Do you mean that there is no test program to verify the implementation is correct ?

Some sound card 's hardware has FIFO buffer.

For example , au88x0

according to patent 6167465

Additionally, a first-in-first-out (FIFO) buffer is established for the DMA channel. Then a DMA request is generated in order to begin filling the FIFO buffer with data from the first data segment 150(0) using the information provided by CPU 210.

The FIFO buffer may be any suitable size capable of handling bus latencies while small enough such that the memory requirement is not overly burdensome. In one embodiment of the present invention, the FIFO buffer is 32 samples or words in depth, which would withstand a maximum bus latency of about 4 microseconds. By way of example, if 48 DMA channels trigger simultaneously, with a trigger level set at half the buffer size (16 samples), each DMA channel can wait 16 samples (20 microseconds per sample), or approximately 320 microseconds before running out of data.

It seem that alsa-lib should not allow application to rewind the application pointer to hardware pointer since you cannot write the next sample because of FIFO buffer even using mmap read/write

or bug in *snd_pcm_rewindable() ?*

2009/9/24 Takashi Iwai tiwai@suse.de

At Thu, 24 Sep 2009 13:04:24 +0530, Anuj Aggarwal wrote:

...

Hi,

I want to try the forward/rewind APIs supported by the ALSA library. Can someone point to an ALSA application which exercises these APIs?

Pulseaudio is using snd_pcm_rewind(). About snd_pcm_forward(), I don't know of...

Takashi _______________________________________________ Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

It seem that the current implementation , it seem that the application can rewind the application pointer to hardware pointer , rewrite the next sample.for all drivers

static snd_pcm_sframes_t snd_pcm_hw_rewindable(snd_pcm_t *pcm) { return snd_pcm_mmap_hw_avail(pcm); }

However this is almost impossible for any sound cards since the hardware register just report the number of samples processed. this mean that the sound card is already processing the next sample.

2009/9/24 Anuj Aggarwal anuj.aggarwal@gmail.com

Yeah, I am looking for some sample application to test forward/rewind functionality.

Regards, Anuj

On Thu, Sep 24, 2009 at 1:27 PM, Raymond Yau superquad.vortex2@gmail.com wrote:

...

Do you mean that there is no test program to verify the implementation is correct ?

Some sound card 's hardware has FIFO buffer.

For example , au88x0

according to patent 6167465

Additionally, a first-in-first-out (FIFO) buffer is established for the

DMA

...

channel. Then a DMA request is generated in order to begin filling the

FIFO

...

buffer with data from the first data segment 150(0) using the information provided by CPU 210.

The FIFO buffer may be any suitable size capable of handling bus

latencies

...

while small enough such that the memory requirement is not overly burdensome. In one embodiment of the present invention, the FIFO buffer

is

...

32 samples or words in depth, which would withstand a maximum bus latency

of

...

about 4 microseconds. By way of example, if 48 DMA channels trigger simultaneously, with a trigger level set at half the buffer size (16 samples), each DMA channel can wait 16 samples (20 microseconds per

sample),

...

or approximately 320 microseconds before running out of data.

It seem that alsa-lib should not allow application to rewind the

application

...

pointer to hardware pointer since you cannot write the next sample

because

...

of FIFO buffer even using mmap read/write

or bug in *snd_pcm_rewindable() ?*

2009/9/24 Takashi Iwai tiwai@suse.de

...

At Thu, 24 Sep 2009 13:04:24 +0530, Anuj Aggarwal wrote:

...

Hi,

I want to try the forward/rewind APIs supported by the ALSA library. Can someone point to an ALSA application which exercises these APIs?

Pulseaudio is using snd_pcm_rewind(). About snd_pcm_forward(), I don't know of...

Takashi _______________________________________________ Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

---

Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

-- Best Regards, Anuj Aggarwal

Anuj Aggarwal wrote:

I found the solution, one can use mplayer to test forward/rewind feature of ALSA library.

mplayer disables ALSA's underrun detection and uses snd_pcm_forward() to recover from an underrun, if it happens.

(To force an underrun, press Ctrl+S (then Ctrl+Q) to prevent mplayer from updating its status line.)

HTH Clemens

It seem that the difference between the application pointer and the hardware pointer is just the theoretical value which none of the sound card can achieve

static snd_pcm_sframes_t snd_pcm_hw_rewindable(snd_pcm_t *pcm) { return snd_pcm_mmap_hw_avail(pcm); }

To allow the application rewind the application pointer to hardware pointer and rewrite the next sample immediately at any time

The audio controller have to fetch the audio data as the same rate as the sampling rate.

The practical value may be the difference between the application pointer and the nearest period boundary

2009/9/24 Takashi Iwai tiwai@suse.de

At Thu, 24 Sep 2009 13:04:24 +0530, Anuj Aggarwal wrote:

...

Hi,

I want to try the forward/rewind APIs supported by the ALSA library. Can someone point to an ALSA application which exercises these APIs?

Pulseaudio is using snd_pcm_rewind(). About snd_pcm_forward(), I don't know of...

Takashi _______________________________________________ Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel