On 04/08/2018 03:14 AM, Nicolin Chen wrote:
On Sat, Apr 07, 2018 at 03:02:21PM +0200, Marek Vasut wrote:
When setting xFP directly, set the xPM predivider to 1, otherwise it could remain set to previously set incorrect value and interfere with the correct clocking.
This doesn't sound right to me.
OK
Could you please provide a failed instance? It's been a while since I wrote the code. But I can tell that PM is supposed to be called by set_sysclk() only, while FP is used for bclk. If you clear PM when setting FP, the output of HCK could be messed.
Try feeding it the following values, The codec I use is PCM1808.
fsl_esai_set_dai_sysclk[227] clk_id=0 freq=24576000 dir=1 fsl_esai_divisor_cal[131] tx=0 ratio=2 usefp=0 fp=0 fsl_esai_set_bclk[322] tx=0 freq=3072000 fsl_esai_divisor_cal[131] tx=0 ratio=8 usefp=1 fp=8
Also, I think there is another bug in the fsl_esai_divisor_cal() now that I look at it. If usefp = 0, then maxfp = 1 , then savesub = 0 and then in the loop sub is never < savesub , although the loop will terminate on savesub == 0 immediately with pm = 999 and the driver will fail.
Thanks Nicolin
Signed-off-by: Marek Vasut marex@denx.de Cc: Fabio Estevam fabio.estevam@nxp.com Cc: Gustavo A. R. Silva garsilva@embeddedor.com Cc: Mark Brown broonie@kernel.org
sound/soc/fsl/fsl_esai.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/sound/soc/fsl/fsl_esai.c b/sound/soc/fsl/fsl_esai.c index 40a700493f4c..9f69823b50d7 100644 --- a/sound/soc/fsl/fsl_esai.c +++ b/sound/soc/fsl/fsl_esai.c @@ -128,8 +128,11 @@ static int fsl_esai_divisor_cal(struct snd_soc_dai *dai, bool tx, u32 ratio,
maxfp = usefp ? 16 : 1;
- if (usefp && fp)
if (usefp && fp) {
regmap_update_bits(esai_priv->regmap, REG_ESAI_xCCR(tx),ESAI_xCCR_xPM_MASK, 0);goto out_fp;
}
if (ratio > 2 * 8 * 256 * maxfp || ratio < 2) { dev_err(dai->dev, "the ratio is out of range (2 ~ %d)\n",
-- 2.16.2
Alsa-devel mailing list Alsa-devel@alsa-project.org http://mailman.alsa-project.org/mailman/listinfo/alsa-devel