zephyr: stop using CONFIG_SOF to identify Zephyr build (w/ Zephyr Nov/20 update) #10335

kv2019i · 2025-10-28T15:12:21Z

Prepare for change in Zephyr to drop SOF module from the upstream Zephyr manifest. Remove use of CONFIG_SOF which is defined in work/zephyr/modules/Kconfig.sof, but will now be removed.

Fix this by using CONFIG_ZEPHYR_SOF_MODULE instead. This is defined by the Zephyr build system when using the work/sof/west.yml for building SOF. If SOF is built without Zephyr (e.g. testbench or cmocka build), this define will not be set.

After this change, one must use the SOF work/sof/west.yml manifest to build SOF with Zephyr. This is used by all public instructions and CI jobs already.

Copilot

Pull Request Overview

This PR removes the dependency on CONFIG_SOF (which was defined in Zephyr's SOF module Kconfig) and replaces it with CONFIG_ZEPHYR_SOF_MODULE. This change prepares for the removal of the SOF module from the upstream Zephyr manifest.

Key changes:

Replaced CONFIG_SOF with CONFIG_ZEPHYR_SOF_MODULE in CMakeLists.txt
Removed if SOF wrapper from Kconfig file
Removed CONFIG_SOF=y setting from app/prj.conf

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
zephyr/Kconfig	Removed the conditional `if SOF` wrapper around SOF configuration options
zephyr/CMakeLists.txt	Updated build condition from `CONFIG_SOF` to `CONFIG_ZEPHYR_SOF_MODULE`
app/prj.conf	Removed explicit `CONFIG_SOF=y` setting as it's no longer needed

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

kv2019i · 2025-10-28T15:23:29Z

@nashif Alternative to #10321 . It should be possible to merge this already now, in preparation of your PR. Only side-effect that after we merge this in SOF, then SOF cannot be used with Zephyr manifest anymore (but that's anyways happening in Zephyr 97946).

lgirdwood · 2025-10-28T15:45:25Z

@kv2019i some build issue in CI, do we need any CI changes ?

kv2019i · 2025-10-28T16:19:41Z

There are issues with this still. I can't really fully explain what happens. Some CONFIG_FOO options are missing when other settings (defined in same Kconfig file) are present. E.g.

.config for MTL build:

# CONFIG_MT8196 is not set
CONFIG_MAX_CORE_COUNT=3
CONFIG_INTEL=y
CONFIG_ACE=y
CONFIG_ACE_VERSION_1_5=y

So CONFIG_MAX_CORE_COUNT is set but not CONFG_CORE_COUNT, both Kconfig options defined in the smae file. Hmm. Will debug this more.

kv2019i · 2025-10-28T16:54:26Z

Ok, wow, it seems this this the extra CONFIG_CORE_COUNT definition in arch/host now. I now start to realize why nobody else has cleaned this up. This is quite messed up. Let me put this to draft and try to push this commit together with #10334 .

kv2019i · 2025-10-29T08:10:33Z

See dependencies in #10334 (comment) . PRs #10334 and #10319 need to be merged first, before this PR can go in.

kv2019i · 2025-11-18T14:40:24Z

v2:

rebased to kick the CI again after dependencies merged
the Zephyr side change has been merged, so this PR is needed to build SOF with current Zephyr main
marking as ready for review

lgirdwood

@kv2019i are we pending the Zephyr v4.3 west update before we merge ?

kv2019i · 2025-11-18T16:34:45Z

@lgirdwood wrote:

kv2019i are we pending the Zephyr v4.3 west update before we merge ?

No, this can go in right now if the CI passes.

kv2019i · 2025-11-18T18:04:35Z

@tmleman This now breaks the new ZTests: https://github.com/thesofproject/sof/actions/runs/19469971535/job/55714257536?pr=10335

It seems with the test, we want to not build without any of the SOF stuff, but I couldn't immediately figure out how to differentiate between ztest build and normal SOF build. In both cases CONFIG_ZTEST=y and CONFIG_ZEPHYR_SOF_MODULE=y .

We could bring CONFIG_SOF back, but where to set it? I'll look at this tomorrow... (we can merge #10365 first )

kv2019i · 2025-11-18T19:03:44Z

V3:

Updated` series to address the SOF ztest build failures.
This turned out to be a bit complicated. In addition to 1) the main SOF application builds (SOF build with Zephyr to run on a DSP hw target), we have 2) SOF application builds using Zephyr POSIX target (like the fuzzer), 3) SOF build targets like sof/test/ztest that use Zephyr but do not use any of the SOF side infra like logging and platform definitions, and 4) SOF builds that do not use Zephyr (cmocka, ALSA plugin). To add to the complexity, some targets use west, while some do not. So in this specific case, case (3) was broken my previous version of this PR. Without this PR, (1) + (2) are broken with Zephyr main, but curiously enough not (3). With this PR, 1+2+3+4 should work with current manifest version of Zephyr, plus also with Zephyr main. Party time!

ujfalusi · 2025-11-19T07:06:46Z

zephyr/Kconfig

+	  Top-level build option to enable full build of the SOF application
+	  on top of Zephyr. This should be set to no only for unit test and
+	  such special build targets.



Should you retain the if here and the endif:

if SOF_FULL_ZEPHYR_APPLICATION

or better:

menuconfig SOF_FULL_ZEPHYR_APPLICATION bool "Build full SOF Zephyr application" default y help Top-level build option to enable full build of the SOF application on top of Zephyr. This should be set to no only for unit test and such special build targets. if SOF_FULL_ZEPHYR_APPLICATION ... endif

@ujfalusi I left this out on purposew. There are some Kconfigs that are useful to be availlable for SOF ztest type of uses and with PR, these can now be left defined (but set to "n" in the test configs if not needed).

It seems CI is all good with this as well, so I'd say the less ifdefs the better.

majunkier

I have checked fw building with current and newest Zephyr SHA in west.yaml - both works.

kv2019i · 2025-11-20T12:20:08Z

All tests failing, triggering Intel jenkins CI once more before merge.

kv2019i · 2025-11-20T12:20:32Z

SOFCI TEST

Prepare for change in Zephyr to drop SOF module from the upstream Zephyr manifest. Remove use of CONFIG_SOF which is defined in work/zephyr/modules/Kconfig.sof, but will now be removed. Fix this by adding new CONFIG_SOF_FULL_ZEPHYR_APPLICATION on SOF side and use it similarly to old CONFIG_SOF. If SOF is built without Zephyr (e.g. testbench or cmocka build), this define will not be set. After this change, one must use the SOF work/sof/west.yml manifest to build SOF with Zephyr. This is used by all public instructions and CI jobs already. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

SOF builds previously relied on CONFIG_SOF to be externally defined in zephyr/modules/Kconfig.sof, but set in sof/app/prj.conf. This was removed in Zephyr, and now instead the new CONFIG_SOF_FULL_ZEPHYR_APPLICATION is set on SOF side. This must be set to 'no' for the ztest unit tests to ensure full SOF dependencies are not pulled into unit tests builds. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

kv2019i · 2025-11-20T19:09:54Z

V4 update:

it turns out the jenkins failures are real
- when compiling new SOF (this PR) and old Zephyr, lack of CONFIG_SOF=y disables logic in zephyr/scripts/west_commands/sign.py
- for some reason, some of the CI targets still worked (!), so I didn't immediately realize the problem
solve the issue by first keeping CONFIG_SOF in sof/app/prj.conf and only remove it together with a Zephyr west.yml update (that removes CONFIG_SOF from Zephyr)

lyakh · 2025-11-21T07:34:25Z

solve the issue by first keeping CONFIG_SOF in sof/app/prj.conf and only remove it together with a Zephyr west.yml update (that removes CONFIG_SOF from Zephyr)

@kv2019i sorry, not sure I understand: CI only tests the merged result of all the changes in the PR, what difference does it make for the CI if you remove CONFIG_SOF in the first or in the last commit?

lyakh · 2025-11-21T07:30:03Z

app/prj.conf

@@ -1,5 +1,5 @@
-CONFIG_SOF=y
 CONFIG_BUILD_OUTPUT_BIN=n
+CONFIG_SOF=y


looks like this hunk can be dropped, and the commit title should be updated?

@lyakh It's just moved to the last patch. I have to split this into multiple patches not to break bisect. If I remove CONFIG_SOF in a different commit than the zepyr west.yaml, I break bisect.

@kv2019i I understand why you keep CONFIG_SOF=y but I don't think swapping the lines in the file is needed? ;-) So you could just drop this hunk and leave the file unchanged - first line CONFIG_SOF=y, second line CONFIG_BUILD_OUTPUT_BIN=n

Total of 798 commits. Includes removal of CONFIG_SOF in Zephyr. This commit has the matching change to sof/app/prj.conf. Other changes include: 3f6ef5043ce5 soc: intel_adsp: ace30: allow userspace to execute cold functions 18b34bb3d29d drivers: power_domain: intel_adsp: Refactor power management initialization 27180d2fc573 arch: riscv + xtensa + x86: workaround needed for LLVM linker 95b48cd5ba7a west_commands: do not depend on CONFIG_SOF 71ec5df8f31f intel_adsp: remove workaround for SOF setting core count 1a780f933e1d manifest: optional: remove sof from optional manifest 65d31ea8609e tests: intel_adsp/smoke: cast to signed ints before abs() e45808b9cd71 xtensa: mmu: add page table usage statistics a48345fccfdb xtensa: mmu/ptables: conserve memory by using COW on L2 tables 1e41db3ddd84 xtensa: mmu: no need for cache ops if page tables are not cached 501368601dc7 xtensa: mmu/ptables: rework TLB invalidation on L2 unmap 259be3d55929 xtensa: mmu: remove XTENSA_ prefix for page table array macros 1470d9ef7460 xtensa: mmu: move PTE macros into source file b6713c0145e8 xtensa: mmu: skip PTE SW field redirection 2bfcb20258f5 xtensa: mmu: unify PTE macros 11dca1208aa3 boards: xtensa: add support for DOIT ESP32 DevKit V1 3b2b513be37e drivers: wifi: nxp: Raise Wi-Fi mgmt events for soft AP start/stop. Signed-off-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>

kv2019i · 2025-11-21T08:41:47Z

V5 update:

C17 is now enforced in Zephyr upstream
to give time to remove old targets from SOF CI that do not work with C17, I updated this PR to the last commit before C17 requirement (zephyrproject-rtos/zephyr@db0c34d836ae) in Zephyr
the Zephyr SDK will continue to support these targets in upstream (e.g. github workflows will continue to build these targets)

kv2019i · 2025-11-24T09:42:12Z

@lrudyX Can you check? All tests that are run are passing, but a few tests fail to execute ("ConnectionRefusedError: [Errno 111] Connection refused" in the logs). This PR would be good to get in to allow other Zephyr PR's through. FYI @abonislawski @tmleman

lrudyX · 2025-11-24T10:16:02Z

@lrudyX Can you check? All tests that are run are passing, but a few tests fail to execute ("ConnectionRefusedError: [Errno 111] Connection refused" in the logs). This PR would be good to get in to allow other Zephyr PR's through. FYI @abonislawski @tmleman

Scheduled for rerun.

kv2019i requested review from Copilot and nashif October 28, 2025 15:12

kv2019i requested review from abonislawski, dbaluta, iuliana-prodan, lbetlej, lgirdwood, lyakh, mmaka1 and plbossart as code owners October 28, 2025 15:12

Copilot AI reviewed Oct 28, 2025

View reviewed changes

nashif approved these changes Oct 28, 2025

View reviewed changes

lgirdwood approved these changes Oct 28, 2025

View reviewed changes

kv2019i marked this pull request as draft October 28, 2025 16:54

kv2019i mentioned this pull request Oct 28, 2025

platform: simplify setting CORE_COUNT and MAX_CORE_COUNT #10334

Merged

tmleman approved these changes Oct 29, 2025

View reviewed changes

kv2019i force-pushed the 202510-config-sof-refactor branch from 95f5897 to 84592ab Compare November 18, 2025 14:39

kv2019i marked this pull request as ready for review November 18, 2025 14:39

ujfalusi approved these changes Nov 18, 2025

View reviewed changes

lgirdwood approved these changes Nov 18, 2025

View reviewed changes

kv2019i force-pushed the 202510-config-sof-refactor branch from 84592ab to a8c7a86 Compare November 18, 2025 18:56

ujfalusi reviewed Nov 19, 2025

View reviewed changes

majunkier approved these changes Nov 19, 2025

View reviewed changes

tmleman mentioned this pull request Nov 20, 2025

west.yml: update zephyr to v4.3.0 release #10365

Merged

kv2019i added 2 commits November 20, 2025 20:53

kv2019i force-pushed the 202510-config-sof-refactor branch from a8c7a86 to 47c5e42 Compare November 20, 2025 19:04

kv2019i changed the title ~~zephyr: stop using CONFIG_SOF to identify Zephyr build~~ zephyr: stop using CONFIG_SOF to identify Zephyr build (w/ Zephyr Nov/20 update) Nov 20, 2025

lyakh reviewed Nov 21, 2025

View reviewed changes

kv2019i force-pushed the 202510-config-sof-refactor branch from 47c5e42 to 628afcd Compare November 21, 2025 08:39

kv2019i merged commit 60780a6 into thesofproject:main Nov 24, 2025
39 of 42 checks passed

zephyr: stop using CONFIG_SOF to identify Zephyr build (w/ Zephyr Nov/20 update) #10335

zephyr: stop using CONFIG_SOF to identify Zephyr build (w/ Zephyr Nov/20 update) #10335

Uh oh!

Conversation

kv2019i commented Oct 28, 2025

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Reviewed Changes

Uh oh!

kv2019i commented Oct 28, 2025

Uh oh!

lgirdwood commented Oct 28, 2025

Uh oh!

kv2019i commented Oct 28, 2025

Uh oh!

kv2019i commented Oct 28, 2025

Uh oh!

kv2019i commented Oct 29, 2025

Uh oh!

kv2019i commented Nov 18, 2025

Uh oh!

lgirdwood left a comment

Choose a reason for hiding this comment

Uh oh!

kv2019i commented Nov 18, 2025

Uh oh!

kv2019i commented Nov 18, 2025

Uh oh!

kv2019i commented Nov 18, 2025

Uh oh!

ujfalusi Nov 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

kv2019i Nov 19, 2025

Choose a reason for hiding this comment

Uh oh!

majunkier left a comment

Choose a reason for hiding this comment

Uh oh!

kv2019i commented Nov 20, 2025

Uh oh!

kv2019i commented Nov 20, 2025

Uh oh!

kv2019i commented Nov 20, 2025

Uh oh!

lyakh commented Nov 21, 2025

Uh oh!

lyakh Nov 21, 2025

Choose a reason for hiding this comment

Uh oh!

kv2019i Nov 21, 2025

Choose a reason for hiding this comment

Uh oh!

lyakh Nov 24, 2025

Choose a reason for hiding this comment

Uh oh!

kv2019i commented Nov 21, 2025

Uh oh!

kv2019i commented Nov 24, 2025

Uh oh!

lrudyX commented Nov 24, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

ujfalusi Nov 19, 2025 •

edited

Loading