questions?

I replied to your mail on the day it was originally posted.

Re: Exporting playback and capture capability to user space

[Rajeev kumar]

Hello Vinod, Mark


On 11/21/2011 11:53 AM, Vinod Koul wrote:
> On Mon, 2011-11-21 at 11:19 +0530, Rajeev kumar wrote:
>> Hello Mark
>>
>> Could you please help me by answering these questions?
>>
>> Best Regards
>> Rajeev
>>
> Pls don't top-post!!
>> On 11/17/2011 7:05 PM, Rajeev kumar wrote:
>>> Hello Mark
>>>
>>> I have a single codec with playback and capture capability. We are using
>>> it for different i2s controller. Below are the details for that.
>>>
>>> case 1: I2S controller-1: with playback and capture capability
>>> case 2: I2S controller-2: with playback capability only
>>> case 3: I2S controller-3: with capture capability only.
>>>
>>> We are registering all the I2S controller with there respective capability.
>>>
>>> In command 'aplay -l' or 'cat /proc/asound/pcm', among playback devices,
>>> capture device is also listed. This is true in case-1 but for case-2 and
>>> case-3 also it is showing the same behavior. According to my
>>> understanding it should not show the list of all devices i.e. in case of
>>> controller having playback capability it should show only playback
>>> devices not capture devices and vice versa. Please correct me if I am
>>> wrong.
>>>
>>> So my questions are
>>> 1. How to separate playback and capture devices in case-2 and case-3.
>>> Are we need to register different codec for different i2s controller?
> How asoc builds this is based on dailinks you have provided. If for the
> device 2 and device 3 the cpu and codec dais support both playback and
> capture devices, it will go ahead and register these device.
> The fact that for device 2 and device 3 capture is also registered
> suggests me that your cpu dais are not correct.
>
> Would help if you post your code, esp the codec, cpu dais and dialinks.
>

Codec dai:

static struct snd_soc_dai_driver sta529_dai = {
	.name = "sta529-audio",
	.playback = {
		.stream_name = "Playback",
		.channels_min = 2,
		.channels_max = 2,
		.rates = SPEAR_PCM_RATES,
		.formats = SPEAR_PCM_FORMAT,
	},
	.capture = {
		.stream_name = "Capture",
		.channels_min = 2,
		.channels_max = 2,
		.rates = SPEAR_PCM_RATES,
		.formats = SPEAR_PCM_FORMAT,
	},
	.ops	= &sta529_dai_ops,
};

cpu dai: In case of cpu dai, we are passing the playback and capture 
capability from the platform code and depending on the capability we are 
filling the cpu dai structure in probe function, like

if (cap & PLAY) {

	/* Fill Playback capability */
}

if (cap & RECORD) {
	/* Fill Capture capability */
}

dai links:

static struct snd_soc_dai_link spear_evb_dai[] = {
	 {
		.name		= "sta529-pcm0",
		.stream_name	= "I2S Playback",
		.cpu_dai_name	= "designware-i2s.0",
		.platform_name	= "spear-pcm-audio",
		.codec_dai_name	= "sta529-audio",
		.codec_name	= "sta529-codec.0-001a",
		.ops		= &sta529_ops,
	}, {
		.name		= "sta529-pcm1",
		.stream_name	= "I2S Capture",
		.cpu_dai_name	= "designware-i2s.1",
		.platform_name	= "spear-pcm-audio",
		.codec_dai_name	= "sta529-audio",
		.codec_name	= "sta529-codec.0-001a",
		.ops		= &sta529_ops,
	},
};

static struct snd_soc_card spear_snd_card = {
	.name		= "spear-evb",
	.dai_link	= spear_evb_dai,
	.num_links	= ARRAY_SIZE(spear_evb_dai),
};

Best Regards
~Rajeev

Re: Exporting playback and capture capability to user space

[Mark Brown]

On Tue, Nov 22, 2011 at 09:57:51AM +0530, Rajeev kumar wrote:

> 
> static struct snd_soc_dai_driver sta529_dai = {

> 		.rates = SPEAR_PCM_RATES,
> 		.formats = SPEAR_PCM_FORMAT,

This looks wrong...

> cpu dai: In case of cpu dai, we are passing the playback and capture
> capability from the platform code and depending on the capability we
> are filling the cpu dai structure in probe function, like

Why are you doing this?  Your driver should export the capabilities of
the hardware and let the machine driver pick any specific configuration
that's needed.

Re: Exporting playback and capture capability to user space

[Rajeev kumar]

Hello Mark

On 11/22/2011 6:38 PM, Mark Brown wrote:
> On Tue, Nov 22, 2011 at 09:57:51AM +0530, Rajeev kumar wrote:
>
>>
>> static struct snd_soc_dai_driver sta529_dai = {
>
>> 		.rates = SPEAR_PCM_RATES,
>> 		.formats = SPEAR_PCM_FORMAT,
>
> This looks wrong...
>

Agreed, This should be STA529_RATES and STA529_FORMATS.

>> cpu dai: In case of cpu dai, we are passing the playback and capture
>> capability from the platform code and depending on the capability we
>> are filling the cpu dai structure in probe function, like
>
> Why are you doing this?  Your driver should export the capabilities of
> the hardware and let the machine driver pick any specific configuration
> that's needed.
>

This is needed because we have a single cpu driver for three different 
I2S controller having different capability(as I explained in the first 
mail.). So capabilities are passed from the platform code and get filled 
in the probe part of the i2s controller.


Best Regards
~Rajeev

Re: Exporting playback and capture capability to user space

[Rajeev kumar]

Hello Vinod,

On 11/22/2011 9:57 AM, Rajeev kumar wrote:
> Hello Vinod, Mark
>
>
> On 11/21/2011 11:53 AM, Vinod Koul wrote:
>> On Mon, 2011-11-21 at 11:19 +0530, Rajeev kumar wrote:
>>> Hello Mark
>>>
>>> Could you please help me by answering these questions?
>>>
>>> Best Regards
>>> Rajeev
>>>
>> Pls don't top-post!!
>>> On 11/17/2011 7:05 PM, Rajeev kumar wrote:
>>>> Hello Mark
>>>>
>>>> I have a single codec with playback and capture capability. We are
>>>> using
>>>> it for different i2s controller. Below are the details for that.
>>>>
>>>> case 1: I2S controller-1: with playback and capture capability
>>>> case 2: I2S controller-2: with playback capability only
>>>> case 3: I2S controller-3: with capture capability only.
>>>>
>>>> We are registering all the I2S controller with there respective
>>>> capability.
>>>>
>>>> In command 'aplay -l' or 'cat /proc/asound/pcm', among playback
>>>> devices,
>>>> capture device is also listed. This is true in case-1 but for case-2
>>>> and
>>>> case-3 also it is showing the same behavior. According to my
>>>> understanding it should not show the list of all devices i.e. in
>>>> case of
>>>> controller having playback capability it should show only playback
>>>> devices not capture devices and vice versa. Please correct me if I am
>>>> wrong.
>>>>
>>>> So my questions are
>>>> 1. How to separate playback and capture devices in case-2 and case-3.
>>>> Are we need to register different codec for different i2s controller?
>> How asoc builds this is based on dailinks you have provided. If for the
>> device 2 and device 3 the cpu and codec dais support both playback and
>> capture devices, it will go ahead and register these device.
>> The fact that for device 2 and device 3 capture is also registered
>> suggests me that your cpu dais are not correct.
>>
>> Would help if you post your code, esp the codec, cpu dais and dialinks.
>>
>
> Codec dai:
>
> static struct snd_soc_dai_driver sta529_dai = {
> .name = "sta529-audio",
> .playback = {
> .stream_name = "Playback",
> .channels_min = 2,
> .channels_max = 2,
> .rates = SPEAR_PCM_RATES,
> .formats = SPEAR_PCM_FORMAT,
> },
> .capture = {
> .stream_name = "Capture",
> .channels_min = 2,
> .channels_max = 2,
> .rates = SPEAR_PCM_RATES,
> .formats = SPEAR_PCM_FORMAT,
> },
> .ops = &sta529_dai_ops,
> };
>
> cpu dai: In case of cpu dai, we are passing the playback and capture
> capability from the platform code and depending on the capability we are
> filling the cpu dai structure in probe function, like
>
> if (cap & PLAY) {
>
> /* Fill Playback capability */
> }
>
> if (cap & RECORD) {
> /* Fill Capture capability */
> }
>
> dai links:
>
> static struct snd_soc_dai_link spear_evb_dai[] = {
> {
> .name = "sta529-pcm0",
> .stream_name = "I2S Playback",
> .cpu_dai_name = "designware-i2s.0",
> .platform_name = "spear-pcm-audio",
> .codec_dai_name = "sta529-audio",
> .codec_name = "sta529-codec.0-001a",
> .ops = &sta529_ops,
> }, {
> .name = "sta529-pcm1",
> .stream_name = "I2S Capture",
> .cpu_dai_name = "designware-i2s.1",
> .platform_name = "spear-pcm-audio",
> .codec_dai_name = "sta529-audio",
> .codec_name = "sta529-codec.0-001a",
> .ops = &sta529_ops,
> },
> };
>
> static struct snd_soc_card spear_snd_card = {
> .name = "spear-evb",
> .dai_link = spear_evb_dai,
> .num_links = ARRAY_SIZE(spear_evb_dai),
> };
>

I had actually missed the last mail by mark. I dont know why? . So sorry 
for replying on the top of this mail.

Vinod, have you gone through  the cpu_dai, codec_dai and dailink ? Could 
you please make any suggestion?

Best Regards
~Rajeev



> Best Regards
> ~Rajeev

Re: Exporting playback and capture capability to user space

[Vinod Koul]

On Fri, 2011-12-02 at 16:20 +0530, Rajeev kumar wrote:
> > cpu dai: In case of cpu dai, we are passing the playback and capture
> > capability from the platform code and depending on the capability we
> are
> > filling the cpu dai structure in probe function, like
Unless we find what is being passed, its hard to comment.

As I said previously this will happen when you are reporting cpu dai
wrongly, so  hack there to fix your issue

-- 
~Vinod

Re: Exporting playback and capture capability to user space

[Rajeev kumar]

On 12/2/2011 4:20 PM, Rajeev kumar wrote:
> Hello Vinod,
>
> On 11/22/2011 9:57 AM, Rajeev kumar wrote:
>> Hello Vinod, Mark
>>
>>
>> On 11/21/2011 11:53 AM, Vinod Koul wrote:
>>> On Mon, 2011-11-21 at 11:19 +0530, Rajeev kumar wrote:
>>>> Hello Mark
>>>>
>>>> Could you please help me by answering these questions?
>>>>
>>>> Best Regards
>>>> Rajeev
>>>>
>>> Pls don't top-post!!
>>>> On 11/17/2011 7:05 PM, Rajeev kumar wrote:
>>>>> Hello Mark
>>>>>
>>>>> I have a single codec with playback and capture capability. We are
>>>>> using
>>>>> it for different i2s controller. Below are the details for that.
>>>>>
>>>>> case 1: I2S controller-1: with playback and capture capability
>>>>> case 2: I2S controller-2: with playback capability only
>>>>> case 3: I2S controller-3: with capture capability only.
>>>>>
>>>>> We are registering all the I2S controller with there respective
>>>>> capability.
>>>>>
>>>>> In command 'aplay -l' or 'cat /proc/asound/pcm', among playback
>>>>> devices,
>>>>> capture device is also listed. This is true in case-1 but for case-2
>>>>> and
>>>>> case-3 also it is showing the same behavior. According to my
>>>>> understanding it should not show the list of all devices i.e. in
>>>>> case of
>>>>> controller having playback capability it should show only playback
>>>>> devices not capture devices and vice versa. Please correct me if I am
>>>>> wrong.
>>>>>
>>>>> So my questions are
>>>>> 1. How to separate playback and capture devices in case-2 and case-3.
>>>>> Are we need to register different codec for different i2s controller?
>>> How asoc builds this is based on dailinks you have provided. If for the
>>> device 2 and device 3 the cpu and codec dais support both playback and
>>> capture devices, it will go ahead and register these device.
>>> The fact that for device 2 and device 3 capture is also registered
>>> suggests me that your cpu dais are not correct.
>>>
>>> Would help if you post your code, esp the codec, cpu dais and dialinks.
>>>
>>
>> Codec dai:
>>
>> static struct snd_soc_dai_driver sta529_dai = {
>> .name = "sta529-audio",
>> .playback = {
>> .stream_name = "Playback",
>> .channels_min = 2,
>> .channels_max = 2,
>> .rates = SPEAR_PCM_RATES,
>> .formats = SPEAR_PCM_FORMAT,
>> },
>> .capture = {
>> .stream_name = "Capture",
>> .channels_min = 2,
>> .channels_max = 2,
>> .rates = SPEAR_PCM_RATES,
>> .formats = SPEAR_PCM_FORMAT,
>> },
>> .ops = &sta529_dai_ops,
>> };
>>
>> cpu dai: In case of cpu dai, we are passing the playback and capture
>> capability from the platform code and depending on the capability we are
>> filling the cpu dai structure in probe function, like
>>
>> if (cap & PLAY) {
>>
>> /* Fill Playback capability */
>> }
>>
>> if (cap & RECORD) {
>> /* Fill Capture capability */
>> }
>>
>> dai links:
>>
>> static struct snd_soc_dai_link spear_evb_dai[] = {
>> {
>> .name = "sta529-pcm0",
>> .stream_name = "I2S Playback",
>> .cpu_dai_name = "designware-i2s.0",
>> .platform_name = "spear-pcm-audio",
>> .codec_dai_name = "sta529-audio",
>> .codec_name = "sta529-codec.0-001a",
>> .ops = &sta529_ops,
>> }, {
>> .name = "sta529-pcm1",
>> .stream_name = "I2S Capture",
>> .cpu_dai_name = "designware-i2s.1",
>> .platform_name = "spear-pcm-audio",
>> .codec_dai_name = "sta529-audio",
>> .codec_name = "sta529-codec.0-001a",
>> .ops = &sta529_ops,
>> },
>> };
>>
>> static struct snd_soc_card spear_snd_card = {
>> .name = "spear-evb",
>> .dai_link = spear_evb_dai,
>> .num_links = ARRAY_SIZE(spear_evb_dai),
>> };
>>
>
> I had actually missed the last mail by mark. I dont know why? . So sorry
> for replying on the top of this mail.
>
> Vinod, have you gone through the cpu_dai, codec_dai and dailink ? Could
> you please make any suggestion?
>

Ping

> Best Regards
> ~Rajeev
>
>
>
>> Best Regards
>> ~Rajeev
>

Re: Exporting playback and capture capability to user space

[Rajeev kumar]

Hello Mark

On 1/9/2012 12:01 PM, Rajeev kumar wrote:
> On 12/2/2011 4:20 PM, Rajeev kumar wrote:
>> Hello Vinod,
>>
>> On 11/22/2011 9:57 AM, Rajeev kumar wrote:
>>> Hello Vinod, Mark
>>>
>>>
>>> On 11/21/2011 11:53 AM, Vinod Koul wrote:
>>>> On Mon, 2011-11-21 at 11:19 +0530, Rajeev kumar wrote:
>>>>> Hello Mark
>>>>>
>>>>> Could you please help me by answering these questions?
>>>>>
>>>>> Best Regards
>>>>> Rajeev
>>>>>
>>>> Pls don't top-post!!
>>>>> On 11/17/2011 7:05 PM, Rajeev kumar wrote:
>>>>>> Hello Mark
>>>>>>
>>>>>> I have a single codec with playback and capture capability. We are
>>>>>> using
>>>>>> it for different i2s controller. Below are the details for that.
>>>>>>
>>>>>> case 1: I2S controller-1: with playback and capture capability
>>>>>> case 2: I2S controller-2: with playback capability only
>>>>>> case 3: I2S controller-3: with capture capability only.
>>>>>>
>>>>>> We are registering all the I2S controller with there respective
>>>>>> capability.
>>>>>>
>>>>>> In command 'aplay -l' or 'cat /proc/asound/pcm', among playback
>>>>>> devices,
>>>>>> capture device is also listed. This is true in case-1 but for case-2
>>>>>> and
>>>>>> case-3 also it is showing the same behavior. According to my
>>>>>> understanding it should not show the list of all devices i.e. in
>>>>>> case of
>>>>>> controller having playback capability it should show only playback
>>>>>> devices not capture devices and vice versa. Please correct me if I am
>>>>>> wrong.
>>>>>>
>>>>>> So my questions are
>>>>>> 1. How to separate playback and capture devices in case-2 and case-3.
>>>>>> Are we need to register different codec for different i2s controller?
>>>> How asoc builds this is based on dailinks you have provided. If for the
>>>> device 2 and device 3 the cpu and codec dais support both playback and
>>>> capture devices, it will go ahead and register these device.
>>>> The fact that for device 2 and device 3 capture is also registered
>>>> suggests me that your cpu dais are not correct.
>>>>
>>>> Would help if you post your code, esp the codec, cpu dais and dialinks.
>>>>
>>>
>>> Codec dai:
>>>
>>> static struct snd_soc_dai_driver sta529_dai = {
>>> .name = "sta529-audio",
>>> .playback = {
>>> .stream_name = "Playback",
>>> .channels_min = 2,
>>> .channels_max = 2,
>>> .rates = SPEAR_PCM_RATES,
>>> .formats = SPEAR_PCM_FORMAT,
>>> },
>>> .capture = {
>>> .stream_name = "Capture",
>>> .channels_min = 2,
>>> .channels_max = 2,
>>> .rates = SPEAR_PCM_RATES,
>>> .formats = SPEAR_PCM_FORMAT,
>>> },
>>> .ops = &sta529_dai_ops,
>>> };
>>>
>>> cpu dai: In case of cpu dai, we are passing the playback and capture
>>> capability from the platform code and depending on the capability we are
>>> filling the cpu dai structure in probe function, like
>>>
>>> if (cap & PLAY) {
>>>
>>> /* Fill Playback capability */
>>> }
>>>
>>> if (cap & RECORD) {
>>> /* Fill Capture capability */
>>> }
>>>
>>> dai links:
>>>
>>> static struct snd_soc_dai_link spear_evb_dai[] = {
>>> {
>>> .name = "sta529-pcm0",
>>> .stream_name = "I2S Playback",
>>> .cpu_dai_name = "designware-i2s.0",
>>> .platform_name = "spear-pcm-audio",
>>> .codec_dai_name = "sta529-audio",
>>> .codec_name = "sta529-codec.0-001a",
>>> .ops = &sta529_ops,
>>> }, {
>>> .name = "sta529-pcm1",
>>> .stream_name = "I2S Capture",
>>> .cpu_dai_name = "designware-i2s.1",
>>> .platform_name = "spear-pcm-audio",
>>> .codec_dai_name = "sta529-audio",
>>> .codec_name = "sta529-codec.0-001a",
>>> .ops = &sta529_ops,
>>> },
>>> };
>>>
>>> static struct snd_soc_card spear_snd_card = {
>>> .name = "spear-evb",
>>> .dai_link = spear_evb_dai,
>>> .num_links = ARRAY_SIZE(spear_evb_dai),
>>> };
>>>
>>
>> I had actually missed the last mail by mark. I dont know why? . So sorry
>> for replying on the top of this mail.
>>
>> Vinod, have you gone through the cpu_dai, codec_dai and dailink ? Could
>> you please make any suggestion?
>>
>
> Ping
>

The o/p of the cat /proc/asound/pcm does not depends on the &ing of the 
capability of i2s controller and codec. I was going through the 
soc-core.c and it seems that the capability exposed to user space only 
depends on the codec. If codec has both playback and capture capability 
,then in user space both will be exported and hence cat /proc/asound/pcm 
will show two different devices one for playback and other for capture. 
In case codec has only playback capability, the o/p of the cat 
/proc/asound/pcm will contain only playback device.

So in case of three different i2s controller with different capability, 
separate codec are required.

Please correct me if I am wrong.

Best Regards
Rajeev

>> Best Regards
>> ~Rajeev
>>
>>
>>
>>> Best Regards
>>> ~Rajeev
>>
>