Guennadi Liakhovetski

> DONE !!!! > > * [[SKIP CI] Add sparse build to Github Actions #6311](https://github.com/thesofproject/sof/pull/6311) @marc-hb great, approved, thanks for investigating and fixing it! After that we should add checks...

> > Intel test plan 14732 started > > The only major fail is for ADLP_RVP_NOCODEC_CI but that's a different topology. It seems like a known issue. @keqiaozhang @lyakh can...

@lgirdwood @plbossart to reply to several comments, made so far in this PR: I don't think handling of the `source_c->source` (and similar) pointer has anything to do with `buffer_acquire()` /...

> Let's try differently @lyakh. > > What is the rule telling us when source_c->source or sink_c->sink can be safely dereferenced? @plbossart I don't think there are any special rules...

> > What are the rules of handling `source` and `sink` component objects, or any other objects to which buffer points - is a different question. So it is up...

> Who's the consumer of the notifier message ? @lgirdwood `dai_atomic_trigger()` - the notification is used to handle DAI groups. When all DAIs have handled the trigger, that callback is...

> > > Who's the consumer of the notifier message ? > > > > > > @lgirdwood `dai_atomic_trigger()` - the notification is used to handle DAI groups. When all...

> The callback is run for each single physical DAI instance in a loop. How would moving IRQs handling inside the callback help to protect the entire loop for all...

@lgirdwood we're waiting for a reply from @mmaka1 here

@lgirdwood I cannot claim that I fully understand what the present code is trying to achieve with the `.walking` flag, but it is my impression that it's exactly what it...