[PATCH v2 0/2] ASoC: simple-card-utils: support system-clock-id DT property

[PATCH v2 0/2] ASoC: simple-card-utils: support system-clock-id DT property

[Sen Wang]

Adding an optional, u32 system-clock-id DT property into the generic machine
driver so clocking topology info can be exposed in DT.

Sen Wang (2):
  dt-bindings: sound: simple-card: add system-clock-id property
  ASoC: simple-card-utils: support system-clock-id DT property

 Documentation/devicetree/bindings/sound/audio-graph-port.yaml |  2 +
 Documentation/devicetree/bindings/sound/simple-card.yaml      | 10 +++++++
 include/sound/simple_card_utils.h                              |  1 +
 sound/soc/generic/simple-card-utils.c                         | 13 +++++----
 4 files changed, 21 insertions(+), 5 deletions(-)

---
base-commit: f5ffe03e0678 (linux-next)

[PATCH v2 1/2] dt-bindings: sound: simple-card: add system-clock-id property

[Sen Wang]

Document the new optional "system-clock-id" u32 property for cpu/codec
sub-nodes in simple-audio-card and audio-graph-card.

The property maps directly to the clk_id argument of
snd_soc_dai_set_sysclk(). When absent the existing default of 0 is
used, so no existing DT is affected. CPU and codec sub-nodes are
parsed independently; each side may carry a different value to select
different clock sources on the two DAIs within the same dai-link.

Signed-off-by: Sen Wang <sen@ti.com>
---
changes in V2:
  - Newly added (V1 is RFC)

diff --git a/Documentation/devicetree/bindings/sound/audio-graph-port.yaml b/Documentation/devicetree/bindings/sound/audio-graph-port.yaml
index d1cbfc5edd3a..8094a76b9b4a 100644
--- a/Documentation/devicetree/bindings/sound/audio-graph-port.yaml
+++ b/Documentation/devicetree/bindings/sound/audio-graph-port.yaml
@@ -67,6 +67,8 @@ definitions:
         $ref: simple-card.yaml#/definitions/system-clock-direction-out
       system-clock-fixed:
         $ref: simple-card.yaml#/definitions/system-clock-fixed
+      system-clock-id:
+        $ref: simple-card.yaml#/definitions/system-clock-id
 
       dai-format:
         description: audio format.
diff --git a/Documentation/devicetree/bindings/sound/simple-card.yaml b/Documentation/devicetree/bindings/sound/simple-card.yaml
index a14716b2732f..47a02d340d5e 100644
--- a/Documentation/devicetree/bindings/sound/simple-card.yaml
+++ b/Documentation/devicetree/bindings/sound/simple-card.yaml
@@ -49,6 +49,14 @@ definitions:
       single fixed sampling rate.
     $ref: /schemas/types.yaml#/definitions/flag
 
+  system-clock-id:
+    description:
+      Selects which system clock to configure via set_sysclk(). Corresponds
+      to the clk_id argument of snd_soc_dai_set_sysclk(). When absent the
+      driver default (0) is used. CPU and codec sub-nodes are parsed
+      independently so each side may carry a different value.
+    $ref: /schemas/types.yaml#/definitions/uint32
+
   mclk-fs:
     description: |
       Multiplication factor between stream rate and codec mclk.
@@ -135,6 +143,8 @@ definitions:
         $ref: "#/definitions/system-clock-direction-out"
       system-clock-fixed:
         $ref: "#/definitions/system-clock-fixed"
+      system-clock-id:
+        $ref: "#/definitions/system-clock-id"
     required:
       - sound-dai
 

Re: [PATCH v2 1/2] dt-bindings: sound: simple-card: add system-clock-id property

[Krzysztof Kozlowski]

On Fri, May 15, 2026 at 11:13:57AM -0500, Sen Wang wrote:
> Document the new optional "system-clock-id" u32 property for cpu/codec
> sub-nodes in simple-audio-card and audio-graph-card.
> 
> The property maps directly to the clk_id argument of
> snd_soc_dai_set_sysclk(). When absent the existing default of 0 is

That's driver argument in bindings. Poor way to start discussion. Please
describe the hardware instead.


> used, so no existing DT is affected. CPU and codec sub-nodes are
> parsed independently; each side may carry a different value to select
> different clock sources on the two DAIs within the same dai-link.

Again, nothing about the hardware here.

> 
> Signed-off-by: Sen Wang <sen@ti.com>
> ---
> changes in V2:
>   - Newly added (V1 is RFC)
> 
> diff --git a/Documentation/devicetree/bindings/sound/audio-graph-port.yaml b/Documentation/devicetree/bindings/sound/audio-graph-port.yaml
> index d1cbfc5edd3a..8094a76b9b4a 100644
> --- a/Documentation/devicetree/bindings/sound/audio-graph-port.yaml
> +++ b/Documentation/devicetree/bindings/sound/audio-graph-port.yaml
> @@ -67,6 +67,8 @@ definitions:
>          $ref: simple-card.yaml#/definitions/system-clock-direction-out
>        system-clock-fixed:
>          $ref: simple-card.yaml#/definitions/system-clock-fixed
> +      system-clock-id:
> +        $ref: simple-card.yaml#/definitions/system-clock-id
>  
>        dai-format:
>          description: audio format.
> diff --git a/Documentation/devicetree/bindings/sound/simple-card.yaml b/Documentation/devicetree/bindings/sound/simple-card.yaml
> index a14716b2732f..47a02d340d5e 100644
> --- a/Documentation/devicetree/bindings/sound/simple-card.yaml
> +++ b/Documentation/devicetree/bindings/sound/simple-card.yaml
> @@ -49,6 +49,14 @@ definitions:
>        single fixed sampling rate.
>      $ref: /schemas/types.yaml#/definitions/flag
>  
> +  system-clock-id:
> +    description:
> +      Selects which system clock to configure via set_sysclk(). Corresponds
> +      to the clk_id argument of snd_soc_dai_set_sysclk(). When absent the

So now driver APIs are part of binding :/. No.

You should have built on top or use previous discussion instead of
making new mistakes:
https://lore.kernel.org/linux-devicetree/20221022162742.21671-2-aidanmacdonald.0x0@gmail.com/

Best regards,
Krzysztof

Re: [PATCH v2 1/2] dt-bindings: sound: simple-card: add system-clock-id property

[Wang, Sen]

On 5/16/2026 4:54 AM, Krzysztof Kozlowski wrote:
> On Fri, May 15, 2026 at 11:13:57AM -0500, Sen Wang wrote:
>> Document the new optional "system-clock-id" u32 property for cpu/codec
>> sub-nodes in simple-audio-card and audio-graph-card.

(snip)
> So now driver APIs are part of binding :/. No.
> 
> You should have built on top or use previous discussion instead of
> making new mistakes:
> https://lore.kernel.org/linux-devicetree/20221022162742.21671-2-aidanmacdonald.0x0@gmail.com/
> 

Thanks for your insight Krzysztof, definitely failed my due diligence in 
finding previous discussions :/.

Let me see if I can do some refactor in ASoC clks instead.

Best,
Sen Wang

[PATCH v2 2/2] ASoC: simple-card-utils: support system-clock-id DT property

[Sen Wang]

The generic machine driver never had a mechanism to adjust for various
clk_ids that is tailored for individual codec & cpu peripheral driver.
simple_util_dai_init() has hardcoded 0, thus making non-default clock IDs
unreachable from DTS. Boards needing a specific clk_id have had no choice
but to write dedicated machine drivers.

Add an optional "system-clock-id" u32 property to the cpu/codec
sub-node and use it in all three set_sysclk() call sites. When absent
clk_id stays 0, preserving identical behavior for all existing boards.

And since simple_util_parse_clk() is invoked once per sub-node, per-DAI
independence can be preserved as cpu and codec sub-nodes carry independent
clk_id values.

E.g., the following dts topology would be supported:

  cpu   { system-clock-id = <6>; }
  codec { system-clock-id = <1>; }

Signed-off-by: Sen Wang <sen@ti.com>
---

Apologies Morimoto-san, when finalizing the patch I realize a few more
places need to be changed, can you kindly re-review and give your
acked-by?

Changes in v2:
- Adopted sysclk func changes for mclk cases
- Links to v1 (RFC):
https://lore.kernel.org/all/20260514212802.3569643-1-sen@ti.com/

diff --git a/include/sound/simple_card_utils.h b/include/sound/simple_card_utils.h
index 915e6ae5f68d..f221fba92e5e 100644
--- a/include/sound/simple_card_utils.h
+++ b/include/sound/simple_card_utils.h
@@ -26,6 +26,7 @@ struct simple_util_dai {
 	const char *name;
 	unsigned int sysclk;
 	int clk_direction;
+	int clk_id;
 	int slots;
 	int slot_width;
 	unsigned int tx_slot_mask;
diff --git a/sound/soc/generic/simple-card-utils.c b/sound/soc/generic/simple-card-utils.c
index 4cf3a5395763..ce7e960db5e9 100644
--- a/sound/soc/generic/simple-card-utils.c
+++ b/sound/soc/generic/simple-card-utils.c
@@ -290,6 +290,9 @@ int simple_util_parse_clk(struct device *dev,
 	if (of_property_read_bool(node, "system-clock-direction-out"))
 		simple_dai->clk_direction = SND_SOC_CLOCK_OUT;
 
+	if (!of_property_read_u32(node, "system-clock-id", &val))
+		simple_dai->clk_id = val;
+
 	return 0;
 }
 EXPORT_SYMBOL_GPL(simple_util_parse_clk);
@@ -506,14 +509,16 @@ int simple_util_hw_params(struct snd_pcm_substream *substream,
 			/* CPU first */
 			for_each_rtd_cpu_dais(rtd, i, sdai) {
 				pdai = simple_props_to_dai_cpu(props, i);
-				ret = snd_soc_dai_set_sysclk(sdai, 0, mclk, pdai->clk_direction);
+				ret = snd_soc_dai_set_sysclk(sdai, pdai->clk_id,
+						mclk, pdai->clk_direction);
 				if (ret && ret != -ENOTSUPP)
 					goto end;
 			}
 
 			for_each_rtd_codec_dais(rtd, i, sdai) {
 				pdai = simple_props_to_dai_codec(props, i);
-				ret = snd_soc_dai_set_sysclk(sdai, 0, mclk, pdai->clk_direction);
+				ret = snd_soc_dai_set_sysclk(sdai, pdai->clk_id,
+						mclk, pdai->clk_direction);
 				if (ret && ret != -ENOTSUPP)
 					goto end;
 			}
@@ -521,14 +526,16 @@ int simple_util_hw_params(struct snd_pcm_substream *substream,
 			/* default: codec first */
 			for_each_rtd_codec_dais(rtd, i, sdai) {
 				pdai = simple_props_to_dai_codec(props, i);
-				ret = snd_soc_dai_set_sysclk(sdai, 0, mclk, pdai->clk_direction);
+				ret = snd_soc_dai_set_sysclk(sdai, pdai->clk_id,
+						mclk, pdai->clk_direction);
 				if (ret && ret != -ENOTSUPP)
 					goto end;
 			}
 
 			for_each_rtd_cpu_dais(rtd, i, sdai) {
 				pdai = simple_props_to_dai_cpu(props, i);
-				ret = snd_soc_dai_set_sysclk(sdai, 0, mclk, pdai->clk_direction);
+				ret = snd_soc_dai_set_sysclk(sdai, pdai->clk_id,
+						mclk, pdai->clk_direction);
 				if (ret && ret != -ENOTSUPP)
 					goto end;
 			}
@@ -587,7 +594,8 @@ static int simple_init_dai(struct simple_util_priv *priv,
 		return 0;
 
 	if (simple_dai->sysclk) {
-		ret = snd_soc_dai_set_sysclk(dai, 0, simple_dai->sysclk,
+		ret = snd_soc_dai_set_sysclk(dai, simple_dai->clk_id,
+					     simple_dai->sysclk,
 					     simple_dai->clk_direction);
 		if (ret && ret != -ENOTSUPP) {
 			dev_err(dai->dev, "simple-card: set_sysclk error\n");

Re: [PATCH v2 2/2] ASoC: simple-card-utils: support system-clock-id DT property

[sashiko-bot]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The commit introduces asymmetric sysclk configuration by failing to update `clk_id` in the shutdown path, leading to incorrect hardware states and resource leaks.
--

commit fb04cb3ea63c6e1c03c9835d2530dab5d343394b
Author: Sen Wang <sen@ti.com>

ASoC: simple-card-utils: support system-clock-id DT property

This commit adds an optional system-clock-id DT property to the cpu/codec
sub-nodes. It updates simple_util_parse_clk(), simple_util_hw_params(),
and simple_init_dai() to read and use this clock ID instead of hardcoding 0.

> diff --git a/sound/soc/generic/simple-card-utils.c b/sound/soc/generic/simple-card-utils.c
> index e5cb602fd2487..441f68df85220 100644
> --- a/sound/soc/generic/simple-card-utils.c
> +++ b/sound/soc/generic/simple-card-utils.c

[ ... ]

> @@ -506,14 +509,16 @@ int simple_util_hw_params(struct snd_pcm_substream *substream,
>  			/* CPU first */
>  			for_each_rtd_cpu_dais(rtd, i, sdai) {
>  				pdai = simple_props_to_dai_cpu(props, i);
> -				ret = snd_soc_dai_set_sysclk(sdai, 0, mclk, pdai->clk_direction);
> +				ret = snd_soc_dai_set_sysclk(sdai, pdai->clk_id,
> +						mclk, pdai->clk_direction);

Does the teardown path in simple_util_shutdown() also need to be updated to
use clk_id?

It currently hardcodes 0 when disabling the clock:

sound/soc/generic/simple-card-utils.c:simple_util_shutdown() {
	...
	for_each_prop_dai_cpu(props, i, dai) {
		struct snd_soc_dai *cpu_dai = snd_soc_rtd_to_cpu(rtd, i);

		if (props->mclk_fs && !dai->clk_fixed && !snd_soc_dai_active(cpu_dai))
			snd_soc_dai_set_sysclk(cpu_dai, 0, 0, dai->clk_direction);
	...
}

If a device is configured with a non-zero system-clock-id, will the shutdown
path disable the wrong clock, leaving the intended clock running?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260515161358.1462453-1-sen@ti.com?part=2

Re: [PATCH v2 0/2] ASoC: simple-card-utils: support system-clock-id DT property

[Wang, Sen]

On 5/15/2026 11:13 AM, Sen Wang wrote:
> Adding an optional, u32 system-clock-id DT property into the generic machine
> driver so clocking topology info can be exposed in DT.
> 
Sorry folks, Sashiko has caught a major bug in my patch. No need to 
review this. Will send out a V3.

Best,
Sen Wang