All of lore.kernel.org
 help / color / mirror / Atom feed
From: Takashi Iwai <tiwai@suse.de>
To: twischer@de.adit-jv.com
Cc: alsa-devel@alsa-project.org
Subject: Re: [PATCH - JACK plugin 1/4] jack: Do not Xrun the ALSA buffer
Date: Thu, 01 Mar 2018 16:24:48 +0100	[thread overview]
Message-ID: <s5hfu5jizsv.wl-tiwai@suse.de> (raw)
In-Reply-To: <1519910048-19732-2-git-send-email-twischer@de.adit-jv.com>

On Thu, 01 Mar 2018 14:14:05 +0100,
<twischer@de.adit-jv.com> wrote:
> 
> From: Timo Wischer <twischer@de.adit-jv.com>
> 
> when the JACK thread is requesting too many audio frames
> 
> Playback:
> Without this commit the ALSA audio buffer
> will be played with endless repeats as long as the user application
> has not provided new audio data.
> Therefore this garbage will be played as long as the user application
> has not called snd_pcm_stop() after an Xrun.
> With this fix the rest of the JACK buffer will be filled
> with silence.
> 
> Capture:
> Without this commit the audio data in the ALSA buffer would be
> overwritten.
> With this commit the new data from the JACK buffer
> will not be copied.
> Therefore the existing data in the ALSA buffer
> will not be overwritten.

Please try to format the text a bit nicer.  These line breaks appear
like you're writing a poem :)

Now about the code changes:

> +static inline snd_pcm_uframes_t snd_pcm_jack_playback_avail(const snd_pcm_ioplug_t* const io,
> +                                                            const snd_pcm_uframes_t hw_ptr,
> +                                                            const snd_pcm_uframes_t appl_ptr)

You don't need to put inline unless it's really mandatory.


> +{
> +	const snd_pcm_jack_t* const jack = io->private_data;
> +
> +	/* cannot use io->hw_ptr without calling snd_pcm_avail_update()
> +	 * because it is not guranteed that snd_pcm_jack_pointer() was already
> +	 * called
> +	 */
> +	snd_pcm_sframes_t avail;
> +	avail = hw_ptr + io->buffer_size - appl_ptr;
> +	if (avail < 0)
> +		avail += jack->boundary;
> +	else if ((snd_pcm_uframes_t) avail >= jack->boundary)
> +		avail -= jack->boundary;
> +	return avail;
> +}
> +
> +static inline snd_pcm_uframes_t snd_pcm_jack_capture_avail(const snd_pcm_ioplug_t* const io,
> +                                                           const snd_pcm_uframes_t hw_ptr,
> +                                                           const snd_pcm_uframes_t appl_ptr)
> +{
> +	const snd_pcm_jack_t* const jack = io->private_data;
> +
> +	/* cannot use io->hw_ptr without calling snd_pcm_avail_update()
> +	 * because it is not guranteed that snd_pcm_jack_pointer() was already
> +	 * called
> +	 */
> +	snd_pcm_sframes_t avail;
> +	avail = hw_ptr - appl_ptr;
> +	if (avail < 0)
> +		avail += jack->boundary;
> +	return avail;
> +}
> +
> +static inline snd_pcm_uframes_t snd_pcm_jack_avail(const snd_pcm_ioplug_t* const io,
> +                                                   const snd_pcm_uframes_t hw_ptr,
> +                                                   const snd_pcm_uframes_t appl_ptr)
> +{
> +	return (io->stream == SND_PCM_STREAM_PLAYBACK) ?
> +	                        snd_pcm_jack_playback_avail(io, hw_ptr, appl_ptr) :
> +	                        snd_pcm_jack_capture_avail(io, hw_ptr, appl_ptr);
> +}
> +
> +static inline snd_pcm_uframes_t snd_pcm_jack_hw_avail(const snd_pcm_ioplug_t* const io,
> +                                                      const snd_pcm_uframes_t hw_ptr,
> +                                                      const snd_pcm_uframes_t appl_ptr)
> +{
> +	/* available data/space which can be transfered by the user application */
> +	const snd_pcm_uframes_t user_avail = snd_pcm_jack_avail(io, hw_ptr,
> +	                                                        appl_ptr);
> +
> +	if (user_avail > io->buffer_size) {
> +		/* there was an Xrun */
> +		return 0;
> +	}
> +	/* available data/space which can be transfered by the DMA */
> +	return io->buffer_size - user_avail;
> +}

Hm, this whole stuff would fit better in ioplug code itself instead of
open-coding in each plugin.  Maybe providing a new helper function?


>  static int pcm_poll_block_check(snd_pcm_ioplug_t *io)
>  {
>  	static char buf[32];
> @@ -144,7 +206,6 @@ snd_pcm_jack_process_cb(jack_nframes_t nframes, snd_pcm_ioplug_t *io)
>  {
>  	snd_pcm_jack_t *jack = io->private_data;
>  	snd_pcm_uframes_t hw_ptr;
> -	const snd_pcm_channel_area_t *areas;
>  	snd_pcm_uframes_t xfer = 0;
>  	unsigned int channel;
>  	
> @@ -154,40 +215,57 @@ snd_pcm_jack_process_cb(jack_nframes_t nframes, snd_pcm_ioplug_t *io)
>  		jack->areas[channel].first = 0;
>  		jack->areas[channel].step = jack->sample_bits;
>  	}
> -		
> -	if (io->state != SND_PCM_STATE_RUNNING) {
> -		if (io->stream == SND_PCM_STREAM_PLAYBACK) {
> -			for (channel = 0; channel < io->channels; channel++)
> -				snd_pcm_area_silence(&jack->areas[channel], 0, nframes, io->format);
> -			return 0;
> -		}
> -	}
>  
>  	hw_ptr = jack->hw_ptr;
> -	areas = snd_pcm_ioplug_mmap_areas(io);
> -
> -	while (xfer < nframes) {
> -		snd_pcm_uframes_t frames = nframes - xfer;
> -		snd_pcm_uframes_t offset = hw_ptr % io->buffer_size;
> -		snd_pcm_uframes_t cont = io->buffer_size - offset;
> -
> -		if (cont < frames)
> -			frames = cont;
> +	if (io->state == SND_PCM_STATE_RUNNING) {
> +		const snd_pcm_channel_area_t *areas = snd_pcm_ioplug_mmap_areas(io);
> +
> +		while (xfer < nframes) {
> +			snd_pcm_uframes_t frames = nframes - xfer;
> +			snd_pcm_uframes_t offset = hw_ptr % io->buffer_size;
> +			snd_pcm_uframes_t cont = io->buffer_size - offset;
> +			snd_pcm_uframes_t hw_avail =
> +			                snd_pcm_jack_hw_avail(io, hw_ptr,
> +			                                      io->appl_ptr);
> +
> +			/* stop copying if there is no more data available */
> +			if (hw_avail <= 0)
> +				break;
> +
> +			/* split the snd_pcm_area_copy() function into two parts
> +			 * if the data to copy passes the buffer wrap around
> +			 */
> +			if (cont < frames)
> +				frames = cont;
> +
> +			if (hw_avail < frames)
> +				frames = hw_avail;
> +
> +			for (channel = 0; channel < io->channels; channel++) {
> +				if (io->stream == SND_PCM_STREAM_PLAYBACK)
> +					snd_pcm_area_copy(&jack->areas[channel], xfer, &areas[channel], offset, frames, io->format);
> +				else
> +					snd_pcm_area_copy(&areas[channel], offset, &jack->areas[channel], xfer, frames, io->format);
> +			}
>  
> -		for (channel = 0; channel < io->channels; channel++) {
> -			if (io->stream == SND_PCM_STREAM_PLAYBACK)
> -				snd_pcm_area_copy(&jack->areas[channel], xfer, &areas[channel], offset, frames, io->format);
> -			else
> -				snd_pcm_area_copy(&areas[channel], offset, &jack->areas[channel], xfer, frames, io->format);
> +			hw_ptr += frames;
> +			if (hw_ptr >= jack->boundary)
> +				hw_ptr -= jack->boundary;
> +			xfer += frames;
>  		}
> -		
> -		hw_ptr += frames;
> -		if (hw_ptr >= jack->boundary)
> -			hw_ptr -= jack->boundary;
> -		xfer += frames;
>  	}
>  	jack->hw_ptr = hw_ptr;
>  
> +	/* check if requested frames were copied */
> +	if (xfer < nframes) {
> +		/* always fill the not yet written JACK buffer with silence */
> +		if (io->stream == SND_PCM_STREAM_PLAYBACK) {
> +			const snd_pcm_uframes_t samples = nframes - xfer;
> +			for (channel = 0; channel < io->channels; channel++)
> +				snd_pcm_area_silence(&jack->areas[channel], xfer, samples, io->format);
> +		}
> +	}
> +

Ditto.  Basically the whole procedure here is some common pattern, so
it wouldn't be too bad to have it in ioplug side.


thanks,

Takashi

  reply	other threads:[~2018-03-01 15:24 UTC|newest]

Thread overview: 49+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-01-24 11:53 [PATCH - JACK PCM plugin] jack: Do not Xrun the ALSA buffer twischer
2018-02-26  8:14 ` Wischer, Timo (ADITG/ESB)
2018-02-27  8:48   ` Takashi Iwai
2018-03-01 13:14     ` [PATCH - JACK PCM plugin] Xrun handling twischer
2018-03-01 13:14       ` [PATCH - JACK plugin 1/4] jack: Do not Xrun the ALSA buffer twischer
2018-03-01 15:24         ` Takashi Iwai [this message]
2018-03-02 16:21           ` twischer
2018-03-02 16:21             ` [PATCH - PCM 2/3] pcm: ioplug: Provide hw_avail helper function for plugins twischer
2018-03-02 16:21             ` [PATCH - PCM 3/3] pcm: Provide areas_copy function which handles buffer wrap around twischer
2018-03-02 16:21             ` [PATCH - JACK 1/1] jack: Do not Xrun the ALSA buffer twischer
2018-03-13  8:34               ` [PATCH - JACK plugin] " twischer
2018-03-13  8:34                 ` [PATCH - PCM 1/2] pcm: ioplug: Provide hw_avail helper function for plugins twischer
2018-03-13  8:34                 ` [PATCH - PCM 2/2] pcm: Provide areas_copy function which handles buffer wrap around twischer
2018-03-13  8:34                 ` [PATCH - JACK 1/1] jack: Do not Xrun the ALSA buffer twischer
2018-03-13 21:20                 ` [PATCH - JACK plugin] " Takashi Iwai
2018-03-15 15:09                   ` Wischer, Timo (ADITG/ESB)
2018-03-15 15:13                     ` Takashi Iwai
2018-03-01 13:14       ` [PATCH - JACK plugin 2/4] jack: Use internal snd_pcm_state to reduce amount of additional variables twischer
2018-03-01 15:26         ` Takashi Iwai
2018-03-15 12:56           ` [PATCH - JACK plugin] " twischer
2018-03-15 12:56             ` [PATCH - JACK 1/1] " twischer
2018-03-15 13:14               ` Takashi Iwai
2018-03-15 14:05                 ` Wischer, Timo (ADITG/ESB)
2018-03-15 14:20                   ` Takashi Iwai
2018-03-15 15:07                     ` Wischer, Timo (ADITG/ESB)
2018-03-15 15:49                       ` Takashi Iwai
2018-03-16 10:02                         ` [PATCH - IOPLUG] Update prepare and draining state correctly twischer
2018-03-16 10:02                           ` [PATCH - IOPLUG 1/1] pcm: ioplug: update Prepare " twischer
2018-03-16 10:08                             ` Takashi Iwai
2018-03-16 10:20                               ` [PATCH - IOPLUG 1/1] pcm: ioplug: update prepare " twischer
2018-03-16 10:30                                 ` Takashi Iwai
2018-03-01 13:14       ` [PATCH - JACK plugin 3/4] jack: Report Xruns to user application twischer
2018-03-16 14:23         ` twischer
2018-03-16 14:23           ` [PATCH - JACK 1/1] " twischer
2018-03-16 14:41             ` Takashi Iwai
2018-03-16 14:47               ` Wischer, Timo (ADITG/ESB)
2018-03-16 14:57                 ` Takashi Iwai
2018-03-01 13:14       ` [PATCH - JACK plugin 4/4] jack: Support snd_pcm_drain() twischer
2018-03-01 13:41         ` Jaroslav Kysela
2018-03-16 14:44           ` Wischer, Timo (ADITG/ESB)
2018-03-16 15:11             ` Jaroslav Kysela
2018-03-16 15:52               ` Wischer, Timo (ADITG/ESB)
2018-03-21 16:22                 ` Wischer, Timo (ADITG/ESB)
2018-03-21 16:34                   ` Takashi Iwai
2018-03-21 16:47                     ` Wischer, Timo (ADITG/ESB)
2018-03-21 16:52                       ` Takashi Iwai
2018-03-21 17:02                         ` Wischer, Timo (ADITG/ESB)
2018-03-21 17:07                           ` Takashi Iwai
2018-03-01 15:19       ` [PATCH - JACK PCM plugin] Xrun handling Takashi Iwai

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=s5hfu5jizsv.wl-tiwai@suse.de \
    --to=tiwai@suse.de \
    --cc=alsa-devel@alsa-project.org \
    --cc=twischer@de.adit-jv.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.