* [PATCH v7 01/18] ASoC: dt-bindings: sound: Add DT binding for RZ/G3E sound
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:23 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 02/18] ASoC: rsnd: Fix RSND_SOC_MASK width to single nibble John Madieu
` (17 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
Add a standalone device tree binding for the Renesas RZ/G3E (R9A09G047)
sound controller.
The RZ/G3E sound IP is based on R-Car Sound but differs in several ways:
- Uses unprefixed sub-node names (ssi, ssiu, src, dvc, mix, ctu) instead
of R-Car's rcar_sound,xxx prefixed names.
- Supports up to 5 DMA controllers per direction, allowing multiple DMA
entries with repeated channel names in SSIU, SRC and DVC sub-nodes.
- Has 47 clocks including per-SSI ADG clocks (adg-ssi-[0-9]), SCU clocks
(scu, scu_x2, scu_supply), SSIF supply clock, AUDMAC peri-peri clock,
and ADG clock.
- Has 14 reset lines including SCU, ADG and AUDMAC peri-peri resets.
- SSI operates exclusively in BUSIF mode.
These differences make the RZ/G3E binding incompatible with the existing
renesas,rsnd.yaml, so it is added as a separate standalone binding with
its own $ref to dai-common.yaml.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7:
- Add per-property descriptions for the DMA specifier lists in the
dvc, ssiu and src sub-nodes, explaining that the dma-engine core
falls through the repeated entries to find a free channel
(Rob Herring).
- Add descriptions for the property-less mix-N and ctu-N sub-nodes,
explaining they exist only so the driver can enumerate the
instances and so DT labels can be attached for the dai/playback/
capture phandle routing arrays (Rob Herring).
- Add descriptions for the dai playback and capture phandle arrays
(Rob Herring).
- Drop the dmas / dma-names blanket "true" allowance on the ssi
sub-node; the SSI always operates in BUSIF mode (sashiko-bot).
- Add "#sound-dai-cells" and "#clock-cells" to the required list
(sashiko-bot).
- Add DT labels (dvc0, dvc1, src0, src1, ssi3, ssi4) to the example
so the playback/capture phandle references resolve and
dt_binding_check passes (sashiko-bot).
v6:
- Rename all indexed clock-names and reset-names from the dotted
form (ssi.0, src.0, adg.ssi.0, clk_a, clk_b, clk_c, clk_i) to
the hyphenated form (ssi-0, src-0, adg-ssi-0, audio-clka,
audio-clkb, audio-clkc, audio-clki) so the new binding follows
the standard DT naming convention.
- Tighten #sound-dai-cells to const: 1.
- Drop unused properties: clock-frequency, clkout-lr-asynchronous.
- Simplify the ports/endpoint schema (single ports object with
port@N children referencing audio-graph-port.yaml), drop the
separate top-level dai patternProperties block.
- Move additionalProperties: false to the top of each sub-object
(dvc, mix, ctu, src, ssiu, ssi).
- Reorder example clocks/resets to match the new ordinal-ascending
name order.
v5:
- Drop the two-patch rsnd.yaml split approach from v4. Replace
with a single self-contained standalone binding that does not
touch renesas,rsnd.yaml at all.
- Remove select: false, redundant blanket properties
(compatible: true, reg: true, etc.) and pointless
patternProperties per Krzysztof's review.
- Add missing #clock-cells and #sound-dai-cells constraints.
- Add hardware description text instead of "Binding for ..."
phrasing.
- Move G3E-specific DMA comment into the binding itself rather
than relying on a shared schema.
- Use unprefixed sub-node names (ssi, ssiu, src, dvc, mix, ctu)
to reflect the actual RZ/G3E DT binding.
v4: No changes
v3: No changes
v2:
- Introduce RZ/G3E sound binding as a standalone schema.
.../sound/renesas,r9a09g047-sound.yaml | 800 ++++++++++++++++++
1 file changed, 800 insertions(+)
create mode 100644 Documentation/devicetree/bindings/sound/renesas,r9a09g047-sound.yaml
diff --git a/Documentation/devicetree/bindings/sound/renesas,r9a09g047-sound.yaml b/Documentation/devicetree/bindings/sound/renesas,r9a09g047-sound.yaml
new file mode 100644
index 000000000000..d7fa16554698
--- /dev/null
+++ b/Documentation/devicetree/bindings/sound/renesas,r9a09g047-sound.yaml
@@ -0,0 +1,800 @@
+# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/sound/renesas,r9a09g047-sound.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: Renesas RZ/G3E Sound Controller
+
+maintainers:
+ - Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
+ - John Madieu <john.madieu.xa@bp.renesas.com>
+
+description:
+ The RZ/G3E (R9A09G047) sound controller is based on R-Car Sound IP
+ with extended DMA channel support (up to 5 DMACs per direction),
+ additional clock domains (47 clocks including per-SSI ADG clocks),
+ and additional reset lines (14 including SCU, ADG and Audio DMAC
+ peri-peri resets). SSI operates exclusively in BUSIF mode with
+ 2-4 BUSIF channels per SSI.
+
+allOf:
+ - $ref: dai-common.yaml#
+
+properties:
+ compatible:
+ const: renesas,r9a09g047-sound
+
+ reg:
+ maxItems: 5
+
+ reg-names:
+ items:
+ - const: scu
+ - const: adg
+ - const: ssiu
+ - const: ssi
+ - const: audmapp
+
+ "#sound-dai-cells":
+ const: 1
+
+ "#clock-cells":
+ const: 0
+
+ "#address-cells":
+ const: 1
+
+ "#size-cells":
+ const: 0
+
+ clocks:
+ maxItems: 47
+
+ clock-names:
+ items:
+ - const: ssi-all
+ - const: ssi-0
+ - const: ssi-1
+ - const: ssi-2
+ - const: ssi-3
+ - const: ssi-4
+ - const: ssi-5
+ - const: ssi-6
+ - const: ssi-7
+ - const: ssi-8
+ - const: ssi-9
+ - const: src-0
+ - const: src-1
+ - const: src-2
+ - const: src-3
+ - const: src-4
+ - const: src-5
+ - const: src-6
+ - const: src-7
+ - const: src-8
+ - const: src-9
+ - const: mix-0
+ - const: mix-1
+ - const: ctu-0
+ - const: ctu-1
+ - const: dvc-0
+ - const: dvc-1
+ - const: audio-clka
+ - const: audio-clkb
+ - const: audio-clkc
+ - const: audio-clki
+ - const: ssif_supply
+ - const: scu
+ - const: scu_x2
+ - const: scu_supply
+ - const: adg-ssi-0
+ - const: adg-ssi-1
+ - const: adg-ssi-2
+ - const: adg-ssi-3
+ - const: adg-ssi-4
+ - const: adg-ssi-5
+ - const: adg-ssi-6
+ - const: adg-ssi-7
+ - const: adg-ssi-8
+ - const: adg-ssi-9
+ - const: audmapp
+ - const: adg
+
+ power-domains:
+ maxItems: 1
+
+ resets:
+ maxItems: 14
+
+ reset-names:
+ items:
+ - const: ssi-all
+ - const: ssi-0
+ - const: ssi-1
+ - const: ssi-2
+ - const: ssi-3
+ - const: ssi-4
+ - const: ssi-5
+ - const: ssi-6
+ - const: ssi-7
+ - const: ssi-8
+ - const: ssi-9
+ - const: scu
+ - const: adg
+ - const: audmapp
+
+ dvc:
+ type: object
+ additionalProperties: false
+
+ patternProperties:
+ "^dvc-[0-1]$":
+ type: object
+ additionalProperties: false
+
+ properties:
+ dmas:
+ maxItems: 5
+ description:
+ List of references to DMA specifiers, one per DMA
+ controller, all for the transmission direction
+ (DVC is playback-only). The dma-engine core falls
+ through the list to find a free channel.
+
+ dma-names:
+ maxItems: 5
+ allOf:
+ - items:
+ enum:
+ - tx
+
+ required:
+ - dmas
+ - dma-names
+
+ mix:
+ type: object
+ additionalProperties: false
+ description:
+ Per-channel Mixer (MIX) sub-nodes. Each mix-N node has no
+ properties of its own. It exists so the driver can enumerate
+ the MIX instances and so that DT labels can be attached to it
+ for the dai/playback/capture phandle routing arrays.
+
+ patternProperties:
+ "^mix-[0-1]$":
+ type: object
+ additionalProperties: false
+
+ ctu:
+ type: object
+ additionalProperties: false
+ description:
+ Per-channel Channel Transfer Unit (CTU) sub-nodes. Each ctu-N
+ node has no properties of its own. It exists so the driver
+ can enumerate the CTU instances and so that DT labels can be
+ attached to it for the dai/playback/capture phandle routing arrays.
+
+ patternProperties:
+ "^ctu-[0-7]$":
+ type: object
+ additionalProperties: false
+
+ src:
+ type: object
+ additionalProperties: false
+
+ patternProperties:
+ "^src-[0-9]$":
+ type: object
+ additionalProperties: false
+
+ properties:
+ interrupts:
+ maxItems: 1
+
+ dmas:
+ maxItems: 10
+ description:
+ Must contain a list of pairs of references to DMA
+ specifiers, one for transmission and one for reception,
+ repeated for each DMA controller. The dma-engine core
+ falls through the list to find a free channel.
+
+ dma-names:
+ maxItems: 10
+ allOf:
+ - items:
+ enum:
+ - tx
+ - rx
+
+ ssiu:
+ type: object
+ additionalProperties: false
+
+ patternProperties:
+ "^ssiu-[0-9]+$":
+ type: object
+ additionalProperties: false
+
+ properties:
+ dmas:
+ maxItems: 10
+ description:
+ Must contain a list of pairs of references to DMA
+ specifiers, one for transmission and one for reception,
+ repeated for each DMA controller. The dma-engine core
+ falls through the list to find a free channel.
+
+ dma-names:
+ maxItems: 10
+ allOf:
+ - items:
+ enum:
+ - tx
+ - rx
+
+ required:
+ - dmas
+ - dma-names
+
+ ssi:
+ type: object
+ additionalProperties: false
+
+ patternProperties:
+ "^ssi-[0-9]$":
+ type: object
+ additionalProperties: false
+
+ properties:
+ interrupts:
+ maxItems: 1
+
+ shared-pin:
+ description: Shared clock pin.
+ $ref: /schemas/types.yaml#/definitions/flag
+
+ required:
+ - interrupts
+
+ ports:
+ $ref: audio-graph-port.yaml#/definitions/port-base
+ unevaluatedProperties: false
+ patternProperties:
+ '^port@[0-9a-f]+$':
+ $ref: audio-graph-port.yaml#/definitions/port-base
+ unevaluatedProperties: false
+ properties:
+ reg:
+ maxItems: 1
+ endpoint:
+ $ref: audio-graph-port.yaml#/definitions/endpoint-base
+ unevaluatedProperties: false
+ properties:
+ playback:
+ $ref: /schemas/types.yaml#/definitions/phandle-array
+ description:
+ Ordered list of phandles to the in-SoC modules used
+ by this DAI in the playback direction. Each phandle
+ must reference one of the ssi-N, src-N, ctu-N,
+ mix-N or dvc-N sub-nodes of the parent sound
+ controller. The list order is the pipeline order
+ from CPU to off-SoC endpoint.
+ capture:
+ $ref: /schemas/types.yaml#/definitions/phandle-array
+ description:
+ Ordered list of phandles to the in-SoC modules used
+ by this DAI in the capture direction. Each phandle
+ must reference one of the ssi-N, src-N, ctu-N,
+ mix-N or dvc-N sub-nodes of the parent sound
+ controller. The list order is the pipeline order
+ from off-SoC endpoint to CPU.
+
+required:
+ - compatible
+ - reg
+ - reg-names
+ - "#sound-dai-cells"
+ - "#clock-cells"
+ - clocks
+ - clock-names
+ - resets
+ - reset-names
+
+unevaluatedProperties: false
+
+examples:
+ - |
+ #include <dt-bindings/interrupt-controller/arm-gic.h>
+
+ sound@13c00000 {
+ #sound-dai-cells = <1>;
+ #clock-cells = <0>;
+ compatible = "renesas,r9a09g047-sound";
+ reg = <0x13c00000 0x10000>,
+ <0x13c20000 0x10000>,
+ <0x13c30000 0x1000>,
+ <0x13c31000 0x1f000>,
+ <0x13c50000 0x10000>;
+ reg-names = "scu", "adg", "ssiu", "ssi", "audmapp";
+ clocks = <&cpg 245>,
+ <&cpg 385>, <&cpg 386>,
+ <&cpg 387>, <&cpg 388>,
+ <&cpg 389>, <&cpg 390>,
+ <&cpg 391>, <&cpg 392>,
+ <&cpg 393>, <&cpg 394>,
+ <&cpg 372>, <&cpg 373>,
+ <&cpg 374>, <&cpg 375>,
+ <&cpg 376>, <&cpg 377>,
+ <&cpg 378>, <&cpg 379>,
+ <&cpg 380>, <&cpg 381>,
+ <&cpg 370>, <&cpg 371>,
+ <&cpg 370>, <&cpg 371>,
+ <&cpg 368>, <&cpg 369>,
+ <&cpg 251>, <&cpg 252>,
+ <&cpg 253>, <&cpg 250>,
+ <&cpg 384>,
+ <&cpg 246>, <&cpg 247>,
+ <&cpg 382>,
+ <&cpg 352>, <&cpg 353>,
+ <&cpg 354>, <&cpg 355>,
+ <&cpg 356>, <&cpg 357>,
+ <&cpg 358>, <&cpg 359>,
+ <&cpg 360>, <&cpg 361>,
+ <&cpg 248>, <&cpg 249>;
+ clock-names = "ssi-all",
+ "ssi-0", "ssi-1",
+ "ssi-2", "ssi-3",
+ "ssi-4", "ssi-5",
+ "ssi-6", "ssi-7",
+ "ssi-8", "ssi-9",
+ "src-0", "src-1",
+ "src-2", "src-3",
+ "src-4", "src-5",
+ "src-6", "src-7",
+ "src-8", "src-9",
+ "mix-0", "mix-1",
+ "ctu-0", "ctu-1",
+ "dvc-0", "dvc-1",
+ "audio-clka", "audio-clkb",
+ "audio-clkc", "audio-clki",
+ "ssif_supply",
+ "scu", "scu_x2",
+ "scu_supply",
+ "adg-ssi-0", "adg-ssi-1",
+ "adg-ssi-2", "adg-ssi-3",
+ "adg-ssi-4", "adg-ssi-5",
+ "adg-ssi-6", "adg-ssi-7",
+ "adg-ssi-8", "adg-ssi-9",
+ "audmapp", "adg";
+ power-domains = <&cpg>;
+ resets = <&cpg 225>,
+ <&cpg 226>, <&cpg 227>,
+ <&cpg 228>, <&cpg 229>,
+ <&cpg 230>, <&cpg 231>,
+ <&cpg 232>, <&cpg 233>,
+ <&cpg 234>, <&cpg 235>,
+ <&cpg 236>, <&cpg 238>, <&cpg 237>;
+ reset-names = "ssi-all",
+ "ssi-0", "ssi-1",
+ "ssi-2", "ssi-3",
+ "ssi-4", "ssi-5",
+ "ssi-6", "ssi-7",
+ "ssi-8", "ssi-9",
+ "scu", "adg",
+ "audmapp";
+
+ ctu {
+ ctu-0 { };
+ ctu-1 { };
+ ctu-2 { };
+ ctu-3 { };
+ ctu-4 { };
+ ctu-5 { };
+ ctu-6 { };
+ ctu-7 { };
+ };
+
+ dvc {
+ dvc0: dvc-0 {
+ dmas = <&dmac0 0x1db3>, <&dmac1 0x1db3>,
+ <&dmac2 0x1db3>, <&dmac3 0x1db3>,
+ <&dmac4 0x1db3>;
+ dma-names = "tx", "tx", "tx", "tx", "tx";
+ };
+ dvc1: dvc-1 {
+ dmas = <&dmac0 0x1db4>, <&dmac1 0x1db4>,
+ <&dmac2 0x1db4>, <&dmac3 0x1db4>,
+ <&dmac4 0x1db4>;
+ dma-names = "tx", "tx", "tx", "tx", "tx";
+ };
+ };
+
+ mix {
+ mix-0 { };
+ mix-1 { };
+ };
+
+ src {
+ src0: src-0 {
+ interrupts = <GIC_SPI 902 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1d9f>, <&dmac0 0x1da9>,
+ <&dmac1 0x1d9f>, <&dmac1 0x1da9>,
+ <&dmac2 0x1d9f>, <&dmac2 0x1da9>,
+ <&dmac3 0x1d9f>, <&dmac3 0x1da9>,
+ <&dmac4 0x1d9f>, <&dmac4 0x1da9>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src1: src-1 {
+ interrupts = <GIC_SPI 903 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da0>, <&dmac0 0x1daa>,
+ <&dmac1 0x1da0>, <&dmac1 0x1daa>,
+ <&dmac2 0x1da0>, <&dmac2 0x1daa>,
+ <&dmac3 0x1da0>, <&dmac3 0x1daa>,
+ <&dmac4 0x1da0>, <&dmac4 0x1daa>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src-2 {
+ interrupts = <GIC_SPI 904 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da1>, <&dmac0 0x1dab>,
+ <&dmac1 0x1da1>, <&dmac1 0x1dab>,
+ <&dmac2 0x1da1>, <&dmac2 0x1dab>,
+ <&dmac3 0x1da1>, <&dmac3 0x1dab>,
+ <&dmac4 0x1da1>, <&dmac4 0x1dab>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src-3 {
+ interrupts = <GIC_SPI 905 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da2>, <&dmac0 0x1dac>,
+ <&dmac1 0x1da2>, <&dmac1 0x1dac>,
+ <&dmac2 0x1da2>, <&dmac2 0x1dac>,
+ <&dmac3 0x1da2>, <&dmac3 0x1dac>,
+ <&dmac4 0x1da2>, <&dmac4 0x1dac>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src-4 {
+ interrupts = <GIC_SPI 906 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da3>, <&dmac0 0x1dad>,
+ <&dmac1 0x1da3>, <&dmac1 0x1dad>,
+ <&dmac2 0x1da3>, <&dmac2 0x1dad>,
+ <&dmac3 0x1da3>, <&dmac3 0x1dad>,
+ <&dmac4 0x1da3>, <&dmac4 0x1dad>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src-5 {
+ interrupts = <GIC_SPI 907 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da4>, <&dmac0 0x1dae>,
+ <&dmac1 0x1da4>, <&dmac1 0x1dae>,
+ <&dmac2 0x1da4>, <&dmac2 0x1dae>,
+ <&dmac3 0x1da4>, <&dmac3 0x1dae>,
+ <&dmac4 0x1da4>, <&dmac4 0x1dae>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src-6 {
+ interrupts = <GIC_SPI 908 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da5>, <&dmac0 0x1daf>,
+ <&dmac1 0x1da5>, <&dmac1 0x1daf>,
+ <&dmac2 0x1da5>, <&dmac2 0x1daf>,
+ <&dmac3 0x1da5>, <&dmac3 0x1daf>,
+ <&dmac4 0x1da5>, <&dmac4 0x1daf>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src-7 {
+ interrupts = <GIC_SPI 909 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da6>, <&dmac0 0x1db0>,
+ <&dmac1 0x1da6>, <&dmac1 0x1db0>,
+ <&dmac2 0x1da6>, <&dmac2 0x1db0>,
+ <&dmac3 0x1da6>, <&dmac3 0x1db0>,
+ <&dmac4 0x1da6>, <&dmac4 0x1db0>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src-8 {
+ interrupts = <GIC_SPI 910 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da7>, <&dmac0 0x1db1>,
+ <&dmac1 0x1da7>, <&dmac1 0x1db1>,
+ <&dmac2 0x1da7>, <&dmac2 0x1db1>,
+ <&dmac3 0x1da7>, <&dmac3 0x1db1>,
+ <&dmac4 0x1da7>, <&dmac4 0x1db1>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ src-9 {
+ interrupts = <GIC_SPI 911 IRQ_TYPE_LEVEL_HIGH>;
+ dmas = <&dmac0 0x1da8>, <&dmac0 0x1db2>,
+ <&dmac1 0x1da8>, <&dmac1 0x1db2>,
+ <&dmac2 0x1da8>, <&dmac2 0x1db2>,
+ <&dmac3 0x1da8>, <&dmac3 0x1db2>,
+ <&dmac4 0x1da8>, <&dmac4 0x1db2>;
+ dma-names = "rx", "tx", "rx", "tx", "rx", "tx",
+ "rx", "tx", "rx", "tx";
+ };
+ };
+
+ ssi {
+ ssi-0 {
+ interrupts = <GIC_SPI 889 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ ssi-1 {
+ interrupts = <GIC_SPI 890 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ ssi-2 {
+ interrupts = <GIC_SPI 891 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ ssi3: ssi-3 {
+ interrupts = <GIC_SPI 892 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ ssi4: ssi-4 {
+ interrupts = <GIC_SPI 893 IRQ_TYPE_LEVEL_HIGH>;
+ shared-pin;
+ };
+ ssi-5 {
+ interrupts = <GIC_SPI 894 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ ssi-6 {
+ interrupts = <GIC_SPI 895 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ ssi-7 {
+ interrupts = <GIC_SPI 896 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ ssi-8 {
+ interrupts = <GIC_SPI 897 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ ssi-9 {
+ interrupts = <GIC_SPI 898 IRQ_TYPE_LEVEL_HIGH>;
+ };
+ };
+
+ ssiu {
+ ssiu-0 {
+ dmas = <&dmac0 0x1d61>, <&dmac0 0x1d62>,
+ <&dmac1 0x1d61>, <&dmac1 0x1d62>,
+ <&dmac2 0x1d61>, <&dmac2 0x1d62>,
+ <&dmac3 0x1d61>, <&dmac3 0x1d62>,
+ <&dmac4 0x1d61>, <&dmac4 0x1d62>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-1 {
+ dmas = <&dmac0 0x1d63>, <&dmac0 0x1d64>,
+ <&dmac1 0x1d63>, <&dmac1 0x1d64>,
+ <&dmac2 0x1d63>, <&dmac2 0x1d64>,
+ <&dmac3 0x1d63>, <&dmac3 0x1d64>,
+ <&dmac4 0x1d63>, <&dmac4 0x1d64>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-2 {
+ dmas = <&dmac0 0x1d65>, <&dmac0 0x1d66>,
+ <&dmac1 0x1d65>, <&dmac1 0x1d66>,
+ <&dmac2 0x1d65>, <&dmac2 0x1d66>,
+ <&dmac3 0x1d65>, <&dmac3 0x1d66>,
+ <&dmac4 0x1d65>, <&dmac4 0x1d66>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-3 {
+ dmas = <&dmac0 0x1d67>, <&dmac0 0x1d68>,
+ <&dmac1 0x1d67>, <&dmac1 0x1d68>,
+ <&dmac2 0x1d67>, <&dmac2 0x1d68>,
+ <&dmac3 0x1d67>, <&dmac3 0x1d68>,
+ <&dmac4 0x1d67>, <&dmac4 0x1d68>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-4 {
+ dmas = <&dmac0 0x1d69>, <&dmac0 0x1d6a>,
+ <&dmac1 0x1d69>, <&dmac1 0x1d6a>,
+ <&dmac2 0x1d69>, <&dmac2 0x1d6a>,
+ <&dmac3 0x1d69>, <&dmac3 0x1d6a>,
+ <&dmac4 0x1d69>, <&dmac4 0x1d6a>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-5 {
+ dmas = <&dmac0 0x1d6b>, <&dmac0 0x1d6c>,
+ <&dmac1 0x1d6b>, <&dmac1 0x1d6c>,
+ <&dmac2 0x1d6b>, <&dmac2 0x1d6c>,
+ <&dmac3 0x1d6b>, <&dmac3 0x1d6c>,
+ <&dmac4 0x1d6b>, <&dmac4 0x1d6c>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-6 {
+ dmas = <&dmac0 0x1d6d>, <&dmac0 0x1d6e>,
+ <&dmac1 0x1d6d>, <&dmac1 0x1d6e>,
+ <&dmac2 0x1d6d>, <&dmac2 0x1d6e>,
+ <&dmac3 0x1d6d>, <&dmac3 0x1d6e>,
+ <&dmac4 0x1d6d>, <&dmac4 0x1d6e>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-7 {
+ dmas = <&dmac0 0x1d6f>, <&dmac0 0x1d70>,
+ <&dmac1 0x1d6f>, <&dmac1 0x1d70>,
+ <&dmac2 0x1d6f>, <&dmac2 0x1d70>,
+ <&dmac3 0x1d6f>, <&dmac3 0x1d70>,
+ <&dmac4 0x1d6f>, <&dmac4 0x1d70>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-8 {
+ dmas = <&dmac0 0x1d71>, <&dmac0 0x1d72>,
+ <&dmac1 0x1d71>, <&dmac1 0x1d72>,
+ <&dmac2 0x1d71>, <&dmac2 0x1d72>,
+ <&dmac3 0x1d71>, <&dmac3 0x1d72>,
+ <&dmac4 0x1d71>, <&dmac4 0x1d72>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-9 {
+ dmas = <&dmac0 0x1d73>, <&dmac0 0x1d74>,
+ <&dmac1 0x1d73>, <&dmac1 0x1d74>,
+ <&dmac2 0x1d73>, <&dmac2 0x1d74>,
+ <&dmac3 0x1d73>, <&dmac3 0x1d74>,
+ <&dmac4 0x1d73>, <&dmac4 0x1d74>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-10 {
+ dmas = <&dmac0 0x1d75>, <&dmac0 0x1d76>,
+ <&dmac1 0x1d75>, <&dmac1 0x1d76>,
+ <&dmac2 0x1d75>, <&dmac2 0x1d76>,
+ <&dmac3 0x1d75>, <&dmac3 0x1d76>,
+ <&dmac4 0x1d75>, <&dmac4 0x1d76>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-11 {
+ dmas = <&dmac0 0x1d77>, <&dmac0 0x1d78>,
+ <&dmac1 0x1d77>, <&dmac1 0x1d78>,
+ <&dmac2 0x1d77>, <&dmac2 0x1d78>,
+ <&dmac3 0x1d77>, <&dmac3 0x1d78>,
+ <&dmac4 0x1d77>, <&dmac4 0x1d78>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-12 {
+ dmas = <&dmac0 0x1d79>, <&dmac0 0x1d7a>,
+ <&dmac1 0x1d79>, <&dmac1 0x1d7a>,
+ <&dmac2 0x1d79>, <&dmac2 0x1d7a>,
+ <&dmac3 0x1d79>, <&dmac3 0x1d7a>,
+ <&dmac4 0x1d79>, <&dmac4 0x1d7a>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-13 {
+ dmas = <&dmac0 0x1d7b>, <&dmac0 0x1d7c>,
+ <&dmac1 0x1d7b>, <&dmac1 0x1d7c>,
+ <&dmac2 0x1d7b>, <&dmac2 0x1d7c>,
+ <&dmac3 0x1d7b>, <&dmac3 0x1d7c>,
+ <&dmac4 0x1d7b>, <&dmac4 0x1d7c>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-14 {
+ dmas = <&dmac0 0x1d7d>, <&dmac0 0x1d7e>,
+ <&dmac1 0x1d7d>, <&dmac1 0x1d7e>,
+ <&dmac2 0x1d7d>, <&dmac2 0x1d7e>,
+ <&dmac3 0x1d7d>, <&dmac3 0x1d7e>,
+ <&dmac4 0x1d7d>, <&dmac4 0x1d7e>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-15 {
+ dmas = <&dmac0 0x1d7f>, <&dmac0 0x1d80>,
+ <&dmac1 0x1d7f>, <&dmac1 0x1d80>,
+ <&dmac2 0x1d7f>, <&dmac2 0x1d80>,
+ <&dmac3 0x1d7f>, <&dmac3 0x1d80>,
+ <&dmac4 0x1d7f>, <&dmac4 0x1d80>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-16 {
+ dmas = <&dmac0 0x1d81>, <&dmac0 0x1d82>,
+ <&dmac1 0x1d81>, <&dmac1 0x1d82>,
+ <&dmac2 0x1d81>, <&dmac2 0x1d82>,
+ <&dmac3 0x1d81>, <&dmac3 0x1d82>,
+ <&dmac4 0x1d81>, <&dmac4 0x1d82>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-17 {
+ dmas = <&dmac0 0x1d83>, <&dmac0 0x1d84>,
+ <&dmac1 0x1d83>, <&dmac1 0x1d84>,
+ <&dmac2 0x1d83>, <&dmac2 0x1d84>,
+ <&dmac3 0x1d83>, <&dmac3 0x1d84>,
+ <&dmac4 0x1d83>, <&dmac4 0x1d84>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-18 {
+ dmas = <&dmac0 0x1d85>, <&dmac0 0x1d86>,
+ <&dmac1 0x1d85>, <&dmac1 0x1d86>,
+ <&dmac2 0x1d85>, <&dmac2 0x1d86>,
+ <&dmac3 0x1d85>, <&dmac3 0x1d86>,
+ <&dmac4 0x1d85>, <&dmac4 0x1d86>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-19 {
+ dmas = <&dmac0 0x1d87>, <&dmac0 0x1d88>,
+ <&dmac1 0x1d87>, <&dmac1 0x1d88>,
+ <&dmac2 0x1d87>, <&dmac2 0x1d88>,
+ <&dmac3 0x1d87>, <&dmac3 0x1d88>,
+ <&dmac4 0x1d87>, <&dmac4 0x1d88>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-20 {
+ dmas = <&dmac0 0x1d89>, <&dmac0 0x1d8a>,
+ <&dmac1 0x1d89>, <&dmac1 0x1d8a>,
+ <&dmac2 0x1d89>, <&dmac2 0x1d8a>,
+ <&dmac3 0x1d89>, <&dmac3 0x1d8a>,
+ <&dmac4 0x1d89>, <&dmac4 0x1d8a>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-21 {
+ dmas = <&dmac0 0x1d8b>, <&dmac0 0x1d8c>,
+ <&dmac1 0x1d8b>, <&dmac1 0x1d8c>,
+ <&dmac2 0x1d8b>, <&dmac2 0x1d8c>,
+ <&dmac3 0x1d8b>, <&dmac3 0x1d8c>,
+ <&dmac4 0x1d8b>, <&dmac4 0x1d8c>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-22 {
+ dmas = <&dmac0 0x1d8d>, <&dmac0 0x1d8e>,
+ <&dmac1 0x1d8d>, <&dmac1 0x1d8e>,
+ <&dmac2 0x1d8d>, <&dmac2 0x1d8e>,
+ <&dmac3 0x1d8d>, <&dmac3 0x1d8e>,
+ <&dmac4 0x1d8d>, <&dmac4 0x1d8e>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-23 {
+ dmas = <&dmac0 0x1d8f>, <&dmac0 0x1d90>,
+ <&dmac1 0x1d8f>, <&dmac1 0x1d90>,
+ <&dmac2 0x1d8f>, <&dmac2 0x1d90>,
+ <&dmac3 0x1d8f>, <&dmac3 0x1d90>,
+ <&dmac4 0x1d8f>, <&dmac4 0x1d90>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-24 {
+ dmas = <&dmac0 0x1d91>, <&dmac0 0x1d92>,
+ <&dmac1 0x1d91>, <&dmac1 0x1d92>,
+ <&dmac2 0x1d91>, <&dmac2 0x1d92>,
+ <&dmac3 0x1d91>, <&dmac3 0x1d92>,
+ <&dmac4 0x1d91>, <&dmac4 0x1d92>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-25 {
+ dmas = <&dmac0 0x1d93>, <&dmac0 0x1d94>,
+ <&dmac1 0x1d93>, <&dmac1 0x1d94>,
+ <&dmac2 0x1d93>, <&dmac2 0x1d94>,
+ <&dmac3 0x1d93>, <&dmac3 0x1d94>,
+ <&dmac4 0x1d93>, <&dmac4 0x1d94>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-26 {
+ dmas = <&dmac0 0x1d95>, <&dmac0 0x1d96>,
+ <&dmac1 0x1d95>, <&dmac1 0x1d96>,
+ <&dmac2 0x1d95>, <&dmac2 0x1d96>,
+ <&dmac3 0x1d95>, <&dmac3 0x1d96>,
+ <&dmac4 0x1d95>, <&dmac4 0x1d96>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ ssiu-27 {
+ dmas = <&dmac0 0x1d97>, <&dmac0 0x1d98>,
+ <&dmac1 0x1d97>, <&dmac1 0x1d98>,
+ <&dmac2 0x1d97>, <&dmac2 0x1d98>,
+ <&dmac3 0x1d97>, <&dmac3 0x1d98>,
+ <&dmac4 0x1d97>, <&dmac4 0x1d98>;
+ dma-names = "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx", "tx", "rx";
+ };
+ };
+
+ ports {
+ #address-cells = <1>;
+ #size-cells = <0>;
+ port@0 {
+ reg = <0>;
+ rsnd_endpoint0: endpoint {
+ remote-endpoint = <&codec_endpoint>;
+ dai-format = "i2s";
+ bitclock-master = <&rsnd_endpoint0>;
+ frame-master = <&rsnd_endpoint0>;
+ playback = <&ssi3>, <&src1>, <&dvc1>;
+ capture = <&ssi4>, <&src0>, <&dvc0>;
+ };
+ };
+ };
+ };
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 01/18] ASoC: dt-bindings: sound: Add DT binding for RZ/G3E sound
2026-05-25 11:02 ` [PATCH v7 01/18] ASoC: dt-bindings: sound: Add DT binding for RZ/G3E sound John Madieu
@ 2026-05-25 11:23 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 11:23 UTC (permalink / raw)
To: John Madieu; +Cc: conor+dt, devicetree, robh, wsa+renesas
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Low] Missing `required` constraints for `interrupts`, `dmas`, and `dma-names` in `src` pattern properties.
- [Low] Top-level `"#address-cells"` and `"#size-cells"` are defined but unused and invalid for the direct child nodes.
--
commit 608c00e35f32aa411be040aeecd6e9cb09cb8264
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: dt-bindings: sound: Add DT binding for RZ/G3E sound
Add a standalone device tree binding for the Renesas RZ/G3E (R9A09G047)
sound controller.
> diff --git a/Documentation/devicetree/bindings/sound/renesas,r9a09g047-sound.yaml b/Documentation/devicetree/bindings/sound/renesas,r9a09g047-sound.yaml
> new file mode 100644
> index 0000000000000..d7fa165546985
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/sound/renesas,r9a09g047-sound.yaml
[ ... ]
> + "#sound-dai-cells":
> + const: 1
> +
> + "#clock-cells":
> + const: 0
> +
> + "#address-cells":
> + const: 1
> +
> + "#size-cells":
> + const: 0
> +
> + clocks:
> + maxItems: 47
[Severity: Low]
Are these top-level address and size cells necessary?
It appears no direct child nodes of the sound controller (such as ctu, dvc,
mix, src, ssi, or ssiu) use unit addresses or reg properties. The ports
node correctly defines its own cells for its children. Leaving these at the
top level might lead to dtc warnings about avoiding empty address-cells
if device tree writers blindly include them.
[ ... ]
> + src:
> + type: object
> + additionalProperties: false
> +
> + patternProperties:
> + "^src-[0-9]$":
> + type: object
> + additionalProperties: false
> +
> + properties:
> + interrupts:
> + maxItems: 1
> +
> + dmas:
> + maxItems: 10
> + description:
> + Must contain a list of pairs of references to DMA
> + specifiers, one for transmission and one for reception,
> + repeated for each DMA controller. The dma-engine core
> + falls through the list to find a free channel.
> +
> + dma-names:
> + maxItems: 10
> + allOf:
> + - items:
> + enum:
> + - tx
> + - rx
[Severity: Low]
Should there be a required list for interrupts, dmas, and dma-names here?
The schema currently allows src subnodes to pass validation even if they
are missing the interrupts property. However, it looks like the driver
strictly requires an interrupt to be present for an active src node, and
it will fail to probe without it. Should the schema enforce these
properties for src subnodes similar to how it does for ssi subnodes?
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=1
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 02/18] ASoC: rsnd: Fix RSND_SOC_MASK width to single nibble
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
2026-05-25 11:02 ` [PATCH v7 01/18] ASoC: dt-bindings: sound: Add DT binding for RZ/G3E sound John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:02 ` [PATCH v7 03/18] ASoC: rsnd: Add reset controller support to rsnd_mod John Madieu
` (16 subsequent siblings)
18 siblings, 0 replies; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
RSND_SOC_MASK was defined as (0xFF << 4), spanning bits 4-11. This is
wider than needed since only nibble B (bits 7:4) is used for SoC
identifiers. Narrow it to (0xF << 4) to match the intended single-nibble
allocation and prevent overlap with bits 8-11 which will be used by
upcoming RZ series flags.
No functional change, since the only current user (RSND_SOC_E) fits
within a single nibble.
Fixes: ba164a49f8f7 ("ASoC: rsnd: src: Avoid a potential deadlock")
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: No changes
v6: No changes
v5:
- New patch. Extracted as a standalone bug-fix patch per
Kuninori's request (previously embedded in patch 04/12 of v4).
- Add Fixes: tag referencing the commit that introduced the mask.
sound/soc/renesas/rcar/rsnd.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/sound/soc/renesas/rcar/rsnd.h b/sound/soc/renesas/rcar/rsnd.h
index 04c70690f7a2..3e666125959b 100644
--- a/sound/soc/renesas/rcar/rsnd.h
+++ b/sound/soc/renesas/rcar/rsnd.h
@@ -624,7 +624,7 @@ struct rsnd_priv {
#define RSND_GEN2 (2 << 0)
#define RSND_GEN3 (3 << 0)
#define RSND_GEN4 (4 << 0)
-#define RSND_SOC_MASK (0xFF << 4)
+#define RSND_SOC_MASK (0xF << 4)
#define RSND_SOC_E (1 << 4) /* E1/E2/E3 */
/*
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* [PATCH v7 03/18] ASoC: rsnd: Add reset controller support to rsnd_mod
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
2026-05-25 11:02 ` [PATCH v7 01/18] ASoC: dt-bindings: sound: Add DT binding for RZ/G3E sound John Madieu
2026-05-25 11:02 ` [PATCH v7 02/18] ASoC: rsnd: Fix RSND_SOC_MASK width to single nibble John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:02 ` [PATCH v7 04/18] ASoC: rsnd: Support hyphen or dot in indexed clock and reset names John Madieu
` (15 subsequent siblings)
18 siblings, 0 replies; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
The RZ/G3E SoC requires per-module reset control for the audio subsystem.
Add reset controller support to struct rsnd_mod and update rsnd_mod_init()
to accept and handle a reset_control parameter and mirror it in
rsnd_mod_quit().
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7:
- rsnd_mod_quit(): assert the reset before unpreparing the clock,
mirroring rsnd_mod_init() which enables the clock before
deasserting the reset, so the teardown is the exact inverse of
init (sashiko-bot).
- Drop the unused <linux/delay.h> include added to core.c
(sashiko-bot).
v6: No changes
v5: No changes
v4:
- Add reset_control_assert() in rsnd_mod_quit() for symmetry with
deassert in rsnd_mod_init().
v3: No changes
v2: No changes
sound/soc/renesas/rcar/adg.c | 2 +-
sound/soc/renesas/rcar/cmd.c | 2 +-
sound/soc/renesas/rcar/core.c | 15 ++++++++++++++-
sound/soc/renesas/rcar/ctu.c | 2 +-
sound/soc/renesas/rcar/dma.c | 4 ++--
sound/soc/renesas/rcar/dvc.c | 2 +-
sound/soc/renesas/rcar/mix.c | 2 +-
sound/soc/renesas/rcar/rsnd.h | 3 +++
sound/soc/renesas/rcar/src.c | 2 +-
sound/soc/renesas/rcar/ssi.c | 2 +-
sound/soc/renesas/rcar/ssiu.c | 2 +-
11 files changed, 27 insertions(+), 11 deletions(-)
diff --git a/sound/soc/renesas/rcar/adg.c b/sound/soc/renesas/rcar/adg.c
index 8641b73d1f77..0105c60a144e 100644
--- a/sound/soc/renesas/rcar/adg.c
+++ b/sound/soc/renesas/rcar/adg.c
@@ -780,7 +780,7 @@ int rsnd_adg_probe(struct rsnd_priv *priv)
return -ENOMEM;
ret = rsnd_mod_init(priv, &adg->mod, &adg_ops,
- NULL, 0, 0);
+ NULL, NULL, 0, 0);
if (ret)
return ret;
diff --git a/sound/soc/renesas/rcar/cmd.c b/sound/soc/renesas/rcar/cmd.c
index 8d9a1e345a22..13beef389797 100644
--- a/sound/soc/renesas/rcar/cmd.c
+++ b/sound/soc/renesas/rcar/cmd.c
@@ -171,7 +171,7 @@ int rsnd_cmd_probe(struct rsnd_priv *priv)
for_each_rsnd_cmd(cmd, priv, i) {
int ret = rsnd_mod_init(priv, rsnd_mod_get(cmd),
- &rsnd_cmd_ops, NULL,
+ &rsnd_cmd_ops, NULL, NULL,
RSND_MOD_CMD, i);
if (ret)
return ret;
diff --git a/sound/soc/renesas/rcar/core.c b/sound/soc/renesas/rcar/core.c
index 2dc078358612..2038e3079a9a 100644
--- a/sound/soc/renesas/rcar/core.c
+++ b/sound/soc/renesas/rcar/core.c
@@ -196,18 +196,29 @@ int rsnd_mod_init(struct rsnd_priv *priv,
struct rsnd_mod *mod,
struct rsnd_mod_ops *ops,
struct clk *clk,
+ struct reset_control *rstc,
enum rsnd_mod_type type,
int id)
{
- int ret = clk_prepare(clk);
+ int ret;
+ ret = clk_prepare_enable(clk);
if (ret)
return ret;
+ ret = reset_control_deassert(rstc);
+ if (ret) {
+ clk_disable_unprepare(clk);
+ return ret;
+ }
+
+ clk_disable(clk);
+
mod->id = id;
mod->ops = ops;
mod->type = type;
mod->clk = clk;
+ mod->rstc = rstc;
mod->priv = priv;
return 0;
@@ -215,6 +226,8 @@ int rsnd_mod_init(struct rsnd_priv *priv,
void rsnd_mod_quit(struct rsnd_mod *mod)
{
+ reset_control_assert(mod->rstc);
+ mod->rstc = NULL;
clk_unprepare(mod->clk);
mod->clk = NULL;
}
diff --git a/sound/soc/renesas/rcar/ctu.c b/sound/soc/renesas/rcar/ctu.c
index bd4c61f9fb3c..81bba6a1af6e 100644
--- a/sound/soc/renesas/rcar/ctu.c
+++ b/sound/soc/renesas/rcar/ctu.c
@@ -360,7 +360,7 @@ int rsnd_ctu_probe(struct rsnd_priv *priv)
}
ret = rsnd_mod_init(priv, rsnd_mod_get(ctu), &rsnd_ctu_ops,
- clk, RSND_MOD_CTU, i);
+ clk, NULL, RSND_MOD_CTU, i);
if (ret)
goto rsnd_ctu_probe_done;
diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
index 2035ce06fe4c..68c859897e68 100644
--- a/sound/soc/renesas/rcar/dma.c
+++ b/sound/soc/renesas/rcar/dma.c
@@ -803,7 +803,7 @@ static int rsnd_dma_alloc(struct rsnd_dai_stream *io, struct rsnd_mod *mod,
*dma_mod = rsnd_mod_get(dma);
- ret = rsnd_mod_init(priv, *dma_mod, ops, NULL,
+ ret = rsnd_mod_init(priv, *dma_mod, ops, NULL, NULL,
type, dma_id);
if (ret < 0)
return ret;
@@ -879,5 +879,5 @@ int rsnd_dma_probe(struct rsnd_priv *priv)
priv->dma = dmac;
/* dummy mem mod for debug */
- return rsnd_mod_init(NULL, &mem, &mem_ops, NULL, 0, 0);
+ return rsnd_mod_init(NULL, &mem, &mem_ops, NULL, NULL, 0, 0);
}
diff --git a/sound/soc/renesas/rcar/dvc.c b/sound/soc/renesas/rcar/dvc.c
index 988cbddbc611..bf7146ceb5f6 100644
--- a/sound/soc/renesas/rcar/dvc.c
+++ b/sound/soc/renesas/rcar/dvc.c
@@ -364,7 +364,7 @@ int rsnd_dvc_probe(struct rsnd_priv *priv)
}
ret = rsnd_mod_init(priv, rsnd_mod_get(dvc), &rsnd_dvc_ops,
- clk, RSND_MOD_DVC, i);
+ clk, NULL, RSND_MOD_DVC, i);
if (ret)
goto rsnd_dvc_probe_done;
diff --git a/sound/soc/renesas/rcar/mix.c b/sound/soc/renesas/rcar/mix.c
index aea74e703305..566e9b2a488c 100644
--- a/sound/soc/renesas/rcar/mix.c
+++ b/sound/soc/renesas/rcar/mix.c
@@ -328,7 +328,7 @@ int rsnd_mix_probe(struct rsnd_priv *priv)
}
ret = rsnd_mod_init(priv, rsnd_mod_get(mix), &rsnd_mix_ops,
- clk, RSND_MOD_MIX, i);
+ clk, NULL, RSND_MOD_MIX, i);
if (ret)
goto rsnd_mix_probe_done;
diff --git a/sound/soc/renesas/rcar/rsnd.h b/sound/soc/renesas/rcar/rsnd.h
index 3e666125959b..3d419b31cf40 100644
--- a/sound/soc/renesas/rcar/rsnd.h
+++ b/sound/soc/renesas/rcar/rsnd.h
@@ -15,6 +15,7 @@
#include <linux/list.h>
#include <linux/module.h>
#include <linux/of.h>
+#include <linux/reset.h>
#include <linux/sh_dma.h>
#include <linux/workqueue.h>
#include <sound/soc.h>
@@ -353,6 +354,7 @@ struct rsnd_mod {
struct rsnd_mod_ops *ops;
struct rsnd_priv *priv;
struct clk *clk;
+ struct reset_control *rstc;
u32 status;
};
/*
@@ -420,6 +422,7 @@ int rsnd_mod_init(struct rsnd_priv *priv,
struct rsnd_mod *mod,
struct rsnd_mod_ops *ops,
struct clk *clk,
+ struct reset_control *rstc,
enum rsnd_mod_type type,
int id);
void rsnd_mod_quit(struct rsnd_mod *mod);
diff --git a/sound/soc/renesas/rcar/src.c b/sound/soc/renesas/rcar/src.c
index 6a3dbc84f474..8b58cc20e7a8 100644
--- a/sound/soc/renesas/rcar/src.c
+++ b/sound/soc/renesas/rcar/src.c
@@ -766,7 +766,7 @@ int rsnd_src_probe(struct rsnd_priv *priv)
}
ret = rsnd_mod_init(priv, rsnd_mod_get(src),
- &rsnd_src_ops, clk, RSND_MOD_SRC, i);
+ &rsnd_src_ops, clk, NULL, RSND_MOD_SRC, i);
if (ret)
goto rsnd_src_probe_done;
diff --git a/sound/soc/renesas/rcar/ssi.c b/sound/soc/renesas/rcar/ssi.c
index 0420041e282c..c06cebb36170 100644
--- a/sound/soc/renesas/rcar/ssi.c
+++ b/sound/soc/renesas/rcar/ssi.c
@@ -1225,7 +1225,7 @@ int rsnd_ssi_probe(struct rsnd_priv *priv)
ops = &rsnd_ssi_dma_ops;
ret = rsnd_mod_init(priv, rsnd_mod_get(ssi), ops, clk,
- RSND_MOD_SSI, i);
+ NULL, RSND_MOD_SSI, i);
if (ret)
goto rsnd_ssi_probe_done;
diff --git a/sound/soc/renesas/rcar/ssiu.c b/sound/soc/renesas/rcar/ssiu.c
index 244fb833292a..0cfa84fe5ea8 100644
--- a/sound/soc/renesas/rcar/ssiu.c
+++ b/sound/soc/renesas/rcar/ssiu.c
@@ -586,7 +586,7 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
}
ret = rsnd_mod_init(priv, rsnd_mod_get(ssiu),
- ops, NULL, RSND_MOD_SSIU, i);
+ ops, NULL, NULL, RSND_MOD_SSIU, i);
if (ret)
return ret;
}
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* [PATCH v7 04/18] ASoC: rsnd: Support hyphen or dot in indexed clock and reset names
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (2 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 03/18] ASoC: rsnd: Add reset controller support to rsnd_mod John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:02 ` [PATCH v7 05/18] ASoC: rsnd: Add RZ/G3E SoC probing and register map John Madieu
` (14 subsequent siblings)
18 siblings, 0 replies; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
The rsnd driver historically looks up per-instance clocks and resets
using dot-separated names matching the ones declared in R-Car device
tree bindings ("ssi.0", "src.0", "adg.ssi.0", ...). The dot separator
is unusual for device tree clock-names / reset-names and newer
Renesas SoC bindings (RZ/G3E and later) use the more standard hyphen
form ("ssi-0", "src-0", ...).
Rather than force every existing R-Car user to rename their DT entries,
add a small set of helpers that try the hyphen form first and fall
back to the dot form. While at it, convert the existing indexed
devm_clk_get() call sites in the SSI, SRC, CTU, DVC and MIX probes to use
the new helpers and drop the now unused per-module name buffers and
NAME_SIZE defines.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7:
- Fix the kernel-doc comment on rsnd_devm_clk_get_indexed(): the
fallback to the dot form triggers on -ENOENT, not ENODEV, to
match the actual PTR_ERR(clk) != -ENOENT check (Mark Brown).
v6: New patch
sound/soc/renesas/rcar/core.c | 67 +++++++++++++++++++++++++++++++++++
sound/soc/renesas/rcar/ctu.c | 7 +---
sound/soc/renesas/rcar/dvc.c | 7 +---
sound/soc/renesas/rcar/mix.c | 7 +---
sound/soc/renesas/rcar/rsnd.h | 19 ++++++++++
sound/soc/renesas/rcar/src.c | 7 +---
sound/soc/renesas/rcar/ssi.c | 7 +---
7 files changed, 91 insertions(+), 30 deletions(-)
diff --git a/sound/soc/renesas/rcar/core.c b/sound/soc/renesas/rcar/core.c
index 2038e3079a9a..a60b7a5d90ad 100644
--- a/sound/soc/renesas/rcar/core.c
+++ b/sound/soc/renesas/rcar/core.c
@@ -1232,6 +1232,73 @@ int rsnd_node_count(struct rsnd_priv *priv, struct device_node *node, char *name
return i;
}
+/*
+ * Build "<base>-<index>" or "<base>.<index>" and try the hyphen form first,
+ * falling back to the dot form if the hyphen form is not present. This lets
+ * the driver accept both the new DT convention ("ssi-0", "src-0", ...) and
+ * the legacy R-Car convention ("ssi.0", "src.0", ...) transparently.
+ *
+ * @base: name prefix ("ssi", "src", "ctu", "mix", "dvc", "adg.ssi", ...)
+ * @index: integer suffix
+ *
+ * On -ENOENT from the hyphen form, the dot form is tried. All other errors
+ * (including -EPROBE_DEFER) are returned to the caller unchanged, so
+ * behaviour against the clock and reset frameworks is preserved.
+ */
+#define RSND_INDEXED_NAME_MAX 32
+
+static void rsnd_format_indexed_name(char *buf, size_t buflen, char sep,
+ const char *base, int index)
+{
+ snprintf(buf, buflen, "%s%c%d", base, sep, index);
+}
+
+struct clk *rsnd_devm_clk_get_indexed(struct device *dev,
+ const char *base, int index)
+{
+ char name[RSND_INDEXED_NAME_MAX];
+ struct clk *clk;
+
+ rsnd_format_indexed_name(name, sizeof(name), '-', base, index);
+ clk = devm_clk_get(dev, name);
+ if (!IS_ERR(clk) || PTR_ERR(clk) != -ENOENT)
+ return clk;
+
+ rsnd_format_indexed_name(name, sizeof(name), '.', base, index);
+ return devm_clk_get(dev, name);
+}
+
+struct clk *rsnd_devm_clk_get_optional_indexed(struct device *dev,
+ const char *base, int index)
+{
+ char name[RSND_INDEXED_NAME_MAX];
+ struct clk *clk;
+
+ rsnd_format_indexed_name(name, sizeof(name), '-', base, index);
+ clk = devm_clk_get_optional(dev, name);
+ if (IS_ERR(clk) || clk)
+ return clk;
+
+ rsnd_format_indexed_name(name, sizeof(name), '.', base, index);
+ return devm_clk_get_optional(dev, name);
+}
+
+struct reset_control *
+rsnd_devm_reset_control_get_optional_indexed(struct device *dev,
+ const char *base, int index)
+{
+ char name[RSND_INDEXED_NAME_MAX];
+ struct reset_control *rstc;
+
+ rsnd_format_indexed_name(name, sizeof(name), '-', base, index);
+ rstc = devm_reset_control_get_optional(dev, name);
+ if (IS_ERR(rstc) || rstc)
+ return rstc;
+
+ rsnd_format_indexed_name(name, sizeof(name), '.', base, index);
+ return devm_reset_control_get_optional(dev, name);
+}
+
static struct device_node*
rsnd_pick_endpoint_node_for_ports(struct device_node *e_ports,
struct device_node *e_port)
diff --git a/sound/soc/renesas/rcar/ctu.c b/sound/soc/renesas/rcar/ctu.c
index 81bba6a1af6e..293b0eec1ded 100644
--- a/sound/soc/renesas/rcar/ctu.c
+++ b/sound/soc/renesas/rcar/ctu.c
@@ -6,7 +6,6 @@
#include "rsnd.h"
-#define CTU_NAME_SIZE 16
#define CTU_NAME "ctu"
/*
@@ -319,7 +318,6 @@ int rsnd_ctu_probe(struct rsnd_priv *priv)
struct device *dev = rsnd_priv_to_dev(priv);
struct rsnd_ctu *ctu;
struct clk *clk;
- char name[CTU_NAME_SIZE];
int i, nr, ret;
node = rsnd_ctu_of_node(priv);
@@ -350,10 +348,7 @@ int rsnd_ctu_probe(struct rsnd_priv *priv)
* CTU00, CTU01, CTU02, CTU03 => CTU0
* CTU10, CTU11, CTU12, CTU13 => CTU1
*/
- snprintf(name, CTU_NAME_SIZE, "%s.%d",
- CTU_NAME, i / 4);
-
- clk = devm_clk_get(dev, name);
+ clk = rsnd_devm_clk_get_indexed(dev, CTU_NAME, i / 4);
if (IS_ERR(clk)) {
ret = PTR_ERR(clk);
goto rsnd_ctu_probe_done;
diff --git a/sound/soc/renesas/rcar/dvc.c b/sound/soc/renesas/rcar/dvc.c
index bf7146ceb5f6..26f80d542da8 100644
--- a/sound/soc/renesas/rcar/dvc.c
+++ b/sound/soc/renesas/rcar/dvc.c
@@ -29,7 +29,6 @@
#include "rsnd.h"
-#define RSND_DVC_NAME_SIZE 16
#define DVC_NAME "dvc"
@@ -327,7 +326,6 @@ int rsnd_dvc_probe(struct rsnd_priv *priv)
struct device *dev = rsnd_priv_to_dev(priv);
struct rsnd_dvc *dvc;
struct clk *clk;
- char name[RSND_DVC_NAME_SIZE];
int i, nr, ret;
node = rsnd_dvc_of_node(priv);
@@ -354,10 +352,7 @@ int rsnd_dvc_probe(struct rsnd_priv *priv)
for_each_child_of_node_scoped(node, np) {
dvc = rsnd_dvc_get(priv, i);
- snprintf(name, RSND_DVC_NAME_SIZE, "%s.%d",
- DVC_NAME, i);
-
- clk = devm_clk_get(dev, name);
+ clk = rsnd_devm_clk_get_indexed(dev, DVC_NAME, i);
if (IS_ERR(clk)) {
ret = PTR_ERR(clk);
goto rsnd_dvc_probe_done;
diff --git a/sound/soc/renesas/rcar/mix.c b/sound/soc/renesas/rcar/mix.c
index 566e9b2a488c..9ffa591aa4a4 100644
--- a/sound/soc/renesas/rcar/mix.c
+++ b/sound/soc/renesas/rcar/mix.c
@@ -32,7 +32,6 @@
#include "rsnd.h"
-#define MIX_NAME_SIZE 16
#define MIX_NAME "mix"
struct rsnd_mix {
@@ -291,7 +290,6 @@ int rsnd_mix_probe(struct rsnd_priv *priv)
struct device *dev = rsnd_priv_to_dev(priv);
struct rsnd_mix *mix;
struct clk *clk;
- char name[MIX_NAME_SIZE];
int i, nr, ret;
node = rsnd_mix_of_node(priv);
@@ -318,10 +316,7 @@ int rsnd_mix_probe(struct rsnd_priv *priv)
for_each_child_of_node_scoped(node, np) {
mix = rsnd_mix_get(priv, i);
- snprintf(name, MIX_NAME_SIZE, "%s.%d",
- MIX_NAME, i);
-
- clk = devm_clk_get(dev, name);
+ clk = rsnd_devm_clk_get_indexed(dev, MIX_NAME, i);
if (IS_ERR(clk)) {
ret = PTR_ERR(clk);
goto rsnd_mix_probe_done;
diff --git a/sound/soc/renesas/rcar/rsnd.h b/sound/soc/renesas/rcar/rsnd.h
index 3d419b31cf40..f3f1ad1180f8 100644
--- a/sound/soc/renesas/rcar/rsnd.h
+++ b/sound/soc/renesas/rcar/rsnd.h
@@ -476,6 +476,25 @@ int rsnd_runtime_is_multi_ssi(struct rsnd_dai_stream *io);
int rsnd_runtime_is_tdm(struct rsnd_dai_stream *io);
int rsnd_runtime_is_tdm_split(struct rsnd_dai_stream *io);
+/*
+ * Indexed clock and reset name helpers.
+ *
+ * Historically the rsnd driver has looked up per-instance clocks and
+ * resets using dot-separated names (e.g. "ssi.0", "src.0", "adg.ssi.0").
+ * Newer Renesas SoC bindings (RZ/G3E and later) use hyphen-separated
+ * names ("ssi-0", "src-0", ...) to follow the standard Device Tree
+ * naming convention. These helpers look up the hyphenated name first
+ * and transparently fall back to the dotted name, so a single driver
+ * build supports both conventions.
+ */
+struct clk *rsnd_devm_clk_get_indexed(struct device *dev,
+ const char *base, int index);
+struct clk *rsnd_devm_clk_get_optional_indexed(struct device *dev,
+ const char *base, int index);
+struct reset_control *
+rsnd_devm_reset_control_get_optional_indexed(struct device *dev,
+ const char *base, int index);
+
/*
* DT
*/
diff --git a/sound/soc/renesas/rcar/src.c b/sound/soc/renesas/rcar/src.c
index 8b58cc20e7a8..43abe13137bf 100644
--- a/sound/soc/renesas/rcar/src.c
+++ b/sound/soc/renesas/rcar/src.c
@@ -39,7 +39,6 @@ struct rsnd_src {
int irq;
};
-#define RSND_SRC_NAME_SIZE 16
#define rsnd_src_get(priv, id) ((struct rsnd_src *)(priv->src) + id)
#define rsnd_src_nr(priv) ((priv)->src_nr)
@@ -715,7 +714,6 @@ int rsnd_src_probe(struct rsnd_priv *priv)
struct device *dev = rsnd_priv_to_dev(priv);
struct rsnd_src *src;
struct clk *clk;
- char name[RSND_SRC_NAME_SIZE];
int i, nr, ret;
node = rsnd_src_of_node(priv);
@@ -750,16 +748,13 @@ int rsnd_src_probe(struct rsnd_priv *priv)
src = rsnd_src_get(priv, i);
- snprintf(name, RSND_SRC_NAME_SIZE, "%s.%d",
- SRC_NAME, i);
-
src->irq = irq_of_parse_and_map(np, 0);
if (!src->irq) {
ret = -EINVAL;
goto rsnd_src_probe_done;
}
- clk = devm_clk_get(dev, name);
+ clk = rsnd_devm_clk_get_indexed(dev, SRC_NAME, i);
if (IS_ERR(clk)) {
ret = PTR_ERR(clk);
goto rsnd_src_probe_done;
diff --git a/sound/soc/renesas/rcar/ssi.c b/sound/soc/renesas/rcar/ssi.c
index c06cebb36170..cf98cc9ee44c 100644
--- a/sound/soc/renesas/rcar/ssi.c
+++ b/sound/soc/renesas/rcar/ssi.c
@@ -21,7 +21,6 @@
#include <linux/of_irq.h>
#include <linux/delay.h>
#include "rsnd.h"
-#define RSND_SSI_NAME_SIZE 16
/*
* SSICR
@@ -1163,7 +1162,6 @@ int rsnd_ssi_probe(struct rsnd_priv *priv)
struct rsnd_mod_ops *ops;
struct clk *clk;
struct rsnd_ssi *ssi;
- char name[RSND_SSI_NAME_SIZE];
int i, nr, ret;
node = rsnd_ssi_of_node(priv);
@@ -1198,10 +1196,7 @@ int rsnd_ssi_probe(struct rsnd_priv *priv)
ssi = rsnd_ssi_get(priv, i);
- snprintf(name, RSND_SSI_NAME_SIZE, "%s.%d",
- SSI_NAME, i);
-
- clk = devm_clk_get(dev, name);
+ clk = rsnd_devm_clk_get_indexed(dev, SSI_NAME, i);
if (IS_ERR(clk)) {
ret = PTR_ERR(clk);
goto rsnd_ssi_probe_done;
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* [PATCH v7 05/18] ASoC: rsnd: Add RZ/G3E SoC probing and register map
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (3 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 04/18] ASoC: rsnd: Support hyphen or dot in indexed clock and reset names John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 12:23 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 06/18] ASoC: rsnd: Add audmapp clock and reset support for RZ/G3E John Madieu
` (13 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
RZ/G3E audio subsystem has a different register layout compared to
R-Car Gen2/Gen3/Gen4, as described below:
- Different base address organization (SCU, ADG, SSIU, SSI as
separate regions accessed by name)
- Additional registers: AUDIO_CLK_SEL3, SSI_MODE3, SSI_CONTROL2
- Different register offsets within each region
Add RZ/G3E SoC's audio subsystem register layouts and probe support.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: No changes
v6: No changes
v5:
- Simplify flags layout comment per Kuninori's feedback.
- RSND_SOC_MASK fix moved to dedicated patch 02/14.
v4:
- Fix RSND_SOC_MASK to (0xF << 4) to avoid overlap with
RSND_RZ_MASK.
- Add comment documenting flag nibble layout.
v3: No changes
v2: No changes
sound/soc/renesas/rcar/core.c | 1 +
sound/soc/renesas/rcar/gen.c | 180 ++++++++++++++++++++++++++++++++++
sound/soc/renesas/rcar/rsnd.h | 26 ++++-
3 files changed, 204 insertions(+), 3 deletions(-)
diff --git a/sound/soc/renesas/rcar/core.c b/sound/soc/renesas/rcar/core.c
index a60b7a5d90ad..7169d0ef8f90 100644
--- a/sound/soc/renesas/rcar/core.c
+++ b/sound/soc/renesas/rcar/core.c
@@ -106,6 +106,7 @@ static const struct of_device_id rsnd_of_match[] = {
{ .compatible = "renesas,rcar_sound-gen4", .data = (void *)RSND_GEN4 },
/* Special Handling */
{ .compatible = "renesas,rcar_sound-r8a77990", .data = (void *)(RSND_GEN3 | RSND_SOC_E) },
+ { .compatible = "renesas,r9a09g047-sound", .data = (void *)(RSND_RZ3 | RSND_RZG3E) },
{},
};
MODULE_DEVICE_TABLE(of, rsnd_of_match);
diff --git a/sound/soc/renesas/rcar/gen.c b/sound/soc/renesas/rcar/gen.c
index d1f20cde66be..05d5f656fb01 100644
--- a/sound/soc/renesas/rcar/gen.c
+++ b/sound/soc/renesas/rcar/gen.c
@@ -464,6 +464,184 @@ static int rsnd_gen1_probe(struct rsnd_priv *priv)
return ret_adg | ret_ssi;
}
+/*
+ * RZ/G3E Generation
+ */
+static int rsnd_rzg3e_probe(struct rsnd_priv *priv)
+{
+ static const struct rsnd_regmap_field_conf conf_ssiu[] = {
+ RSND_GEN_S_REG(SSI_MODE1, 0x804),
+ RSND_GEN_S_REG(SSI_MODE2, 0x808),
+ RSND_GEN_S_REG(SSI_MODE3, 0x80c),
+ RSND_GEN_S_REG(SSI_CONTROL, 0x810),
+ RSND_GEN_S_REG(SSI_CONTROL2, 0x814),
+ RSND_GEN_S_REG(SSI_SYS_STATUS0, 0x840),
+ RSND_GEN_S_REG(SSI_SYS_STATUS1, 0x844),
+ RSND_GEN_S_REG(SSI_SYS_STATUS2, 0x848),
+ RSND_GEN_S_REG(SSI_SYS_STATUS3, 0x84c),
+ RSND_GEN_S_REG(SSI_SYS_INT_ENABLE0, 0x850),
+ RSND_GEN_S_REG(SSI_SYS_INT_ENABLE1, 0x854),
+ RSND_GEN_S_REG(SSI_SYS_INT_ENABLE2, 0x858),
+ RSND_GEN_S_REG(SSI_SYS_INT_ENABLE3, 0x85c),
+ RSND_GEN_M_REG(SSI_BUSIF0_MODE, 0x0, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF0_ADINR, 0x4, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF0_DALIGN, 0x8, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF1_MODE, 0x20, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF1_ADINR, 0x24, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF1_DALIGN, 0x28, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF2_MODE, 0x40, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF2_ADINR, 0x44, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF2_DALIGN, 0x48, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF3_MODE, 0x60, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF3_ADINR, 0x64, 0x80),
+ RSND_GEN_M_REG(SSI_BUSIF3_DALIGN, 0x68, 0x80),
+ RSND_GEN_M_REG(SSI_MODE, 0xc, 0x80),
+ RSND_GEN_M_REG(SSI_CTRL, 0x10, 0x80),
+ RSND_GEN_M_REG(SSI_INT_ENABLE, 0x18, 0x80),
+ RSND_GEN_S_REG(SSI9_BUSIF0_MODE, 0x480),
+ RSND_GEN_S_REG(SSI9_BUSIF0_ADINR, 0x484),
+ RSND_GEN_S_REG(SSI9_BUSIF0_DALIGN, 0x488),
+ RSND_GEN_S_REG(SSI9_BUSIF1_MODE, 0x4a0),
+ RSND_GEN_S_REG(SSI9_BUSIF1_ADINR, 0x4a4),
+ RSND_GEN_S_REG(SSI9_BUSIF1_DALIGN, 0x4a8),
+ RSND_GEN_S_REG(SSI9_BUSIF2_MODE, 0x4c0),
+ RSND_GEN_S_REG(SSI9_BUSIF2_ADINR, 0x4c4),
+ RSND_GEN_S_REG(SSI9_BUSIF2_DALIGN, 0x4c8),
+ RSND_GEN_S_REG(SSI9_BUSIF3_MODE, 0x4e0),
+ RSND_GEN_S_REG(SSI9_BUSIF3_ADINR, 0x4e4),
+ RSND_GEN_S_REG(SSI9_BUSIF3_DALIGN, 0x4e8),
+ };
+ static const struct rsnd_regmap_field_conf conf_scu[] = {
+ RSND_GEN_M_REG(SRC_I_BUSIF_MODE, 0x0, 0x20),
+ RSND_GEN_M_REG(SRC_O_BUSIF_MODE, 0x4, 0x20),
+ RSND_GEN_M_REG(SRC_BUSIF_DALIGN, 0x8, 0x20),
+ RSND_GEN_M_REG(SRC_ROUTE_MODE0, 0xc, 0x20),
+ RSND_GEN_M_REG(SRC_CTRL, 0x10, 0x20),
+ RSND_GEN_M_REG(SRC_INT_ENABLE0, 0x18, 0x20),
+ RSND_GEN_M_REG(CMD_BUSIF_MODE, 0x184, 0x20),
+ RSND_GEN_M_REG(CMD_BUSIF_DALIGN, 0x188, 0x20),
+ RSND_GEN_M_REG(CMD_ROUTE_SLCT, 0x18c, 0x20),
+ RSND_GEN_M_REG(CMD_CTRL, 0x190, 0x20),
+ RSND_GEN_S_REG(SCU_SYS_STATUS0, 0x1c8),
+ RSND_GEN_S_REG(SCU_SYS_INT_EN0, 0x1cc),
+ RSND_GEN_S_REG(SCU_SYS_STATUS1, 0x1d0),
+ RSND_GEN_S_REG(SCU_SYS_INT_EN1, 0x1d4),
+ RSND_GEN_M_REG(SRC_SWRSR, 0x200, 0x40),
+ RSND_GEN_M_REG(SRC_SRCIR, 0x204, 0x40),
+ RSND_GEN_M_REG(SRC_ADINR, 0x214, 0x40),
+ RSND_GEN_M_REG(SRC_IFSCR, 0x21c, 0x40),
+ RSND_GEN_M_REG(SRC_IFSVR, 0x220, 0x40),
+ RSND_GEN_M_REG(SRC_SRCCR, 0x224, 0x40),
+ RSND_GEN_M_REG(SRC_BSDSR, 0x22c, 0x40),
+ RSND_GEN_M_REG(SRC_BSISR, 0x238, 0x40),
+ RSND_GEN_M_REG(CTU_SWRSR, 0x500, 0x100),
+ RSND_GEN_M_REG(CTU_CTUIR, 0x504, 0x100),
+ RSND_GEN_M_REG(CTU_ADINR, 0x508, 0x100),
+ RSND_GEN_M_REG(CTU_CPMDR, 0x510, 0x100),
+ RSND_GEN_M_REG(CTU_SCMDR, 0x514, 0x100),
+ RSND_GEN_M_REG(CTU_SV00R, 0x518, 0x100),
+ RSND_GEN_M_REG(CTU_SV01R, 0x51c, 0x100),
+ RSND_GEN_M_REG(CTU_SV02R, 0x520, 0x100),
+ RSND_GEN_M_REG(CTU_SV03R, 0x524, 0x100),
+ RSND_GEN_M_REG(CTU_SV04R, 0x528, 0x100),
+ RSND_GEN_M_REG(CTU_SV05R, 0x52c, 0x100),
+ RSND_GEN_M_REG(CTU_SV06R, 0x530, 0x100),
+ RSND_GEN_M_REG(CTU_SV07R, 0x534, 0x100),
+ RSND_GEN_M_REG(CTU_SV10R, 0x538, 0x100),
+ RSND_GEN_M_REG(CTU_SV11R, 0x53c, 0x100),
+ RSND_GEN_M_REG(CTU_SV12R, 0x540, 0x100),
+ RSND_GEN_M_REG(CTU_SV13R, 0x544, 0x100),
+ RSND_GEN_M_REG(CTU_SV14R, 0x548, 0x100),
+ RSND_GEN_M_REG(CTU_SV15R, 0x54c, 0x100),
+ RSND_GEN_M_REG(CTU_SV16R, 0x550, 0x100),
+ RSND_GEN_M_REG(CTU_SV17R, 0x554, 0x100),
+ RSND_GEN_M_REG(CTU_SV20R, 0x558, 0x100),
+ RSND_GEN_M_REG(CTU_SV21R, 0x55c, 0x100),
+ RSND_GEN_M_REG(CTU_SV22R, 0x560, 0x100),
+ RSND_GEN_M_REG(CTU_SV23R, 0x564, 0x100),
+ RSND_GEN_M_REG(CTU_SV24R, 0x568, 0x100),
+ RSND_GEN_M_REG(CTU_SV25R, 0x56c, 0x100),
+ RSND_GEN_M_REG(CTU_SV26R, 0x570, 0x100),
+ RSND_GEN_M_REG(CTU_SV27R, 0x574, 0x100),
+ RSND_GEN_M_REG(CTU_SV30R, 0x578, 0x100),
+ RSND_GEN_M_REG(CTU_SV31R, 0x57c, 0x100),
+ RSND_GEN_M_REG(CTU_SV32R, 0x580, 0x100),
+ RSND_GEN_M_REG(CTU_SV33R, 0x584, 0x100),
+ RSND_GEN_M_REG(CTU_SV34R, 0x588, 0x100),
+ RSND_GEN_M_REG(CTU_SV35R, 0x58c, 0x100),
+ RSND_GEN_M_REG(CTU_SV36R, 0x590, 0x100),
+ RSND_GEN_M_REG(CTU_SV37R, 0x594, 0x100),
+ RSND_GEN_M_REG(MIX_SWRSR, 0xd00, 0x40),
+ RSND_GEN_M_REG(MIX_MIXIR, 0xd04, 0x40),
+ RSND_GEN_M_REG(MIX_ADINR, 0xd08, 0x40),
+ RSND_GEN_M_REG(MIX_MIXMR, 0xd10, 0x40),
+ RSND_GEN_M_REG(MIX_MVPDR, 0xd14, 0x40),
+ RSND_GEN_M_REG(MIX_MDBAR, 0xd18, 0x40),
+ RSND_GEN_M_REG(MIX_MDBBR, 0xd1c, 0x40),
+ RSND_GEN_M_REG(MIX_MDBCR, 0xd20, 0x40),
+ RSND_GEN_M_REG(MIX_MDBDR, 0xd24, 0x40),
+ RSND_GEN_M_REG(MIX_MDBER, 0xd28, 0x40),
+ RSND_GEN_M_REG(DVC_SWRSR, 0xe00, 0x100),
+ RSND_GEN_M_REG(DVC_DVUIR, 0xe04, 0x100),
+ RSND_GEN_M_REG(DVC_ADINR, 0xe08, 0x100),
+ RSND_GEN_M_REG(DVC_DVUCR, 0xe10, 0x100),
+ RSND_GEN_M_REG(DVC_ZCMCR, 0xe14, 0x100),
+ RSND_GEN_M_REG(DVC_VRCTR, 0xe18, 0x100),
+ RSND_GEN_M_REG(DVC_VRPDR, 0xe1c, 0x100),
+ RSND_GEN_M_REG(DVC_VRDBR, 0xe20, 0x100),
+ RSND_GEN_M_REG(DVC_VOL0R, 0xe28, 0x100),
+ RSND_GEN_M_REG(DVC_VOL1R, 0xe2c, 0x100),
+ RSND_GEN_M_REG(DVC_VOL2R, 0xe30, 0x100),
+ RSND_GEN_M_REG(DVC_VOL3R, 0xe34, 0x100),
+ RSND_GEN_M_REG(DVC_VOL4R, 0xe38, 0x100),
+ RSND_GEN_M_REG(DVC_VOL5R, 0xe3c, 0x100),
+ RSND_GEN_M_REG(DVC_VOL6R, 0xe40, 0x100),
+ RSND_GEN_M_REG(DVC_VOL7R, 0xe44, 0x100),
+ RSND_GEN_M_REG(DVC_DVUER, 0xe48, 0x100),
+ };
+ static const struct rsnd_regmap_field_conf conf_adg[] = {
+ RSND_GEN_S_REG(BRRA, 0x00),
+ RSND_GEN_S_REG(BRRB, 0x04),
+ RSND_GEN_S_REG(BRGCKR, 0x08),
+ RSND_GEN_S_REG(AUDIO_CLK_SEL0, 0x0c),
+ RSND_GEN_S_REG(AUDIO_CLK_SEL1, 0x10),
+ RSND_GEN_S_REG(AUDIO_CLK_SEL2, 0x14),
+ RSND_GEN_S_REG(AUDIO_CLK_SEL3, 0x18),
+ RSND_GEN_S_REG(DIV_EN, 0x30),
+ RSND_GEN_S_REG(SRCIN_TIMSEL0, 0x34),
+ RSND_GEN_S_REG(SRCIN_TIMSEL1, 0x38),
+ RSND_GEN_S_REG(SRCIN_TIMSEL2, 0x3c),
+ RSND_GEN_S_REG(SRCIN_TIMSEL3, 0x40),
+ RSND_GEN_S_REG(SRCIN_TIMSEL4, 0x44),
+ RSND_GEN_S_REG(SRCOUT_TIMSEL0, 0x48),
+ RSND_GEN_S_REG(SRCOUT_TIMSEL1, 0x4c),
+ RSND_GEN_S_REG(SRCOUT_TIMSEL2, 0x50),
+ RSND_GEN_S_REG(SRCOUT_TIMSEL3, 0x54),
+ RSND_GEN_S_REG(SRCOUT_TIMSEL4, 0x58),
+ RSND_GEN_S_REG(CMDOUT_TIMSEL, 0x5c),
+ };
+ static const struct rsnd_regmap_field_conf conf_ssi[] = {
+ RSND_GEN_M_REG(SSICR, 0x00, 0x40),
+ RSND_GEN_M_REG(SSISR, 0x04, 0x40),
+ RSND_GEN_M_REG(SSIWSR, 0x20, 0x40),
+ };
+ int ret;
+
+ ret = rsnd_gen_regmap_init(priv, 10, RSND_BASE_SCU, "scu", conf_scu);
+ if (ret < 0)
+ return ret;
+
+ ret = rsnd_gen_regmap_init(priv, 1, RSND_BASE_ADG, "adg", conf_adg);
+ if (ret < 0)
+ return ret;
+
+ ret = rsnd_gen_regmap_init(priv, 10, RSND_BASE_SSIU, "ssiu", conf_ssiu);
+ if (ret < 0)
+ return ret;
+
+ return rsnd_gen_regmap_init(priv, 10, RSND_BASE_SSI, "ssi", conf_ssi);
+}
+
/*
* Gen
*/
@@ -487,6 +665,8 @@ int rsnd_gen_probe(struct rsnd_priv *priv)
ret = rsnd_gen2_probe(priv);
else if (rsnd_is_gen4(priv))
ret = rsnd_gen4_probe(priv);
+ else if (rsnd_is_rzg3e(priv))
+ ret = rsnd_rzg3e_probe(priv);
if (ret < 0)
dev_err(dev, "unknown generation R-Car sound device\n");
diff --git a/sound/soc/renesas/rcar/rsnd.h b/sound/soc/renesas/rcar/rsnd.h
index f3f1ad1180f8..e917aa12fa80 100644
--- a/sound/soc/renesas/rcar/rsnd.h
+++ b/sound/soc/renesas/rcar/rsnd.h
@@ -143,13 +143,16 @@ enum rsnd_reg {
AUDIO_CLK_SEL0,
AUDIO_CLK_SEL1,
AUDIO_CLK_SEL2,
+ AUDIO_CLK_SEL3,
/* SSIU */
SSI_MODE,
SSI_MODE0,
SSI_MODE1,
SSI_MODE2,
+ SSI_MODE3,
SSI_CONTROL,
+ SSI_CONTROL2,
SSI_CTRL,
SSI_BUSIF0_MODE,
SSI_BUSIF1_MODE,
@@ -641,14 +644,28 @@ struct rsnd_priv {
struct platform_device *pdev;
spinlock_t lock;
unsigned long flags;
+
+ /*
+ * Flags layout: 0xDCBA
+ *
+ * A: R-Car generation (Gen1/Gen2/Gen3/Gen4)
+ * B: R-Car SoC variant (e.g. SOC_E for E1/E2/E3)
+ * C: RZ series generation
+ * D: RZ series SoC identifier (e.g. RZG3E)
+ *
+ * Bits 16+ are used for capability flags.
+ */
#define RSND_GEN_MASK (0xF << 0)
#define RSND_GEN1 (1 << 0)
#define RSND_GEN2 (2 << 0)
#define RSND_GEN3 (3 << 0)
#define RSND_GEN4 (4 << 0)
-#define RSND_SOC_MASK (0xF << 4)
-#define RSND_SOC_E (1 << 4) /* E1/E2/E3 */
-
+#define RSND_SOC_MASK (0xF << 4) /* nibble B */
+#define RSND_SOC_E (1 << 4) /* E1/E2/E3 */
+#define RSND_RZ_MASK (0xF << 8) /* nibble C */
+#define RSND_RZ3 (3 << 8)
+#define RSND_RZ_ID_MASK (0xF << 12) /* nibble D */
+#define RSND_RZG3E (1 << 12)
/*
* below value will be filled on rsnd_gen_probe()
*/
@@ -727,6 +744,9 @@ struct rsnd_priv {
#define rsnd_is_gen3_e3(priv) (((priv)->flags & \
(RSND_GEN_MASK | RSND_SOC_MASK)) == \
(RSND_GEN3 | RSND_SOC_E))
+#define rsnd_is_rzg3e(priv) (((priv)->flags & \
+ (RSND_RZ_MASK | RSND_RZ_ID_MASK)) == \
+ (RSND_RZ3 | RSND_RZG3E))
#define rsnd_flags_has(p, f) ((p)->flags & (f))
#define rsnd_flags_set(p, f) ((p)->flags |= (f))
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 05/18] ASoC: rsnd: Add RZ/G3E SoC probing and register map
2026-05-25 11:02 ` [PATCH v7 05/18] ASoC: rsnd: Add RZ/G3E SoC probing and register map John Madieu
@ 2026-05-25 12:23 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 12:23 UTC (permalink / raw)
To: John Madieu; +Cc: robh, wsa+renesas, conor+dt, devicetree
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Critical] The driver computes an incorrect physical address for DMA transfers on RZ/G3E by mistakenly falling back to R-Car Gen2/Gen3 hardcoded offsets.
- [Medium] The driver unconditionally attempts to write to `SSI_MODE0` during stream initialization for RZ/G3E, but this register is omitted from the RZ/G3E register map, causing continuous error logs.
--
commit 15c83e0793c513d3877a09357bf6e83e9029bbdb
Author: John Madieu <john.madieu.xa@bp.renesas.com>
Date: Mon May 25 11:02:17 2026 +0000
ASoC: rsnd: Add RZ/G3E SoC probing and register map
This commit adds probe support and register layouts for the RZ/G3E audio
subsystem. It introduces specific register maps to accommodate the different
base address organization and register offsets compared to R-Car Gen2/3/4.
> diff --git a/sound/soc/renesas/rcar/gen.c b/sound/soc/renesas/rcar/gen.c
> index d1f20cde66be1..05d5f656fb01c 100644
> --- a/sound/soc/renesas/rcar/gen.c
> +++ b/sound/soc/renesas/rcar/gen.c
> @@ -464,6 +464,184 @@ static int rsnd_gen1_probe(struct rsnd_priv *priv)
> return ret_adg | ret_ssi;
> }
>
> +/*
> + * RZ/G3E Generation
> + */
> +static int rsnd_rzg3e_probe(struct rsnd_priv *priv)
> +{
> + static const struct rsnd_regmap_field_conf conf_ssiu[] = {
> + RSND_GEN_S_REG(SSI_MODE1, 0x804),
[Severity: Medium]
Will this cause unsupported register access errors during stream initialization?
The conf_ssiu register map omits SSI_MODE0, but rsnd_ssiu_init() will
unconditionally attempt to write to it using rsnd_mod_bset(). Without this
register in the map, rsnd_is_accessible_reg() might fail, potentially
leading to error logs every time a stream starts.
I note this regression is addressed in the subsequent commit
"ASoC: rsnd: ssiu: Add RZ/G3E BUSIF support". Could this cause issues for
users who bisect to this specific commit?
> + RSND_GEN_S_REG(SSI_MODE2, 0x808),
> + RSND_GEN_S_REG(SSI_MODE3, 0x80c),
[ ... ]
> @@ -487,6 +665,8 @@ int rsnd_gen_probe(struct rsnd_priv *priv)
> ret = rsnd_gen2_probe(priv);
> else if (rsnd_is_gen4(priv))
> ret = rsnd_gen4_probe(priv);
> + else if (rsnd_is_rzg3e(priv))
> + ret = rsnd_rzg3e_probe(priv);
[Severity: Critical]
Does this new RZ/G3E generation cause a regression in rsnd_dma_addr()?
When setting up a DMA transfer, there doesn't seem to be a specific branch
for RZ/G3E address calculations. Because rsnd_is_gen1() and rsnd_is_gen4()
evaluate to false, execution appears to default to rsnd_gen2_dma_addr().
This fallback subtracts hardcoded offsets valid only for R-Car Gen2/Gen3,
which could yield an incorrect physical address for RZ/G3E and potentially
cause incorrect memory access during DMA transfers.
I note this regression is fixed later in the series by the commit
"ASoC: rsnd: Add RZ/G3E DMA address calculation support". Should the DMA
calculation support be included in this patch to maintain bisectability?
>
> if (ret < 0)
> dev_err(dev, "unknown generation R-Car sound device\n");
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=5
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 06/18] ASoC: rsnd: Add audmapp clock and reset support for RZ/G3E
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (4 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 05/18] ASoC: rsnd: Add RZ/G3E SoC probing and register map John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:02 ` [PATCH v7 07/18] ASoC: rsnd: Refactor DMA address tables with named structs John Madieu
` (12 subsequent siblings)
18 siblings, 0 replies; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
RZ/G3E requires additional audmapp clock and reset lines for
Audio DMA-PP operation.
Add global audmapp clock/reset management in rsnd_dma_probe()
using optional APIs to remain transparent to other platforms.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7:
- Fix the "audmacpp" spelling to "audmapp" in the patch subject
and commit message, matching the clock/reset name and the rest
of the driver. No code change.
v6:
- Enable the audmapp clock before deasserting the audmapp reset
so the block sees a stable clock on the way out of reset.
- Drop redundant braces around single-statement if blocks (style).
v5:
- Add comment on audmapp clock/reset acquisition clarifying these
are optional and transparent to platforms that don't have them
in DT, per Kuninori's request.
- Drop spurious blank line added to struct rsnd_priv in rsnd.h.
v4:
- Move audmapp_clk and audmapp_rstc from struct rsnd_priv into
struct rsnd_dma_ctrl.
v3: No changes
v2: No changes
sound/soc/renesas/rcar/dma.c | 22 ++++++++++++++++++++++
1 file changed, 22 insertions(+)
diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
index 68c859897e68..d52cada711e5 100644
--- a/sound/soc/renesas/rcar/dma.c
+++ b/sound/soc/renesas/rcar/dma.c
@@ -47,6 +47,9 @@ struct rsnd_dma_ctrl {
phys_addr_t ppres;
int dmaen_num;
int dmapp_num;
+ /* RZ/G3E: Audio DMAC peri-peri clock and reset */
+ struct clk *audmapp_clk;
+ struct reset_control *audmapp_rstc;
};
#define rsnd_priv_to_dmac(p) ((struct rsnd_dma_ctrl *)(p)->dma)
@@ -870,6 +873,25 @@ int rsnd_dma_probe(struct rsnd_priv *priv)
return 0; /* it will be PIO mode */
}
+ /*
+ * Audio DMAC peri-peri clock and reset for RZ/G3E.
+ * These use optional APIs, so they gracefully return NULL
+ * (no error) on platforms whose DT does not provide them.
+ *
+ * Enable the clock first so the block sees a stable clock on
+ * the way out of reset, then deassert the reset line.
+ */
+ dmac->audmapp_clk = devm_clk_get_optional_enabled(dev, "audmapp");
+ if (IS_ERR(dmac->audmapp_clk))
+ return dev_err_probe(dev, PTR_ERR(dmac->audmapp_clk),
+ "failed to get audmapp clock\n");
+
+ dmac->audmapp_rstc =
+ devm_reset_control_get_optional_exclusive_deasserted(dev, "audmapp");
+ if (IS_ERR(dmac->audmapp_rstc))
+ return dev_err_probe(dev, PTR_ERR(dmac->audmapp_rstc),
+ "failed to get audmapp reset\n");
+
dmac->dmapp_num = 0;
dmac->ppres = res->start;
dmac->ppbase = devm_ioremap_resource(dev, res);
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* [PATCH v7 07/18] ASoC: rsnd: Refactor DMA address tables with named structs
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (5 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 06/18] ASoC: rsnd: Add audmapp clock and reset support for RZ/G3E John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:23 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 08/18] ASoC: rsnd: Add RZ/G3E DMA address calculation support John Madieu
` (11 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
Replace the raw multi-dimensional array used for DMA address lookup in
rsnd_gen2_dma_addr() with properly named structs: rsnd_dma_addr (in/out
pair), rsnd_dma_addr_dir (capture/playback arrays), and
rsnd_dma_addr_map (src/ssi/ssiu module sets).
While at it, extract the common lookup logic (is_ssi / use_src / use_cmd
evaluation and table indexing) into a shared rsnd_dma_addr_lookup()
function.
No functional change. This is a preparatory refactor for upcoming RZ/G3E
support which will add its own DMA address map using the same struct and
lookup function.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: No changes
v6: No changes
v5:
- New patch, extracted from v4 patch 06/12 as a standalone
preparatory refactor per Kuninori's request.
- No RZ/G3E content; purely the struct introduction and
rsnd_gen2 conversion.
sound/soc/renesas/rcar/dma.c | 147 +++++++++++++++++++++++------------
1 file changed, 99 insertions(+), 48 deletions(-)
diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
index d52cada711e5..0ce77eee338d 100644
--- a/sound/soc/renesas/rcar/dma.c
+++ b/sound/soc/renesas/rcar/dma.c
@@ -481,6 +481,69 @@ static struct rsnd_mod_ops rsnd_dmapp_ops = {
DEBUG_INFO
};
+struct rsnd_dma_addr {
+ dma_addr_t out_addr;
+ dma_addr_t in_addr;
+};
+
+struct rsnd_dma_addr_dir {
+ struct rsnd_dma_addr capture[3];
+ struct rsnd_dma_addr playback[3];
+};
+
+struct rsnd_dma_addr_map {
+ struct rsnd_dma_addr_dir src;
+ struct rsnd_dma_addr_dir ssi;
+ struct rsnd_dma_addr_dir ssiu;
+};
+
+static dma_addr_t
+rsnd_dma_addr_lookup(struct rsnd_dai_stream *io,
+ struct rsnd_mod *mod,
+ struct rsnd_priv *priv,
+ const struct rsnd_dma_addr_map *map,
+ int is_play, int is_from)
+{
+ struct device *dev = rsnd_priv_to_dev(priv);
+ int is_ssi = !!(rsnd_io_to_mod_ssi(io) == mod) ||
+ !!(rsnd_io_to_mod_ssiu(io) == mod);
+ int use_src = !!rsnd_io_to_mod_src(io);
+ int use_cmd = !!rsnd_io_to_mod_dvc(io) ||
+ !!rsnd_io_to_mod_mix(io) ||
+ !!rsnd_io_to_mod_ctu(io);
+ int id = rsnd_mod_id(mod);
+ const struct rsnd_dma_addr_dir *dir;
+ const struct rsnd_dma_addr *addr;
+
+ /* it shouldn't happen */
+ if (use_cmd && !use_src)
+ dev_err(dev, "DVC is selected without SRC\n");
+
+ /* use SSIU or SSI? */
+ if (is_ssi && rsnd_ssi_use_busif(io))
+ is_ssi++;
+
+ dev_dbg(dev, "dma%d addr : is_ssi=%d use_src=%d use_cmd=%d\n",
+ id, is_ssi, use_src, use_cmd);
+
+ switch (is_ssi) {
+ case 2:
+ dir = &map->ssiu;
+ break;
+ case 1:
+ dir = &map->ssi;
+ break;
+ default:
+ dir = &map->src;
+ break;
+ }
+
+ addr = is_play ? &dir->playback[use_src + use_cmd]
+ : &dir->capture[use_src + use_cmd];
+
+ return is_from ? addr->out_addr : addr->in_addr;
+}
+
/*
* Common DMAC Interface
*/
@@ -527,47 +590,45 @@ rsnd_gen2_dma_addr(struct rsnd_dai_stream *io,
struct device *dev = rsnd_priv_to_dev(priv);
phys_addr_t ssi_reg = rsnd_gen_get_phy_addr(priv, RSND_BASE_SSI);
phys_addr_t src_reg = rsnd_gen_get_phy_addr(priv, RSND_BASE_SCU);
- int is_ssi = !!(rsnd_io_to_mod_ssi(io) == mod) ||
- !!(rsnd_io_to_mod_ssiu(io) == mod);
- int use_src = !!rsnd_io_to_mod_src(io);
- int use_cmd = !!rsnd_io_to_mod_dvc(io) ||
- !!rsnd_io_to_mod_mix(io) ||
- !!rsnd_io_to_mod_ctu(io);
int id = rsnd_mod_id(mod);
int busif = rsnd_mod_id_sub(rsnd_io_to_mod_ssiu(io));
- struct dma_addr {
- dma_addr_t out_addr;
- dma_addr_t in_addr;
- } dma_addrs[3][2][3] = {
- /* SRC */
- /* Capture */
- {{{ 0, 0 },
- { RDMA_SRC_O_N(src, id), RDMA_SRC_I_P(src, id) },
- { RDMA_CMD_O_N(src, id), RDMA_SRC_I_P(src, id) } },
- /* Playback */
- {{ 0, 0, },
- { RDMA_SRC_O_P(src, id), RDMA_SRC_I_N(src, id) },
- { RDMA_CMD_O_P(src, id), RDMA_SRC_I_N(src, id) } }
+ const struct rsnd_dma_addr_map map = {
+ .src = {
+ .capture = {
+ { 0, 0 },
+ { RDMA_SRC_O_N(src, id), RDMA_SRC_I_P(src, id) },
+ { RDMA_CMD_O_N(src, id), RDMA_SRC_I_P(src, id) },
+ },
+ .playback = {
+ { 0, 0 },
+ { RDMA_SRC_O_P(src, id), RDMA_SRC_I_N(src, id) },
+ { RDMA_CMD_O_P(src, id), RDMA_SRC_I_N(src, id) },
+ },
+ },
+ .ssi = {
+ .capture = {
+ { RDMA_SSI_O_N(ssi, id), 0 },
+ { RDMA_SSIU_O_P(ssi, id, busif), 0 },
+ { RDMA_SSIU_O_P(ssi, id, busif), 0 },
+ },
+ .playback = {
+ { 0, RDMA_SSI_I_N(ssi, id) },
+ { 0, RDMA_SSIU_I_P(ssi, id, busif) },
+ { 0, RDMA_SSIU_I_P(ssi, id, busif) },
+ },
},
- /* SSI */
- /* Capture */
- {{{ RDMA_SSI_O_N(ssi, id), 0 },
- { RDMA_SSIU_O_P(ssi, id, busif), 0 },
- { RDMA_SSIU_O_P(ssi, id, busif), 0 } },
- /* Playback */
- {{ 0, RDMA_SSI_I_N(ssi, id) },
- { 0, RDMA_SSIU_I_P(ssi, id, busif) },
- { 0, RDMA_SSIU_I_P(ssi, id, busif) } }
+ .ssiu = {
+ .capture = {
+ { RDMA_SSIU_O_N(ssi, id, busif), 0 },
+ { RDMA_SSIU_O_P(ssi, id, busif), 0 },
+ { RDMA_SSIU_O_P(ssi, id, busif), 0 },
+ },
+ .playback = {
+ { 0, RDMA_SSIU_I_N(ssi, id, busif) },
+ { 0, RDMA_SSIU_I_P(ssi, id, busif) },
+ { 0, RDMA_SSIU_I_P(ssi, id, busif) },
+ },
},
- /* SSIU */
- /* Capture */
- {{{ RDMA_SSIU_O_N(ssi, id, busif), 0 },
- { RDMA_SSIU_O_P(ssi, id, busif), 0 },
- { RDMA_SSIU_O_P(ssi, id, busif), 0 } },
- /* Playback */
- {{ 0, RDMA_SSIU_I_N(ssi, id, busif) },
- { 0, RDMA_SSIU_I_P(ssi, id, busif) },
- { 0, RDMA_SSIU_I_P(ssi, id, busif) } } },
};
/*
@@ -580,17 +641,7 @@ rsnd_gen2_dma_addr(struct rsnd_dai_stream *io,
dev_err(dev, "This driver doesn't support SSI%d-%d, so far",
id, busif);
- /* it shouldn't happen */
- if (use_cmd && !use_src)
- dev_err(dev, "DVC is selected without SRC\n");
-
- /* use SSIU or SSI ? */
- if (is_ssi && rsnd_ssi_use_busif(io))
- is_ssi++;
-
- return (is_from) ?
- dma_addrs[is_ssi][is_play][use_src + use_cmd].out_addr :
- dma_addrs[is_ssi][is_play][use_src + use_cmd].in_addr;
+ return rsnd_dma_addr_lookup(io, mod, priv, &map, is_play, is_from);
}
/*
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 07/18] ASoC: rsnd: Refactor DMA address tables with named structs
2026-05-25 11:02 ` [PATCH v7 07/18] ASoC: rsnd: Refactor DMA address tables with named structs John Madieu
@ 2026-05-25 11:23 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 11:23 UTC (permalink / raw)
To: John Madieu; +Cc: conor+dt, wsa+renesas, devicetree, robh
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [High] Potential NULL pointer dereference in rsnd_ssi_use_busif if the stream lacks an SSI component.
--
commit c7d64665ffb93009db59da73396bbe58b794f096
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: rsnd: Refactor DMA address tables with named structs
This commit replaces the raw multi-dimensional array used for DMA address
lookup in rsnd_gen2_dma_addr() with named structs. It also extracts the
common lookup logic into a shared rsnd_dma_addr_lookup() function to
prepare for upcoming RZ/G3E support.
> diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
> index d52cada711e50..0ce77eee338da 100644
> --- a/sound/soc/renesas/rcar/dma.c
> +++ b/sound/soc/renesas/rcar/dma.c
[ ... ]
> @@ -481,6 +481,69 @@ static struct rsnd_mod_ops rsnd_dmapp_ops = {
> DEBUG_INFO
> };
>
> +struct rsnd_dma_addr {
[ ... ]
> +static dma_addr_t
> +rsnd_dma_addr_lookup(struct rsnd_dai_stream *io,
> + struct rsnd_mod *mod,
> + struct rsnd_priv *priv,
> + const struct rsnd_dma_addr_map *map,
> + int is_play, int is_from)
> +{
[ ... ]
> + /* use SSIU or SSI? */
> + if (is_ssi && rsnd_ssi_use_busif(io))
[Severity: High]
This is a pre-existing issue, but could this lead to a NULL pointer
dereference if the stream lacks an SSI component?
If a device tree specifies a DAI link that includes an SSIU module but
omits an SSI module, rsnd_io_to_mod_ssi() will return NULL.
When rsnd_ssi_use_busif() is called, it unconditionally passes this NULL
module to rsnd_ssi_is_dma_mode(), which dereferences the ops struct:
sound/soc/renesas/rcar/ssi.c:rsnd_ssi_use_busif() {
struct rsnd_mod *mod = rsnd_io_to_mod_ssi(io);
...
if (!rsnd_ssi_is_dma_mode(mod))
return 0;
...
}
Would it be safer for rsnd_ssi_use_busif() to verify the module exists
before checking its DMA mode, allowing a malformed device tree to
fail gracefully instead of causing a panic?
> + is_ssi++;
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=7
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 08/18] ASoC: rsnd: Add RZ/G3E DMA address calculation support
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (6 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 07/18] ASoC: rsnd: Refactor DMA address tables with named structs John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:29 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 09/18] ASoC: rsnd: ssiu: Add shared SSI reset controller support John Madieu
` (10 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
RZ/G3E has different DMA register base addresses and offset
calculations compared to R-Car platforms.
Add dedicated rsnd_rzg3e_dma_addr() function with dispatch from
rsnd_dma_addr(), following the existing per-generation pattern.
The function reuses rsnd_dma_addr_lookup() and rsnd_dma_addr_map.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7:
- Wrap the i and j arguments of the RDMA_SSI_*_G3E,
RDMA_SSIU_*_G3E, RDMA_SRC_*_G3E and RDMA_CMD_*_G3E address macros
in parentheses to guard against operator-precedence surprises if
a non-trivial expression is ever passed (sashiko-bot).
v6: No changes
v5:
- Split from v4 patch 06/12 per Kuninori's request.
- Move G3E DMA address comment block into rsnd_rzg3e_dma_addr()
per Kuninori's request.
v4:
- Replace raw [3][2][3] DMA address array with named structs
rsnd_dma_addr_dir and rsnd_dma_addr_map. Use designated
initializers in both rsnd_gen2_dma_addr() and
rsnd_rzg3e_dma_addr().
v3: No changes
v2: No changes
sound/soc/renesas/rcar/dma.c | 81 ++++++++++++++++++++++++++++++++++++
1 file changed, 81 insertions(+)
diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
index 0ce77eee338d..92974610ac15 100644
--- a/sound/soc/renesas/rcar/dma.c
+++ b/sound/soc/renesas/rcar/dma.c
@@ -644,6 +644,85 @@ rsnd_gen2_dma_addr(struct rsnd_dai_stream *io,
return rsnd_dma_addr_lookup(io, mod, priv, &map, is_play, is_from);
}
+/*
+ * ex) G3E case
+ * mod / DMAC in / DMAC out / DMAC PP in / DMAC pp out
+ * SSI : 0x13C31000 / 0x13C40000 / 0x13C40000
+ * SSIU: 0x13C31000 / 0x13C40000 / 0x13C40000 / 0xEC400000 / 0xEC400000
+ * SCU : 0x13C00000 / 0x13C10000 / 0x13C14000 / 0xEC300000 / 0xEC304000
+ * CMD : 0x13C00000 / / 0x13C18000 0xEC308000
+ */
+
+/* RZ/G3E DMA address macros */
+#define RDMA_SSI_I_N_G3E(addr, i) (addr ##_reg + 0x0000F000 + (0x1000 * (i)))
+#define RDMA_SSI_O_N_G3E(addr, i) (addr ##_reg + 0x0000F000 + (0x1000 * (i)))
+
+#define RDMA_SSIU_I_N_G3E(addr, i, j) (addr ##_reg + 0x0000F000 + (0x1000 * (i)) + (((j) / 4) * 0xA000) + (((j) % 4) * 0x400) - (0x4000 * ((i) / 9) * ((j) / 4)))
+#define RDMA_SSIU_O_N_G3E(addr, i, j) RDMA_SSIU_I_N_G3E(addr, i, j)
+
+#define RDMA_SSIU_I_P_G3E(addr, i, j) (addr ##_reg + 0xD87CF000 + (0x1000 * (i)) + (((j) / 4) * 0xA000) + (((j) % 4) * 0x400) - (0x4000 * ((i) / 9) * ((j) / 4)))
+#define RDMA_SSIU_O_P_G3E(addr, i, j) RDMA_SSIU_I_P_G3E(addr, i, j)
+
+#define RDMA_SRC_I_N_G3E(addr, i) (addr ##_reg + 0x00010000 + (0x400 * (i)))
+#define RDMA_SRC_O_N_G3E(addr, i) (addr ##_reg + 0x00014000 + (0x400 * (i)))
+
+#define RDMA_SRC_I_P_G3E(addr, i) (addr ##_reg + 0xD8700000 + (0x400 * (i)))
+#define RDMA_SRC_O_P_G3E(addr, i) (addr ##_reg + 0xD8704000 + (0x400 * (i)))
+
+#define RDMA_CMD_O_N_G3E(addr, i) (addr ##_reg + 0x00018000 + (0x400 * (i)))
+#define RDMA_CMD_O_P_G3E(addr, i) (addr ##_reg + 0xD8708000 + (0x400 * (i)))
+
+static dma_addr_t
+rsnd_rzg3e_dma_addr(struct rsnd_dai_stream *io,
+ struct rsnd_mod *mod, int is_play, int is_from)
+{
+ struct rsnd_priv *priv = rsnd_io_to_priv(io);
+ phys_addr_t ssi_reg = rsnd_gen_get_phy_addr(priv, RSND_BASE_SSI);
+ phys_addr_t src_reg = rsnd_gen_get_phy_addr(priv, RSND_BASE_SCU);
+ int id = rsnd_mod_id(mod);
+ int busif = rsnd_mod_id_sub(rsnd_io_to_mod_ssiu(io));
+ const struct rsnd_dma_addr_map map = {
+ .src = {
+ .capture = {
+ { 0, 0 },
+ { RDMA_SRC_O_N_G3E(src, id), RDMA_SRC_I_P_G3E(src, id) },
+ { RDMA_CMD_O_N_G3E(src, id), RDMA_SRC_I_P_G3E(src, id) },
+ },
+ .playback = {
+ { 0, 0 },
+ { RDMA_SRC_O_P_G3E(src, id), RDMA_SRC_I_N_G3E(src, id) },
+ { RDMA_CMD_O_P_G3E(src, id), RDMA_SRC_I_N_G3E(src, id) },
+ },
+ },
+ .ssi = {
+ .capture = {
+ { RDMA_SSI_O_N_G3E(ssi, id), 0 },
+ { RDMA_SSIU_O_P_G3E(ssi, id, busif), 0 },
+ { RDMA_SSIU_O_P_G3E(ssi, id, busif), 0 },
+ },
+ .playback = {
+ { 0, RDMA_SSI_I_N_G3E(ssi, id) },
+ { 0, RDMA_SSIU_I_P_G3E(ssi, id, busif) },
+ { 0, RDMA_SSIU_I_P_G3E(ssi, id, busif) },
+ },
+ },
+ .ssiu = {
+ .capture = {
+ { RDMA_SSIU_O_N_G3E(ssi, id, busif), 0 },
+ { RDMA_SSIU_O_P_G3E(ssi, id, busif), 0 },
+ { RDMA_SSIU_O_P_G3E(ssi, id, busif), 0 },
+ },
+ .playback = {
+ { 0, RDMA_SSIU_I_N_G3E(ssi, id, busif) },
+ { 0, RDMA_SSIU_I_P_G3E(ssi, id, busif) },
+ { 0, RDMA_SSIU_I_P_G3E(ssi, id, busif) },
+ },
+ },
+ };
+
+ return rsnd_dma_addr_lookup(io, mod, priv, &map, is_play, is_from);
+}
+
/*
* Gen4 DMA read/write register offset
*
@@ -690,6 +769,8 @@ static dma_addr_t rsnd_dma_addr(struct rsnd_dai_stream *io,
return 0;
else if (rsnd_is_gen4(priv))
return rsnd_gen4_dma_addr(io, mod, is_play, is_from);
+ else if (rsnd_is_rzg3e(priv))
+ return rsnd_rzg3e_dma_addr(io, mod, is_play, is_from);
else
return rsnd_gen2_dma_addr(io, mod, is_play, is_from);
}
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 08/18] ASoC: rsnd: Add RZ/G3E DMA address calculation support
2026-05-25 11:02 ` [PATCH v7 08/18] ASoC: rsnd: Add RZ/G3E DMA address calculation support John Madieu
@ 2026-05-25 11:29 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 11:29 UTC (permalink / raw)
To: John Madieu; +Cc: wsa+renesas, conor+dt, devicetree, robh
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [High] Potential NULL pointer dereference in rsnd_rzg3e_dma_addr() when evaluating the bus interface (busif) if the SSIU module is absent. This problem wasn't introduced by this patch, but was copied from the pre-existing rsnd_gen2_dma_addr() function and propagated to the new RZ/G3E platform.
--
commit 81338bf6d6c4e6f1f0cb642937dcc5c87e6a300d
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: rsnd: Add RZ/G3E DMA address calculation support
This commit adds RZ/G3E DMA address calculation support to the rsnd
driver. It implements a dedicated rsnd_rzg3e_dma_addr function to
handle the different register base addresses and offsets for this SoC.
> diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
> index 0ce77eee338da..92974610ac153 100644
> --- a/sound/soc/renesas/rcar/dma.c
> +++ b/sound/soc/renesas/rcar/dma.c
[ ... ]
> @@ -644,6 +644,85 @@ rsnd_gen2_dma_addr(struct rsnd_dai_stream *io,
[ ... ]
> +static dma_addr_t
> +rsnd_rzg3e_dma_addr(struct rsnd_dai_stream *io,
> + struct rsnd_mod *mod, int is_play, int is_from)
> +{
> + struct rsnd_priv *priv = rsnd_io_to_priv(io);
> + phys_addr_t ssi_reg = rsnd_gen_get_phy_addr(priv, RSND_BASE_SSI);
> + phys_addr_t src_reg = rsnd_gen_get_phy_addr(priv, RSND_BASE_SCU);
> + int id = rsnd_mod_id(mod);
> + int busif = rsnd_mod_id_sub(rsnd_io_to_mod_ssiu(io));
[Severity: High]
This is a pre-existing issue, but can this lead to a NULL pointer dereference
if the SSIU module is absent?
If the device tree configures the SSI module for PIO mode, the SSIU module
might not be connected to the audio stream, meaning rsnd_io_to_mod_ssiu(io)
evaluates to NULL.
If another module in the pipeline, such as SRC, unconditionally requests DMA
during probe, rsnd_dma_alloc() will determine the DMA path and call
rsnd_rzg3e_dma_addr().
Since rsnd_mod_id_sub() does not check for a NULL mod pointer:
sound/soc/renesas/rcar/core.c:rsnd_mod_id_sub() {
if ((mod)->ops->id_sub)
return (mod)->ops->id_sub(mod);
return 0;
}
Could evaluating rsnd_mod_id_sub(rsnd_io_to_mod_ssiu(io)) here trigger a kernel
panic on probe with a misconfigured device tree? Should this be handled
gracefully instead?
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=8
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 09/18] ASoC: rsnd: ssiu: Add shared SSI reset controller support
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (7 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 08/18] ASoC: rsnd: Add RZ/G3E DMA address calculation support John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:41 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 10/18] ASoC: rsnd: ssiu: Add RZ/G3E BUSIF support John Madieu
` (9 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
The RZ/G3E SoC exposes a single shared "ssi-all" reset that gates all
SSI/SSIU modules. Acquire it at SSIU probe and pass it through
rsnd_mod_init() so it is wired into the rsnd_mod->rstc plumbing.
devm_reset_control_get_optional_shared() returns NULL when no reset is
described in DT, leaving existing R-Car generations unaffected.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: New patch. Split out of [PATCH v6 09/16] "ASoC: rsnd: ssui: Add
RZ/G3E SSIU BUSIF support" per Kuninori Morimoto's request that
the reset, the BUSIF layout and busif_status_count be separate
changes. This patch carries only the shared "ssi-all" reset
acquisition in rsnd_ssiu_probe() and its rsnd_mod_init()
plumbing; it is a no-op for non-RZ/G3E DTs.
sound/soc/renesas/rcar/ssiu.c | 9 ++++++++-
1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/sound/soc/renesas/rcar/ssiu.c b/sound/soc/renesas/rcar/ssiu.c
index 0cfa84fe5ea8..2b922ead62d0 100644
--- a/sound/soc/renesas/rcar/ssiu.c
+++ b/sound/soc/renesas/rcar/ssiu.c
@@ -510,6 +510,7 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
{
struct device *dev = rsnd_priv_to_dev(priv);
struct device_node *node __free(device_node) = rsnd_ssiu_of_node(priv);
+ struct reset_control *rstc;
struct rsnd_ssiu *ssiu;
struct rsnd_mod_ops *ops;
const int *list = NULL;
@@ -564,6 +565,12 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
}
}
+ /* Acquire shared reset once for all SSIU modules */
+ rstc = devm_reset_control_get_optional_shared(dev, "ssi-all");
+ if (IS_ERR(rstc))
+ return dev_err_probe(dev, PTR_ERR(rstc),
+ "failed to get ssi-all reset\n");
+
for_each_rsnd_ssiu(ssiu, priv, i) {
int ret;
@@ -586,7 +593,7 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
}
ret = rsnd_mod_init(priv, rsnd_mod_get(ssiu),
- ops, NULL, NULL, RSND_MOD_SSIU, i);
+ ops, NULL, rstc, RSND_MOD_SSIU, i);
if (ret)
return ret;
}
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 09/18] ASoC: rsnd: ssiu: Add shared SSI reset controller support
2026-05-25 11:02 ` [PATCH v7 09/18] ASoC: rsnd: ssiu: Add shared SSI reset controller support John Madieu
@ 2026-05-25 11:41 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 11:41 UTC (permalink / raw)
To: John Madieu; +Cc: wsa+renesas, conor+dt, devicetree, robh
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [High] Missing rollback on probe error paths leaks the shared reset control's deassert count.
--
commit 75b9405455f9c9d54972ded411c26d45e6a17c73
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: rsnd: ssiu: Add shared SSI reset controller support
The RZ/G3E SoC exposes a single shared "ssi-all" reset that gates all
SSI/SSIU modules. Acquire it at SSIU probe and pass it through
rsnd_mod_init() so it is wired into the rsnd_mod->rstc plumbing.
> diff --git a/sound/soc/renesas/rcar/ssiu.c b/sound/soc/renesas/rcar/ssiu.c
> --- a/sound/soc/renesas/rcar/ssiu.c
> +++ b/sound/soc/renesas/rcar/ssiu.c
[ ... ]
> @@ -586,7 +593,7 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
> }
>
> ret = rsnd_mod_init(priv, rsnd_mod_get(ssiu),
> - ops, NULL, NULL, RSND_MOD_SSIU, i);
> + ops, NULL, rstc, RSND_MOD_SSIU, i);
> if (ret)
> return ret;
[Severity: High]
This is a pre-existing issue, but does this error path leak the shared reset
control's deassert count?
If probe fails on a subsequent module (for example, returning -EPROBE_DEFER
from rsnd_mod_init), the function returns directly without calling
rsnd_mod_quit() for the modules initialized in previous iterations of the loop.
Since rsnd_mod_init() calls reset_control_deassert(), the reset remains
deasserted for those modules.
While devres cleanup will call reset_control_put() on the shared reset acquired
via devm_reset_control_get_optional_shared(), this only drops the kref. It does
not balance the atomic deassert_count or assert the hardware line.
Across common initialization events like -EPROBE_DEFER retries, will the
deassert count grow unbounded and leave the hardware permanently deasserted?
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=9
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 10/18] ASoC: rsnd: ssiu: Add RZ/G3E BUSIF support
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (8 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 09/18] ASoC: rsnd: ssiu: Add shared SSI reset controller support John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:47 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 11/18] ASoC: rsnd: Add SSI reset support for RZ/G3E platform John Madieu
` (8 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
Add support for the SSIU found on the Renesas RZ/G3E SoC, which
provides a different BUSIF layout compared to earlier generations:
- SSI0-SSI4: 4 BUSIF instances each (BUSIF0-3)
- SSI5-SSI8: 1 BUSIF instance each (BUSIF0 only)
- SSI9: 4 BUSIF instances (BUSIF0-3)
- Total: 28 BUSIFs
The RZ/G3E also has only two pairs of BUSIF error-status registers
instead of four, and the SSI always operates in BUSIF mode: the
SSI_MODE0 BUSIF/PIO select bit is not implemented and must not be written.
While at it, add RSND_SSIU_BUSIF_STATUS_COUNT_2 as a capability flag in
the match data, consumed via struct rsnd_ssiu_ctrl, to parametrise the two
BUSIF error-status loops.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: New patch. Split out of [PATCH v6 09/16] "ASoC: rsnd: ssui: Add
RZ/G3E SSIU BUSIF support" per Kuninori Morimoto's request. This
patch carries the RZ/G3E BUSIF layout: the rzg3e_id[] index
table, struct rsnd_ssiu_ctrl with busif_status_count, the
RSND_SSIU_BUSIF_STATUS_COUNT_2 capability flag and the SSI_MODE0
guard.
- Fix the "ssui" typo in the subject to "ssiu".
- Split the r9a09g047 match-data initialiser line after the comma
instead of after the pipe character.
sound/soc/renesas/rcar/core.c | 3 +-
sound/soc/renesas/rcar/rsnd.h | 2 ++
sound/soc/renesas/rcar/ssiu.c | 55 +++++++++++++++++++++++------------
3 files changed, 41 insertions(+), 19 deletions(-)
diff --git a/sound/soc/renesas/rcar/core.c b/sound/soc/renesas/rcar/core.c
index 7169d0ef8f90..f5c8ba8c5d56 100644
--- a/sound/soc/renesas/rcar/core.c
+++ b/sound/soc/renesas/rcar/core.c
@@ -106,7 +106,8 @@ static const struct of_device_id rsnd_of_match[] = {
{ .compatible = "renesas,rcar_sound-gen4", .data = (void *)RSND_GEN4 },
/* Special Handling */
{ .compatible = "renesas,rcar_sound-r8a77990", .data = (void *)(RSND_GEN3 | RSND_SOC_E) },
- { .compatible = "renesas,r9a09g047-sound", .data = (void *)(RSND_RZ3 | RSND_RZG3E) },
+ { .compatible = "renesas,r9a09g047-sound",
+ .data = (void *)(RSND_RZ3 | RSND_RZG3E | RSND_SSIU_BUSIF_STATUS_COUNT_2) },
{},
};
MODULE_DEVICE_TABLE(of, rsnd_of_match);
diff --git a/sound/soc/renesas/rcar/rsnd.h b/sound/soc/renesas/rcar/rsnd.h
index e917aa12fa80..95843a20c43c 100644
--- a/sound/soc/renesas/rcar/rsnd.h
+++ b/sound/soc/renesas/rcar/rsnd.h
@@ -666,6 +666,7 @@ struct rsnd_priv {
#define RSND_RZ3 (3 << 8)
#define RSND_RZ_ID_MASK (0xF << 12) /* nibble D */
#define RSND_RZG3E (1 << 12)
+#define RSND_SSIU_BUSIF_STATUS_COUNT_2 BIT(16) /* Only 2 BUSIF error-status register pairs */
/*
* below value will be filled on rsnd_gen_probe()
*/
@@ -690,6 +691,7 @@ struct rsnd_priv {
/*
* below value will be filled on rsnd_ssiu_probe()
*/
+ void *ssiu_ctrl;
void *ssiu;
int ssiu_nr;
diff --git a/sound/soc/renesas/rcar/ssiu.c b/sound/soc/renesas/rcar/ssiu.c
index 2b922ead62d0..8d4ce9d35e9e 100644
--- a/sound/soc/renesas/rcar/ssiu.c
+++ b/sound/soc/renesas/rcar/ssiu.c
@@ -29,31 +29,39 @@ struct rsnd_ssiu {
i++)
/*
- * SSI Gen2 Gen3 Gen4
- * 0 BUSIF0-3 BUSIF0-7 BUSIF0-7
- * 1 BUSIF0-3 BUSIF0-7
- * 2 BUSIF0-3 BUSIF0-7
- * 3 BUSIF0 BUSIF0-7
- * 4 BUSIF0 BUSIF0-7
- * 5 BUSIF0 BUSIF0
- * 6 BUSIF0 BUSIF0
- * 7 BUSIF0 BUSIF0
- * 8 BUSIF0 BUSIF0
- * 9 BUSIF0-3 BUSIF0-7
- * total 22 52 8
+ * SSI Gen2 Gen3 Gen4 RZ/G3E
+ * 0 BUSIF0-3 BUSIF0-7 BUSIF0-7 BUSIF0-3
+ * 1 BUSIF0-3 BUSIF0-7 BUSIF0-3
+ * 2 BUSIF0-3 BUSIF0-7 BUSIF0-3
+ * 3 BUSIF0 BUSIF0-7 BUSIF0-3
+ * 4 BUSIF0 BUSIF0-7 BUSIF0-3
+ * 5 BUSIF0 BUSIF0 BUSIF0
+ * 6 BUSIF0 BUSIF0 BUSIF0
+ * 7 BUSIF0 BUSIF0 BUSIF0
+ * 8 BUSIF0 BUSIF0 BUSIF0
+ * 9 BUSIF0-3 BUSIF0-7 BUSIF0-3
+ * total 22 52 8 28
*/
static const int gen2_id[] = { 0, 4, 8, 12, 13, 14, 15, 16, 17, 18 };
static const int gen3_id[] = { 0, 8, 16, 24, 32, 40, 41, 42, 43, 44 };
static const int gen4_id[] = { 0 };
+static const int rzg3e_id[] = { 0, 4, 8, 12, 16, 20, 21, 22, 23, 24 };
+
+struct rsnd_ssiu_ctrl {
+ unsigned int busif_status_count;
+};
+
+#define rsnd_priv_to_ssiu_ctrl(priv) \
+ ((struct rsnd_ssiu_ctrl *)(priv)->ssiu_ctrl)
/* enable busif buffer over/under run interrupt. */
#define rsnd_ssiu_busif_err_irq_enable(mod) rsnd_ssiu_busif_err_irq_ctrl(mod, 1)
#define rsnd_ssiu_busif_err_irq_disable(mod) rsnd_ssiu_busif_err_irq_ctrl(mod, 0)
static void rsnd_ssiu_busif_err_irq_ctrl(struct rsnd_mod *mod, int enable)
{
+ struct rsnd_priv *priv = rsnd_mod_to_priv(mod);
int id = rsnd_mod_id(mod);
int shift, offset;
- int i;
switch (id) {
case 0:
@@ -72,7 +80,7 @@ static void rsnd_ssiu_busif_err_irq_ctrl(struct rsnd_mod *mod, int enable)
return;
}
- for (i = 0; i < 4; i++) {
+ for (unsigned int i = 0; i < rsnd_priv_to_ssiu_ctrl(priv)->busif_status_count; i++) {
enum rsnd_reg reg = SSI_SYS_INT_ENABLE((i * 2) + offset);
u32 val = 0xf << (shift * 4);
u32 sys_int_enable = rsnd_mod_read(mod, reg);
@@ -87,10 +95,10 @@ static void rsnd_ssiu_busif_err_irq_ctrl(struct rsnd_mod *mod, int enable)
bool rsnd_ssiu_busif_err_status_clear(struct rsnd_mod *mod)
{
+ struct rsnd_priv *priv = rsnd_mod_to_priv(mod);
bool error = false;
int id = rsnd_mod_id(mod);
int shift, offset;
- int i;
switch (id) {
case 0:
@@ -109,14 +117,13 @@ bool rsnd_ssiu_busif_err_status_clear(struct rsnd_mod *mod)
goto out;
}
- for (i = 0; i < 4; i++) {
+ for (unsigned int i = 0; i < rsnd_priv_to_ssiu_ctrl(priv)->busif_status_count; i++) {
u32 reg = SSI_SYS_STATUS(i * 2) + offset;
u32 status = rsnd_mod_read(mod, reg);
u32 val = 0xf << (shift * 4);
status &= val;
if (status) {
- struct rsnd_priv *priv = rsnd_mod_to_priv(mod);
struct device *dev = rsnd_priv_to_dev(priv);
rsnd_print_irq_status(dev, "%s err status : 0x%08x\n",
@@ -160,7 +167,8 @@ static int rsnd_ssiu_init(struct rsnd_mod *mod,
/*
* SSI_MODE0
*/
- rsnd_mod_bset(mod, SSI_MODE0, (1 << id), !use_busif << id);
+ if (!rsnd_is_rzg3e(priv))
+ rsnd_mod_bset(mod, SSI_MODE0, (1 << id), !use_busif << id);
/*
* SSI_MODE1 / SSI_MODE2
@@ -511,6 +519,7 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
struct device *dev = rsnd_priv_to_dev(priv);
struct device_node *node __free(device_node) = rsnd_ssiu_of_node(priv);
struct reset_control *rstc;
+ struct rsnd_ssiu_ctrl *ctrl;
struct rsnd_ssiu *ssiu;
struct rsnd_mod_ops *ops;
const int *list = NULL;
@@ -535,8 +544,15 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
if (!ssiu)
return -ENOMEM;
+ ctrl = devm_kzalloc(dev, sizeof(*ctrl), GFP_KERNEL);
+ if (!ctrl)
+ return -ENOMEM;
+
+ ctrl->busif_status_count = rsnd_flags_has(priv, RSND_SSIU_BUSIF_STATUS_COUNT_2) ? 2 : 4;
+
priv->ssiu = ssiu;
priv->ssiu_nr = nr;
+ priv->ssiu_ctrl = ctrl;
if (rsnd_is_gen1(priv))
ops = &rsnd_ssiu_ops_gen1;
@@ -559,6 +575,9 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
} else if (rsnd_is_gen4(priv)) {
list = gen4_id;
nr = ARRAY_SIZE(gen4_id);
+ } else if (rsnd_is_rzg3e(priv)) {
+ list = rzg3e_id;
+ nr = ARRAY_SIZE(rzg3e_id);
} else {
dev_err(dev, "unknown SSIU\n");
return -ENODEV;
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 10/18] ASoC: rsnd: ssiu: Add RZ/G3E BUSIF support
2026-05-25 11:02 ` [PATCH v7 10/18] ASoC: rsnd: ssiu: Add RZ/G3E BUSIF support John Madieu
@ 2026-05-25 11:47 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 11:47 UTC (permalink / raw)
To: John Madieu; +Cc: robh, devicetree, wsa+renesas, conor+dt
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [High] Out-of-bounds array access and undefined bit shifts due to unvalidated device tree node count.
--
commit cd52ab06a4f096e9c70d4a69a4f934f03c3b0b19
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: rsnd: ssiu: Add RZ/G3E BUSIF support
This commit adds support for the SSIU on the Renesas RZ/G3E SoC, which has a
different BUSIF layout compared to earlier generations. It also introduces
a capability flag to parametrize the BUSIF error-status loops, as the
RZ/G3E only has two pairs of error-status registers.
> diff --git a/sound/soc/renesas/rcar/ssiu.c b/sound/soc/renesas/rcar/ssiu.c
> index 2b922ead62d09..8d4ce9d35e9e7 100644
> --- a/sound/soc/renesas/rcar/ssiu.c
> +++ b/sound/soc/renesas/rcar/ssiu.c
[ ... ]
> @@ -559,6 +575,9 @@ int rsnd_ssiu_probe(struct rsnd_priv *priv)
> } else if (rsnd_is_gen4(priv)) {
> list = gen4_id;
> nr = ARRAY_SIZE(gen4_id);
> + } else if (rsnd_is_rzg3e(priv)) {
> + list = rzg3e_id;
> + nr = ARRAY_SIZE(rzg3e_id);
> } else {
> dev_err(dev, "unknown SSIU\n");
> return -ENODEV;
[Severity: High]
This is a pre-existing issue, but can an excessive number of SSIU nodes
from the Device Tree cause out-of-bounds array accesses and undefined
bit shifts?
In rsnd_ssiu_probe(), priv->ssiu_nr is initialized to the total number of
SSIU nodes from the DT without validating against the maximum supported bus
interface index.
During initialization, ssiu->id_sub is computed as i - list[ssiu->id].
If the DT provides too many SSIU nodes (e.g., >= 33 for RZ/G3E), id_sub
will become 8 or greater.
This out-of-bounds id_sub value is later used in rsnd_ssiu_get_status() to
return &ssiu->busif_status[busif], which can lead to kernel memory corruption
when the caller dereferences and modifies it.
Additionally, rsnd_ssiu_start_gen2() uses busif as a shift amount via
1 << (busif * 4), which can result in undefined behavior when shifting a
32-bit integer by 32 or more.
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=10
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 11/18] ASoC: rsnd: Add SSI reset support for RZ/G3E platform
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (9 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 10/18] ASoC: rsnd: ssiu: Add RZ/G3E BUSIF support John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:48 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 12/18] ASoC: rsnd: Add ADG reset support for RZ/G3E John Madieu
` (7 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
Acquire the per-SSI reset controller and pass it through
rsnd_mod_init() so it is wired into the rsnd_mod->rstc plumbing.
The RZ/G3E SoC exposes one reset line per SSI instance. Use the
indexed-name rsnd_devm_reset_control_get_optional_indexed() helper
so the same code accepts both the hyphenated RZ/G3E names
("ssi-0", "ssi-1", ...) and the legacy dotted names used by R-Car
("ssi.0", ...).
The helper returns NULL when no reset is described in DT, leaving
existing R-Car generations unaffected.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: No changes
v6:
- Acquire the per-SSI reset through the new
rsnd_devm_reset_control_get_optional_indexed() helper from
patch 04/16, so the same code accepts both the hyphenated
RZ/G3E names ("ssi-0") and the legacy dotted names ("ssi.0").
- Rewrite the commit message accordingly.
v5: No changes
v4:
- Clarify in commit message that PIO mode remains available on
R-Car Gen2/Gen3/Gen4 platforms.
v3: No changes
v2: No changes
sound/soc/renesas/rcar/ssi.c | 14 +++++++++++++-
1 file changed, 13 insertions(+), 1 deletion(-)
diff --git a/sound/soc/renesas/rcar/ssi.c b/sound/soc/renesas/rcar/ssi.c
index cf98cc9ee44c..e6734671328c 100644
--- a/sound/soc/renesas/rcar/ssi.c
+++ b/sound/soc/renesas/rcar/ssi.c
@@ -1157,6 +1157,7 @@ int __rsnd_ssi_is_pin_sharing(struct rsnd_mod *mod)
int rsnd_ssi_probe(struct rsnd_priv *priv)
{
+ struct reset_control *rstc;
struct device_node *node;
struct device *dev = rsnd_priv_to_dev(priv);
struct rsnd_mod_ops *ops;
@@ -1202,6 +1203,17 @@ int rsnd_ssi_probe(struct rsnd_priv *priv)
goto rsnd_ssi_probe_done;
}
+ /*
+ * RZ/G3E uses per-SSI reset controllers.
+ * R-Car platforms typically don't have SSI reset controls.
+ */
+ rstc = rsnd_devm_reset_control_get_optional_indexed(dev,
+ SSI_NAME, i);
+ if (IS_ERR(rstc)) {
+ ret = PTR_ERR(rstc);
+ goto rsnd_ssi_probe_done;
+ }
+
if (of_property_read_bool(np, "shared-pin"))
rsnd_flags_set(ssi, RSND_SSI_CLK_PIN_SHARE);
@@ -1220,7 +1232,7 @@ int rsnd_ssi_probe(struct rsnd_priv *priv)
ops = &rsnd_ssi_dma_ops;
ret = rsnd_mod_init(priv, rsnd_mod_get(ssi), ops, clk,
- NULL, RSND_MOD_SSI, i);
+ rstc, RSND_MOD_SSI, i);
if (ret)
goto rsnd_ssi_probe_done;
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 11/18] ASoC: rsnd: Add SSI reset support for RZ/G3E platform
2026-05-25 11:02 ` [PATCH v7 11/18] ASoC: rsnd: Add SSI reset support for RZ/G3E platform John Madieu
@ 2026-05-25 11:48 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 11:48 UTC (permalink / raw)
To: John Madieu; +Cc: robh, conor+dt, devicetree, wsa+renesas
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] The module clock is not enabled before calling reset_control_assert() during module teardown, which can cause the reset propagation to stall and time out.
--
commit ef6d940426384aba7286c552a35eb9b1119adf84
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: rsnd: Add SSI reset support for RZ/G3E platform
This commit adds SSI reset support for the RZ/G3E platform by acquiring the
per-SSI reset controller and passing it to rsnd_mod_init(). It uses the
rsnd_devm_reset_control_get_optional_indexed() helper to support both new
hyphenated and legacy dotted names.
> diff --git a/sound/soc/renesas/rcar/ssi.c b/sound/soc/renesas/rcar/ssi.c
> index cf98cc9ee44c3..e6734671328c8 100644
> --- a/sound/soc/renesas/rcar/ssi.c
> +++ b/sound/soc/renesas/rcar/ssi.c
[ ... ]
> @@ -1220,7 +1232,7 @@ int rsnd_ssi_probe(struct rsnd_priv *priv)
> ops = &rsnd_ssi_dma_ops;
>
> ret = rsnd_mod_init(priv, rsnd_mod_get(ssi), ops, clk,
> - NULL, RSND_MOD_SSI, i);
> + rstc, RSND_MOD_SSI, i);
[Severity: High]
Does this code stall or panic during module teardown?
While rsnd_mod_init() correctly enables the clock before calling
reset_control_deassert(), the teardown path in rsnd_mod_quit() calls
reset_control_assert() without first enabling the module clock.
Because rsnd_mod_quit() is executed during module teardown when streams are
stopped and module clocks are disabled:
rsnd_mod_quit()
reset_control_assert(mod->rstc);
Asserting the reset will fail to propagate through the unclocked IP block.
Will this cause the SoC's reset controller driver to time out while polling
the reset monitor bit, leaving the hardware in an inconsistent state?
> if (ret)
> goto rsnd_ssi_probe_done;
>
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=11
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 12/18] ASoC: rsnd: Add ADG reset support for RZ/G3E
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (10 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 11/18] ASoC: rsnd: Add SSI reset support for RZ/G3E platform John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:49 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 13/18] ASoC: rsnd: adg: Add per-SSI ADG and SSIF supply clock management John Madieu
` (6 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
RZ/G3E requires the ADG reset line to be deasserted for the audio
subsystem to operate. The ADG module clock is already managed via
rsnd_adg_clk_enable/disable() through adg->adg, so no additional
clock handling is needed.
Add support for the optional "adg" reset control on Renesas RZ/G3E SoC.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: No changes
v6: No changes
v5: No changes
v4:
- Collapse dev_err_probe() and rsnd_mod_init() calls to single
lines.
v3: No changes
v2: No changes
sound/soc/renesas/rcar/adg.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
diff --git a/sound/soc/renesas/rcar/adg.c b/sound/soc/renesas/rcar/adg.c
index 0105c60a144e..813ad5eabba6 100644
--- a/sound/soc/renesas/rcar/adg.c
+++ b/sound/soc/renesas/rcar/adg.c
@@ -771,6 +771,7 @@ void rsnd_adg_clk_dbg_info(struct rsnd_priv *priv, struct seq_file *m)
int rsnd_adg_probe(struct rsnd_priv *priv)
{
+ struct reset_control *rstc;
struct rsnd_adg *adg;
struct device *dev = rsnd_priv_to_dev(priv);
int ret;
@@ -779,8 +780,11 @@ int rsnd_adg_probe(struct rsnd_priv *priv)
if (!adg)
return -ENOMEM;
- ret = rsnd_mod_init(priv, &adg->mod, &adg_ops,
- NULL, NULL, 0, 0);
+ rstc = devm_reset_control_get_optional_exclusive(dev, "adg");
+ if (IS_ERR(rstc))
+ return dev_err_probe(dev, PTR_ERR(rstc), "failed to get adg reset\n");
+
+ ret = rsnd_mod_init(priv, &adg->mod, &adg_ops, NULL, rstc, 0, 0);
if (ret)
return ret;
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 12/18] ASoC: rsnd: Add ADG reset support for RZ/G3E
2026-05-25 11:02 ` [PATCH v7 12/18] ASoC: rsnd: Add ADG reset support for RZ/G3E John Madieu
@ 2026-05-25 11:49 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 11:49 UTC (permalink / raw)
To: John Madieu; +Cc: conor+dt, devicetree, wsa+renesas, robh
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The ADG module reset is deasserted before its module clock is requested or enabled, violating hardware sequencing rules.
- [Medium] The newly introduced ADG reset control is never asserted on driver unbind or probe error paths, leaking the hardware state.
--
commit d6340a23ee109a2c61a0071ea77ab96a55373b74
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: rsnd: Add ADG reset support for RZ/G3E
This commit adds support for an optional ADG reset line for the RZ/G3E SoC.
It fetches the reset control in rsnd_adg_probe() and passes it to
rsnd_mod_init() during initialization.
> diff --git a/sound/soc/renesas/rcar/adg.c b/sound/soc/renesas/rcar/adg.c
> index 0105c60a144e0..813ad5eabba61 100644
> --- a/sound/soc/renesas/rcar/adg.c
> +++ b/sound/soc/renesas/rcar/adg.c
[ ... ]
> @@ -779,8 +780,11 @@ int rsnd_adg_probe(struct rsnd_priv *priv)
> if (!adg)
> return -ENOMEM;
>
> - ret = rsnd_mod_init(priv, &adg->mod, &adg_ops,
> - NULL, NULL, 0, 0);
> + rstc = devm_reset_control_get_optional_exclusive(dev, "adg");
> + if (IS_ERR(rstc))
> + return dev_err_probe(dev, PTR_ERR(rstc), "failed to get adg reset\n");
> +
> + ret = rsnd_mod_init(priv, &adg->mod, &adg_ops, NULL, rstc, 0, 0);
[Severity: High]
Does this sequence deassert the ADG module reset before its module clock is
enabled?
When rsnd_mod_init() is called here, the clk argument is passed as NULL.
Inside rsnd_mod_init(), clk_prepare_enable() does nothing for a NULL clock,
and then reset_control_deassert() is executed immediately.
Since the ADG clock is not fetched until the subsequent call to
rsnd_adg_get_clkin() and is only enabled later in rsnd_adg_clk_enable(),
this appears to deassert a synchronous hardware reset without an active
module clock. Can this invalid sequence fail to propagate the reset state
properly or lead to bus lockups?
> if (ret)
> return ret;
>
[Severity: Medium]
Are there missing error handling and teardown paths for this new reset?
If rsnd_adg_probe() fails after rsnd_mod_init() completes successfully
(for example, if rsnd_adg_get_clkin() returns an error), the function
returns directly. It appears there is no call to rsnd_mod_quit() to unwind
the state and assert the reset line on this error path.
While the devm framework will free the reset control reference on probe
failure, it does not automatically assert the line, leaving the hardware
permanently out of reset. Is an update to rsnd_adg_remove() also needed to
ensure the reset is properly asserted during driver unbind?
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=12
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 13/18] ASoC: rsnd: adg: Add per-SSI ADG and SSIF supply clock management
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (11 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 12/18] ASoC: rsnd: Add ADG reset support for RZ/G3E John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 12:14 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 14/18] ASoC: rsnd: adg: Look up RZ/G3E clkin under audio-clk{a,b,c,i} John Madieu
` (5 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
RZ/G3E's ADG module requires explicit clock management for SSI audio
interfaces that differs from R-Car Gen2/Gen3/Gen4:
- Per-SSI ADG clocks (adg-ssi-N, or adg.ssi.N in legacy bindings)
for each SSI module
- A shared SSIF supply clock for the SSI subsystem
These clocks are acquired using optional APIs, making them transparent
to platforms that do not require them.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7:
- Replace the inline "ret |= clk_prepare(...)" accumulation in
rsnd_adg_clk_control() with dedicated rsnd_adg_ssi_clk_prepare()
and rsnd_adg_ssi_clk_unprepare() helpers. The prepare helper
stops at the first failure and unwinds only the clocks it
already prepared, instead of unconditionally unpreparing every
clock (which would underflow the state of any clock that never
prepared).
- Track prepared state with a new bool ssi_clk_prepared in
struct rsnd_adg so prepare and unprepare are idempotent.
- Preserve the first error from the surrounding for_each_rsnd_clkin
loop: adopt the SSI-clock helper's return value only when ret was
still zero, instead of OR-ing error codes together (sashiko-bot).
v6:
- Rename the per-SSI clock lookup string in code from
"adg.ssi.%d" to "adg-ssi-%d", matching the new DT binding. The
clock is RZ/G3E-only and has no legacy dotted form, so the
indexed helper from patch 04/16 is not needed here.
- Simplify rsnd_adg_clk_control(): collapse the per-SSI ADG and
SSIF supply prepare/unprepare path into the accumulating
"ret |= clk_prepare(...)" style already used for the regular
clkin loop above, restructured as a single if (enable) / else
block.
- Update the in-driver comment to refer to the hyphenated name.
v5: No changes
v4:
- Move clk_prepare/unprepare for per-SSI ADG and SSIF supply
clocks into rsnd_adg_clk_control() instead of separate
prepare/unprepare functions, centralising clock lifecycle
management.
- Return proper errors on clk_enable() failure instead of
dev_warn().
- Eliminates hw_params prepare leak concern since prepare now
happens once at probe/resume.
v3: No changes
v2:
- Split clock handling into prepare/enable phases for atomic
context safety.
sound/soc/renesas/rcar/adg.c | 124 ++++++++++++++++++++++++++++++++++-
1 file changed, 123 insertions(+), 1 deletion(-)
diff --git a/sound/soc/renesas/rcar/adg.c b/sound/soc/renesas/rcar/adg.c
index 813ad5eabba6..5dce62287d20 100644
--- a/sound/soc/renesas/rcar/adg.c
+++ b/sound/soc/renesas/rcar/adg.c
@@ -19,6 +19,9 @@
#define CLKOUT3 3
#define CLKOUTMAX 4
+/* Maximum SSI count for per-SSI clocks */
+#define ADG_SSI_MAX 10
+
#define BRGCKR_31 (1 << 31)
#define BRRx_MASK(x) (0x3FF & x)
@@ -34,10 +37,14 @@ struct rsnd_adg {
struct clk *adg;
struct clk *clkin[CLKINMAX];
struct clk *clkout[CLKOUTMAX];
+ /* RZ/G3E: per-SSI ADG clocks (adg-ssi-0 through adg-ssi-9) */
+ struct clk *clk_adg_ssi[ADG_SSI_MAX];
+ struct clk *clk_ssif_supply;
struct clk *null_clk;
struct clk_onecell_data onecell;
struct rsnd_mod mod;
int clkin_rate[CLKINMAX];
+ bool ssi_clk_prepared;
int clkin_size;
int clkout_size;
u32 ckr;
@@ -343,8 +350,16 @@ int rsnd_adg_clk_query(struct rsnd_priv *priv, unsigned int rate)
int rsnd_adg_ssi_clk_stop(struct rsnd_mod *ssi_mod)
{
+ struct rsnd_priv *priv = rsnd_mod_to_priv(ssi_mod);
+ struct rsnd_adg *adg = rsnd_priv_to_adg(priv);
+ int id = rsnd_mod_id(ssi_mod);
+
rsnd_adg_set_ssi_clk(ssi_mod, 0);
+ /* RZ/G3E: only disable here, unprepare is done in hw_free */
+ clk_disable(adg->clk_adg_ssi[id]);
+ clk_disable(adg->clk_ssif_supply);
+
return 0;
}
@@ -354,7 +369,8 @@ int rsnd_adg_ssi_clk_try_start(struct rsnd_mod *ssi_mod, unsigned int rate)
struct rsnd_adg *adg = rsnd_priv_to_adg(priv);
struct device *dev = rsnd_priv_to_dev(priv);
struct rsnd_mod *adg_mod = rsnd_mod_get(adg);
- int data;
+ int id = rsnd_mod_id(ssi_mod);
+ int ret, data;
u32 ckr = 0;
data = rsnd_adg_clk_query(priv, rate);
@@ -376,9 +392,63 @@ int rsnd_adg_ssi_clk_try_start(struct rsnd_mod *ssi_mod, unsigned int rate)
(ckr) ? adg->brg_rate[ADG_HZ_48] :
adg->brg_rate[ADG_HZ_441]);
+ /*
+ * RZ/G3E: enable per-SSI and supply clocks
+ */
+ ret = clk_enable(adg->clk_adg_ssi[id]);
+ if (ret) {
+ dev_err(dev, "Cannot enable adg-ssi-%d ADG clock\n", id);
+ return ret;
+ }
+
+ ret = clk_enable(adg->clk_ssif_supply);
+ if (ret) {
+ dev_err(dev, "Cannot enable SSIF supply clock\n");
+ clk_disable(adg->clk_adg_ssi[id]);
+ return ret;
+ }
+
return 0;
}
+static int rsnd_adg_ssi_clk_prepare(struct rsnd_adg *adg)
+{
+ int i, ret;
+
+ if (adg->ssi_clk_prepared)
+ return 0;
+
+ for (i = 0; i < ADG_SSI_MAX; i++) {
+ ret = clk_prepare(adg->clk_adg_ssi[i]);
+ if (ret)
+ goto unwind;
+ }
+ ret = clk_prepare(adg->clk_ssif_supply);
+ if (ret)
+ goto unwind;
+
+ adg->ssi_clk_prepared = true;
+ return 0;
+
+unwind:
+ while (i--)
+ clk_unprepare(adg->clk_adg_ssi[i]);
+ return ret;
+}
+
+static void rsnd_adg_ssi_clk_unprepare(struct rsnd_adg *adg)
+{
+ int i;
+
+ if (!adg->ssi_clk_prepared)
+ return;
+ adg->ssi_clk_prepared = false;
+
+ clk_unprepare(adg->clk_ssif_supply);
+ for (i = 0; i < ADG_SSI_MAX; i++)
+ clk_unprepare(adg->clk_adg_ssi[i]);
+}
+
int rsnd_adg_clk_control(struct rsnd_priv *priv, int enable)
{
struct rsnd_adg *adg = rsnd_priv_to_adg(priv);
@@ -417,6 +487,28 @@ int rsnd_adg_clk_control(struct rsnd_priv *priv, int enable)
}
}
+ /*
+ * rsnd_adg_clk_enable() might return error (_disable() will not).
+ * We need to rollback in such case
+ */
+ /*
+ * RZ/G3E per-SSI ADG and SSIF supply clocks.
+ *
+ * Follow the same style as for_each_rsnd_clkin() above: on enable,
+ * try to prepare every clock and accumulate the error. On disable,
+ * unprepare every clock. Absent optional clocks are NULL, for
+ * which clk_prepare() and clk_unprepare() are no-ops.
+ */
+ if (enable) {
+ int sub_ret = rsnd_adg_ssi_clk_prepare(adg);
+
+ /* Preserve the first error from the clkin loop above. */
+ if (sub_ret && !ret)
+ ret = sub_ret;
+ } else {
+ rsnd_adg_ssi_clk_unprepare(adg);
+ }
+
/*
* rsnd_adg_clk_enable() might return error (_disable() will not).
* We need to rollback in such case
@@ -769,6 +861,31 @@ void rsnd_adg_clk_dbg_info(struct rsnd_priv *priv, struct seq_file *m)
#define rsnd_adg_clk_dbg_info(priv, m)
#endif
+static int rsnd_adg_get_ssi_clks(struct rsnd_priv *priv)
+{
+ struct rsnd_adg *adg = rsnd_priv_to_adg(priv);
+ struct device *dev = rsnd_priv_to_dev(priv);
+ char name[16];
+ int i;
+
+ /* SSIF supply clock */
+ adg->clk_ssif_supply = devm_clk_get_optional(dev, "ssif_supply");
+ if (IS_ERR(adg->clk_ssif_supply))
+ return dev_err_probe(dev, PTR_ERR(adg->clk_ssif_supply),
+ "failed to get ssif_supply clock\n");
+
+ /* Per-SSI ADG clocks (RZ/G3E-only; no legacy dotted form exists) */
+ for (i = 0; i < ADG_SSI_MAX; i++) {
+ snprintf(name, sizeof(name), "adg-ssi-%d", i);
+ adg->clk_adg_ssi[i] = devm_clk_get_optional(dev, name);
+ if (IS_ERR(adg->clk_adg_ssi[i]))
+ return dev_err_probe(dev, PTR_ERR(adg->clk_adg_ssi[i]),
+ "failed to get %s clock\n", name);
+ }
+
+ return 0;
+}
+
int rsnd_adg_probe(struct rsnd_priv *priv)
{
struct reset_control *rstc;
@@ -798,6 +915,11 @@ int rsnd_adg_probe(struct rsnd_priv *priv)
if (ret)
return ret;
+ /* RZ/G3E-specific: per-SSI ADG and SSIF supply clocks */
+ ret = rsnd_adg_get_ssi_clks(priv);
+ if (ret)
+ return ret;
+
ret = rsnd_adg_clk_enable(priv);
if (ret)
return ret;
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 13/18] ASoC: rsnd: adg: Add per-SSI ADG and SSIF supply clock management
2026-05-25 11:02 ` [PATCH v7 13/18] ASoC: rsnd: adg: Add per-SSI ADG and SSIF supply clock management John Madieu
@ 2026-05-25 12:14 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 12:14 UTC (permalink / raw)
To: John Madieu; +Cc: wsa+renesas, devicetree, robh, conor+dt
Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [High] The SSI module ID can exceed the bounds of the `adg->clk_adg_ssi` array, leading to out-of-bounds memory access.
- [Low] Inaccurate comment states unprepare happens in `hw_free`.
--
commit 1339aa202161336aea7400a2edeb8b4b15c203d4
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: rsnd: adg: Add per-SSI ADG and SSIF supply clock management
This commit adds support for per-SSI ADG clocks and a shared SSIF supply
clock for RZ/G3E SoCs. These clocks are acquired and managed as optional
resources transparently for platforms that require them.
> diff --git a/sound/soc/renesas/rcar/adg.c b/sound/soc/renesas/rcar/adg.c
> index 813ad5eabba61..5dce62287d200 100644
> --- a/sound/soc/renesas/rcar/adg.c
> +++ b/sound/soc/renesas/rcar/adg.c
[ ... ]
> @@ -343,8 +350,16 @@ int rsnd_adg_clk_query(struct rsnd_priv *priv, unsigned int rate)
>
> int rsnd_adg_ssi_clk_stop(struct rsnd_mod *ssi_mod)
> {
> + struct rsnd_priv *priv = rsnd_mod_to_priv(ssi_mod);
> + struct rsnd_adg *adg = rsnd_priv_to_adg(priv);
> + int id = rsnd_mod_id(ssi_mod);
> +
> rsnd_adg_set_ssi_clk(ssi_mod, 0);
>
> + /* RZ/G3E: only disable here, unprepare is done in hw_free */
[Severity: Low]
This isn't a bug, but is this comment accurate regarding hw_free?
Looking at rsnd_adg_ssi_clk_unprepare(), it is only called from
rsnd_adg_clk_control(priv, 0), which runs during the driver's remove and
suspend phases, not during the PCM hw_free callback.
> + clk_disable(adg->clk_adg_ssi[id]);
[Severity: High]
Does this code overflow the adg->clk_adg_ssi array?
The array is defined with a fixed size of ADG_SSI_MAX (10), but the id
variable is dynamically derived from the Device Tree node name via
rsnd_mod_id(). This can yield indices up to 63 via rsnd_node_fixed_index().
If a Device Tree contains an SSI index of 10 or greater, won't this read
past the end of the array and pass an invalid pointer to clk_disable()?
The same issue appears to exist in rsnd_adg_ssi_clk_try_start() with
clk_enable().
> + clk_disable(adg->clk_ssif_supply);
> +
> return 0;
> }
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=13
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 14/18] ASoC: rsnd: adg: Look up RZ/G3E clkin under audio-clk{a,b,c,i}
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (12 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 13/18] ASoC: rsnd: adg: Add per-SSI ADG and SSIF supply clock management John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:53 ` sashiko-bot
2026-05-25 11:02 ` [PATCH v7 15/18] ASoC: rsnd: src: Acquire shared SCU clocks for RZ/G3E John Madieu
` (4 subsequent siblings)
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
The R-Car Sound ADG block has up to four external master-clock inputs
named CLKA, CLKB, CLKC and CLKI by the silicon. On Gen2 R-Car these
come from DT under the legacy names "clk_a", "clk_b", "clk_c", "clk_i"
defined by renesas,rsnd.yaml. Gen4 collapses them to a single "clkin".
The new standalone RZ/G3E sound binding (renesas,r9a09g047-sound.yaml)
uses the standard DT naming convention with a vendor-meaningful prefix
that matches the SoC datasheet pin labels: "audio-clka", "audio-clkb",
"audio-clkc", "audio-clki".
Add a third clkin name table for RZ/G3E and dispatch to it from
rsnd_adg_get_clkin() in the same style as the existing Gen4 branch.
The CLKA/B/C/I enum values, the clkin[] array, and the BRGA/BRGB
derivation are unchanged - only the DT lookup names differ.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: No changes
v6: New patch
sound/soc/renesas/rcar/adg.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/sound/soc/renesas/rcar/adg.c b/sound/soc/renesas/rcar/adg.c
index 5dce62287d20..483979c24319 100644
--- a/sound/soc/renesas/rcar/adg.c
+++ b/sound/soc/renesas/rcar/adg.c
@@ -77,6 +77,13 @@ static const char * const clkin_name_gen2[] = {
[CLKI] = "clk_i",
};
+static const char * const clkin_name_rzg3e[] = {
+ [CLKA] = "audio-clka",
+ [CLKB] = "audio-clkb",
+ [CLKC] = "audio-clkc",
+ [CLKI] = "audio-clki",
+};
+
static const char * const clkout_name_gen2[] = {
[CLKOUT] = "audio_clkout",
[CLKOUT1] = "audio_clkout1",
@@ -574,6 +581,9 @@ static int rsnd_adg_get_clkin(struct rsnd_priv *priv)
if (rsnd_is_gen4(priv)) {
clkin_name = clkin_name_gen4;
clkin_size = ARRAY_SIZE(clkin_name_gen4);
+ } else if (rsnd_is_rzg3e(priv)) {
+ clkin_name = clkin_name_rzg3e;
+ clkin_size = ARRAY_SIZE(clkin_name_rzg3e);
}
/*
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 14/18] ASoC: rsnd: adg: Look up RZ/G3E clkin under audio-clk{a,b,c,i}
2026-05-25 11:02 ` [PATCH v7 14/18] ASoC: rsnd: adg: Look up RZ/G3E clkin under audio-clk{a,b,c,i} John Madieu
@ 2026-05-25 11:53 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 11:53 UTC (permalink / raw)
To: John Madieu; +Cc: conor+dt, robh, wsa+renesas, devicetree
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
Pre-existing issues:
- [High] Error handling in `rsnd_adg_null_clk_clean` blindly dereferences an `ERR_PTR`, leading to a guaranteed kernel panic on deferred probe.
--
commit cc9d8467b86b1274ca6e7e69addd9762d7c23073
Author: John Madieu <john.madieu.xa@bp.renesas.com>
ASoC: rsnd: adg: Look up RZ/G3E clkin under audio-clk{a,b,c,i}
This patch updates the R-Car Sound ADG block driver to support the RZ/G3E
sound binding. It introduces a new array mapping the master clock inputs
to their DT naming conventions. The driver is also modified to dispatch to
this new table when the RZ/G3E SoC is detected.
> diff --git a/sound/soc/renesas/rcar/adg.c b/sound/soc/renesas/rcar/adg.c
> index 5dce62287d200..483979c243198 100644
> --- a/sound/soc/renesas/rcar/adg.c
> +++ b/sound/soc/renesas/rcar/adg.c
[ ... ]
> @@ -574,6 +581,9 @@ static int rsnd_adg_get_clkin(struct rsnd_priv *priv)
> if (rsnd_is_gen4(priv)) {
> clkin_name = clkin_name_gen4;
> clkin_size = ARRAY_SIZE(clkin_name_gen4);
> + } else if (rsnd_is_rzg3e(priv)) {
> + clkin_name = clkin_name_rzg3e;
> + clkin_size = ARRAY_SIZE(clkin_name_rzg3e);
> }
>
> /*
[Severity: High]
This is a pre-existing issue, but does this addition make the error handling
more likely to trigger a boot crash?
If clk_register_fixed_rate() fails, it returns an ERR_PTR which is cached in
adg->null_clk. The cleanup function rsnd_adg_null_clk_clean() checks
if (adg->null_clk) (which evaluates to true for an ERR_PTR like -EEXIST or
-ENOMEM) and blindly passes it to clk_unregister_fixed_rate().
Inside the clock framework, __clk_get_hw() dereferences the invalid pointer,
causing an immediate kernel panic.
This reliably crashes the system on deferred probe: if rsnd_probe() returns an
error after rsnd_adg_probe() successfully registers the rsnd_adg_null clock,
the driver aborts without calling rsnd_adg_remove().
This leaks the registered clock. On the next deferred probe attempt,
clk_register_fixed_rate() fails with -EEXIST (returning ERR_PTR(-EEXIST)).
The driver then jumps to the error path, passes this ERR_PTR to
clk_unregister_fixed_rate(), and panics.
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=14
^ permalink raw reply [flat|nested] 31+ messages in thread
* [PATCH v7 15/18] ASoC: rsnd: src: Acquire shared SCU clocks for RZ/G3E
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (13 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 14/18] ASoC: rsnd: adg: Look up RZ/G3E clkin under audio-clk{a,b,c,i} John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:02 ` [PATCH v7 16/18] ASoC: rsnd: src: Add SRC reset support " John Madieu
` (3 subsequent siblings)
18 siblings, 0 replies; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
The RZ/G3E SoC requires explicit SCU (Sampling Rate Converter Unit)
clock management:
- scu: SCU top-level module clock (CPG_CLKON_15.CLK6_ON)
- scu_x2: SCU top-level double-rate clock (CPG_CLKON_15.CLK7_ON)
- scu_supply: SCU register-access / housekeeping clock
(CPG_CLKON_23.CLK14_ON)
Without every one of them enabled, no SCU register is reachable.
Hold these in a new struct rsnd_src_ctrl and acquire them with
devm_clk_get_optional_enabled(). scu_supply is intentionally left
untouched by the system suspend/resume path added later in the
series, so SCU registers stay reachable across PM transitions.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: New patch. Split out of [PATCH v6 14/16] "ASoC: rsnd: src: Add
SRC reset and clock support for RZ/G3E" per Kuninori Morimoto's
request that the SCU clock and the SCU reset be independent
changes. This patch carries only struct rsnd_src_ctrl and the
scu / scu_x2 / scu_supply acquisition via
devm_clk_get_optional_enabled(), so non-RZ/G3E DTs are
unaffected.
sound/soc/renesas/rcar/rsnd.h | 1 +
sound/soc/renesas/rcar/src.c | 37 +++++++++++++++++++++++++++++++++++
2 files changed, 38 insertions(+)
diff --git a/sound/soc/renesas/rcar/rsnd.h b/sound/soc/renesas/rcar/rsnd.h
index 95843a20c43c..7d7da6cecf09 100644
--- a/sound/soc/renesas/rcar/rsnd.h
+++ b/sound/soc/renesas/rcar/rsnd.h
@@ -698,6 +698,7 @@ struct rsnd_priv {
/*
* below value will be filled on rsnd_src_probe()
*/
+ void *src_ctrl;
void *src;
int src_nr;
diff --git a/sound/soc/renesas/rcar/src.c b/sound/soc/renesas/rcar/src.c
index 43abe13137bf..88ea9aad5cae 100644
--- a/sound/soc/renesas/rcar/src.c
+++ b/sound/soc/renesas/rcar/src.c
@@ -53,6 +53,14 @@ struct rsnd_src {
((pos) = (struct rsnd_src *)(priv)->src + i); \
i++)
+struct rsnd_src_ctrl {
+ struct clk *scu;
+ struct clk *scu_x2;
+ struct clk *scu_supply;
+};
+
+#define rsnd_priv_to_src_ctrl(priv) \
+ ((struct rsnd_src_ctrl *)(priv)->src_ctrl)
/*
* image of SRC (Sampling Rate Converter)
@@ -712,6 +720,7 @@ int rsnd_src_probe(struct rsnd_priv *priv)
{
struct device_node *node;
struct device *dev = rsnd_priv_to_dev(priv);
+ struct rsnd_src_ctrl *src_ctrl;
struct rsnd_src *src;
struct clk *clk;
int i, nr, ret;
@@ -726,6 +735,12 @@ int rsnd_src_probe(struct rsnd_priv *priv)
goto rsnd_src_probe_done;
}
+ src_ctrl = devm_kzalloc(dev, sizeof(*src_ctrl), GFP_KERNEL);
+ if (!src_ctrl) {
+ ret = -ENOMEM;
+ goto rsnd_src_probe_done;
+ }
+
src = devm_kcalloc(dev, nr, sizeof(*src), GFP_KERNEL);
if (!src) {
ret = -ENOMEM;
@@ -734,6 +749,28 @@ int rsnd_src_probe(struct rsnd_priv *priv)
priv->src_nr = nr;
priv->src = src;
+ priv->src_ctrl = src_ctrl;
+
+ src_ctrl->scu = devm_clk_get_optional_enabled(dev, "scu");
+ if (IS_ERR(src_ctrl->scu)) {
+ ret = dev_err_probe(dev, PTR_ERR(src_ctrl->scu),
+ "failed to get scu clock\n");
+ goto rsnd_src_probe_done;
+ }
+
+ src_ctrl->scu_x2 = devm_clk_get_optional_enabled(dev, "scu_x2");
+ if (IS_ERR(src_ctrl->scu_x2)) {
+ ret = dev_err_probe(dev, PTR_ERR(src_ctrl->scu_x2),
+ "failed to get scu_x2 clock\n");
+ goto rsnd_src_probe_done;
+ }
+
+ src_ctrl->scu_supply = devm_clk_get_optional_enabled(dev, "scu_supply");
+ if (IS_ERR(src_ctrl->scu_supply)) {
+ ret = dev_err_probe(dev, PTR_ERR(src_ctrl->scu_supply),
+ "failed to get scu_supply clock\n");
+ goto rsnd_src_probe_done;
+ }
i = 0;
for_each_child_of_node_scoped(node, np) {
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* [PATCH v7 16/18] ASoC: rsnd: src: Add SRC reset support for RZ/G3E
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (14 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 15/18] ASoC: rsnd: src: Acquire shared SCU clocks for RZ/G3E John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:02 ` [PATCH v7 17/18] ASoC: rsnd: Support unprefixed DT node names " John Madieu
` (2 subsequent siblings)
18 siblings, 0 replies; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
The RZ/G3E SoC exposes a shared SCU reset controller used by all SRC
modules. Acquire it once and pass it through per-instance's
rsnd_mod_init() so it is wired into the rsnd_mod->rstc plumbing.
devm_reset_control_get_optional_shared() returns NULL when no reset
is described in DT, leaving existing R-Car generations unaffected.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: New patch. Split out of [PATCH v6 14/16] "ASoC: rsnd: src: Add
SRC reset and clock support for RZ/G3E" per Kuninori Morimoto's
request. This patch carries only the shared "scu" reset
acquisition and its rsnd_mod_init() plumbing.
- Acquire the shared "scu" reset once, before the per-SRC probe
loop, instead of inside it.
sound/soc/renesas/rcar/src.c | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/sound/soc/renesas/rcar/src.c b/sound/soc/renesas/rcar/src.c
index 88ea9aad5cae..cad15fcc37f2 100644
--- a/sound/soc/renesas/rcar/src.c
+++ b/sound/soc/renesas/rcar/src.c
@@ -720,6 +720,7 @@ int rsnd_src_probe(struct rsnd_priv *priv)
{
struct device_node *node;
struct device *dev = rsnd_priv_to_dev(priv);
+ struct reset_control *rstc;
struct rsnd_src_ctrl *src_ctrl;
struct rsnd_src *src;
struct clk *clk;
@@ -772,6 +773,16 @@ int rsnd_src_probe(struct rsnd_priv *priv)
goto rsnd_src_probe_done;
}
+ /*
+ * Shared SCU reset for every SRC module; acquire once.
+ * R-Car platforms typically don't have SRC reset controls.
+ */
+ rstc = devm_reset_control_get_optional_shared(dev, "scu");
+ if (IS_ERR(rstc)) {
+ ret = PTR_ERR(rstc);
+ goto rsnd_src_probe_done;
+ }
+
i = 0;
for_each_child_of_node_scoped(node, np) {
if (!of_device_is_available(np))
@@ -798,7 +809,7 @@ int rsnd_src_probe(struct rsnd_priv *priv)
}
ret = rsnd_mod_init(priv, rsnd_mod_get(src),
- &rsnd_src_ops, clk, NULL, RSND_MOD_SRC, i);
+ &rsnd_src_ops, clk, rstc, RSND_MOD_SRC, i);
if (ret)
goto rsnd_src_probe_done;
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* [PATCH v7 17/18] ASoC: rsnd: Support unprefixed DT node names for RZ/G3E
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (15 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 16/18] ASoC: rsnd: src: Add SRC reset support " John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 11:02 ` [PATCH v7 18/18] ASoC: rsnd: Add system suspend/resume support John Madieu
2026-05-25 22:45 ` [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support Kuninori Morimoto
18 siblings, 0 replies; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
The RZ/G3E device tree binding uses standard unprefixed node names
("ssi", "ssiu", "src", "dvc", "mix", "ctu") instead of the legacy
"rcar_sound," prefixed names used by R-Car bindings.
Convert rsnd_parse_of_node() from a macro into a function that tries
the legacy prefixed name first, then falls back to the unprefixed name
by stripping the "rcar_sound," prefix. This makes the driver work
transparently with both old and new bindings.
While at it, update the related comments in dma.c, ssi.c and ssiu.c
that reference the hardcoded "rcar_sound,ssiu" / "rcar_sound,ssi"
names to note that the driver now accepts both the prefixed and the
unprefixed forms.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: No changes
v6:
- Factor the "rcar_sound," prefix-strip into a small static
helper rsnd_node_name_strip_prefix() and call it from
rsnd_parse_of_node(), so the convention is in one place.
- Drop the of_node_name_eq() "dai" fallback that v5 added to
rsnd_dai_of_node() and rsnd_dai_probe(); not needed once
rsnd_parse_of_node() handles both forms uniformly.
v5: New patch.
sound/soc/renesas/rcar/core.c | 34 ++++++++++++++++++++++++++++++++++
sound/soc/renesas/rcar/dma.c | 8 ++++----
sound/soc/renesas/rcar/rsnd.h | 4 ++--
sound/soc/renesas/rcar/ssi.c | 8 ++++----
sound/soc/renesas/rcar/ssiu.c | 8 ++++----
5 files changed, 48 insertions(+), 14 deletions(-)
diff --git a/sound/soc/renesas/rcar/core.c b/sound/soc/renesas/rcar/core.c
index f5c8ba8c5d56..fbf7f7236460 100644
--- a/sound/soc/renesas/rcar/core.c
+++ b/sound/soc/renesas/rcar/core.c
@@ -1301,6 +1301,40 @@ rsnd_devm_reset_control_get_optional_indexed(struct device *dev,
return devm_reset_control_get_optional(dev, name);
}
+/*
+ * Strip the "rcar_sound," prefix from a legacy node name.
+ *
+ * The RZ/G3E binding uses unprefixed sub-node names (e.g. "ssi",
+ * "ssiu") while earlier R-Car bindings use the legacy "rcar_sound,*"
+ * form. This helper returns the unprefixed portion (the part after
+ * the comma) or NULL if there is no prefix.
+ *
+ * Centralising the convention here keeps every call site consistent.
+ */
+static const char *rsnd_node_name_strip_prefix(const char *name)
+{
+ const char *comma = strchr(name, ',');
+
+ return comma ? comma + 1 : NULL;
+}
+
+struct device_node *rsnd_parse_of_node(struct rsnd_priv *priv, const char *name)
+{
+ struct device_node *np = rsnd_priv_to_dev(priv)->of_node;
+ struct device_node *node;
+ const char *unprefixed;
+
+ node = of_get_child_by_name(np, name);
+ if (node)
+ return node;
+
+ unprefixed = rsnd_node_name_strip_prefix(name);
+ if (unprefixed)
+ node = of_get_child_by_name(np, unprefixed);
+
+ return node;
+}
+
static struct device_node*
rsnd_pick_endpoint_node_for_ports(struct device_node *e_ports,
struct device_node *e_port)
diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
index 92974610ac15..537b71841f8e 100644
--- a/sound/soc/renesas/rcar/dma.c
+++ b/sound/soc/renesas/rcar/dma.c
@@ -794,11 +794,11 @@ static void rsnd_dma_of_path(struct rsnd_mod *this,
int nr, i, idx;
/*
- * It should use "rcar_sound,ssiu" on DT.
- * But, we need to keep compatibility for old version.
+ * It should use "rcar_sound,ssiu" (R-Car) or "ssiu" (RZ/G3E) on DT.
+ * We need to keep compatibility for old version.
*
- * If it has "rcar_sound.ssiu", it will be used.
- * If not, "rcar_sound.ssi" will be used.
+ * If it has "rcar_sound.ssiu" or "ssiu", it will be used.
+ * If not, "rcar_sound.ssi" or "ssi" will be used.
* see
* rsnd_ssiu_dma_req()
* rsnd_ssi_dma_req()
diff --git a/sound/soc/renesas/rcar/rsnd.h b/sound/soc/renesas/rcar/rsnd.h
index 7d7da6cecf09..f38bd92d4faf 100644
--- a/sound/soc/renesas/rcar/rsnd.h
+++ b/sound/soc/renesas/rcar/rsnd.h
@@ -501,8 +501,8 @@ rsnd_devm_reset_control_get_optional_indexed(struct device *dev,
/*
* DT
*/
-#define rsnd_parse_of_node(priv, node) \
- of_get_child_by_name(rsnd_priv_to_dev(priv)->of_node, node)
+struct device_node *rsnd_parse_of_node(struct rsnd_priv *priv, const char *name);
+
#define RSND_NODE_DAI "rcar_sound,dai"
#define RSND_NODE_SSI "rcar_sound,ssi"
#define RSND_NODE_SSIU "rcar_sound,ssiu"
diff --git a/sound/soc/renesas/rcar/ssi.c b/sound/soc/renesas/rcar/ssi.c
index e6734671328c..007a7c91d470 100644
--- a/sound/soc/renesas/rcar/ssi.c
+++ b/sound/soc/renesas/rcar/ssi.c
@@ -1009,11 +1009,11 @@ static struct dma_chan *rsnd_ssi_dma_req(struct rsnd_dai_stream *io,
char *name;
/*
- * It should use "rcar_sound,ssiu" on DT.
- * But, we need to keep compatibility for old version.
+ * It should use "rcar_sound,ssiu" (R-Car) or "ssiu" (RZ/G3E) on DT.
+ * We need to keep compatibility for old version.
*
- * If it has "rcar_sound.ssiu", it will be used.
- * If not, "rcar_sound.ssi" will be used.
+ * If it has "rcar_sound.ssiu" or "ssiu", it will be used.
+ * If not, "rcar_sound.ssi" or "ssi" will be used.
* see
* rsnd_ssiu_dma_req()
* rsnd_dma_of_path()
diff --git a/sound/soc/renesas/rcar/ssiu.c b/sound/soc/renesas/rcar/ssiu.c
index 8d4ce9d35e9e..7d3d463c21bf 100644
--- a/sound/soc/renesas/rcar/ssiu.c
+++ b/sound/soc/renesas/rcar/ssiu.c
@@ -400,11 +400,11 @@ static struct dma_chan *rsnd_ssiu_dma_req(struct rsnd_dai_stream *io,
char *name;
/*
- * It should use "rcar_sound,ssiu" on DT.
- * But, we need to keep compatibility for old version.
+ * It should use "rcar_sound,ssiu" (R-Car) or "ssiu" (RZ/G3E) on DT.
+ * We need to keep compatibility for old versions.
*
- * If it has "rcar_sound.ssiu", it will be used.
- * If not, "rcar_sound.ssi" will be used.
+ * If it has "rcar_sound.ssiu" or "ssiu", it will be used.
+ * If not, "rcar_sound.ssi" or "ssi" will be used.
* see
* rsnd_ssi_dma_req()
* rsnd_dma_of_path()
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* [PATCH v7 18/18] ASoC: rsnd: Add system suspend/resume support
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (16 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 17/18] ASoC: rsnd: Support unprefixed DT node names " John Madieu
@ 2026-05-25 11:02 ` John Madieu
2026-05-25 12:23 ` sashiko-bot
2026-05-25 22:45 ` [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support Kuninori Morimoto
18 siblings, 1 reply; 31+ messages in thread
From: John Madieu @ 2026-05-25 11:02 UTC (permalink / raw)
To: Kuninori Morimoto, Mark Brown, Liam Girdwood, Rob Herring,
Krzysztof Kozlowski, Conor Dooley, Geert Uytterhoeven
Cc: Jaroslav Kysela, Takashi Iwai, Magnus Damm, Philipp Zabel,
Claudiu Beznea, Biju Das, john.madieu, linux-sound,
linux-renesas-soc, devicetree, linux-kernel, John Madieu
Add system suspend/resume support for the ASoC rsnd driver, required
for RZ/G3E platforms. Distribute the per-module suspend/resume work
across the relevant files (adg.c, ssi.c, ssiu.c, src.c, ctu.c, mix.c,
dvc.c, dma.c) rather than centralising it in core.c.
Signed-off-by: John Madieu <john.madieu.xa@bp.renesas.com>
---
Changes:
v7: No changes
v6:
- rsnd_dma_suspend()/_resume(): respect the audmapp probe
ordering. On suspend, assert the reset before disabling the
clock; on resume, enable the clock before deasserting the
reset.
- rsnd_src_suspend()/_resume(): cache src_ctrl in a local and
add an "if (!src_ctrl) return;" early-out so platforms that
do not instantiate the SRC subsystem are safe.
v5: No changes
v4:
- Absorb rsnd_adg_mod_get() helper directly instead of a
separate preparatory patch.
- Distribute suspend/resume declarations into their respective
IP sections in rsnd.h.
v3: No changes
v2:
- Distribute suspend/resume into per-module files (ssi.c,
ssiu.c, src.c, ctu.c, mix.c, dvc.c, adg.c, dma.c) instead of
monolithic loops in core.c, following Morimoto-san's
architecture suggestion.
sound/soc/renesas/rcar/adg.c | 26 +++++++++++++++++++++
sound/soc/renesas/rcar/core.c | 43 +++++++++++++++++++++++++++++++++--
sound/soc/renesas/rcar/ctu.c | 20 ++++++++++++++++
sound/soc/renesas/rcar/dma.c | 22 ++++++++++++++++++
sound/soc/renesas/rcar/dvc.c | 20 ++++++++++++++++
sound/soc/renesas/rcar/mix.c | 20 ++++++++++++++++
sound/soc/renesas/rcar/rsnd.h | 18 +++++++++++++++
sound/soc/renesas/rcar/src.c | 34 +++++++++++++++++++++++++++
sound/soc/renesas/rcar/ssi.c | 20 ++++++++++++++++
sound/soc/renesas/rcar/ssiu.c | 20 ++++++++++++++++
10 files changed, 241 insertions(+), 2 deletions(-)
diff --git a/sound/soc/renesas/rcar/adg.c b/sound/soc/renesas/rcar/adg.c
index 483979c24319..5479cefb6dbe 100644
--- a/sound/soc/renesas/rcar/adg.c
+++ b/sound/soc/renesas/rcar/adg.c
@@ -953,3 +953,29 @@ void rsnd_adg_remove(struct rsnd_priv *priv)
/* It should be called after rsnd_adg_clk_disable() */
rsnd_adg_null_clk_clean(priv);
}
+
+static struct rsnd_mod *rsnd_adg_mod_get(struct rsnd_priv *priv)
+{
+ struct rsnd_adg *adg = rsnd_priv_to_adg(priv);
+
+ if (!adg)
+ return NULL;
+
+ return rsnd_mod_get(adg);
+}
+
+void rsnd_adg_suspend(struct rsnd_priv *priv)
+{
+ struct rsnd_mod *mod = rsnd_adg_mod_get(priv);
+
+ if (mod)
+ rsnd_suspend_clk_reset(mod->clk, mod->rstc);
+}
+
+void rsnd_adg_resume(struct rsnd_priv *priv)
+{
+ struct rsnd_mod *mod = rsnd_adg_mod_get(priv);
+
+ if (mod)
+ rsnd_resume_clk_reset(mod->clk, mod->rstc);
+}
diff --git a/sound/soc/renesas/rcar/core.c b/sound/soc/renesas/rcar/core.c
index fbf7f7236460..9ce56cd84f46 100644
--- a/sound/soc/renesas/rcar/core.c
+++ b/sound/soc/renesas/rcar/core.c
@@ -962,7 +962,8 @@ static int rsnd_soc_hw_rule_channels(struct snd_pcm_hw_params *params,
static const struct snd_pcm_hardware rsnd_pcm_hardware = {
.info = SNDRV_PCM_INFO_INTERLEAVED |
SNDRV_PCM_INFO_MMAP |
- SNDRV_PCM_INFO_MMAP_VALID,
+ SNDRV_PCM_INFO_MMAP_VALID |
+ SNDRV_PCM_INFO_RESUME,
.buffer_bytes_max = 64 * 1024,
.period_bytes_min = 32,
.period_bytes_max = 8192,
@@ -2159,11 +2160,35 @@ static void rsnd_remove(struct platform_device *pdev)
remove_func[i](priv);
}
+void rsnd_suspend_clk_reset(struct clk *clk, struct reset_control *rstc)
+{
+ clk_unprepare(clk);
+ reset_control_assert(rstc);
+}
+
+void rsnd_resume_clk_reset(struct clk *clk, struct reset_control *rstc)
+{
+ reset_control_deassert(rstc);
+ clk_prepare(clk);
+}
+
static int rsnd_suspend(struct device *dev)
{
struct rsnd_priv *priv = dev_get_drvdata(dev);
+ /*
+ * Reverse order of probe:
+ * ADG -> DVC -> MIX -> CTU -> SRC -> SSIU -> SSI -> DMA
+ */
rsnd_adg_clk_disable(priv);
+ rsnd_adg_suspend(priv);
+ rsnd_dvc_suspend(priv);
+ rsnd_mix_suspend(priv);
+ rsnd_ctu_suspend(priv);
+ rsnd_src_suspend(priv);
+ rsnd_ssiu_suspend(priv);
+ rsnd_ssi_suspend(priv);
+ rsnd_dma_suspend(priv);
return 0;
}
@@ -2172,7 +2197,21 @@ static int rsnd_resume(struct device *dev)
{
struct rsnd_priv *priv = dev_get_drvdata(dev);
- return rsnd_adg_clk_enable(priv);
+ /*
+ * Same order as probe:
+ * DMA -> SSI -> SSIU -> SRC -> CTU -> MIX -> DVC -> ADG
+ */
+ rsnd_dma_resume(priv);
+ rsnd_ssi_resume(priv);
+ rsnd_ssiu_resume(priv);
+ rsnd_src_resume(priv);
+ rsnd_ctu_resume(priv);
+ rsnd_mix_resume(priv);
+ rsnd_dvc_resume(priv);
+ rsnd_adg_resume(priv);
+ rsnd_adg_clk_enable(priv);
+
+ return 0;
}
static const struct dev_pm_ops rsnd_pm_ops = {
diff --git a/sound/soc/renesas/rcar/ctu.c b/sound/soc/renesas/rcar/ctu.c
index 293b0eec1ded..7db0fb3612bc 100644
--- a/sound/soc/renesas/rcar/ctu.c
+++ b/sound/soc/renesas/rcar/ctu.c
@@ -378,3 +378,23 @@ void rsnd_ctu_remove(struct rsnd_priv *priv)
rsnd_mod_quit(rsnd_mod_get(ctu));
}
}
+
+void rsnd_ctu_suspend(struct rsnd_priv *priv)
+{
+ struct rsnd_ctu *ctu;
+ int i;
+
+ for_each_rsnd_ctu(ctu, priv, i)
+ rsnd_suspend_clk_reset(rsnd_mod_get(ctu)->clk,
+ rsnd_mod_get(ctu)->rstc);
+}
+
+void rsnd_ctu_resume(struct rsnd_priv *priv)
+{
+ struct rsnd_ctu *ctu;
+ int i;
+
+ for_each_rsnd_ctu(ctu, priv, i)
+ rsnd_resume_clk_reset(rsnd_mod_get(ctu)->clk,
+ rsnd_mod_get(ctu)->rstc);
+}
diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
index 537b71841f8e..793dd4adbe5c 100644
--- a/sound/soc/renesas/rcar/dma.c
+++ b/sound/soc/renesas/rcar/dma.c
@@ -1035,3 +1035,25 @@ int rsnd_dma_probe(struct rsnd_priv *priv)
/* dummy mem mod for debug */
return rsnd_mod_init(NULL, &mem, &mem_ops, NULL, NULL, 0, 0);
}
+
+void rsnd_dma_suspend(struct rsnd_priv *priv)
+{
+ struct rsnd_dma_ctrl *dmac = rsnd_priv_to_dmac(priv);
+
+ if (dmac) {
+ /* Mirror probe (which enables clk before deasserting reset) */
+ rsnd_suspend_clk_reset(NULL, dmac->audmapp_rstc);
+ clk_disable_unprepare(dmac->audmapp_clk);
+ }
+}
+
+void rsnd_dma_resume(struct rsnd_priv *priv)
+{
+ struct rsnd_dma_ctrl *dmac = rsnd_priv_to_dmac(priv);
+
+ if (dmac) {
+ /* Clock must be stable before reset is deasserted */
+ clk_prepare_enable(dmac->audmapp_clk);
+ rsnd_resume_clk_reset(NULL, dmac->audmapp_rstc);
+ }
+}
diff --git a/sound/soc/renesas/rcar/dvc.c b/sound/soc/renesas/rcar/dvc.c
index 26f80d542da8..7601dfb0810a 100644
--- a/sound/soc/renesas/rcar/dvc.c
+++ b/sound/soc/renesas/rcar/dvc.c
@@ -381,3 +381,23 @@ void rsnd_dvc_remove(struct rsnd_priv *priv)
rsnd_mod_quit(rsnd_mod_get(dvc));
}
}
+
+void rsnd_dvc_suspend(struct rsnd_priv *priv)
+{
+ struct rsnd_dvc *dvc;
+ int i;
+
+ for_each_rsnd_dvc(dvc, priv, i)
+ rsnd_suspend_clk_reset(rsnd_mod_get(dvc)->clk,
+ rsnd_mod_get(dvc)->rstc);
+}
+
+void rsnd_dvc_resume(struct rsnd_priv *priv)
+{
+ struct rsnd_dvc *dvc;
+ int i;
+
+ for_each_rsnd_dvc(dvc, priv, i)
+ rsnd_resume_clk_reset(rsnd_mod_get(dvc)->clk,
+ rsnd_mod_get(dvc)->rstc);
+}
diff --git a/sound/soc/renesas/rcar/mix.c b/sound/soc/renesas/rcar/mix.c
index 9ffa591aa4a4..c4da4c4bedb3 100644
--- a/sound/soc/renesas/rcar/mix.c
+++ b/sound/soc/renesas/rcar/mix.c
@@ -345,3 +345,23 @@ void rsnd_mix_remove(struct rsnd_priv *priv)
rsnd_mod_quit(rsnd_mod_get(mix));
}
}
+
+void rsnd_mix_suspend(struct rsnd_priv *priv)
+{
+ struct rsnd_mix *mix;
+ int i;
+
+ for_each_rsnd_mix(mix, priv, i)
+ rsnd_suspend_clk_reset(rsnd_mod_get(mix)->clk,
+ rsnd_mod_get(mix)->rstc);
+}
+
+void rsnd_mix_resume(struct rsnd_priv *priv)
+{
+ struct rsnd_mix *mix;
+ int i;
+
+ for_each_rsnd_mix(mix, priv, i)
+ rsnd_resume_clk_reset(rsnd_mod_get(mix)->clk,
+ rsnd_mod_get(mix)->rstc);
+}
diff --git a/sound/soc/renesas/rcar/rsnd.h b/sound/soc/renesas/rcar/rsnd.h
index f38bd92d4faf..b480085fb0e7 100644
--- a/sound/soc/renesas/rcar/rsnd.h
+++ b/sound/soc/renesas/rcar/rsnd.h
@@ -267,6 +267,8 @@ u32 rsnd_get_busif_shift(struct rsnd_dai_stream *io, struct rsnd_mod *mod);
int rsnd_dma_attach(struct rsnd_dai_stream *io,
struct rsnd_mod *mod, struct rsnd_mod **dma_mod);
int rsnd_dma_probe(struct rsnd_priv *priv);
+void rsnd_dma_suspend(struct rsnd_priv *priv);
+void rsnd_dma_resume(struct rsnd_priv *priv);
struct dma_chan *rsnd_dma_request_channel(struct device_node *of_node, char *name,
struct rsnd_mod *mod, char *x);
@@ -429,6 +431,8 @@ int rsnd_mod_init(struct rsnd_priv *priv,
enum rsnd_mod_type type,
int id);
void rsnd_mod_quit(struct rsnd_mod *mod);
+void rsnd_suspend_clk_reset(struct clk *clk, struct reset_control *rstc);
+void rsnd_resume_clk_reset(struct clk *clk, struct reset_control *rstc);
struct dma_chan *rsnd_mod_dma_req(struct rsnd_dai_stream *io,
struct rsnd_mod *mod);
void rsnd_mod_interrupt(struct rsnd_mod *mod,
@@ -625,6 +629,8 @@ int rsnd_adg_ssi_clk_stop(struct rsnd_mod *ssi_mod);
int rsnd_adg_ssi_clk_try_start(struct rsnd_mod *ssi_mod, unsigned int rate);
int rsnd_adg_probe(struct rsnd_priv *priv);
void rsnd_adg_remove(struct rsnd_priv *priv);
+void rsnd_adg_suspend(struct rsnd_priv *priv);
+void rsnd_adg_resume(struct rsnd_priv *priv);
int rsnd_adg_set_src_timesel_gen2(struct rsnd_mod *src_mod,
struct rsnd_dai_stream *io,
unsigned int in_rate,
@@ -822,6 +828,8 @@ extern const char * const volume_ramp_rate[];
*/
int rsnd_ssi_probe(struct rsnd_priv *priv);
void rsnd_ssi_remove(struct rsnd_priv *priv);
+void rsnd_ssi_suspend(struct rsnd_priv *priv);
+void rsnd_ssi_resume(struct rsnd_priv *priv);
struct rsnd_mod *rsnd_ssi_mod_get(struct rsnd_priv *priv, int id);
int rsnd_ssi_use_busif(struct rsnd_dai_stream *io);
u32 rsnd_ssi_multi_secondaries_runtime(struct rsnd_dai_stream *io);
@@ -845,6 +853,8 @@ int rsnd_ssiu_attach(struct rsnd_dai_stream *io,
struct rsnd_mod *mod);
int rsnd_ssiu_probe(struct rsnd_priv *priv);
void rsnd_ssiu_remove(struct rsnd_priv *priv);
+void rsnd_ssiu_suspend(struct rsnd_priv *priv);
+void rsnd_ssiu_resume(struct rsnd_priv *priv);
void rsnd_parse_connect_ssiu(struct rsnd_dai *rdai,
struct device_node *playback,
struct device_node *capture);
@@ -856,6 +866,8 @@ bool rsnd_ssiu_busif_err_status_clear(struct rsnd_mod *mod);
*/
int rsnd_src_probe(struct rsnd_priv *priv);
void rsnd_src_remove(struct rsnd_priv *priv);
+void rsnd_src_suspend(struct rsnd_priv *priv);
+void rsnd_src_resume(struct rsnd_priv *priv);
struct rsnd_mod *rsnd_src_mod_get(struct rsnd_priv *priv, int id);
#define rsnd_src_get_in_rate(priv, io) rsnd_src_get_rate(priv, io, 1)
@@ -875,6 +887,8 @@ unsigned int rsnd_src_get_rate(struct rsnd_priv *priv,
*/
int rsnd_ctu_probe(struct rsnd_priv *priv);
void rsnd_ctu_remove(struct rsnd_priv *priv);
+void rsnd_ctu_suspend(struct rsnd_priv *priv);
+void rsnd_ctu_resume(struct rsnd_priv *priv);
struct rsnd_mod *rsnd_ctu_mod_get(struct rsnd_priv *priv, int id);
#define rsnd_ctu_of_node(priv) rsnd_parse_of_node(priv, RSND_NODE_CTU)
#define rsnd_parse_connect_ctu(rdai, playback, capture) \
@@ -887,6 +901,8 @@ struct rsnd_mod *rsnd_ctu_mod_get(struct rsnd_priv *priv, int id);
*/
int rsnd_mix_probe(struct rsnd_priv *priv);
void rsnd_mix_remove(struct rsnd_priv *priv);
+void rsnd_mix_suspend(struct rsnd_priv *priv);
+void rsnd_mix_resume(struct rsnd_priv *priv);
struct rsnd_mod *rsnd_mix_mod_get(struct rsnd_priv *priv, int id);
#define rsnd_mix_of_node(priv) rsnd_parse_of_node(priv, RSND_NODE_MIX)
#define rsnd_parse_connect_mix(rdai, playback, capture) \
@@ -899,6 +915,8 @@ struct rsnd_mod *rsnd_mix_mod_get(struct rsnd_priv *priv, int id);
*/
int rsnd_dvc_probe(struct rsnd_priv *priv);
void rsnd_dvc_remove(struct rsnd_priv *priv);
+void rsnd_dvc_suspend(struct rsnd_priv *priv);
+void rsnd_dvc_resume(struct rsnd_priv *priv);
struct rsnd_mod *rsnd_dvc_mod_get(struct rsnd_priv *priv, int id);
#define rsnd_dvc_of_node(priv) rsnd_parse_of_node(priv, RSND_NODE_DVC)
#define rsnd_parse_connect_dvc(rdai, playback, capture) \
diff --git a/sound/soc/renesas/rcar/src.c b/sound/soc/renesas/rcar/src.c
index cad15fcc37f2..ac806bdc96d9 100644
--- a/sound/soc/renesas/rcar/src.c
+++ b/sound/soc/renesas/rcar/src.c
@@ -834,3 +834,37 @@ void rsnd_src_remove(struct rsnd_priv *priv)
rsnd_mod_quit(rsnd_mod_get(src));
}
}
+
+void rsnd_src_suspend(struct rsnd_priv *priv)
+{
+ struct rsnd_src_ctrl *src_ctrl = rsnd_priv_to_src_ctrl(priv);
+ struct rsnd_src *src;
+ int i;
+
+ if (!src_ctrl)
+ return;
+
+ for_each_rsnd_src(src, priv, i)
+ rsnd_suspend_clk_reset(rsnd_mod_get(src)->clk,
+ rsnd_mod_get(src)->rstc);
+
+ clk_disable_unprepare(src_ctrl->scu_x2);
+ clk_disable_unprepare(src_ctrl->scu);
+}
+
+void rsnd_src_resume(struct rsnd_priv *priv)
+{
+ struct rsnd_src_ctrl *src_ctrl = rsnd_priv_to_src_ctrl(priv);
+ struct rsnd_src *src;
+ int i;
+
+ if (!src_ctrl)
+ return;
+
+ clk_prepare_enable(src_ctrl->scu);
+ clk_prepare_enable(src_ctrl->scu_x2);
+
+ for_each_rsnd_src(src, priv, i)
+ rsnd_resume_clk_reset(rsnd_mod_get(src)->clk,
+ rsnd_mod_get(src)->rstc);
+}
diff --git a/sound/soc/renesas/rcar/ssi.c b/sound/soc/renesas/rcar/ssi.c
index 007a7c91d470..2fa76a079982 100644
--- a/sound/soc/renesas/rcar/ssi.c
+++ b/sound/soc/renesas/rcar/ssi.c
@@ -1257,3 +1257,23 @@ void rsnd_ssi_remove(struct rsnd_priv *priv)
rsnd_mod_quit(rsnd_mod_get(ssi));
}
}
+
+void rsnd_ssi_suspend(struct rsnd_priv *priv)
+{
+ struct rsnd_ssi *ssi;
+ int i;
+
+ for_each_rsnd_ssi(ssi, priv, i)
+ rsnd_suspend_clk_reset(rsnd_mod_get(ssi)->clk,
+ rsnd_mod_get(ssi)->rstc);
+}
+
+void rsnd_ssi_resume(struct rsnd_priv *priv)
+{
+ struct rsnd_ssi *ssi;
+ int i;
+
+ for_each_rsnd_ssi(ssi, priv, i)
+ rsnd_resume_clk_reset(rsnd_mod_get(ssi)->clk,
+ rsnd_mod_get(ssi)->rstc);
+}
diff --git a/sound/soc/renesas/rcar/ssiu.c b/sound/soc/renesas/rcar/ssiu.c
index 7d3d463c21bf..2a8593a5d4a6 100644
--- a/sound/soc/renesas/rcar/ssiu.c
+++ b/sound/soc/renesas/rcar/ssiu.c
@@ -629,3 +629,23 @@ void rsnd_ssiu_remove(struct rsnd_priv *priv)
rsnd_mod_quit(rsnd_mod_get(ssiu));
}
}
+
+void rsnd_ssiu_suspend(struct rsnd_priv *priv)
+{
+ struct rsnd_ssiu *ssiu;
+ int i;
+
+ for_each_rsnd_ssiu(ssiu, priv, i)
+ rsnd_suspend_clk_reset(rsnd_mod_get(ssiu)->clk,
+ rsnd_mod_get(ssiu)->rstc);
+}
+
+void rsnd_ssiu_resume(struct rsnd_priv *priv)
+{
+ struct rsnd_ssiu *ssiu;
+ int i;
+
+ for_each_rsnd_ssiu(ssiu, priv, i)
+ rsnd_resume_clk_reset(rsnd_mod_get(ssiu)->clk,
+ rsnd_mod_get(ssiu)->rstc);
+}
--
2.25.1
^ permalink raw reply related [flat|nested] 31+ messages in thread* Re: [PATCH v7 18/18] ASoC: rsnd: Add system suspend/resume support
2026-05-25 11:02 ` [PATCH v7 18/18] ASoC: rsnd: Add system suspend/resume support John Madieu
@ 2026-05-25 12:23 ` sashiko-bot
0 siblings, 0 replies; 31+ messages in thread
From: sashiko-bot @ 2026-05-25 12:23 UTC (permalink / raw)
To: John Madieu; +Cc: robh, devicetree, wsa+renesas, conor+dt
Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:
- [High] Incorrect clock and reset ordering in `rsnd_suspend_clk_reset` and `rsnd_resume_clk_reset`.
- [Medium] The `scu_supply` clock is omitted from suspend/resume management in the SRC module.
- [Medium] Return values from critical resource management functions are silently ignored during system resume.
--
commit 0690a34131ef80f23f3d83d8daa691b816da9afc
Author: John Madieu <john.madieu.xa@bp.renesas.com>
Date: Mon May 25 11:02:30 2026 +0000
ASoC: rsnd: Add system suspend/resume support
This commit adds system suspend and resume support for the ASoC rsnd driver,
distributing the module-specific suspend/resume operations across individual
files instead of centralizing them in core.c.
> diff --git a/sound/soc/renesas/rcar/core.c b/sound/soc/renesas/rcar/core.c
> index fbf7f72364600..9ce56cd84f46d 100644
> --- a/sound/soc/renesas/rcar/core.c
> +++ b/sound/soc/renesas/rcar/core.c
[ ... ]
> @@ -2159,11 +2160,35 @@ static void rsnd_remove(struct platform_device *pdev)
> remove_func[i](priv);
> }
>
> +void rsnd_suspend_clk_reset(struct clk *clk, struct reset_control *rstc)
> +{
> + clk_unprepare(clk);
> + reset_control_assert(rstc);
[Severity: High]
Will this sequence send the reset command to an unclocked IP block?
Looking at the driver's normal module quit sequence in rsnd_mod_quit(), the
reset is asserted before unpreparing the clock. Could asserting the reset
after unpreparing the clock here cause the IP blocks to hang?
> +}
> +
> +void rsnd_resume_clk_reset(struct clk *clk, struct reset_control *rstc)
> +{
> + reset_control_deassert(rstc);
> + clk_prepare(clk);
[Severity: High]
Is it safe to deassert the reset before the clock is prepared?
The hardware typically requires the clock to be stable before deasserting
the reset, which is explicitly noted in rsnd_dma_resume() and done correctly
in rsnd_mod_init(). Will deasserting early here cause issues for the
modules using this helper (SSI, SRC, CTU, MIX, DVC, and ADG)?
[Severity: Medium]
Should the return values of clk_prepare() and reset_control_deassert() be
checked here?
If a clock fails to prepare or a reset fails to deassert, these failures
are silently ignored. Could subsequent register accesses to these IP blocks
trigger a synchronous external abort if they didn't actually resume?
> +}
[ ... ]
> diff --git a/sound/soc/renesas/rcar/dma.c b/sound/soc/renesas/rcar/dma.c
> index 537b71841f8e2..793dd4adbe5c6 100644
> --- a/sound/soc/renesas/rcar/dma.c
> +++ b/sound/soc/renesas/rcar/dma.c
[ ... ]
> @@ -1035,3 +1035,25 @@ int rsnd_dma_probe(struct rsnd_priv *priv)
> +void rsnd_dma_resume(struct rsnd_priv *priv)
> +{
> + struct rsnd_dma_ctrl *dmac = rsnd_priv_to_dmac(priv);
> +
> + if (dmac) {
> + /* Clock must be stable before reset is deasserted */
> + clk_prepare_enable(dmac->audmapp_clk);
[Severity: Medium]
Is it necessary to check the return value of clk_prepare_enable() here?
If enabling the DMA clock fails, the hardware won't be accessible, and the
error will be silently dropped, potentially leading to panics when the block
is accessed.
> + rsnd_resume_clk_reset(NULL, dmac->audmapp_rstc);
> + }
> +}
[ ... ]
> diff --git a/sound/soc/renesas/rcar/src.c b/sound/soc/renesas/rcar/src.c
> index cad15fcc37f22..ac806bdc96d9e 100644
> --- a/sound/soc/renesas/rcar/src.c
> +++ b/sound/soc/renesas/rcar/src.c
[ ... ]
> @@ -834,3 +834,37 @@ void rsnd_src_remove(struct rsnd_priv *priv)
> +void rsnd_src_suspend(struct rsnd_priv *priv)
> +{
> + struct rsnd_src_ctrl *src_ctrl = rsnd_priv_to_src_ctrl(priv);
> + struct rsnd_src *src;
> + int i;
> +
> + if (!src_ctrl)
> + return;
> +
> + for_each_rsnd_src(src, priv, i)
> + rsnd_suspend_clk_reset(rsnd_mod_get(src)->clk,
> + rsnd_mod_get(src)->rstc);
> +
> + clk_disable_unprepare(src_ctrl->scu_x2);
> + clk_disable_unprepare(src_ctrl->scu);
[Severity: Medium]
Does this skip disabling the scu_supply clock during suspend?
During rsnd_src_probe(), scu, scu_x2, and scu_supply are all acquired and
enabled. If scu_supply is omitted here, will it remain enabled throughout
system suspend and prevent the clock domain from entering a low-power state?
> +}
> +
> +void rsnd_src_resume(struct rsnd_priv *priv)
> +{
> + struct rsnd_src_ctrl *src_ctrl = rsnd_priv_to_src_ctrl(priv);
> + struct rsnd_src *src;
> + int i;
> +
> + if (!src_ctrl)
> + return;
> +
> + clk_prepare_enable(src_ctrl->scu);
> + clk_prepare_enable(src_ctrl->scu_x2);
[Severity: Medium]
Should the return values of these clk_prepare_enable() calls be checked?
If the clocks fail to prepare, the error is discarded.
Also, similar to the suspend path, should scu_supply be re-enabled here?
> +
> + for_each_rsnd_src(src, priv, i)
> + rsnd_resume_clk_reset(rsnd_mod_get(src)->clk,
> + rsnd_mod_get(src)->rstc);
> +}
--
Sashiko AI review · https://sashiko.dev/#/patchset/20260525110230.4014435-1-john.madieu.xa@bp.renesas.com?part=18
^ permalink raw reply [flat|nested] 31+ messages in thread
* Re: [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support
2026-05-25 11:02 [PATCH v7 00/18] ASoC: rsnd: Add RZ/G3E audio driver support John Madieu
` (17 preceding siblings ...)
2026-05-25 11:02 ` [PATCH v7 18/18] ASoC: rsnd: Add system suspend/resume support John Madieu
@ 2026-05-25 22:45 ` Kuninori Morimoto
18 siblings, 0 replies; 31+ messages in thread
From: Kuninori Morimoto @ 2026-05-25 22:45 UTC (permalink / raw)
To: John Madieu
Cc: Mark Brown, Liam Girdwood, Rob Herring, Krzysztof Kozlowski,
Conor Dooley, Geert Uytterhoeven, Jaroslav Kysela, Takashi Iwai,
Magnus Damm, Philipp Zabel, Claudiu Beznea, Biju Das, john.madieu,
linux-sound, linux-renesas-soc, devicetree, linux-kernel
Hi John
> Add audio support for the Renesas RZ/G3E SoC to the R-Car Sound
> driver. The RZ/G3E audio subsystem is based on R-Car Sound IP but
> has several differences requiring dedicated handling:
>
> - SSI operates exclusively in BUSIF mode (no PIO)
> - 2-4 BUSIF channels per SSI (layout differs from R-Car)
> - Separate register regions for SCU, ADG, SSIU, SSI accessed by name
> - Per-SSI ADG and SSIF supply clocks
> - Dedicated audmapp clock/reset for Audio DMAC peri-peri
> - Per-SSI and per-module reset controllers via CPG
> - Unprefixed DT sub-node names (ssi, ssiu, src, ...) instead of
> rcar_sound,xxx
> - Hyphenated indexed clock/reset names (ssi-0, src-0, adg-ssi-0,
> audio-clka, ...) instead of the legacy dotted form
Thank you for the patches.
For all patches (except DT bindings)
Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Thank you for your help !!
Best regards
---
Kuninori Morimoto
^ permalink raw reply [flat|nested] 31+ messages in thread