* [PATCH] ASoC: Improve EP93xx I2S clocks management.
@ 2011-03-05 0:57 Alexander Sverdlin
2011-03-06 22:02 ` Ryan Mallon
0 siblings, 1 reply; 3+ messages in thread
From: Alexander Sverdlin @ 2011-03-05 0:57 UTC (permalink / raw)
To: alsa-devel; +Cc: Ryan Mallon
From: Alexander Sverdlin <subaparts@yandex.ru>
Improve EP93xx I2S clocks management.
Some freqs values are set not exact as they requested for MCLK and
original code was not able to find divisors for SCLK and LRCLK.
This code just picks up nearest value from 3 possible variants.
This patch makes 44100 and 192000 rates working and fixes
capture function.
Signed-off-by: Alexander Sverdlin <subaparts@yandex.ru>
---
sound/soc/ep93xx/ep93xx-i2s.c | 19 +++++++++----------
1 files changed, 9 insertions(+), 10 deletions(-)
diff --git a/sound/soc/ep93xx/ep93xx-i2s.c b/sound/soc/ep93xx/ep93xx-i2s.c
index 0458dd4..3d4c086 100644
--- a/sound/soc/ep93xx/ep93xx-i2s.c
+++ b/sound/soc/ep93xx/ep93xx-i2s.c
@@ -242,7 +242,7 @@ static int ep93xx_i2s_hw_params(struct snd_pcm_substream *substream,
{
struct ep93xx_i2s_info *info = snd_soc_dai_get_drvdata(dai);
unsigned word_len, div, sdiv, lrdiv;
- int found = 0, err;
+ int err;
switch (params_format(params)) {
case SNDRV_PCM_FORMAT_S16_LE:
@@ -275,15 +275,14 @@ static int ep93xx_i2s_hw_params(struct snd_pcm_substream *substream,
* the codec uses.
*/
div = clk_get_rate(info->mclk) / params_rate(params);
- for (sdiv = 2; sdiv <= 4; sdiv += 2)
- for (lrdiv = 64; lrdiv <= 128; lrdiv <<= 1)
- if (sdiv * lrdiv == div) {
- found = 1;
- goto out;
- }
-out:
- if (!found)
- return -EINVAL;
+ sdiv = 4;
+ if (div > (256 + 512) / 2) {
+ lrdiv = 128;
+ } else {
+ lrdiv = 64;
+ if (div < (128 + 256) / 2)
+ sdiv = 2;
+ }
err = clk_set_rate(info->sclk, clk_get_rate(info->mclk) / sdiv);
if (err)
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] ASoC: Improve EP93xx I2S clocks management.
2011-03-05 0:57 [PATCH] ASoC: Improve EP93xx I2S clocks management Alexander Sverdlin
@ 2011-03-06 22:02 ` Ryan Mallon
2011-03-07 5:39 ` Alexander Sverdlin
0 siblings, 1 reply; 3+ messages in thread
From: Ryan Mallon @ 2011-03-06 22:02 UTC (permalink / raw)
To: Alexander Sverdlin; +Cc: alsa-devel
On 03/05/2011 01:57 PM, Alexander Sverdlin wrote:
> From: Alexander Sverdlin <subaparts@yandex.ru>
>
> Improve EP93xx I2S clocks management.
> Some freqs values are set not exact as they requested for MCLK and
> original code was not able to find divisors for SCLK and LRCLK.
> This code just picks up nearest value from 3 possible variants.
> This patch makes 44100 and 192000 rates working and fixes
> capture function.
>
> Signed-off-by: Alexander Sverdlin <subaparts@yandex.ru>
> ---
> sound/soc/ep93xx/ep93xx-i2s.c | 19 +++++++++----------
> 1 files changed, 9 insertions(+), 10 deletions(-)
>
> diff --git a/sound/soc/ep93xx/ep93xx-i2s.c b/sound/soc/ep93xx/ep93xx-i2s.c
> index 0458dd4..3d4c086 100644
> --- a/sound/soc/ep93xx/ep93xx-i2s.c
> +++ b/sound/soc/ep93xx/ep93xx-i2s.c
> @@ -242,7 +242,7 @@ static int ep93xx_i2s_hw_params(struct snd_pcm_substream *substream,
> {
> struct ep93xx_i2s_info *info = snd_soc_dai_get_drvdata(dai);
> unsigned word_len, div, sdiv, lrdiv;
> - int found = 0, err;
> + int err;
>
> switch (params_format(params)) {
> case SNDRV_PCM_FORMAT_S16_LE:
> @@ -275,15 +275,14 @@ static int ep93xx_i2s_hw_params(struct snd_pcm_substream *substream,
> * the codec uses.
> */
> div = clk_get_rate(info->mclk) / params_rate(params);
> - for (sdiv = 2; sdiv <= 4; sdiv += 2)
> - for (lrdiv = 64; lrdiv <= 128; lrdiv <<= 1)
> - if (sdiv * lrdiv == div) {
> - found = 1;
> - goto out;
> - }
> -out:
> - if (!found)
> - return -EINVAL;
> + sdiv = 4;
> + if (div > (256 + 512) / 2) {
> + lrdiv = 128;
> + } else {
> + lrdiv = 64;
> + if (div < (128 + 256) / 2)
> + sdiv = 2;
> + }
I don't understand where the magic numbers above come from? Does it work
correctly for all rates? I think a comment is needed to explain how this
works.
An alternative would be to keep the existing loop form, but try and find
the closest match rather than an exact match. Something like this
(untested):
unsigned long diff, lowest_diff = ULONG_MAX;
/* Find the closest match for sdiv/lrdiv */
best_sdiv = 2;
best_lrdiv = 64;
for (sdiv = 2; sdiv <= 4; sdiv += 2) {
for (lrdiv = 64; lrdiv <= 128; lrdiv <<= 1) {
diff = abs(div - (sdiv * lrdiv));
if (diff < lowest_diff) {
lowest_diff = diff;
best_sdiv = sdiv;
best_lrdiv = lrdiv;
}
}
}
sdiv = best_sdiv;
lrdiv = best_lrdiv;
~Ryan
--
Bluewater Systems Ltd - ARM Technology Solution Centre
Ryan Mallon 5 Amuri Park, 404 Barbadoes St
ryan@bluewatersys.com PO Box 13 889, Christchurch 8013
http://www.bluewatersys.com New Zealand
Phone: +64 3 3779127 Freecall: Australia 1800 148 751
Fax: +64 3 3779135 USA 1800 261 2934
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] ASoC: Improve EP93xx I2S clocks management.
2011-03-06 22:02 ` Ryan Mallon
@ 2011-03-07 5:39 ` Alexander Sverdlin
0 siblings, 0 replies; 3+ messages in thread
From: Alexander Sverdlin @ 2011-03-07 5:39 UTC (permalink / raw)
To: Ryan Mallon; +Cc: alsa-devel
Dear Ryan,
On Mon, 2011-03-07 at 11:02 +1300, Ryan Mallon wrote:
> > + sdiv = 4;
> > + if (div > (256 + 512) / 2) {
> > + lrdiv = 128;
> > + } else {
> > + lrdiv = 64;
> > + if (div < (128 + 256) / 2)
> > + sdiv = 2;
> > + }
>
> I don't understand where the magic numbers above come from? Does it work
> correctly for all rates? I think a comment is needed to explain how this
> works.
The Magic is simply centers between divisors. The loop produced only 3
variants of divisors. So do this code. But it's simpler and it prefers
LRDIV = 64. I do not know what do you mean by all rates, with fixed MCLK
there is no way to produce all the rates. With variable MCLK I'm having
full 8000-192000 range now.
>
> An alternative would be to keep the existing loop form, but try and find
> the closest match rather than an exact match. Something like this
> (untested):
This is huge comparing to 2 ifs... But the main task of keeping LRCLK=64
where possible still can be achieved if we swap to lines:
> for (sdiv = 2; sdiv <= 4; sdiv += 2) {
> for (lrdiv = 64; lrdiv <= 128; lrdiv <<= 1) {
But I'm sure code proposed by me should work just like previous variant
for you. Could you try it with your configuration please?
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2011-03-07 5:39 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-03-05 0:57 [PATCH] ASoC: Improve EP93xx I2S clocks management Alexander Sverdlin
2011-03-06 22:02 ` Ryan Mallon
2011-03-07 5:39 ` Alexander Sverdlin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).