On Mon, 2022-03-21 at 17:21 +0800, Tzung-Bi Shih wrote:
On Mon, Mar 21, 2022 at 05:14:08PM +0800, Jiaxin Yu wrote:
On Mon, 2022-03-21 at 11:59 +0800, Tzung-Bi Shih wrote:
On Sat, Mar 19, 2022 at 07:41:10PM +0800, Jiaxin Yu wrote:
@@ -1145,6 +1140,13 @@ static int mt8192_mt6359_dev_probe(struct platform_device *pdev) goto err_speaker_codec; }
- headset_codec = of_get_child_by_name(pdev->dev.of_node,
"mediatek,headset-codec");
- if (!headset_codec) {
ret = -EINVAL;dev_err_probe(&pdev->dev, ret, "Property'headset_codec' missing or invalid\n");
goto err_headset_codec;- }
(to be neat) Does it have any reason to prevent from using of_parse_phandle() but of_get_child_by_name()?
"mediatek,headset-codec" is a child node of pdev->dev.of_node, so I use of_get_child_by_name() to get and pass it to snd_soc_of_get_dai_link_codecs().
"mediatek,platform" and "mediatek,hdmi-codec" are also children of pdev->dev.of_node. I guess my question is: why doesn't it also use of_parse_phandle() for "mediatek,headset-codec"? Did I misunderstand?
Hi Tzung-Bi,
The following is from bindings, "mediatek,speaker-codec" and "mediatek,headset-codec" are sub nodes of sound but "mediatek,platform" and "mediatek,hdmi-codec" are the name of properties. So we can't get the sub node pointer through of_parse_phandle().
sound: mt8192-sound { compatible = "mediatek,mt8192_mt6359_rt1015_rt5682"; mediatek,platform = <&afe>; mediatek,hdmi-codec = <&anx_bridge_dp>; pinctrl-names = "aud_clk_mosi_off", "aud_clk_mosi_on"; pinctrl-0 = <&aud_clk_mosi_off>; pinctrl-1 = <&aud_clk_mosi_on>;
mediatek,headset-codec { sound-dai = <&rt5682>; };
mediatek,speaker-codec { sound-dai = <&rt1015_l>, <&rt1015_r>; }; };