Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Thu, 18 Oct 2018 13:02:06 +0200
Message-ID: <s5hefcnv7pt.wl-tiwai@suse.de>
From:   Takashi Iwai <tiwai@suse.de>
To:     Mike Brady <mikebrady@eircom.net>
Cc:     eric@anholt.net, stefan.wahren@i2se.com,
        gregkh@linuxfoundation.org, f.fainelli@gmail.com,
        nishka.dasgupta_ug18@ashoka.edu.in, julia.lawall@lip6.fr,
        k.marinushkin@gmail.com, linux-rpi-kernel@lists.infradead.org,
        linux-arm-kernel@lists.infradead.org, devel@driverdev.osuosl.org,
        linux-kernel@vger.kernel.org
Subject: Re: [PATCH] staging: bcm2835-audio: interpolate audio delay
In-Reply-To: <20181018105715.GA6660@ubuntu>
References: <20181018105715.GA6660@ubuntu>
User-Agent: Wanderlust/2.15.9 (Almost Unreal) SEMI/1.14.6 (Maruoka)
 FLIM/1.14.9 (=?UTF-8?B?R29qxY0=?=) APEL/10.8 Emacs/26
 (x86_64-suse-linux-gnu) MULE/6.0 (HANACHIRUSATO)
MIME-Version: 1.0 (generated by SEMI 1.14.6 - "Maruoka")
Content-Type: text/plain; charset=US-ASCII
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Thu, 18 Oct 2018 12:57:15 +0200,
Mike Brady wrote:
> 
> When the BCM2835 audio output is used, userspace sees a jitter up to 10ms
> in the audio position, aka "delay" -- the number of frames that must
> be output before a new frame would be played.
> Make this a bit nicer for userspace by interpolating the position
> using the CPU clock.
> The overhead is small -- an extra ktime_get() every time a GPU message
> is sent -- and another call and a few calculations whenever the delay
> is sought from userland.
> At 48,000 frames per second, i.e. approximately 20 microseconds per
> frame, it would take a clock inaccuracy of
> 20 microseconds in 10 milliseconds -- 2,000 parts per million --
> to result in an inaccurate estimate, whereas
> crystal- or resonator-based clocks typically have an
> inaccuracy of 10s to 100s of parts per million.
> 
> Signed-off-by: Mike Brady <mikebrady@eircom.net>

This is rather about ALSA stuff, so please put alsa-devel to Cc.

And the addition of BATCH flag has nothing to do with this change.
If it's really needed, split the changes and give rationale.


thanks,

Takashi


> ---
>  .../vc04_services/bcm2835-audio/bcm2835-pcm.c | 23 ++++++++++++++++++-
>  .../vc04_services/bcm2835-audio/bcm2835.h     |  1 +
>  2 files changed, 23 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/staging/vc04_services/bcm2835-audio/bcm2835-pcm.c b/drivers/staging/vc04_services/bcm2835-audio/bcm2835-pcm.c
> index e66da11af5cf..250adf82385a 100644
> --- a/drivers/staging/vc04_services/bcm2835-audio/bcm2835-pcm.c
> +++ b/drivers/staging/vc04_services/bcm2835-audio/bcm2835-pcm.c
> @@ -12,7 +12,8 @@
>  static const struct snd_pcm_hardware snd_bcm2835_playback_hw = {
>  	.info = (SNDRV_PCM_INFO_INTERLEAVED | SNDRV_PCM_INFO_BLOCK_TRANSFER |
>  		 SNDRV_PCM_INFO_MMAP | SNDRV_PCM_INFO_MMAP_VALID |
> -		 SNDRV_PCM_INFO_DRAIN_TRIGGER | SNDRV_PCM_INFO_SYNC_APPLPTR),
> +		 SNDRV_PCM_INFO_BATCH | SNDRV_PCM_INFO_DRAIN_TRIGGER |
> +		 SNDRV_PCM_INFO_SYNC_APPLPTR),
>  	.formats = SNDRV_PCM_FMTBIT_U8 | SNDRV_PCM_FMTBIT_S16_LE,
>  	.rates = SNDRV_PCM_RATE_CONTINUOUS | SNDRV_PCM_RATE_8000_48000,
>  	.rate_min = 8000,
> @@ -74,6 +75,7 @@ void bcm2835_playback_fifo(struct bcm2835_alsa_stream *alsa_stream,
>  	atomic_set(&alsa_stream->pos, pos);
>  
>  	alsa_stream->period_offset += bytes;
> +	alsa_stream->interpolate_start = ktime_get();
>  	if (alsa_stream->period_offset >= alsa_stream->period_size) {
>  		alsa_stream->period_offset %= alsa_stream->period_size;
>  		snd_pcm_period_elapsed(substream);
> @@ -243,6 +245,7 @@ static int snd_bcm2835_pcm_prepare(struct snd_pcm_substream *substream)
>  	atomic_set(&alsa_stream->pos, 0);
>  	alsa_stream->period_offset = 0;
>  	alsa_stream->draining = false;
> +	alsa_stream->interpolate_start = ktime_get();
>  
>  	return 0;
>  }
> @@ -292,6 +295,24 @@ snd_bcm2835_pcm_pointer(struct snd_pcm_substream *substream)
>  {
>  	struct snd_pcm_runtime *runtime = substream->runtime;
>  	struct bcm2835_alsa_stream *alsa_stream = runtime->private_data;
> +	ktime_t now = ktime_get();
> +
> +	/* Give userspace better delay reporting by interpolating between GPU
> +	 * notifications, assuming audio speed is close enough to the clock
> +	 * used for ktime
> +	 */
> +
> +	if ((ktime_to_ns(alsa_stream->interpolate_start)) &&
> +	    (ktime_compare(alsa_stream->interpolate_start, now) < 0)) {
> +		u64 interval =
> +			(ktime_to_ns(ktime_sub(now,
> +				alsa_stream->interpolate_start)));
> +		u64 frames_output_in_interval =
> +			div_u64((interval * runtime->rate), 1000000000);
> +		snd_pcm_sframes_t frames_output_in_interval_sized =
> +			-frames_output_in_interval;
> +		runtime->delay = frames_output_in_interval_sized;
> +	}
>  
>  	return snd_pcm_indirect_playback_pointer(substream,
>  		&alsa_stream->pcm_indirect,
> diff --git a/drivers/staging/vc04_services/bcm2835-audio/bcm2835.h b/drivers/staging/vc04_services/bcm2835-audio/bcm2835.h
> index e13435d1c205..595ad584243f 100644
> --- a/drivers/staging/vc04_services/bcm2835-audio/bcm2835.h
> +++ b/drivers/staging/vc04_services/bcm2835-audio/bcm2835.h
> @@ -78,6 +78,7 @@ struct bcm2835_alsa_stream {
>  	unsigned int period_offset;
>  	unsigned int buffer_size;
>  	unsigned int period_size;
> +	ktime_t interpolate_start;
>  
>  	struct bcm2835_audio_instance *instance;
>  	int idx;
> -- 
> 2.17.1
>