Skip to contentSkip to conversation

sound/soc/codecs/wm8978: add missing BCLK divider setup

PATCH Superseded Brian Sune 3 participants 22 messages Oct 3, 2025 – Oct 7, 2025

Subsystems (3)

v1 of 6 Superseded Latest is v6 →

lore
Share & tools
lore

Brian Sune

author
#0 lore raw

PATCH 1/1 sound/soc/codecs/wm8978: add missing BCLK divider setup

The original WM8978 codec driver did not set the BCLK (bit clock) divider, which can cause audio clocks to be incorrect or unstable depending on the sample rate and word length.

This patch adds proper calculation and configuration of the BCLK divider based on the sample rate and word width, ensuring the codec generates the correct bit clock for all supported rates.

---
 sound/soc/codecs/wm8978.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
sound/soc/codecs/wm8978.c +20 -0
index 8c45ba6fc4c3..2109c84f33df 100644
@@ -717,6 +717,11 @@ static int wm8978_hw_params(struct snd_pcm_substream *substream,
 			    struct snd_pcm_hw_params *params,
 			    struct snd_soc_dai *dai)
 {
+	unsigned int bclk, bclkdiv = 0, min_diff = UINT_MAX;
+	unsigned int target_bclk = params_rate(params) * params_width(params) * 2;
+	/* WM8978 supports divisors */
+	static const int bclk_divs[] = {1, 2, 4, 8, 16, 32};
+
 	struct snd_soc_component *component = dai->component;
 	struct wm8978_priv *wm8978 = snd_soc_component_get_drvdata(component);
 	/* Word length mask = 0x60 */
@@ -820,6 +825,21 @@ static int wm8978_hw_params(struct snd_pcm_substream *substream,
 	/* MCLK divisor mask = 0xe0 */
 	snd_soc_component_update_bits(component, WM8978_CLOCKING, 0xe0, best << 5);
 
+	for (i = 0; i < ARRAY_SIZE(bclk_divs); i++) {
+		bclk = wm8978->f_256fs / bclk_divs[i];
+		if (abs(bclk - target_bclk) < min_diff) {
+			min_diff = abs(bclk - target_bclk);
+			bclkdiv = i;
+		}
+	}
+
+	dev_dbg(component->dev, "%s: fs=%u width=%u -> target BCLK=%u, using div #%u\n",
+		__func__, params_rate(params), params_width(params), target_bclk,
+		bclk_divs[bclkdiv]);
+
+	/* BCLKDIV divisor mask = 0x1c */
+	snd_soc_component_update_bits(component, WM8978_CLOCKING, 0x1c, bclkdiv << 2);
+
 	snd_soc_component_write(component, WM8978_AUDIO_INTERFACE, iface_ctl);
 	snd_soc_component_write(component, WM8978_ADDITIONAL_CONTROL, add_ctl);
 
Signature · 1 line
2.47.1.windows.1
#1 lore raw

On Fri, Oct 03, 2025 at 05:13:04PM +0800, Brian Sune wrote:

Patch title should be tweaked to match the other patches to this driver:

ASoC: wm8978: ...

Quoted 3 lines

The original WM8978 codec driver did not set the BCLK (bit clock) divider, which can cause audio clocks to be incorrect or unstable depending on the sample rate and word length.

This isn't totally accurate, the driver expects it will be set through the set_clkdiv callback. Which one could probably argue is a bit of a more legacy approach, but probably worth calling that out here.

Quoted 14 lines

This patch adds proper calculation and configuration of the BCLK divider based on the sample rate and word width, ensuring the codec generates the correct bit clock for all supported rates.

---
 sound/soc/codecs/wm8978.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
sound/soc/codecs/wm8978.c +0 -0
index 8c45ba6fc4c3..2109c84f33df 100644

sound/soc/codecs/wm8978.c static int wm8978_hw_params(struct snd_pcm_substream *substream, · line 721

Quoted 5 lines
sound/soc/codecs/wm8978.c +2 -0
@@ -717,6 +717,11 @@ static int wm8978_hw_params(struct snd_pcm_substream *substream,
 			    struct snd_pcm_hw_params *params,
 			    struct snd_soc_dai *dai)
 {
+	unsigned int bclk, bclkdiv = 0, min_diff = UINT_MAX;
+	unsigned int target_bclk = params_rate(params) * params_width(params) * 2;

Probably better to use snd_soc_params_to_bclk or similar helper.

Quoted 14 lines
sound/soc/codecs/wm8978.c +7 -0
+	/* WM8978 supports divisors */
+	static const int bclk_divs[] = {1, 2, 4, 8, 16, 32};
+
 	struct snd_soc_component *component = dai->component;
 	struct wm8978_priv *wm8978 = snd_soc_component_get_drvdata(component);
 	/* Word length mask = 0x60 */
@@ -820,6 +825,21 @@ static int wm8978_hw_params(struct snd_pcm_substream *substream,
 	/* MCLK divisor mask = 0xe0 */
 	snd_soc_component_update_bits(component, WM8978_CLOCKING, 0xe0, best << 5);
 
+	for (i = 0; i < ARRAY_SIZE(bclk_divs); i++) {
+		bclk = wm8978->f_256fs / bclk_divs[i];
+		if (abs(bclk - target_bclk) < min_diff) {
+			min_diff = abs(bclk - target_bclk);

sound/soc/codecs/wm8978.c static int wm8978_hw_params(struct snd_pcm_substream *substream, · line 841

Quoted 10 lines
sound/soc/codecs/wm8978.c +10 -0
+			bclkdiv = i;
+		}
+	}
+
+	dev_dbg(component->dev, "%s: fs=%u width=%u -> target BCLK=%u, using div #%u\n",
+		__func__, params_rate(params), params_width(params), target_bclk,
+		bclk_divs[bclkdiv]);
+
+	/* BCLKDIV divisor mask = 0x1c */
+	snd_soc_component_update_bits(component, WM8978_CLOCKING, 0x1c, bclkdiv << 2);

We should probably add something to handle the interaction with set_clkdiv and skip this if a fixed divider has been set. Well either that or remove the set_clkdiv option, although that is a little braver.

Thanks,
Charles

author
#2 lore raw

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午6:30寫道:

Quoted 7 lines · Brian Sune

Patch title should be tweaked to match the other patches to this driver:

ASoC: wm8978: ...

Will submit update patch, thank you.

Quoted 10 lines
Quoted 4 lines

The original WM8978 codec driver did not set the BCLK (bit clock) divider, which can cause audio clocks to be incorrect or unstable depending on the sample rate and word length.

This isn't totally accurate, the driver expects it will be set through the set_clkdiv callback. Which one could probably argue is a bit of a more legacy approach, but probably worth calling that out here.

According to actual hardware tests and the WM8978 driver study. There are no bclk register setup in the entire driver. So I am not sure How could this even set through the callback? The IC itself requires 2-wires register load and this can't be done via software level.

Quoted 43 lines
Quoted 14 lines

This patch adds proper calculation and configuration of the BCLK divider based on the sample rate and word width, ensuring the codec generates the correct bit clock for all supported rates.

---
 sound/soc/codecs/wm8978.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
sound/soc/codecs/wm8978.c +0 -0
index 8c45ba6fc4c3..2109c84f33df 100644

sound/soc/codecs/wm8978.c static int wm8978_hw_params(struct snd_pcm_substream *substream, · line 721

Quoted 5 lines
sound/soc/codecs/wm8978.c +2 -0
@@ -717,6 +717,11 @@ static int wm8978_hw_params(struct snd_pcm_substream *substream,
                          struct snd_pcm_hw_params *params,
                          struct snd_soc_dai *dai)
 {
+     unsigned int bclk, bclkdiv = 0, min_diff = UINT_MAX;
+     unsigned int target_bclk = params_rate(params) * params_width(params) * 2;

Probably better to use snd_soc_params_to_bclk or similar helper.

Quoted 21 lines
sound/soc/codecs/wm8978.c +14 -0
+     /* WM8978 supports divisors */
+     static const int bclk_divs[] = {1, 2, 4, 8, 16, 32};
+
      struct snd_soc_component *component = dai->component;
      struct wm8978_priv *wm8978 = snd_soc_component_get_drvdata(component);
      /* Word length mask = 0x60 */
@@ -820,6 +825,21 @@ static int wm8978_hw_params(struct snd_pcm_substream *substream,
      /* MCLK divisor mask = 0xe0 */
      snd_soc_component_update_bits(component, WM8978_CLOCKING, 0xe0, best << 5);
 
+     for (i = 0; i < ARRAY_SIZE(bclk_divs); i++) {
+             bclk = wm8978->f_256fs / bclk_divs[i];
+             if (abs(bclk - target_bclk) < min_diff) {
+                     min_diff = abs(bclk - target_bclk);
+                     bclkdiv = i;
+             }
+     }
+
+     dev_dbg(component->dev, "%s: fs=%u width=%u -> target BCLK=%u, using div #%u\n",
+             __func__, params_rate(params), params_width(params), target_bclk,
+             bclk_divs[bclkdiv]);
Quoted 8 lines

sound/soc/codecs/wm8978.c static int wm8978_hw_params(struct snd_pcm_substream *substream, · line 841

Quoted 3 lines
sound/soc/codecs/wm8978.c +3 -0
+
+     /* BCLKDIV divisor mask = 0x1c */
+     snd_soc_component_update_bits(component, WM8978_CLOCKING, 0x1c, bclkdiv << 2);

We should probably add something to handle the interaction with set_clkdiv and skip this if a fixed divider has been set. Well either that or remove the set_clkdiv option, although that is a little braver.

Not too understand this idea but the current problem or bug is just the bclk is missing.
Which had tested via this patch and provide good result. Meanwhile, the mclk divider
is restricted to actual IC clock design so this is why it is a fixed table.

Quoted 3 lines

Thanks,
Charles

author
#13 lore raw

PATCH v2 1/1 ASoC: wm8978: add missing BCLK divider setup

In reply to Brian Sune

The original WM8978 codec driver did not set the BCLK (bit clock) divider, which can cause audio clocks to be incorrect or unstable depending on the sample rate and word length.

This patch adds proper calculation and configuration of the BCLK divider based on the sample rate and word width, ensuring the codec generates the correct bit clock for all supported rates.

---
 sound/soc/codecs/wm8978.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)
sound/soc/codecs/wm8978.c +20 -0
index 8c45ba6fc4c3..2109c84f33df 100644
@@ -717,6 +717,11 @@ static int wm8978_hw_params(struct snd_pcm_substream *substream,
 			    struct snd_pcm_hw_params *params,
 			    struct snd_soc_dai *dai)
 {
+	unsigned int bclk, bclkdiv = 0, min_diff = UINT_MAX;
+	unsigned int target_bclk = params_rate(params) * params_width(params) * 2;
+	/* WM8978 supports divisors */
+	static const int bclk_divs[] = {1, 2, 4, 8, 16, 32};
+
 	struct snd_soc_component *component = dai->component;
 	struct wm8978_priv *wm8978 = snd_soc_component_get_drvdata(component);
 	/* Word length mask = 0x60 */
@@ -820,6 +825,21 @@ static int wm8978_hw_params(struct snd_pcm_substream *substream,
 	/* MCLK divisor mask = 0xe0 */
 	snd_soc_component_update_bits(component, WM8978_CLOCKING, 0xe0, best << 5);
 
+	for (i = 0; i < ARRAY_SIZE(bclk_divs); i++) {
+		bclk = wm8978->f_256fs / bclk_divs[i];
+		if (abs(bclk - target_bclk) < min_diff) {
+			min_diff = abs(bclk - target_bclk);
+			bclkdiv = i;
+		}
+	}
+
+	dev_dbg(component->dev, "%s: fs=%u width=%u -> target BCLK=%u, using div #%u\n",
+		__func__, params_rate(params), params_width(params), target_bclk,
+		bclk_divs[bclkdiv]);
+
+	/* BCLKDIV divisor mask = 0x1c */
+	snd_soc_component_update_bits(component, WM8978_CLOCKING, 0x1c, bclkdiv << 2);
+
 	snd_soc_component_write(component, WM8978_AUDIO_INTERFACE, iface_ctl);
 	snd_soc_component_write(component, WM8978_ADDITIONAL_CONTROL, add_ctl);
 
Signature · 1 line
2.47.1.windows.1
maintainer
#14 lore raw

Re: [PATCH v2] ASoC: wm8978: add missing BCLK divider setup

Quoted 3 lines · Brian Sune

The original WM8978 codec driver did not set the BCLK (bit clock) divider, which can cause audio clocks to be incorrect or unstable depending on the sample rate and word length.

Please don't send new patches in reply to old patches or serieses, this makes it harder for both people and tools to understand what is going on - it can bury things in mailboxes and make it difficult to keep track of what current patches are, both for the new patches and the old ones.

author
#15 lore raw

Re: [PATCH v2] ASoC: wm8978: add missing BCLK divider setup

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午7:44寫道:

Quoted 10 lines · Brian Sune
Quoted 3 lines

The original WM8978 codec driver did not set the BCLK (bit clock) divider, which can cause audio clocks to be incorrect or unstable depending on the sample rate and word length.

Please don't send new patches in reply to old patches or serieses, this makes it harder for both people and tools to understand what is going on - it can bury things in mailboxes and make it difficult to keep track of what current patches are, both for the new patches and the old ones.

Sorry for this action. But this patch is just a title fixe according to previous comment.
Patch body is fully aligned to previous revision.

May I know what is the proper way to amend the title?

maintainer
#16 lore raw

Re: [PATCH v2] ASoC: wm8978: add missing BCLK divider setup

Quoted 7 lines · Sune Brian

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午7:44寫道:

Quoted 4 lines

Please don't send new patches in reply to old patches or serieses, this makes it harder for both people and tools to understand what is going on - it can bury things in mailboxes and make it difficult to keep track of what current patches are, both for the new patches and the old ones.

Quoted 5 lines

Sorry for this action. But this patch is just a title fixe according to previous comment.
Patch body is fully aligned to previous revision.

May I know what is the proper way to amend the title?

The title is fine, the above is about sending the patch as a reply.

author
#17 lore raw

Re: [PATCH v2] ASoC: wm8978: add missing BCLK divider setup

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午7:52寫道:

Quoted 9 lines · Sune Brian
Quoted 6 lines

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午7:44寫道:

Quoted 4 lines

Please don't send new patches in reply to old patches or serieses, this makes it harder for both people and tools to understand what is going on - it can bury things in mailboxes and make it difficult to keep track of what current patches are, both for the new patches and the old ones.

Quoted 7 lines
Quoted 5 lines

Sorry for this action. But this patch is just a title fixe according to previous comment.
Patch body is fully aligned to previous revision.

May I know what is the proper way to amend the title?

The title is fine, the above is about sending the patch as a reply.

Got it, when title amended version is sent, need no reply with previous revision ID.
Thank you

maintainer
#18 lore raw

Re: [PATCH v2] ASoC: wm8978: add missing BCLK divider setup

Quoted 11 lines · Sune Brian

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午7:52寫道:

Quoted 6 lines
Quoted 4 lines
Quoted 4 lines

Please don't send new patches in reply to old patches or serieses, this makes it harder for both people and tools to understand what is going on - it can bury things in mailboxes and make it difficult to keep track of what current patches are, both for the new patches and the old ones.

The title is fine, the above is about sending the patch as a reply.

Got it, when title amended version is sent, need no reply with previous revision ID.

No. You sent the new version as a reply to the prior verison which nearly got it binned. You should send new versions as new threads, not as replies to other serieses.

#3 lore raw
In reply to Sune Brian
Quoted 11 lines · Sune Brian

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午6:30寫道:

Quoted 9 lines · Brian Sune
Quoted 3 lines

The original WM8978 codec driver did not set the BCLK (bit clock) divider, which can cause audio clocks to be incorrect or unstable depending on the sample rate and word length.

This isn't totally accurate, the driver expects it will be set through the set_clkdiv callback. Which one could probably argue is a bit of a more legacy approach, but probably worth calling that out here.

Quoted 4 lines

According to actual hardware tests and the WM8978 driver study. There are no bclk register setup in the entire driver. So I am not sure How could this even set through the callback? The IC itself requires 2-wires register load and this can't be done via software level.

/*
* Configure WM8978 clock dividers.
*/
static int wm8978_set_dai_clkdiv(struct snd_soc_dai *codec_dai,
int div_id, int div)

{
	struct snd_soc_component *component = codec_dai->component;
	struct wm8978_priv *wm8978 = snd_soc_component_get_drvdata(component);
	int ret = 0;

	switch (div_id) {
	case WM8978_OPCLKRATE:
		...
	case WM8978_BCLKDIV:
		if (div & ~0x1c)
			return -EINVAL;
		snd_soc_component_update_bits(component, WM8978_CLOCKING, 0x1c, div); <<---- HERE
		break;
	default:
		return -EINVAL;
	}

	dev_dbg(component->dev, "%s: ID %d, value %u\n", __func__, div_id, div);

	return ret;
}

Its not missing its right there. That said your way is probably slightly more standard these days, but we should take care of the interaction between the two.

Thanks,
Charles

maintainer
#4 lore raw
Quoted 3 lines · Charles Keepax

Its not missing its right there. That said your way is probably slightly more standard these days, but we should take care of the interaction between the two.

Where we've had users using the manual divider configuration that we cared about in the past we've suppressed automatic configuration if manual configuration has ever been done by the machine driver.

#5 lore raw
Quoted 9 lines · Mark Brown
Quoted 3 lines · Charles Keepax

Its not missing its right there. That said your way is probably slightly more standard these days, but we should take care of the interaction between the two.

Where we've had users using the manual divider configuration that we cared about in the past we've suppressed automatic configuration if manual configuration has ever been done by the machine driver.

I don't see any current users for the manual divider configuration at least in mainline, so we could probably just drop the manual stuff. I don't think I have a strong opinion between that and blocking the automatic configuration if it was previously set manually, but we should do one of them to keep things clear.

Thanks,
Charles

maintainer
#6 lore raw
Quoted 5 lines · Charles Keepax

I don't see any current users for the manual divider configuration at least in mainline, so we could probably just drop the manual stuff. I don't think I have a strong opinion between that and blocking the automatic configuration if it was previously set manually, but we should do one of them to keep things clear.

Yes, definitely.

author
#7 lore raw

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午8:30寫道:

Quoted 10 lines · Charles Keepax
Quoted 6 lines

I don't see any current users for the manual divider configuration at least in mainline, so we could probably just drop the manual stuff. I don't think I have a strong opinion between that and blocking the automatic configuration if it was previously set manually, but we should do one of them to keep things clear.

Yes, definitely.

if you read that code more clearly. It is a case that is select between WM8978_OPCLKRATE and WM8978_BCLKDIV.
With the latest test on the dts file.
When running such codec on master mode the bclk had never set to correct bclk rate. As such even you see that codes exist. It never run as expected. I am not sure any adjust is required on dts but with proper lrclk and mclk output.
The bclk is wrong on that version.

This is why I propose the patch of the bclk.

author
#8 lore raw
In reply to Charles Keepax

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午8:11寫道:

Quoted 45 lines · Sune Brian
Quoted 11 lines

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午6:30寫道:

Quoted 9 lines · Brian Sune
Quoted 3 lines

The original WM8978 codec driver did not set the BCLK (bit clock) divider, which can cause audio clocks to be incorrect or unstable depending on the sample rate and word length.

This isn't totally accurate, the driver expects it will be set through the set_clkdiv callback. Which one could probably argue is a bit of a more legacy approach, but probably worth calling that out here.

Quoted 4 lines

According to actual hardware tests and the WM8978 driver study. There are no bclk register setup in the entire driver. So I am not sure How could this even set through the callback? The IC itself requires 2-wires register load and this can't be done via software level.

/*
* Configure WM8978 clock dividers.
*/
static int wm8978_set_dai_clkdiv(struct snd_soc_dai *codec_dai,
int div_id, int div)

{
        struct snd_soc_component *component = codec_dai->component;
        struct wm8978_priv *wm8978 = snd_soc_component_get_drvdata(component);
        int ret = 0;

        switch (div_id) {
        case WM8978_OPCLKRATE:
                ...
        case WM8978_BCLKDIV:
                if (div & ~0x1c)
                        return -EINVAL;
                snd_soc_component_update_bits(component, WM8978_CLOCKING, 0x1c, div); <<---- HERE
                break;
        default:
                return -EINVAL;
        }

        dev_dbg(component->dev, "%s: ID %d, value %u\n", __func__, div_id, div);

        return ret;
}
Quoted 3 lines

Its not missing its right there. That said your way is probably slightly more standard these days, but we should take care of the interaction between the two.

What my missing meant is if run with DEBUG flag on that case had never behave as expected.
MCLK and LRCLK both is correctly outputted. While the current unpatched version will generate
wrong BCLK complete break the codec. As such I proposed the BCLK patch. I had not investigate deep why it never calls but the "int div" is loaded and computed by where is a bit puzzling.
And the loaded it simply with div on actual mclk/2/bit_per_channel is also incorrect.
As mentioned in previous explanations, the clock register is a fix table on dividing # that is a LUT with restricted # allowed.

Quoted 3 lines

Thanks,
Charles

author
#19 lore raw

Re: [PATCH v2] ASoC: wm8978: add missing BCLK divider setup

In reply to Mark Brown

Understood, so in this situation.
What is the proper way to fix it?
Leave it as such or create a new thread?

Thank you

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午8:10寫道:

Quoted 17 lines · Sune Brian
Quoted 11 lines

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午7:52寫道:

Quoted 6 lines
Quoted 4 lines
Quoted 4 lines

Please don't send new patches in reply to old patches or serieses, this makes it harder for both people and tools to understand what is going on - it can bury things in mailboxes and make it difficult to keep track of what current patches are, both for the new patches and the old ones.

The title is fine, the above is about sending the patch as a reply.

Got it, when title amended version is sent, need no reply with previous revision ID.

No. You sent the new version as a reply to the prior verison which nearly got it binned. You should send new versions as new threads, not as replies to other serieses.

maintainer
#20 lore raw

Re: [PATCH v2] ASoC: wm8978: add missing BCLK divider setup

Quoted 3 lines · Sune Brian

Understood, so in this situation.
What is the proper way to fix it?
Leave it as such or create a new thread?

It's fine, just don't do this for any future versions you post.

Please don't top post, reply in line with needed context. This allows readers to readily follow the flow of conversation and understand what you are talking about and also helps ensure that everything in the discussion is being addressed.

#9 lore raw
In reply to Sune Brian
Quoted 9 lines · Sune Brian

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午8:11寫道:

Quoted 7 lines · Sune Brian
Quoted 3 lines

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午6:30寫道:

Quoted 2 lines · Brian Sune
Quoted 1 line

The original WM8978 codec driver did not set the BCLK (bit clock)

Its not missing its right there. That said your way is probably slightly more standard these days, but we should take care of the interaction between the two.

Quoted 11 lines

What my missing meant is if run with DEBUG flag on that case had never behave as expected.
MCLK and LRCLK both is correctly outputted. While the current unpatched version will generate
wrong BCLK complete break the codec. As such I proposed the BCLK patch. I had not investigate deep why it never calls but the "int div" is loaded and computed by where is a bit puzzling.
And the loaded it simply with div on actual mclk/2/bit_per_channel is also incorrect.
As mentioned in previous explanations, the clock register is a fix table on dividing # that is a LUT with restricted # allowed.

Yeah the existing code expects the machine driver to call snd_soc_dai_set_clkdiv. I am guessing you are using something like simple card that doesn't do this.

To be clear the bulk of your patch is good, updating this in hw_params is probably more normal these days. But we need to make sure the two paths don't interfere with each other. Think of a system that is already calling snd_soc_dai_set_clkdiv() to set BCLKDIV, after your patch BCLKDIV will be set twice potentially to two different values and would generate no error messages.

I think you have two options:

1) Remove WM8978_BCLKDIV from wm8978_set_dai_clkdiv. There are no upstream users that I can see, so this should be fine. This would mean an out of tree user of snd_soc_dai_set_clkdiv would now get an error so they know they need to fix something.
2) Only run your dynamic BCLK code if wm8978_set_dai_clkdiv hasn't been called. This would mean any out of tree users of snd_soc_dai_set_clkdiv would have no problems everything would keep working as before, but at the cost of a little complexity in the code.

I am happy with either approach so which ever you prefer is fine with me.

Thanks,
Charles

author
#10 lore raw

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午9:15寫道:

Quoted 27 lines · Sune Brian
Quoted 9 lines

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午8:11寫道:

Quoted 7 lines · Sune Brian
Quoted 3 lines

Charles Keepax <[email protected]> 於 2025年10月7日 週二 下午6:30寫道:

Quoted 2 lines · Brian Sune
Quoted 1 line

The original WM8978 codec driver did not set the BCLK (bit clock)

Its not missing its right there. That said your way is probably slightly more standard these days, but we should take care of the interaction between the two.

Quoted 11 lines

What my missing meant is if run with DEBUG flag on that case had never behave as expected.
MCLK and LRCLK both is correctly outputted. While the current unpatched version will generate
wrong BCLK complete break the codec. As such I proposed the BCLK patch. I had not investigate deep why it never calls but the "int div" is loaded and computed by where is a bit puzzling.
And the loaded it simply with div on actual mclk/2/bit_per_channel is also incorrect.
As mentioned in previous explanations, the clock register is a fix table on dividing # that is a LUT with restricted # allowed.

Yeah the existing code expects the machine driver to call snd_soc_dai_set_clkdiv. I am guessing you are using something like simple card that doesn't do this.

Quoted 7 lines

To be clear the bulk of your patch is good, updating this in hw_params is probably more normal these days. But we need to make sure the two paths don't interfere with each other. Think of a system that is already calling snd_soc_dai_set_clkdiv() to set BCLKDIV, after your patch BCLKDIV will be set twice potentially to two different values and would generate no error messages.

Before the below action started. I need more background, if possible. If my understand is correct on your ideas. Do you mean the default driver calls the bclkdiv on other places? But how could that bclkdiv # be correct from first place?
As I had mentioned the div # ifself like this codec is a LUT rather than the actual divided #?
For example /32 from LUT is a 3b101?
For example what if MCLK is set higher than /64 could even out of the LUT from first place.
The external div # passed in, how to ensure it is codec LUT compatible from first place?

Quoted 18 lines

I think you have two options:

1) Remove WM8978_BCLKDIV from wm8978_set_dai_clkdiv. There are no upstream users that I can see, so this should be fine. This would mean an out of tree user of snd_soc_dai_set_clkdiv would now get an error so they know they need to fix something.
2) Only run your dynamic BCLK code if wm8978_set_dai_clkdiv hasn't been called. This would mean any out of tree users of snd_soc_dai_set_clkdiv would have no problems everything would keep working as before, but at the cost of a little complexity in the code.

I am happy with either approach so which ever you prefer is fine with me.

Thanks,
Charles

maintainer
#11 lore raw
Quoted 11 lines · Sune Brian

Before the below action started. I need more background, if possible. If my understand is correct on your ideas. Do you mean the default driver calls the bclkdiv on other places? But how could that bclkdiv # be correct from first place?
As I had mentioned the div # ifself like this codec is a LUT rather than the actual divided #?
For example /32 from LUT is a 3b101?
For example what if MCLK is set higher than /64 could even out of the LUT from first place.
The external div # passed in, how to ensure it is codec LUT compatible from first place?

Systems doing this are using a machine driver written for that specific machine and will know exactly which CODEC and other components they are dealing with.

author
#21 lore raw

Re: [PATCH v2] ASoC: wm8978: add missing BCLK divider setup

In reply to Mark Brown

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午9:15寫道:

Quoted 4 lines

Please don't top post, reply in line with needed context. This allows readers to readily follow the flow of conversation and understand what you are talking about and also helps ensure that everything in the discussion is being addressed.

Understood crop previous reply to the content that this is mail replying. Will soon repatch another version on Charles comments.
Thank you

author
#12 lore raw
In reply to Mark Brown

Mark Brown <[email protected]> 於 2025年10月7日 週二 下午9:48寫道:

Quoted 3 lines

Systems doing this are using a machine driver written for that specific machine and will know exactly which CODEC and other components they are dealing with.

With such idea, this is not CODEC controlled from first place to restrict the possible configurations.
This could introduce many possible hazard, but not this patch trying to fix from first place.

As for Charles, I will try to patch with option 2 unless it is too tedious to do so.

I think you have two options:

1) Remove WM8978_BCLKDIV from wm8978_set_dai_clkdiv. There are no upstream users that I can see, so this should be fine. This would mean an out of tree user of snd_soc_dai_set_clkdiv would now get an error so they know they need to fix something.
2) Only run your dynamic BCLK code if wm8978_set_dai_clkdiv hasn't been called. This would mean any out of tree users of snd_soc_dai_set_clkdiv would have no problems everything would keep working as before, but at the cost of a little complexity in the code.