cadence: unification of cadence module naming

[dnikodem]

This change improves and standardizes the naming convention of the cadence module during its creation.

[dnikodem]

Sof CI tests fails - WiP

[dbaluta]

FYI: There is a compilation error problem, existing just before this patch!

 from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/audio_stream.h:23,
from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/buffer.h:11,
from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/component.h:19,
from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/module_adapter/module/generic.h:16,
from /home/nxf25322/work/sof_dir5/sof/src/audio/module_adapter/module/cadence.c:14:
/home/nxf25322/work/sof_dir5/sof/src/include/sof/lib/uuid.h:80:31: error: '_cadence_uuid' undeclared here (not in a function); did you mean 'cadence_api'?
80 | #define _UUID(uuid_name) (&_##uuid_name)
| ^
/home/nxf25322/work/sof_dir5/sof/zephyr/include/sof/trace/../../../../src/include/sof/trace/trace.h:410:27: note: in definition of macro 'DECLARE_TR_CTX'
410 | .uuid_p = uuid, \
| ^~~~
/home/nxf25322/work/sof_dir5/sof/src/include/sof/lib/uuid.h:148:29: note: in expansion of macro '_UUID'
148 | #define SOF_UUID(uuid_name) _UUID(uuid_name)
| ^~~~~
/home/nxf25322/work/sof_dir5/sof/src/audio/module_adapter/module/cadence.c:23:28: note: in expansion of macro 'SOF_UUID'
23 | DECLARE_TR_CTX(cadence_tr, SOF_UUID(cadence_uuid), LOG_LEVEL_INFO);
| ^~~~~~~~
In file included from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/buffer.h:18:
/home/nxf25322/work/sof_dir5/sof/src/audio/module_adapter/module/cadence.c:921:43: error: 'cadence_uuid' undeclared here (not in a function); did you mean 'cadence_api'?
921 | DECLARE_MODULE_ADAPTER(cadence_interface, cadence_uuid, cadence_tr);
| ^~~~~~~~~~~~
/home/nxf25322/work/sof_dir5/sof/src/include/sof/lib/uuid.h:81:31: note: in definition of macro '_RT_UUID'
81 | #define _RT_UUID(uuid_name) (&uuid_name)
| ^~~~~~~~~
/home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/module_adapter/module/generic.h:52:16: note: in expansion of macro 'SOF_RT_UUID'
52 | .uid = SOF_RT_UUID(uuid), \
| ^~~~~~~~~~~
/home/nxf25322/work/sof_dir5/sof/src/audio/module_adapter/module/cadence.c:921:1: note: in expansion of macro 'DECLARE_MODULE_ADAPTER'
921 | DECLARE_MODULE_ADAPTER(cadence_interface, cadence_uuid, cadence_tr);
| ^~~~~~~~~~~~~~~~~~~~~~
[24/68] Building C object modules/hal_nxp/hal_nxp/CMakeFiles/..__modules__hal__nxp.dir/mcux/mcux-sdk/drivers/iuart/fsl_uart.c.obj

I think it was introduce by @andyrossfdb039e / #9261

commit fdb039ef5f77e37e2566e343c8a831d1436e9d91
Author: Andy Ross <andyross@google.com>
Date: Sat Jun 22 13:31:39 2024 -0700
everywhere: Normalize UUID naming

[dbaluta]

@dnikodem Your PR not only unifies the cadence module naming! but also fixes a compilation problem!

[dnikodem]

FYI: There is a compilation error problem, existing just before this patch!

 from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/audio_stream.h:23,
from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/buffer.h:11,
from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/component.h:19,
from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/module_adapter/module/generic.h:16,
from /home/nxf25322/work/sof_dir5/sof/src/audio/module_adapter/module/cadence.c:14:
/home/nxf25322/work/sof_dir5/sof/src/include/sof/lib/uuid.h:80:31: error: '_cadence_uuid' undeclared here (not in a function); did you mean 'cadence_api'?
80 | #define _UUID(uuid_name) (&_##uuid_name)
| ^
/home/nxf25322/work/sof_dir5/sof/zephyr/include/sof/trace/../../../../src/include/sof/trace/trace.h:410:27: note: in definition of macro 'DECLARE_TR_CTX'
410 | .uuid_p = uuid, \
| ^~~~
/home/nxf25322/work/sof_dir5/sof/src/include/sof/lib/uuid.h:148:29: note: in expansion of macro '_UUID'
148 | #define SOF_UUID(uuid_name) _UUID(uuid_name)
| ^~~~~
/home/nxf25322/work/sof_dir5/sof/src/audio/module_adapter/module/cadence.c:23:28: note: in expansion of macro 'SOF_UUID'
23 | DECLARE_TR_CTX(cadence_tr, SOF_UUID(cadence_uuid), LOG_LEVEL_INFO);
| ^~~~~~~~
In file included from /home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/buffer.h:18:
/home/nxf25322/work/sof_dir5/sof/src/audio/module_adapter/module/cadence.c:921:43: error: 'cadence_uuid' undeclared here (not in a function); did you mean 'cadence_api'?
921 | DECLARE_MODULE_ADAPTER(cadence_interface, cadence_uuid, cadence_tr);
| ^~~~~~~~~~~~
/home/nxf25322/work/sof_dir5/sof/src/include/sof/lib/uuid.h:81:31: note: in definition of macro '_RT_UUID'
81 | #define _RT_UUID(uuid_name) (&uuid_name)
| ^~~~~~~~~
/home/nxf25322/work/sof_dir5/sof/src/include/sof/audio/module_adapter/module/generic.h:52:16: note: in expansion of macro 'SOF_RT_UUID'
52 | .uid = SOF_RT_UUID(uuid), \
| ^~~~~~~~~~~
/home/nxf25322/work/sof_dir5/sof/src/audio/module_adapter/module/cadence.c:921:1: note: in expansion of macro 'DECLARE_MODULE_ADAPTER'
921 | DECLARE_MODULE_ADAPTER(cadence_interface, cadence_uuid, cadence_tr);
| ^~~~~~~~~~~~~~~~~~~~~~
[24/68] Building C object modules/hal_nxp/hal_nxp/CMakeFiles/..__modules__hal__nxp.dir/mcux/mcux-sdk/drivers/iuart/fsl_uart.c.obj

I think it was introduce by @andyross

commit fdb039ef5f77e37e2566e343c8a831d1436e9d91
Author: Andy Ross <andyross@google.com>
Date: Sat Jun 22 13:31:39 2024 -0700
everywhere: Normalize UUID naming

correct, this is why I prepared this PR. The problem is that I see a lot of fails in CI (probably not related to this change) - need to verify it.

[andyross]

That's the correct way to adjust the naming. Is this not buildable in CI or otherwise from e.g. xtensa_build_{all,zephyr} or rebuild_testbench.sh? What's the correct way to validate this component? If the validation isn't in-tree, can we add an issue to make sure it gets a test?

(Honestly I sorta worry that there are others like this hiding. SOF has a lot of seemingly dead/zombied code and this was a pretty invasive PR that plays poorly in that kind of bathtub.)

[dnikodem]

Looks like the issues are not related to my PR:

1. sof-logger failed on all PRs
2. LNL SDW's failures are also seen on other PRs

I think that this is ready to merge.

[kv2019i]

@dnikodem wrote:

Looks like the issues are not related to my PR:

1. sof-logger failed on all PRs
2. LNL SDW's failures are also seen on other PRs

I think that this is ready to merge.

Ack, (1) is fixed in main via sof-test PRs and (2) is tracked as thesofproject/linux#5098

Proceeding with merge.

[dbaluta]

That's the correct way to adjust the naming. Is this not buildable in CI or otherwise from e.g. xtensa_build_{all,zephyr} or rebuild_testbench.sh? What's the correct way to validate this component? If the validation isn't in-tree, can we add an issue to make sure it gets a test?

In order to validate this component you will need some proprietary libraries from Cadence. Intel / NXP are using different libraries.

We validate this internally, using our CI - and it usually takes 1 or 2 days until we get signalled about issues.

I think Intel does the same.

[andyross]

In order to validate this component you will need some proprietary libraries from Cadence. Intel / NXP are using different libraries.

The normal pattern for that sort of thing is to provide a "stub" implementation selected via kconfig that emulates the proprietary bits as some kind of noop. Then you build that in a default configuration (the fuzzer seems to be evolving in that direction as the official "enable everything for testing" config) that runs in CI, so you're guaranteed you don't get surprised by this kind of API change.