Repository navigation
[BUG][mtl-002] noise/playback drag issue observed with mtl-002-drop stable release #6727
Description
mtl-002-drop stable release results in playback drag issue.
topology and firmware used from mtl-002-drop stable branch
{{ message }}
Activity
[-][BUG] noise/playback drag issue observed with mtl-002-drop stable release[/-][+][BUG][mtl-002] noise/playback drag issue observed with mtl-002-drop stable release[/+]RDharageswari commentedon Dec 6, 2022
topologyandloopbackfile.zip
This issue is observed with the nocodec topology as well. Please find attached the no_codec topology and the captured loopback file with noise
command used:
aplay -Dhw:0,0 -c2 -r48000 -fS16_LE /home/LRMonoPhase4.wav -vv &
arecord -Dhw:0,0 /home/loop-mtl-002.wav -c2 -r48000 -fs16_le -d10
There are no dmesg error logs
jairaj-arava commentedon Dec 6, 2022
@sathya-nujella @yongzhi1 @udaymb
lgirdwood commentedon Dec 7, 2022
@mwasko @marcinszkudlinski fyi
miRoox commentedon Dec 8, 2022
We can reproduce this issue with both main and mtl-002, but the behavior of main seems a little bit different:
On main branch: If we playback and capture simultaneously at first, we can hear the noise in the captured loopback file; but if we only playback at first, then try playback and capture simultaneously, the captured loopback sounds clear. We guess there might be some conflicts when setting up hw params.
On mtl-002, we can always hear the noise.
aborisovich commentedon Dec 9, 2022
Is it I2S or HDA configuration?
What tolopogy configuration is there?
juimonen commentedon Dec 9, 2022
@aborisovich in the topology zip file there is nocodec tplg -> so I would assume I2S/SSP.
RDharageswari commentedon Dec 9, 2022
driver, firmware logs and the captured loopback file
fw_driver_logs_loopbackdatacaptured_issue.zip
working_logs.zip
mwasko commentedon Dec 12, 2022
@RDharageswari we are still missing logs from the SW that include IPC payloads. This information is critical to reproduce the problem with a FW tests. @juimonen, @lgirdwood can someone from SW team help here?
lgirdwood commentedon Dec 12, 2022
@ranj063 @ujfalusi can someone provide the kernel option to dump all the IPC logs into dmesg. Thanks !
ujfalusi commentedon Dec 12, 2022
options snd_sof sof_debug=0x341should print all dumps and log all messages (even DMA position updates - for dtrace BIT(8)) and their completion.Or do we want to dump the messages itself? They will flood the kernel log pretty fast...
ujfalusi commentedon Dec 12, 2022
Actually, if the kernel is configured with
DEBUG_VERBOSEthen the kernel sent message payload is printed out (normal and large messages as well) for IPC4.abonislawski commentedon Dec 12, 2022
This can be fixed with i2s master link clock selection. Confirmed today with Dhara.
For mtl-002 we can do quick fix with manual selection but for main branch a proper mechanism should be developed using driver config.
mengdonglin commentedon Dec 15, 2022
Closing as being fixed.