Re: [PATCH 1/4] arm64: dts: imx952: Add audio device nodes

On Thu, Mar 12, 2026 at 3:10 PM Daniel Baluta [off-list ref] wrote:

On 3/12/26 08:54, Shengjiu Wang wrote:

quoted

quoted

On 3/12/26 05:49, Shengjiu Wang wrote:

quoted

Add audio device nodes, include SAI, MICFIL, ASRC, Audio Mixer.

Signed-off-by: Shengjiu Wang <redacted>
---
 arch/arm64/boot/dts/freescale/imx952.dtsi | 199 ++++++++++++++++++++++
 1 file changed, 199 insertions(+)

diff --git a/arch/arm64/boot/dts/freescale/imx952.dtsi b/arch/arm64/boot/dts/freescale/imx952.dtsi
index 0f86b5626cdd..5787c12e788f 100644
--- a/arch/arm64/boot/dts/freescale/imx952.dtsi
+++ b/arch/arm64/boot/dts/freescale/imx952.dtsi

@@ -823,6 +823,156 @@ usdhc3: mmc@42c40000 {
                      };
              };

+             aips5: bus@43000000 {
+                     compatible = "fsl,aips-bus", "simple-bus";
+                     reg = <0 0x43000000 0 0x800000>;
+                     #address-cells = <1>;
+                     #size-cells = <1>;
+                     ranges = <0x43000000 0x0 0x43000000 0x800000>;
+
+                     asrc1: asrc@43000000 {
+                             compatible = "fsl,imx952-asrc";
+                             reg = <0x43000000 0x10000>;
+                             interrupts = <GIC_SPI 376 IRQ_TYPE_LEVEL_HIGH>;
+                             clocks = <&scmi_clk IMX952_CLK_BUSWAKEUP>,
+                                      <&scmi_clk IMX952_CLK_BUSWAKEUP>,
+                                      <&scmi_clk IMX952_CLK_ASRC1>,
+                                      <&scmi_clk IMX952_CLK_ASRC2>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>,
+                                      <&clk_dummy>;
+                             clock-names = "mem", "ipg",
+                                           "asrck_0", "asrck_1", "asrck_2", "asrck_3",
+                                           "asrck_4", "asrck_5", "asrck_6", "asrck_7",
+                                           "asrck_8", "asrck_9", "asrck_a", "asrck_b",
+                                           "asrck_c", "asrck_d", "asrck_e", "asrck_f",
+                                           "spba";
+                             dmas = <&edma2 97 0 0>, <&edma2 98 0 0>, <&edma2 99 0 0>,
+                                    <&edma2 100 0 1>, <&edma2 101 0 1>, <&edma2 102 0 1>;

For consistency we should use symbolic macros for directions like you did below.
Use FSL_EDMA_RX for rx direction and also introduce FSL_EDMA_TX for tx direction.

Yes, should use FSL_EDMA_RX.

But introduce FSL_EDMA_TX, it should be zero,  FSL_EDMA_TX = 0.
is it necessary to add it in include/dt-bindings/dma/fsl-edma.h?

I would say yes. For consistency with FSL_EDMA_RX and also for

code readability.

A user doesn't know that 0 means TX.

We could go on and merge this patch as is right now, no problems. And come back later

with a set of cleanup patches that do this for all dtses.

I think it is better to have another patch set to introduce
FSL_EDMA_TX for all dtses.

I will add FSL_EDMA_RX in the next version.

Best regards
shengjiu Wang