Audio: RTNR: Convert to module API #8324
Conversation
|
This is early draft version, shared to find about opens related to module conversion. Currently if fails in init() due to not having configuration blob available, that would be sent after init(), before prepare() by kernel. |
I need to modify further the code to run it, now the problem with binary control arrival after init() prevents the init() to pass. I think it's too early to attempt to initialize the algorithm there. Also, it does not impact mockup, but there's an issue in controlling two internal features with same binary control. It would be simplest to have two binary controls unless I've understood this wrong. |
singalsu
force-pushed
the
rtnr_module_convert
branch
from
33ac228 to
5491cb3
Compare
singalsu
force-pushed
the
rtnr_module_convert
branch
from
5491cb3 to
ed63435
Compare
singalsu
force-pushed
the
rtnr_module_convert
branch
from
ed63435 to
7756bfd
Compare
|
@singalsu still draft ? |
There's some regression, the stub library tries to process 384 channels per frame. Function audio_stream_get_channels() returns it. The function is correct but the stream format gets messed up somewhere. |
singalsu
force-pushed
the
rtnr_module_convert
branch
from
7756bfd to
3c6febe
Compare
src/audio/rtnr/rtnr_stub.c
Outdated
| /* Allocate something, to avoid return NULL and cause error | ||
| * in check of success of this. | ||
| */ | ||
| return rzalloc(SOF_MEM_ZONE_RUNTIME, 0, SOF_MEM_CAPS_RAM, 42); |
There was a problem hiding this comment.
is 42 == sizeof(audio_stream_rtnr)? if so, better use sizeof, or if number are random, you can use 4, 42 is not easy to read.
There was a problem hiding this comment.
I can make it a macro. We don't know the real size used by the closed library, and the run fails if this is NULL. Since I had to choose something, number 42 a reference to plot of https://en.wikipedia.org/wiki/The_Hitchhiker%27s_Guide_to_the_Galaxy . :)
There was a problem hiding this comment.
you may need move 42 definition from:
https://github.com/thesofproject/sof/pull/8301/files#diff-f09311b76861c69cd17f987de389037705c0964aaf78d0d72e0e6da39bcdb2ffR12
to a more common place and reuse it between rtnr and igo_nr module.
There was a problem hiding this comment.
this can be done later, now, I am ok go for merge.
|
@singalsu any ETA on non draft ? |
singalsu
force-pushed
the
rtnr_module_convert
branch
from
3c6febe to
e65a043
Compare
|
Changing to ready for review! The component now works with the stub as ALSA control disable and enable, no more stuck to xrun. I didn't change the rtnr_copy_to/from_sof_stream() from functions where the issue was seen. Apparently there was some regression in SOF that has got fixed. |
This patch updates the RTNR component to module adapter API and IPC4. Signed-off-by: Seppo Ingalsuo <[email protected]>
The "sof," as local sources list is not correct. It causes build fail with x86 testbench when CONFIG_COMP_RTNR_STUB is set. Signed-off-by: Seppo Ingalsuo <[email protected]>
This patch enables RTNR (with stubs) to be built and loaded to testbench. Signed-off-by: Seppo Ingalsuo <[email protected]>
The init code checks for NULL and errors, so need to return from RTKMA_API_Context_Create() a valid allocated address. The free is added to RTKMA_API_Context_Free(). Signed-off-by: Seppo Ingalsuo <[email protected]>
This change avoids with stub library version annoying playback sound from non-updated sink buffer when the processing ALSA switch control is enabled. The audio is passed through from source to sink. Signed-off-by: Seppo Ingalsuo <[email protected]>
This patch allows load of RTNR component with these platforms. Signed-off-by: Seppo Ingalsuo <[email protected]>
singalsu
force-pushed
the
rtnr_module_convert
branch
from
e65a043 to
210899d
Compare
No description provided.