Test user-space and enable CONFIG_USERSPACE on select platforms #10119

lyakh · 2025-07-15T15:09:25Z

Add a boot test to run a Zephyr user-space test

lgirdwood

LGTM, just some opens on how we scale to more user tests.

lgirdwood · 2025-07-16T10:13:37Z

zephyr/sof_syscall.c

@@ -0,0 +1,33 @@
+// SPDX-License-Identifier: BSD-3-Clause


I think we are going to have scheduler, library and llext syscalls so probably best to have a syscall directory here and a file for each subsystem.

lgirdwood · 2025-07-16T10:17:53Z

zephyr/test/userspace.c

@@ -0,0 +1,61 @@
+// SPDX-License-Identifier: BSD-3-Clause


likewise re names - my feeling is we will have more userspace tests so best to say userspace-locking.c

lyakh · 2025-07-16T12:21:11Z

SOFCI TEST

lgirdwood · 2025-07-21T13:35:35Z

SOFCI TEST

lgirdwood · 2025-07-29T10:51:59Z

@lyakh why the DNM - assume we are pending on Zephyr PRs ?

lgirdwood · 2025-07-29T10:52:06Z

SOFCI TEST

lyakh · 2025-07-30T06:01:48Z

@lyakh why the DNM - assume we are pending on Zephyr PRs ?

@lgirdwood yes, because we had problems with upgrading to the newest Zephyr and even now it isn't perfect but now I'll re-push this, dropping the last two commits

kv2019i

Hmm, a bit of a mismatch between "test user-space" and enabling CONFIG_USERSPACE for default for PTL targets....

kv2019i · 2025-08-05T14:22:36Z

app/boards/intel_adsp_ace30_ptl.conf

+
+CONFIG_USERSPACE=y
+CONFIG_APPLICATION_DEFINED_SYSCALL=y
+CONFIG_MAX_THREAD_BYTES=3


This is not really aligned with the PR title "test user-space". This PR in fact enables user-space for all PTL targets including next release. Is this intentional at this point?

@kv2019i yes, just enabling the feature doesn't do much as long as we don't have user-space threads, but you're right, it's still a significant change and should be reflected in the PR title

@lyakh I think enabling is a good first step, but yeah, just looking at ifdefs in Zephyr source, there are a couple of thousand of ifdefs for CONFIG_USERSPACE, so compiled code will be quite differente, whether we user threads in SOF or not.

Lets get CONFIG_USERSPACE tested regularly now in CI. Yes we are not using, but lets get some comfort that enabling does not do any harm.

kv2019i

I modified the title now, good to go now. The CI tests need to pass of course, but codewise this is ok to go.

lyakh · 2025-08-07T06:26:46Z

I modified the title now, good to go now. The CI tests need to pass of course, but codewise this is ok to go.

@kv2019i thanks, yes, looks like we need gperf for Windows - at least that's the failure with the GitHub "build-windows" work flow

kv2019i · 2025-08-07T08:58:10Z

@lyakh wrote:

@kv2019i thanks, yes, looks like we need gperf for Windows - at least that's the failure with the GitHub "build-windows" work
flow

Not sure about that. We've not had a maintainer for the Windows build in a long time, so not sure how we can keep it. If nobody signs up, I'd propose to remove it from CI and move on.

So I was more referring to the required CI checks failing.

lyakh · 2025-10-29T11:40:09Z

Turns out VMH testing on MTL is badly broken https://sof-ci.01.org/sofpr/PR10119/build16827/devicetest/index.html , disable it until it's fixed.

Link: thesofproject#10119

softwarecki · 2025-11-05T14:28:23Z

zephyr/test/CMakeLists.txt

@@ -1,5 +1,11 @@
-if (CONFIG_SOF_BOOT_TEST)
+if (CONFIG_SOF_BOOT_TEST AND NOT CONFIG_METEORLAKE)


Why not mtl?

abonislawski

Looks like there are commits for more than 1 PR

abonislawski · 2025-11-05T15:07:31Z

zephyr/test/CMakeLists.txt

 endif()
+
+if (CONFIG_ACE_VERSION_3_0)
+	zephyr_library_sources_ifdef(CONFIG_SOF_BOOT_TEST


no dependency on CONFIG_USERSPACE?

if (CONFIG_SOF_BOOT_TEST) zephyr_library_sources_ifdef(CONFIG_VIRTUAL_HEAP vmh.c ) zephyr_library_sources_ifdef(CONFIG_USERSPACE userspace/ksem.c ) endif()

PTL supports user-space, enable it by default, which also enabled the boot-time user-space test. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

lyakh · 2025-12-16T10:03:50Z

CI: PTL failure https://sof-ci.01.org/sofpr/PR10119/build18064/devicetest/index.html?model=PTLH_RVP_NOCODEC&testcase=multiple-pipeline-all is I think the known out-of-memory regression

kv2019i · 2025-12-16T12:03:23Z

@lyakh Is it the same #10417 ? The test failing is same and same codec, but at least there seem to be a new way this fails (in terms how this shows up in logs):

https://sof-ci.01.org/sofpr/PR10119/build18064/devicetest/index.html?model=PTLH_RVP_NOCODEC&testcase=multiple-pipeline-all

[ 2847.127690] <inf> ipc: ipc_cmd: rx	: 0x47000000|0x0
[ 2847.128446] <err> schedule: scheduler_register: allocation failed

kv2019i · 2025-12-16T12:05:58Z

Hmm, this now only enables the boot tests, so the commit summary is somewhat misleading now. This does explain why the fails happen differently as there's more use of the system before SOF application logic starts to be run, so we run out of memory sooner. Let's proceed with merge as rest of tests are passing and get some more coverage for this.

lyakh added the DNM Do Not Merge tag label Jul 15, 2025

lyakh force-pushed the user branch from a2c4b32 to 97640c6 Compare July 16, 2025 07:24

lgirdwood reviewed Jul 16, 2025

View reviewed changes

lyakh force-pushed the user branch from 97640c6 to b45a11d Compare July 18, 2025 13:14

lyakh force-pushed the user branch from b45a11d to 877308d Compare July 22, 2025 09:49

lyakh marked this pull request as ready for review July 22, 2025 09:49

lyakh requested review from abonislawski, dbaluta, iuliana-prodan, kv2019i, lbetlej, mmaka1 and plbossart as code owners July 22, 2025 09:49

lyakh force-pushed the user branch from 877308d to 950fddc Compare July 22, 2025 10:01

lyakh force-pushed the user branch 3 times, most recently from 6bf019c to dd42a9a Compare August 1, 2025 14:52

lyakh changed the title ~~[DNM] Test user-space~~ Test user-space Aug 1, 2025

lyakh removed the DNM Do Not Merge tag label Aug 1, 2025

lgirdwood approved these changes Aug 1, 2025

View reviewed changes

kv2019i reviewed Aug 5, 2025

View reviewed changes

kv2019i changed the title ~~Test user-space~~ Test user-space and enable CONFIG_USERSPACE on select platforms Aug 6, 2025

kv2019i approved these changes Aug 6, 2025

View reviewed changes

abonislawski requested a review from softwarecki October 29, 2025 14:01

kv2019i added a commit to kv2019i/sof that referenced this pull request Oct 29, 2025

(baseline from PR10119 - user-space tests)

44558cb

Link: thesofproject#10119

kv2019i mentioned this pull request Oct 29, 2025

lib: dma: add support for user-space usage #10341

Merged

lyakh added the DNM Do Not Merge tag label Oct 30, 2025

lyakh force-pushed the user branch from 32fefb6 to d5566e6 Compare October 31, 2025 08:24

lgirdwood approved these changes Oct 31, 2025

View reviewed changes

lyakh force-pushed the user branch 4 times, most recently from a8d5c25 to aa8828d Compare November 3, 2025 14:36

lgirdwood approved these changes Nov 3, 2025

View reviewed changes

lyakh force-pushed the user branch 2 times, most recently from cf0934e to 2e90058 Compare November 5, 2025 14:16

lyakh requested a review from dabekjakub as a code owner November 5, 2025 14:16

softwarecki reviewed Nov 5, 2025

View reviewed changes

abonislawski reviewed Nov 5, 2025

View reviewed changes

lyakh mentioned this pull request Nov 6, 2025

Test: reenable boot-testing #10355

Merged

lyakh force-pushed the user branch from 2e90058 to a4f5cd6 Compare November 20, 2025 11:55

abonislawski approved these changes Nov 21, 2025

View reviewed changes

kv2019i mentioned this pull request Dec 1, 2025

lib: fast-get: fix userspace build #10406

Merged

lyakh force-pushed the user branch from a4f5cd6 to ba81a32 Compare December 1, 2025 16:35

ptl: enable user-space by default

f490fda

PTL supports user-space, enable it by default, which also enabled the boot-time user-space test. Signed-off-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>

lyakh force-pushed the user branch from ba81a32 to f490fda Compare December 12, 2025 16:12

lyakh removed the DNM Do Not Merge tag label Dec 16, 2025

kv2019i merged commit 055db19 into thesofproject:main Dec 16, 2025
38 of 43 checks passed

lyakh deleted the user branch December 16, 2025 12:58

		@@ -1,5 +1,11 @@
		if (CONFIG_SOF_BOOT_TEST)
		if (CONFIG_SOF_BOOT_TEST AND NOT CONFIG_METEORLAKE)

Test user-space and enable CONFIG_USERSPACE on select platforms #10119

Test user-space and enable CONFIG_USERSPACE on select platforms #10119

Uh oh!

Conversation

lyakh commented Jul 15, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

lgirdwood 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!

lyakh commented Jul 16, 2025

Uh oh!

lgirdwood commented Jul 21, 2025

Uh oh!

lgirdwood commented Jul 29, 2025

Uh oh!

lgirdwood commented Jul 29, 2025

Uh oh!

lyakh commented Jul 30, 2025

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!

kv2019i left a comment

Choose a reason for hiding this comment

Uh oh!

lyakh commented Aug 7, 2025

Uh oh!

kv2019i commented Aug 7, 2025

Uh oh!

lyakh commented Oct 29, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

abonislawski left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

lyakh commented Dec 16, 2025

Uh oh!

kv2019i commented Dec 16, 2025

Uh oh!

kv2019i commented Dec 16, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

lyakh commented Jul 15, 2025 •

edited

Loading