lib.sh: Set ALSA settings from a state file #1294

golowanow · 2025-08-08T07:06:29Z

This PR is to continue with #1284 and PoC #1270

Set ALSA settings from a state file and migrate PTL configurations to the new mechanism:

set_alsa_settings() function now uses alsactl restore command, to change ALSA controls from a custom state file first, if it is present, and next - execute a bash script with amixer commands, if it is also present.
The state file is expected as ./alsa_settings/${PLATFORM}.state whereas the bash script remains as ./alsa_settings/${PLATFORM}.sh
The custom state should be compatible with the platform's default state, see set_alsa() function.
Replace PTL platform ALSA settings done as bash script commands by their equivalent ALSA state files for:

PTLH_HDA_AIOC
PTLH_SDW_RT712
PTLP_RVP_SDW
MTLP_RVP_HDA
MTLP_SDW_AIOC
MTLP_RVP_SDW
LNLM_RVP_HDA
LNLM_SDW_AIOC

The parameters in the state files are explicitly set even if the default values at the appropriate DUT are currently the same.

Log difference between the current ALSA state and the restored ALSA state.
Add README.md documentation for the SOF test ALSA settings.

marc-hb

No time to review and it's still a draft but this looks very useful!

golowanow · 2025-08-12T13:22:42Z

SOFCI TEST

alsa_settings/PTLH_HDA_AIOC.state

majunkier · 2025-08-12T13:58:48Z

alsa_settings/MTLP_RVP_SDW.state

@@ -0,0 +1,35 @@
+state.sofsoundwire {


File MTLP_RVP_SDW.sh contain comment's why we set something, we should keep them

AFAIK, the ALSA state file format has no custom comments, although alsactl seems to ignore them, if added carefully.
I think better to avoid parsing issues and just rely on the git log history to track changes.

In that case, comments should be move to README file to preserve info why specific config was set

In that case, comments should be move to README file to preserve info why specific config was set

in README it will be ~~quickly~~ out of sync .. let's try to add it to the .state file directly

A good git log is important but it should be focused and explaining why things CHANGED. It should ideally not be required to explain why things ARE the way they are.

Agree, git log should focus on explaining why something was changed. But there are cases where leaving comment in code is also important, for changes that could rise questions in future, or like here proposing whole new approach, comment in file can be quicker to find and understand than digging in git history.

alsa_settings/MTLP_SDW_AIOC.state

marc-hb · 2025-08-13T00:24:11Z

For some unknown reason I just remembered that UCM was supposed to be tested interface, not ALSA states...

Add document to clarify BKM of user space packages: alsa-lib, alsa-ucm-conf, PulseAudio, GNOME etc linux#2211

Quoting internal issue sof-framework 635:

Please stop using asound.state and start using UCM files.
If there are no UCM files let's create them
We have to stop the duplication of work, and asound.state files are EVIL.

:-)

golowanow · 2025-08-13T04:33:42Z

Please stop using asound.state and start using UCM files.
If there are no UCM files let's create them

@marc-hb, AFAIK this choice (state vs. ucm) has been discussed several months ago, and #1270 PoC was started focusing on the state method. At this point we need to sort out the currently used mechanism first, with its limitations known.
The UCM method needs its attention too as a next step.

golowanow · 2025-08-13T04:46:05Z

btw, this PR has been already used to identify issues with misaligned configurations, so its merge will improve test execution on our CI.

golowanow · 2025-08-13T11:06:15Z

SOFCI TEST

golowanow · 2025-08-13T11:09:41Z

@majunkier - do you have any other blocking comments ?

marc-hb · 2025-08-13T20:36:56Z

alsa_settings/README.md

+codec and loopback devices.
+
+The `./alsa_settings/` directory contains custom ALSA configurations which
+are currently applied at the SOF CI pipeline's hardware platforms.


Why is the default hardware configuration different and not enough? If this is by design (in other words, testing requires some "special sauce"), very briefly state why. If this is not design, clearly state that all these files are really "workarounds".

Either way, it's not good to be testing something different from regular users. So this deserves a bit of justification in either case.

well, It is a really good question what a 'regular user' can (or should) have as an environment to execute the tests.
I've changed the text a bit, hopefully it becomes more clear why we need these settings here as they currently are.

alsa_settings/README.md

marc-hb · 2025-08-14T06:03:59Z

case-lib/lib.sh

+            if [ -f "${ALSA_SETTINGS_FILE}.sh" ]; then
+                dlogc "${ALSA_SETTINGS_FILE}.sh"
+                "${ALSA_SETTINGS_FILE}".sh 2>&1 || rc=$?
+                [[ "${rc}" -ne 0 ]] && dloge "ALSA settings error=${rc}"


I think you should reset rc to zero first, otherwise this .sh message could be triggered by the previous . .state error despite the .sh running fine...

right, added

marc-hb · 2025-08-14T06:06:59Z

alsa_settings/README.md

+Please note that in some cases the settings are tuned to CI DUT 'harnesses':
+external codecs with their versions, audio cards at loopback connections, etc.
+In the future, most of these settings should be moved to the DUT's default config
+out of the test case scope, so only the test-specific settings will remain here.


There should be a simple way to identify which setting is part of which category in each file using using some "standardized" comment convention and/or sections.

yeah, good idea for future clean-up changes there.

set_alsa_settings() function now uses `alsactl restore` command, to change ALSA controls from a custom state file first, if it is present, and next - execute a bash script with amixer commands if it is also present. The state file is expected as `./alsa_settings/${PLATFORM}.state` whereas the bash script remains as `./alsa_settings/${PLATFORM}.sh` The custom state should be compatible with the platform's default state, see set_alsa() function. Co-authored-by: Lukasz Mrugala <lukaszx.mrugala@intel.com> Co-authored-by: Dmitrii Golovanov <dmitrii.golovanov@intel.com>

Log difference between the current ALSA state and the restored ALSA state. Signed-off-by: Dmitrii Golovanov <dmitrii.golovanov@intel.com>

Replace PTL platform ALSA settings done as bash script commands by their equivalent ALSA state files for: * PTLH_HDA_AIOC * PTLH_SDW_RT712 * PTLP_RVP_SDW The parameters in the state files are explicitly set even if the default values at the appropriate DUT are currently the same. Signed-off-by: Dmitrii Golovanov <dmitrii.golovaniov@intel.com> Co-authored-by: Lukasz Mrugala <lukaszx.mrugala@intel.com>

Replace MTL platform ALSA settings done as bash script commands by their equivalent ALSA state files for: * MTLP_RVP_HDA * MTLP_SDW_AIOC * MTLP_RVP_SDW The parameters in the state files are explicitly set even if the default values at the appropriate DUT are currently the same. Because of that, we can't 'reuse' MTLP_RVP_SDW state for MTLP_SDW_AIOC as before: these configurations have different id's whereas `amixer` had changed them referencing by name. In some cases, USB audio card settings are also included: it is needed to adjust loopback volume levels on a CI DUT. In the future these settings (actually most of the ALSA settings) should be migrated to the DUT's default configs and out of the test case scope, so only test-specific settings remain here. Signed-off-by: Dmitrii Golovanov <dmitrii.golovaniov@intel.com> Co-authored-by: Lukasz Mrugala <lukaszx.mrugala@intel.com>

Replace LNLM platform ALSA settings done as bash script commands by their equivalent ALSA state files for: * LNLM_RVP_HDA * LNLM_SDW_AIOC The parameters in the state files are explicitly set even if the default values at the appropriate DUT are currently the same. Signed-off-by: Dmitrii Golovanov <dmitrii.golovaniov@intel.com> Co-authored-by: Lukasz Mrugala <lukaszx.mrugala@intel.com>

Add README.md documentation for SOF test ALSA settings. Signed-off-by: Dmitrii Golovanov <dmitrii.golovanov@intel.com>

Keep comments from the removed shell scripts adding them to the appropriate .state files: LNLM_SDW_AIOC.state MTLP_SWD_AIOC.state WARNING: these comments might be incompatible with `alsactl` tool in the future. Signed-off-by: Dmitrii Golovanov <dmitrii.golovanov@intel.com>

golowanow · 2025-08-19T07:08:49Z

check-alsabat- test cases are passed on all CI configurations

SzymonRichert

LGTM

golowanow force-pushed the alsa_state_20250807 branch from 780038d to 018d22b Compare August 8, 2025 07:10

golowanow added the improvement The issue or pull request improves already existing functionalities label Aug 8, 2025

golowanow force-pushed the alsa_state_20250807 branch 2 times, most recently from 2851a8c to ed8ff16 Compare August 11, 2025 18:17

marc-hb reviewed Aug 12, 2025

View reviewed changes

golowanow force-pushed the alsa_state_20250807 branch 2 times, most recently from c7b32b6 to 6ad42fa Compare August 12, 2025 12:21

golowanow changed the title ~~[SKIP SOF-TEST] lib.sh: Set ALSA settings from a state file~~ lib.sh: Set ALSA settings from a state file Aug 12, 2025

golowanow marked this pull request as ready for review August 12, 2025 12:25

golowanow requested review from a team and lgirdwood as code owners August 12, 2025 12:25

majunkier requested changes Aug 12, 2025

View reviewed changes

golowanow requested a review from majunkier August 12, 2025 17:56

golowanow force-pushed the alsa_state_20250807 branch from 6ad42fa to 7cc12cc Compare August 13, 2025 09:43

majunkier approved these changes Aug 13, 2025

View reviewed changes

marc-hb reviewed Aug 13, 2025

View reviewed changes

golowanow force-pushed the alsa_state_20250807 branch from 7cc12cc to 4786a83 Compare August 14, 2025 05:58

golowanow requested a review from marc-hb August 14, 2025 06:05

marc-hb reviewed Aug 14, 2025

View reviewed changes

LukaszMrugala and others added 5 commits August 19, 2025 06:32

lib.sh: set_alsa: Log ALSA state difference

97b1aaf

Log difference between the current ALSA state and the restored ALSA state. Signed-off-by: Dmitrii Golovanov <dmitrii.golovanov@intel.com>

golowanow added 2 commits August 19, 2025 06:33

alsa_settings: Add README.md

ccf21e5

Add README.md documentation for SOF test ALSA settings. Signed-off-by: Dmitrii Golovanov <dmitrii.golovanov@intel.com>

golowanow force-pushed the alsa_state_20250807 branch from 4786a83 to 8f201d1 Compare August 19, 2025 04:38

golowanow requested a review from marc-hb August 19, 2025 04:41

SzymonRichert reviewed Aug 19, 2025

View reviewed changes

lgirdwood approved these changes Aug 19, 2025

View reviewed changes

lgirdwood merged commit b5e7bc2 into thesofproject:main Aug 19, 2025
4 of 8 checks passed

majunkier mentioned this pull request Sep 2, 2025

alsa_settings: WCL platform settings #1298

Merged

This was referenced Sep 4, 2025

alsactl DUT restoration #1270

Open

[BUG] Some tests modify machine state #1275

Open

lib.sh: Set ALSA settings from a state file #1294

lib.sh: Set ALSA settings from a state file #1294

Uh oh!

Conversation

golowanow commented Aug 8, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

This PR is to continue with #1284 and PoC #1270

Uh oh!

marc-hb left a comment

Choose a reason for hiding this comment

Uh oh!

golowanow commented Aug 12, 2025

Uh oh!

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!

golowanow Aug 13, 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!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

marc-hb commented Aug 13, 2025

Uh oh!

golowanow commented Aug 13, 2025

Uh oh!

golowanow commented Aug 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

golowanow commented Aug 13, 2025

Uh oh!

golowanow commented Aug 13, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

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!

golowanow commented Aug 19, 2025

Uh oh!

SzymonRichert left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

golowanow commented Aug 8, 2025 •

edited

Loading

golowanow Aug 13, 2025 •

edited

Loading

golowanow commented Aug 13, 2025 •

edited

Loading