ext_manifest: include probe information by CONFIG_PROBE

[Shawn-Huang724]

Put probe information into manifest by CONFIG_PROBE=y would be better.

This change could help to avoid annoying kernel log from SOF loader ("unknown sof_ext_man header type 3 size 0x30") if CONFIG_PROBE=n on most SOF machines.

[sofci]

Can one of the admins verify this patch?

reply test this please to run this test once

[Shawn-Huang724]

Fixes https://github.com/thesofproject
#9674.

[lyakh]

test this please

[kv2019i]

Thanks @Shawn-Huang724 -- it's seems this annoyance has gone unfixed for a very long time.

It would seem thesofproject/linux#5183 is probably the best fix. At least for Intel, the new FW releases use IPC4 so will not trigger the warning in IPC3 loader in kernel.

[Shawn-Huang724]

Thanks @Shawn-Huang724 -- it's seems this annoyance has gone unfixed for a very long time.

It would seem thesofproject/linux#5183 is probably the best fix. At least for Intel, the new FW releases use IPC4 so will not trigger the warning in IPC3 loader in kernel.

Very much agree, it's better for the doer to undo what he has done. [thesofproject/linux#5183] is the best fix of this annoyance in dmesg log.
But from the perspective of firmware, this change could increase the integrity and consistency of SOF firmware. This means that the probe information would be not found since the CONFIG_PROBE is disabled in firmware image.
Thanks again for your valuable feedback and reviewing.

[lgirdwood]

Ack - manifest should only contain extensions if they are enabled in build, I think in this case the author may have been trying to cover both probes and injectors.

[kv2019i]

I was wondering @lgirdwood if some client requires the manifest entry to exist (with just probe and injection points as zero in the manifest), but it seems all CI tests are passing and I can't find any usage. So would seem good to go.