Guennadi Liakhovetski

@wszypelt can this still be reproduced? But this test isn't in the standard CI, so we don't see it in PR testing?

Instead of https://github.com/thesofproject/sof/pull/6030/commits/c9dccff195c16f533b481f25c7d2d397bbbba253 the following diff: ``` diff --git a/src/arch/xtensa/lib/cpu.c b/src/arch/xtensa/lib/cpu.c index 1802384fb..fa9bce42e 100644 --- a/src/arch/xtensa/lib/cpu.c +++ b/src/arch/xtensa/lib/cpu.c @@ -70,12 +70,16 @@ static void unpack_dynamic_vectors(void) } #endif +bool cpu_power_down[CONFIG_CORE_COUNT]; +...

> > +bool cpu_power_down[CONFIG_CORE_COUNT]; > > + > > Not cache coherent, Any array of elems where the elems are not multiples of cache line size are incoherent. @lgirdwood we...

Re-opening because I can reproduce very similar timeout errors on UP-Xtreme i11 with nocodec with two loops: one with aplay and another one with arecord. But even more often the...

@wszypelt could you explain what these tests are - I don't see them in the list of PR QB tests? Are they new or are they only run manually, not...

#4722 *might* help to fix this issue - untested

> If it's alright with you @dbaluta @lgirdwood @lyakh @kv2019i I believe we can merge this PR for now as it seems to work okay without enabling the DMA trace....

@LaurentiuM1234 I agree, that the PR looks rather good now, but I'm going to ask you about one more change, sorry. We discussed again with @lgirdwood and this is how...

Thanks for fixing, looks ok now. For my understanding: until now "simple" module adapters had 1 sink and 1 source. This PR adds support for 1 sink-many source (1:N) modules....

> @lgirdwood @lyakh I like the new suggestion but its kind of hard to look at those configurations now without a module to test with. Can we keep that as...