Re: [PATCH] es1968: fix jitter on some maestro cards

[Takashi Iwai <tiwai@suse.de>]

At Tue, 01 Apr 2008 00:24:43 +0200,
Rene Herman wrote:
> 
> On 30-03-08 00:49, Andreas Mueller wrote:
> 
> > This patch suppresses jitter on several Maestro cards in stereo mode
> > (ALSA of course).
> > 
> > The patch is also incorporated in the *BSD drivers where I "ported" it from.
> > I am barely used to driver-development so please take a look (esp. the channel
> > parameters), nevertheless it completely works for me.
> > 
> > Without this patch most of the stereo audio gets out of sync and really
> > distorted (oss-emulation with mplayer at 48000khz worked somehow).
> > 
> > CCed to those listed in the .c file.
> 
> Added the alsa-devel list (and re-attached the patch). My TerraTec DMX (ESS 
> Canyon3D, ES1970MS-3D) doesn't need this but doesn't seem to suffer from it 
> either. Otherwise no idea about this.

This should depend on the sample rate you use.  48k would work without
the patch.

Anyway, I applied it to ALSA tree now.  Thanks.


Takashi

> 
> Rene.
> [2 es1968_jitter.patch <text/plain (7bit)>]
> --- sound/pci/es1968.c.old	2008-03-29 19:17:16.771116317 +0100
> +++ sound/pci/es1968.c	2008-03-30 00:14:35.363057160 +0100
> @@ -1816,7 +1816,24 @@
>  
>  	return 0;
>  }
> +/*
> + * suppress jitter on some maestros when playing stereo
> + */
> +static void snd_es1968_suppress_jitter(struct es1968 *chip, struct esschan *es)
> +{
> +	unsigned int cp1;
> +	unsigned int cp2;
> +	unsigned int diff;
>  
> +	cp1 = __apu_get_register(chip, 0, 5);
> +	cp2 = __apu_get_register(chip, 1, 5);
> +	diff = (cp1 > cp2 ? cp1 - cp2 : cp2 - cp1);
> +	
> +	if (diff > 1) {
> +		__maestro_write(chip, IDR0_DATA_PORT, cp1);
> +	}
> +}
> +	
>  /*
>   * update pointer
>   */
> @@ -1937,8 +1954,11 @@
>  		struct esschan *es;
>  		spin_lock(&chip->substream_lock);
>  		list_for_each_entry(es, &chip->substream_list, list) {
> -			if (es->running)
> +			if (es->running) {
>  				snd_es1968_update_pcm(chip, es);
> +				if (es->fmt & ESS_FMT_STEREO)
> +					snd_es1968_suppress_jitter(chip, es);
> +			}
>  		}
>  		spin_unlock(&chip->substream_lock);
>  		if (chip->in_measurement) {