[keyonjie]

Add -u option to allow force building xtensa(up) ARCH on platforms which usually building xtensa-smp by default, this is helpful to debug SMP specific issues.

[lgirdwood]

@keyonjie can you redo this after cmake has been merged.

[keyonjie]

@keyonjie can you redo this after cmake has been merged.

sure.

[lgirdwood]

@keyonjie build update merged so you can redo this now.

[keyonjie]

@keyonjie build update merged so you can redo this now.

@lgirdwood done.

[lgirdwood]

@keyonjie can you check this passes internal CI

[keyonjie]

@keyonjie can you check this passes internal CI

it is just finished, all checks passed.

[xiulipan]

@jajanusz
I did not see cmake take any option about arch
how to chose "xtensa-smp" or "xtensa"?
@lgirdwood @keyonjie
I did not see we now use ARCH in cmake configure. Please wait for @jajanusz clarify.

[jajanusz]

@xiulipan
Sorry for no docs for 'how internals work' yet, will do it in free time.
However I wrote about this issue here: #845 (comment)

@lgirdwood @xiulipan @keyonjie
ARCH command line arg is not supported, because we have only 1 FW ARCH, so unless you build for host, your ARCH is xtensa and SMP/UP is just config set from Kconfig.
However there is ARCH variable in cmake buildsystem ready to support others archs (for example 'arm' or whatever) in kernel-like way where you have arch/${ARCH}/ folders, but since we have only 1 arch now, you have no reason to touch it.

[jajanusz]

tldr solution:

	if [[ "x$BUILD_FORCE_UP" == "xyes" ]]
	then
		echo "CONFIG_SMP=n" > override.config
		make overrideconfig
	fi

IMHO we shouldn't keep going for flags, we should just have prepared configs for CIs. I added way for legacy support of ci scripts, but it's not something what you usually see in Kconfig based projects.

[keyonjie]

echo "CONFIG_SMP=n" > override.config

Thanks for clarification, transforming to Kconfig is really good, let me update my PR.

[lgirdwood]

@jajanusz @xiulipan do you guys now approve ?

[jajanusz]

I just noticed it would be better to use > instead of >>. My mistake, cos i misled you. Anyway it doesn't change final result and is good to merge.

[lgirdwood]

@keyonjie can you make the > change and respond to the CI error.

[keyonjie]

I just noticed it would be better to use > instead of >>. My mistake, cos i misled you. Anyway it doesn't change final result and is good to merge.

I got your point, as we "make overrideconfig" for every case, so we actually don't need append mode here, right?

[keyonjie]

@jajanusz updated to address your question, please help review.

[jajanusz]

I just noticed it would be better to use > instead of >>. My mistake, cos i misled you. Anyway it doesn't change final result and is good to merge.

I got your point, as we "make overrideconfig" for every case, so we actually don't need append mode here, right?

Yea, that's what I mean, you made it even a bit better than what I suggested.
Good job.

[xiulipan]

LGTM.