SRC fixes #4341
SRC fixes #4341
Conversation
ranj063
requested review from
ajitkupandey,
bardliao,
dbaluta,
kuanhsuncheng,
kv2019i,
lgirdwood,
plbossart,
ujfalusi and
yaochunhung
as code owners
ranj063
force-pushed
the
fix/src_formats
branch
2 times, most recently
from
82bbaac to
f744f61
Compare
singalsu
left a comment
singalsu
left a comment
There was a problem hiding this comment.
Tested a number of playback and capture SRC rates combinations 32/44.1/48/96 kHz and everything worked. (thesofproject/sof#7547, thesofproject/sof#7440 needed).
I assume similar change will be needed for ASRC.
|
SOFCI TEST |
sound/soc/sof/ipc4-topology.c
Outdated
| out_ref_valid_bits = SOF_IPC4_AUDIO_FORMAT_CFG_V_BIT_DEPTH(in_fmt->fmt_cfg); | ||
| /* | ||
| * For playback, the SRC will only ever have a single output format based on what the DAI | ||
| * supports. The reference params will be ignored when choosing the only available format. |
There was a problem hiding this comment.
it's also wrong to assume that the SRC output is related directly in a 1:1 manner to the DAI. We could perfectly have multiple stages with two SRC, one to go to 48kHz and a second to reach 192kHz.
…ormat Modify the pipeline_params based on the SRC output format and set the sink_rate in the IPC data. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
For playback, the SRC sink rate must be configured based on the requested output format which is restricted to only handle DAI's that support a single audio format for now. For capture, the SRC module should convert the rate to match the rate requested by the PCM hw_params. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
When we walk the list of connected widgets from the source to the sink to prepare all widgets, the pipeline_params must be modified to reflect the output audio format at each widget. But, the copier only modifies the sample format in the pipeline_params. So, fix it to also modify the rate and channels. Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
|
big regression on ADLP_RVP_NOCODEC_IPC4ZPH https://sof-ci.01.org/linuxpr/PR4341/build4885/devicetest/index.html, is this a known issue @ranj063? |
@plbossart I think there's an issue with that particular device. It fails on several PRs |
|
SOFCI TEST |
@plbossart i ran the pr test manually https://sof-ci.ostc.intel.com/#/result/planresultdetail/25789 and it looks good. |
4 participants
No description provided.