[DNM] dai-zephyr: use frames aligned for multi-endpoint copy #9196
Conversation
iganakov
requested review from
RanderWang,
abonislawski,
pblaszko,
lgirdwood,
plbossart,
mmaka1,
lbetlej,
dbaluta and
kv2019i
as code owners
iganakov
force-pushed
the
iganakov/multi_endpoint_copy_fix
branch
from
fc9ad55
to
6bdc7d4
Compare
iganakov
force-pushed
the
iganakov/multi_endpoint_copy_fix
branch
2 times, most recently
from
2f8bcfd
to
26e362a
Compare
iganakov
force-pushed
the
iganakov/multi_endpoint_copy_fix
branch
from
26e362a
to
31bf254
Compare
iganakov
force-pushed
the
iganakov/multi_endpoint_copy_fix
branch
from
31bf254
to
04cc408
Compare
When aggregating streams using dai_zephyr_multi_endpoint_copy() multiple glitches observed in audio streams with odd number of channels. Use audio_stream_avail_frames_aligned() for every dai in a loop to correctly calculate processing frames. Signed-off-by: Ievgen Ganakov <[email protected]>
iganakov
force-pushed
the
iganakov/multi_endpoint_copy_fix
branch
from
04cc408
to
25fbae0
Compare
|
This doesn't look good in the CI results https://sof-ci.01.org/sofpr/PR9196/build5811/devicetest/index.html?model=LNLM_SDW_AIOC&testcase=check-playback-10sec I don't think we have this in baseline. |
@iganakov I dont think any of the CI tests are doing odd number of channels wrt the number of test failures, this looks like we have impact on even channel numbers too. |
|
@kv2019i @lgirdwood I'm able to reproduce the issue with Windows driver also. Looks like something has changed in FW since I implemented this fix some time ago. Let me find the root cause. |
iganakov
changed the title
When aggregating streams using
dai_zephyr_multi_endpoint_copy()multiple glitches observed in audio streams with odd number of channels. Useaudio_stream_avail_frames_aligned()for every dai in a loop to correctly calculate processing frames.