* [PATCH RFC (do not merge)] ASoC: davinci-mcasp: Set rule constraint if implicit bclk divider is used
@ 2015-03-13 11:36 Jyri Sarha
2015-03-13 11:56 ` Lars-Peter Clausen
0 siblings, 1 reply; 3+ messages in thread
From: Jyri Sarha @ 2015-03-13 11:36 UTC (permalink / raw)
To: alsa-devel
Cc: linux-omap, broonie, liam.r.girdwood, peter.ujfalusi, lars, Jyri Sarha
Set a rule constraint to allow only sample-rate and sample-format
combinations that can be played/captured with reasonable accuracy.
Signed-off-by: Jyri Sarha <jsarha@ti.com>
---
In theory this patch does exactly what it is supposed to. It only
allows a sample-rate and sample-format combination if the rate can be
produced with reasonable accuracy. Unfortunately the alsa-lib and
alsa-tools are not able use this information too well. If the requested
sample-rate and sample-format is not available the aplay/arecord
fails, even if plughw is selected, with:
pcm_params.c:170: snd1_pcm_hw_param_get_min: Assertion `!snd_interval_empty(i)' failed.
I think we are better of without this patch until the behavior of ALSA
userspace improved.
sound/soc/davinci/davinci-mcasp.c | 94 ++++++++++++++++++++++++++++++++++-----
1 file changed, 84 insertions(+), 10 deletions(-)
diff --git a/sound/soc/davinci/davinci-mcasp.c b/sound/soc/davinci/davinci-mcasp.c
index de3b155..4ff5820 100644
--- a/sound/soc/davinci/davinci-mcasp.c
+++ b/sound/soc/davinci/davinci-mcasp.c
@@ -855,6 +855,29 @@ static int mcasp_dit_hw_param(struct davinci_mcasp *mcasp,
return 0;
}
+static int davinci_mcasp_calc_clk_div(struct davinci_mcasp *mcasp,
+ unsigned int bclk_freq,
+ int *error_ppm)
+{
+ int div = mcasp->sysclk_freq / bclk_freq;
+ int rem = mcasp->sysclk_freq % bclk_freq;
+
+ if (rem != 0) {
+ if (div == 0 ||
+ ((mcasp->sysclk_freq / div) - bclk_freq) >
+ (bclk_freq - (mcasp->sysclk_freq / (div+1)))) {
+ div++;
+ rem = rem - bclk_freq;
+ }
+ }
+ if (error_ppm)
+ *error_ppm =
+ (div*1000000 + 1000*((rem*1000)/(int)bclk_freq))/div
+ - 1000000;
+
+ return div;
+}
+
static int davinci_mcasp_hw_params(struct snd_pcm_substream *substream,
struct snd_pcm_hw_params *params,
struct snd_soc_dai *cpu_dai)
@@ -872,16 +895,14 @@ static int davinci_mcasp_hw_params(struct snd_pcm_substream *substream,
* the machine driver, we need to calculate the ratio.
*/
if (mcasp->bclk_master && mcasp->bclk_div == 0 && mcasp->sysclk_freq) {
- unsigned int bclk_freq = snd_soc_params_to_bclk(params);
- unsigned int div = mcasp->sysclk_freq / bclk_freq;
- if (mcasp->sysclk_freq % bclk_freq != 0) {
- if (((mcasp->sysclk_freq / div) - bclk_freq) >
- (bclk_freq - (mcasp->sysclk_freq / (div+1))))
- div++;
- dev_warn(mcasp->dev,
- "Inaccurate BCLK: %u Hz / %u != %u Hz\n",
- mcasp->sysclk_freq, div, bclk_freq);
- }
+ int ppm;
+ uint bclk_freq = snd_soc_params_to_bclk(params);
+ int div = davinci_mcasp_calc_clk_div(mcasp, bclk_freq, &ppm);
+
+ if (ppm)
+ dev_info(mcasp->dev, "BCLK rate is off by %d PPM\n",
+ ppm);
+
__davinci_mcasp_set_clkdiv(cpu_dai, 1, div, 0);
}
@@ -973,6 +994,46 @@ static int davinci_mcasp_trigger(struct snd_pcm_substream *substream,
return ret;
}
+static const unsigned int davinci_mcasp_dai_rates[] = {
+ 8000, 11025, 16000, 22050, 32000, 44100, 48000, 64000,
+ 88200, 96000, 176400, 192000,
+};
+
+#define DAVINCI_MAX_RATE_ERROR_PPM 1000
+
+static int davinci_mcasp_hw_rule_rate(struct snd_pcm_hw_params *params,
+ struct snd_pcm_hw_rule *rule)
+{
+ struct davinci_mcasp *mcasp = rule->private;
+ unsigned int list[ARRAY_SIZE(davinci_mcasp_dai_rates)];
+ int frame_size;
+ int i, count = 0;
+
+ frame_size = snd_soc_params_to_frame_size(params);
+ if (frame_size < 0)
+ return frame_size;
+
+ for (i = 0; i < ARRAY_SIZE(davinci_mcasp_dai_rates); i++) {
+ uint bclk_freq = frame_size*davinci_mcasp_dai_rates[i];
+ int ppm;
+
+ davinci_mcasp_calc_clk_div(mcasp, bclk_freq, &ppm);
+
+ dev_dbg(mcasp->dev,
+ "%u bit frames @ %u Hz => %d PPM error\n",
+ frame_size, davinci_mcasp_dai_rates[i], ppm);
+
+ if (abs(ppm) < DAVINCI_MAX_RATE_ERROR_PPM)
+ list[count++] = davinci_mcasp_dai_rates[i];
+
+ }
+ dev_dbg(mcasp->dev, "%d frequencies for %u fsize\n",
+ count, frame_size);
+
+ return snd_interval_list(hw_param_interval(params, rule->var),
+ count, list, 0);
+}
+
static int davinci_mcasp_startup(struct snd_pcm_substream *substream,
struct snd_soc_dai *cpu_dai)
{
@@ -1012,6 +1073,19 @@ static int davinci_mcasp_startup(struct snd_pcm_substream *substream,
snd_pcm_hw_constraint_minmax(substream->runtime,
SNDRV_PCM_HW_PARAM_CHANNELS,
2, max_channels);
+
+ /*
+ * If we rely on implicit BCLK divider setting we should
+ * set constraints based on what we can provide.
+ */
+ if (mcasp->bclk_master && mcasp->bclk_div == 0 && mcasp->sysclk_freq)
+ return snd_pcm_hw_rule_add(substream->runtime, 0,
+ SNDRV_PCM_HW_PARAM_RATE,
+ davinci_mcasp_hw_rule_rate,
+ mcasp,
+ SNDRV_PCM_HW_PARAM_FRAME_BITS,
+ SNDRV_PCM_HW_PARAM_CHANNELS, -1);
+
return 0;
}
--
1.9.1
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH RFC (do not merge)] ASoC: davinci-mcasp: Set rule constraint if implicit bclk divider is used
2015-03-13 11:36 [PATCH RFC (do not merge)] ASoC: davinci-mcasp: Set rule constraint if implicit bclk divider is used Jyri Sarha
@ 2015-03-13 11:56 ` Lars-Peter Clausen
2015-03-13 14:21 ` Jyri Sarha
0 siblings, 1 reply; 3+ messages in thread
From: Lars-Peter Clausen @ 2015-03-13 11:56 UTC (permalink / raw)
To: Jyri Sarha, alsa-devel
Cc: liam.r.girdwood, peter.ujfalusi, broonie, linux-omap
On 03/13/2015 12:36 PM, Jyri Sarha wrote:
[...]
> In theory this patch does exactly what it is supposed to. It only
> allows a sample-rate and sample-format combination if the rate can be
> produced with reasonable accuracy. Unfortunately the alsa-lib and
> alsa-tools are not able use this information too well. If the requested
> sample-rate and sample-format is not available the aplay/arecord
> fails, even if plughw is selected, with:
>
> pcm_params.c:170: snd1_pcm_hw_param_get_min: Assertion `!snd_interval_empty(i)' failed.
[...]
> +
> + /*
> + * If we rely on implicit BCLK divider setting we should
> + * set constraints based on what we can provide.
> + */
> + if (mcasp->bclk_master && mcasp->bclk_div == 0 && mcasp->sysclk_freq)
> + return snd_pcm_hw_rule_add(substream->runtime, 0,
> + SNDRV_PCM_HW_PARAM_RATE,
> + davinci_mcasp_hw_rule_rate,
> + mcasp,
> + SNDRV_PCM_HW_PARAM_FRAME_BITS,
> + SNDRV_PCM_HW_PARAM_CHANNELS, -1);
> +
For things to work correctly you also need reverse rules restricting
CHANNELS and FRAME_BITS based on the RATE. This might fix the issue you are
seeing with the ALSA tools.
- Lars
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH RFC (do not merge)] ASoC: davinci-mcasp: Set rule constraint if implicit bclk divider is used
2015-03-13 11:56 ` Lars-Peter Clausen
@ 2015-03-13 14:21 ` Jyri Sarha
0 siblings, 0 replies; 3+ messages in thread
From: Jyri Sarha @ 2015-03-13 14:21 UTC (permalink / raw)
To: Lars-Peter Clausen, alsa-devel
Cc: linux-omap, broonie, liam.r.girdwood, peter.ujfalusi
On 03/13/15 13:56, Lars-Peter Clausen wrote:
> On 03/13/2015 12:36 PM, Jyri Sarha wrote:
> [...]
>> In theory this patch does exactly what it is supposed to. It only
>> allows a sample-rate and sample-format combination if the rate can be
>> produced with reasonable accuracy. Unfortunately the alsa-lib and
>> alsa-tools are not able use this information too well. If the requested
>> sample-rate and sample-format is not available the aplay/arecord
>> fails, even if plughw is selected, with:
>>
>> pcm_params.c:170: snd1_pcm_hw_param_get_min: Assertion
>> `!snd_interval_empty(i)' failed.
> [...]
>> +
>> + /*
>> + * If we rely on implicit BCLK divider setting we should
>> + * set constraints based on what we can provide.
>> + */
>> + if (mcasp->bclk_master && mcasp->bclk_div == 0 &&
>> mcasp->sysclk_freq)
>> + return snd_pcm_hw_rule_add(substream->runtime, 0,
>> + SNDRV_PCM_HW_PARAM_RATE,
>> + davinci_mcasp_hw_rule_rate,
>> + mcasp,
>> + SNDRV_PCM_HW_PARAM_FRAME_BITS,
>> + SNDRV_PCM_HW_PARAM_CHANNELS, -1);
>> +
>
> For things to work correctly you also need reverse rules restricting
> CHANNELS and FRAME_BITS based on the RATE. This might fix the issue you
> are seeing with the ALSA tools.
>
Thanks, that was indeed the case. I mail a new patch once I get my patch
in order.
Best regards,
Jyri
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2015-03-13 14:21 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2015-03-13 11:36 [PATCH RFC (do not merge)] ASoC: davinci-mcasp: Set rule constraint if implicit bclk divider is used Jyri Sarha
2015-03-13 11:56 ` Lars-Peter Clausen
2015-03-13 14:21 ` Jyri Sarha
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.