Skip to content

mux: consider sinks on same direction only for prepare/reset propagation#6882

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
mwasko merged 1 commit into from
Jan 12, 2023
Merged

mux: consider sinks on same direction only for prepare/reset propagation #6882

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
33 changes: 31 additions & 2 deletions src/audio/mux/mux.c
View file
Open in desktop
Original file line numberDiff line numberDiff line change
Expand Up@@ -629,6 +629,11 @@ static int mux_reset(struct comp_dev *dev)
comp_info(dev, "mux_reset()");

if (dir == SOF_IPC_STREAM_PLAYBACK) {
/*
* On playback path, stop propagating reset signal to downstream if any of the
* source is active. The condition is only applicable to comp MUX (multi-source to
* single-sink).
*/

@lyakhlyakhJan 9, 2023

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

There's no explicit check for the mux / demux type here. We just implicitly use the fact, that in case of a demux we would be coming into reset from the upstream source direction, so the single source is guaranteed to already be inactive? So the below condition if (state > COMP_STATE_READY) can only trigger for a mux?

@johnylin76johnylin76Jan 10, 2023•
edited
Loading

Copy link
Copy Markdown
ContributorAuthor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

AFAIK reset trigger starts from PCM and propagates downstream/upstream for playback/capture. It implies that for DEMUX on playback, reset can be received only when its (single) source got reset. Same concept for MUX on capture.

However I'm not sure if the trigger/propagation rule will be changed in the future. For example of MUX on capture, would it be possible to reset a PCM-less pipeline which presents as a branch of MUX upstream in multiple endpoints scheme? (like PPL#3 in the following graph)

PCMC <-- PPL#1 <--- MUX <-- PPL#2 <--- DAI#2
\--- PPL#3 <--- DAI#3

list_for_item(blist, &dev->bsource_list) {
struct comp_buffer *source = container_of(blist, struct comp_buffer,
sink_list);
Expand All@@ -643,17 +648,34 @@ static int mux_reset(struct comp_dev *dev)
return PPL_STATUS_PATH_STOP;
}
} else {
/*
* On capture path, stop propagating reset signal to upstream if any of the sink is
* active. The condition is only applicable to comp DEMUX (single-source to
* multi-sink).
*
* One exception is that the active sink of DEMUX is on playback path, e.g. DEMUX
* on Echo Reference path (capture) linked to the sink SMART_AMP on Speakers path
* (playback). In practice, the playback sink state should not affect the reset
* process through the capture path DEMUX locates on.
*/
Comment thread
lgirdwood marked this conversation as resolved.
list_for_item(blist, &dev->bsink_list) {
struct comp_buffer *sink = container_of(blist, struct comp_buffer,
source_list);
struct comp_buffer __sparse_cache *sink_c = buffer_acquire(sink);
int state = -1;
int direction = -1;

if (sink_c->sink)
if (sink_c->sink) {
state = sink_c->sink->state;
direction = sink_c->sink->direction;
}
Comment thread
johnylin76 marked this conversation as resolved.
Outdated

buffer_release(sink_c);

/* no need to regard the playback sink state */
if (direction == SOF_IPC_STREAM_PLAYBACK)
continue;

if (state > COMP_STATE_READY)
/* should not reset the upstream components */
return PPL_STATUS_PATH_STOP;
Expand DownExpand Up@@ -719,12 +741,19 @@ static int mux_prepare(struct comp_dev *dev)
source_list);
struct comp_buffer __sparse_cache *sink_c = buffer_acquire(sink);
int state = -1;
int direction = -1;

if (sink_c->sink)
if (sink_c->sink) {
state = sink_c->sink->state;
direction = sink_c->sink->direction;
}

buffer_release(sink_c);

/* no need to regard the playback sink state */
if (direction == SOF_IPC_STREAM_PLAYBACK)
continue;

@plbossartplbossartDec 22, 2022

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

And in this case you only deal loopbacks with the capture path

@lgirdwoodlgirdwoodJan 6, 2023

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So it sounds like rather than use direction as the propagation flag, we should use a custom flags here (for each direction) and it should come from topology ?
Long term we need to move all the pipeline propagation logic up the stack into common pipeline, but this needs o be don after we are using the module API for all processing.

@lgirdwoodlgirdwoodJan 11, 2023

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@johnylin76 any comment here ? am I wrong in my assumptions here ?


/* only prepare upstream if we have no active sinks */
if (state == COMP_STATE_PAUSED || state == COMP_STATE_ACTIVE)
return PPL_STATUS_PATH_STOP;
Expand Down