acp: add native Zephyr drivers for AMD ACP 7.0 (includes Zephyr west.yml update) by Sivasubramanian678 · Pull Request #10571 · thesofproject/sof

Sivasubramanian678 · 2026-02-24T09:12:49Z

This PR adds native Zephyr support for AMD Audio Co-Processor (ACP) version 7.0 for handling Sound Wire IO.

sofci · 2026-02-24T09:16:19Z

Can one of the admins verify this patch?

reply test this please to run this test once

Sivasubramanian678 · 2026-02-24T09:16:24Z

Dependency: This PR depends on zephyrproject-rtos/zephyr#104450 for native Zephyr ACP 7.0 driver enablement.

lgirdwood · 2026-02-26T16:00:40Z

test this please

lgirdwood · 2026-03-03T16:12:50Z

@Sivasubramanian678 can you check the CI, looks like we have some build errors...

Sivasubramanian678 · 2026-03-04T12:37:26Z

@Sivasubramanian678 can you check the CI, looks like we have some build errors...

The error that is happening in CI test is because of the dependency of Zephyr PR - zephyrproject-rtos/zephyr#104450 .In that PR we have included the required macro. I think till that PR gets merged CI test won't pass. Is there anyway to avoid this CI test failure?

lgirdwood · 2026-03-04T16:24:03Z

@Sivasubramanian678 can you check the CI, looks like we have some build errors...

The error that is happening in CI test is because of the dependency of Zephyr PR - zephyrproject-rtos/zephyr#104450 .In that PR we have included the required macro. I think till that PR gets merged CI test won't pass. Is there anyway to avoid this CI test failure?

No, probably best we get the Zephyr PR merged 1st and then we can rerun CI to make sure all is good.

lyakh

some parts of this PR also affect non-AMD platforms, just need to be aware and check potential side-effects

lyakh · 2026-03-06T08:08:14Z


 /* imx currently has no IRQ driver in Zephyr so we force to xtos IRQ */
-#if defined(CONFIG_AMD)
+#ifndef CONFIG_ZEPHYR_NATIVE_DRIVERS


so this now will be enabled for all builds that don't set CONFIG_ZEPHYR_NATIVE_DRIVERS

same alternative as above to limit to just AMD variants

lyakh · 2026-03-06T08:09:07Z

 * to native drivers.
 */
-#if  defined(CONFIG_AMD)
+#ifndef CONFIG_ZEPHYR_NATIVE_DRIVERS


this potentially affects other platforms too

this potentially affects other platforms too

...although currently apparently only AMD has CONFIG_ZEPHYR_NATIVE_DRIVERS=n

This change was done to maintain backward compatibility for previous generation AMD platform. We will check whether we can update the previous generation platform to native zephyr to avoid this change.

This seems ok, can @thesofproject/nxp check?

One alternative is to have check: "#if defined(CONFIG_AMD) && !defined(CONFIG_ZEPHYR_NATIVE_DRIVERS)""

lgirdwood · 2026-03-09T10:42:45Z

@Sivasubramanian678 will you rebase when the Zephyr PR is merged and include the west update as a patch in this PR, this should kickstart all the CI too. Thanks !

lgirdwood · 2026-03-19T12:36:54Z

@Sivasubramanian678 will you rebase when the Zephyr PR is merged and include the west update as a patch in this PR, this should kickstart all the CI too. Thanks !

@Sivasubramanian678 any update from Zephyr side ?

lgirdwood · 2026-03-30T17:17:00Z

@Sivasubramanian678 will you rebase when the Zephyr PR is merged and include the west update as a patch in this PR, this should kickstart all the CI too. Thanks !

@Sivasubramanian678 any update from Zephyr side ?

Ping ?

Sivasubramanian678 · 2026-03-31T07:11:27Z

@Sivasubramanian678 will you rebase when the Zephyr PR is merged and include the west update as a patch in this PR, this should kickstart all the CI too. Thanks !

@Sivasubramanian678 any update from Zephyr side ?

Ping ?

We have received one approval from Zephyr and are currently waiting for one more. Once that approval is received, the code will be merged. After the Zephyr merge, I will rebase this PR and include the west update as a patch.

kv2019i

Looks good, only minor comments below. Let's either have NXP folks ack the ifdef logic (or you change the logic to only apply to AMD configurations).

kv2019i · 2026-04-29T17:16:44Z

 * to native drivers.
 */
-#if  defined(CONFIG_AMD)
+#ifndef CONFIG_ZEPHYR_NATIVE_DRIVERS


This seems ok, can @thesofproject/nxp check?

One alternative is to have check: "#if defined(CONFIG_AMD) && !defined(CONFIG_ZEPHYR_NATIVE_DRIVERS)""

kv2019i · 2026-04-29T17:17:19Z


 /* imx currently has no IRQ driver in Zephyr so we force to xtos IRQ */
-#if defined(CONFIG_AMD)
+#ifndef CONFIG_ZEPHYR_NATIVE_DRIVERS


same alternative as above to limit to just AMD variants

This commit adds support for ACP 7.0 native zephyr and also adds acp_7_0 in zephyr build script. Changes include: - Add changes in platform-specific files for ACP 7.0 for native zephyr support - Add ACP 7.0 in cmakelist under zephyr - Remove DMA based scheduling from kconfig for ACP 7.0 - Add acp_7_0 support in zephyr build script - Add xtensa files idma.h,xtensa-types.h and xt_externalregisters.h - Add UINT32_C macro fallback for stdint.h compatibility in tie.h - Replace addi.a with addi in tie-asm.h for Xtensa assembler compatibility. This enables native Zephyr-based SOF support for AMD ACP 7.0 hardware platform. Co-developed-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: Sivasubramanian <sravisar@amd.com>

Remove legacy chip_offset_byte.h and chip_register.h header files as they are no longer needed with native Zephyr driver support for the 7_0 platform. Co-developed-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: Sivasubramanian <sravisar@amd.com>

Disable COMP_MUX and PROBE and remove AMD_BINARY_BUILD. Co-developed-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: Sivasubramanian <sravisar@amd.com>

This adds SoundWire DAI support for ACP 7.0 native zephyr. Co-developed-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: Sivasubramanian <sravisar@amd.com>

This adds changes for ACP 6.0 non native zephyr support. Co-developed-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: Sivasubramanian <sravisar@amd.com>

This updates the Zephyr revision to include native Zephyr ACP 7.0 drivers (zephyrproject-rtos/zephyr#104450) required for SOF ACP 7.0 platform support. Co-developed-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: Sivasubramanian <sravisar@amd.com>

Sivasubramanian678 · 2026-04-29T18:00:47Z

@kv2019i The requested changes have been completed. Please let me know if any further updates are needed.

kv2019i

Thanks @Sivasubramanian678 , looks good now!

lyakh · 2026-04-30T06:31:37Z

+  idma_hw_error_t  err_type;       /* ErrorCodes field of the Status reg */
+  uint32_t         currDesc;       /* Descriptor causing error           */
+  uint32_t         srcAddr;        /* PIF source address causing error   */
+  uint32_t         dstAddr;        /* PIF dest. address causing error    */


indentation - also below in structures and functions

lyakh · 2026-04-30T06:33:15Z

+ * @retval  None
+ */
+IDMA_API void
+idma_hw_wait_all(int32_t ch);


can we put all of these on single lines? Makes grepping more informative

lyakh · 2026-04-30T06:33:43Z

+ * @retval !IDMA_OK             Error type
+ */
+IDMA_API idma_status_t
+idma_init( int32_t ch,


there was one more case of a space after an opening bracket above

lyakh · 2026-04-30T06:35:04Z

+ *         NOTE: Pointer is valid only if idma task/buffer is in error.
+ */
+IDMA_API idma_error_details_t*   idma_error_details(int32_t ch);
+IDMA_API idma_error_details_t*   idma_buffer_error_details(int32_t ch);


the usual format in SOF is type *function(arguments) i.e. * after a space and no space between it and function

lyakh · 2026-04-30T06:39:36Z

+#endif
+  IDMA_ENABLE_INTS();
+  return IDMA_OK;
+}


oh... ok, it isn't the first time that I see an actual implementation fully in headers, but... SOF doesn't do this usually... @kv2019i @lgirdwood what do you think? Can we put these in a .c file?

lyakh · 2026-04-30T06:44:37Z

+		unsigned int	RO_CLK1_DFS_STATE_IDLE  : 1;
+		unsigned int	CLK1_CURRENT_DFS_DID    : 7;
+		unsigned int	: 5;
+	} bitfields, bits;


those bitfields add up to 32 bits, so I'd use uint32_t here explicitly. Same in other similar bit-field cases

Signed-off-by: Sivasubramanian678 <sravisar@amd.com>

Sivasubramanian678 requested review from RanderWang, abonislawski, bhiregoudar, dbaluta, iuliana-prodan, kv2019i, lbetlej, lgirdwood, lyakh, mmaka1, pblaszko, plbossart, sunilkumardommati and tmleman as code owners February 24, 2026 09:12

Sivasubramanian678 mentioned this pull request Feb 24, 2026

drivers: acp: add native Zephyr drivers for AMD ACP 7.0 zephyrproject-rtos/zephyr#104450

Merged

lgirdwood approved these changes Feb 26, 2026

View reviewed changes

lyakh reviewed Mar 6, 2026

View reviewed changes

Sivasubramanian678 force-pushed the main branch from 2436222 to 3ef8614 Compare March 9, 2026 10:37

Sivasubramanian678 force-pushed the main branch from 3ef8614 to 6b02daa Compare April 29, 2026 14:11

Sivasubramanian678 requested a review from lgirdwood April 29, 2026 14:12

kv2019i changed the title ~~acp: add native Zephyr drivers for AMD ACP 7.0~~ acp: add native Zephyr drivers for AMD ACP 7.0 (includes Zephyr west.yml update) Apr 29, 2026

Sivasubramanian678 force-pushed the main branch from 6b02daa to a531b73 Compare April 29, 2026 17:00

kv2019i reviewed Apr 29, 2026

View reviewed changes

Sivasubramanian678 added 6 commits April 29, 2026 23:18

arch: xtensa: acp_7_0: Update defconfig settings

035c6d4

Disable COMP_MUX and PROBE and remove AMD_BINARY_BUILD. Co-developed-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: Sivasubramanian <sravisar@amd.com>

DAI: Add SoundWire DAI support for native zephyr

d065058

This adds SoundWire DAI support for ACP 7.0 native zephyr. Co-developed-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: DineshKumar Kalva <DineshKumar.Kalva@amd.com> Signed-off-by: Sivasubramanian <sravisar@amd.com>

Sivasubramanian678 force-pushed the main branch from a531b73 to 6453feb Compare April 29, 2026 17:55

kv2019i approved these changes Apr 29, 2026

View reviewed changes

bhiregoudar approved these changes Apr 30, 2026

View reviewed changes

lyakh reviewed Apr 30, 2026

View reviewed changes

Merge branch 'main' into main

597a211

Signed-off-by: Sivasubramanian678 <sravisar@amd.com>

Sivasubramanian678 requested a review from bhiregoudar May 1, 2026 12:17

Conversation

Sivasubramanian678 commented Feb 24, 2026

Uh oh!

sofci commented Feb 24, 2026

Uh oh!

Sivasubramanian678 commented Feb 24, 2026

Uh oh!

lgirdwood commented Feb 26, 2026

Uh oh!

lgirdwood commented Mar 3, 2026

Uh oh!

Sivasubramanian678 commented Mar 4, 2026

Uh oh!

lgirdwood commented Mar 4, 2026

Uh oh!

lyakh left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lgirdwood commented Mar 9, 2026

Uh oh!

lgirdwood commented Mar 19, 2026

Uh oh!

lgirdwood commented Mar 30, 2026

Uh oh!

Sivasubramanian678 commented Mar 31, 2026

Uh oh!

kv2019i left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Sivasubramanian678 commented Apr 29, 2026

Uh oh!

kv2019i left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants