Re: [PATCH v5 3/3] arm64: dts: qcom: sm8650-hdk: Enable I2S for HDMI

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] arm64: dts: qcom: HDMI DAI link breaks sound probe with display-card overlay
--

commit 1d7b2eb9f1841da26b5c6030ed7671341483661b
Author: Neil Armstrong <neil.armstrong@linaro.org>

arm64: dts: qcom: sm8650-hdk: Enable I2S for HDMI

This commit adds the primary MI2S DAI link, Q6APM backend DAI clocks, and
TLMM pin configuration needed to route HDMI audio through the LT9611UXC DSI
bridge on the SM8650 HDK.

> diff --git a/arch/arm64/boot/dts/qcom/sm8650-hdk.dts b/arch/arm64/boot/dts/qcom/sm8650-hdk.dts
> index 6cb0dc089b17a..eba989e6766bf 100644
> --- a/arch/arm64/boot/dts/qcom/sm8650-hdk.dts
> +++ b/arch/arm64/boot/dts/qcom/sm8650-hdk.dts
> @@ -218,6 +218,22 @@ platform {
>  				sound-dai = <&q6apm>;
>  			};
>  		};
> +
> +		pri-mi2s-dai-link {
> +			link-name = "HDMI Playback";
> +
> +			cpu {
> +				sound-dai = <&q6apmbedai PRIMARY_MI2S_RX>;
> +			};
> +
> +			codec {
> +				sound-dai = <&lt9611_codec 0>;
> +			};
> +
> +			platform {
> +				sound-dai = <&q6apm>;
> +			};
> +		};
>  	};

[Severity: High]
Does adding pri-mi2s-dai-link unconditionally here cause a sound card probe
regression when booting with the display-card overlay
(sm8650-hdk-display-card.dtb or
sm8650-hdk-display-card-rear-camera-card.dtb)?

In arch/arm64/boot/dts/qcom/sm8650-hdk-display-card.dtso, both &i2c6 and
&lt9611_codec are disabled because the DSI lanes are routed to the display
daughterboard instead of the HDMI bridge:

    &lt9611_codec {
            status = "disabled";
    };

When that overlay is applied, lt9611_codec never probes or registers its ASoC
component, but pri-mi2s-dai-link remains enabled under /sound.

During sound card probe in sc8280xp_platform_probe() -> qcom_snd_parse_of()
(sound/soc/qcom/common.c), for_each_available_child_of_node_scoped() iterates
over every available child node of /sound and resolves the codec DAI via
snd_soc_of_get_dai_link_codecs() -> snd_soc_of_get_dlc() ->
snd_soc_get_dlc() (sound/soc/soc-core.c):

    if (codec) {
            ret = snd_soc_of_get_dai_link_codecs(dev, codec, link);
            if (ret < 0) {
                    dev_err_probe(dev, ret,
                                  "%s: codec dai not found\n", link->name);
                    return ret;
            }

Because neither of_parse_phandle_with_args() nor snd_soc_get_dlc() checks
of_device_is_available() on the target codec node, snd_soc_get_dlc() returns
-EPROBE_DEFER indefinitely when looking up &lt9611_codec, causing the entire
sound card probe to defer forever and breaking all audio on the board. (The
same issue also appears to affect prim-mi2s-dai-link in
arch/arm64/boot/dts/qcom/sm8550-hdk.dts with
arch/arm64/boot/dts/qcom/sm8550-hdk-display-card.dtso.)

Should pri-mi2s-dai-link be given a label so that
arch/arm64/boot/dts/qcom/sm8650-hdk-display-card.dtso can set its status to
"disabled" alongside &lt9611_codec?

[ ... ]

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261022-topic-sm8x50-next-hdk-i2s-v5-0-aa1030a75e46@linaro.org?part=3