ASoC: SOF: Intel: Fix pipeline state transitions for aggregate DAIs#5765
Open
ujfalusi wants to merge 1 commit into
Open
ASoC: SOF: Intel: Fix pipeline state transitions for aggregate DAIs#5765ujfalusi wants to merge 1 commit into
ujfalusi wants to merge 1 commit into
Conversation
ujfalusi
requested review from
bardliao,
dbaluta,
kv2019i,
lgirdwood,
plbossart and
ranj063
as code owners
There was a problem hiding this comment.
Pull request overview
This PR fixes IPC4 pipeline state transitions for aggregate back-end links (num_cpus > 1) where multiple CPU DAIs may share the same SOF pipeline, which can currently cause premature RUNNING transitions and redundant PAUSED IPCs.
Changes:
- Add a per-pipeline
trigger_countto gate START/PAUSE_RELEASE so RUNNING is sent only after all DAIs sharing a pipeline have completed link DMA triggering. - Deduplicate STOP/SUSPEND/PAUSE_PUSH PAUSED IPCs for shared pipelines by checking the current pipeline state.
- Compute the “DAIs-per-pipeline” threshold dynamically per BE link to support shared/independent/mixed aggregate topologies.
Reviewed changes
Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.
| File | Description |
|---|---|
| sound/soc/sof/sof-audio.h | Adds trigger_count to struct snd_sof_pipeline to track aggregate trigger completion per pipeline. |
| sound/soc/sof/intel/hda-dai-ops.c | Adds per-pipeline DAI counting and uses trigger_count to gate RUNNING/PAUSED IPC transitions for aggregate DAIs. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| if (!w) | ||
| continue; | ||
|
|
||
| sw = w->dobj.private; |
Comment on lines
+443
to
+449
| int spipe_dais = hda_ipc4_count_spipe_dais(substream, | ||
| swidget->spipe); | ||
|
|
||
| swidget->spipe->trigger_count++; | ||
| swidget->spipe->started_count++; | ||
| if (swidget->spipe->trigger_count < spipe_dais) | ||
| break; |
kv2019i
previously approved these changes
May 12, 2026
Collaborator
kv2019i
left a comment
kv2019i
left a comment
There was a problem hiding this comment.
Please check the copilot comment on defensive programming. Otherwise complex, but looks good to me.
| if (!w) | ||
| continue; | ||
|
|
||
| sw = w->dobj.private; |
ujfalusi
force-pushed
the
peter/sof/pr/aggregated-sdw-trigger
branch
from
4f4063d to
22f1037
Compare
Collaborator
Author
|
Changes since v1:
|
For aggregate DAIs (num_cpus > 1) the pre_trigger/post_trigger callbacks send pipeline state IPCs per-DAI without considering that multiple DAIs may share the same pipeline. This causes premature state transitions where the pipeline goes RUNNING before all link DMAs have started, or individual DAIs send redundant IPCs for shared pipelines. Fix this by checking the HDA stream running state in post_trigger: - START/PAUSE_RELEASE: defer RUNNING IPC until all DAIs sharing the same pipeline have their link DMA streams running - STOP/SUSPEND/PAUSE_PUSH: use pipeline state dedup so the PAUSED IPC is sent once regardless of how many DAIs share the pipeline The running-state check naturally handles all aggregate topologies: shared pipelines, independent pipelines, and mixed cases without requiring per-pipeline counters. Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
ujfalusi
force-pushed
the
peter/sof/pr/aggregated-sdw-trigger
branch
from
22f1037 to
763f980
Compare
ujfalusi
changed the title
Collaborator
Author
|
Changes since v2: new approach to drop the changes outside of sof/intel code
|
Comment on lines
+438
to
+446
| /* | ||
| * For aggregated DAIs (num_cpus > 1), defer pipeline RUNNING | ||
| * IPC until all CPU DAIs sharing this pipeline have started | ||
| * their link DMAs via hda_trigger(). The running state of each | ||
| * DAI's HDA stream naturally tracks completion. | ||
| */ | ||
| if (num_cpus > 1 && | ||
| !hda_ipc4_all_spipe_dmas_running(substream, swidget->spipe)) | ||
| break; |
Collaborator
Author
There was a problem hiding this comment.
the PR description has been updated...
Comment on lines
479
to
483
| /* | ||
| * STOP/SUSPEND trigger is invoked only once when all users of this pipeline have | ||
| * been stopped. So, clear the started_count so that the pipeline can be reset | ||
| * been stopped. So, clear the started_count so that the pipeline can be reset. | ||
| */ | ||
| swidget->spipe->started_count = 0; |
Collaborator
Author
There was a problem hiding this comment.
this change should be removed, it is not related to topic.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
For aggregate DAIs (num_cpus > 1) the pre_trigger/post_trigger callbacks
send pipeline state IPCs per-DAI without considering that multiple DAIs
may share the same pipeline. This causes premature state transitions
where the pipeline goes RUNNING before all link DMAs have started, or
individual DAIs send redundant IPCs for shared pipelines.
Fix this by checking the HDA stream running state in post_trigger:
same pipeline have their link DMA streams running
is sent once regardless of how many DAIs share the pipeline
The running-state check naturally handles all aggregate topologies:
shared pipelines, independent pipelines, and mixed cases without
requiring per-pipeline counters.