tests: arch: arm: no_multithreading: Limit platforms allowed

[nordic-krch]

When multithreading is off, kernel source files like sem.c (samphore
implementation) are not present in the build. Some platforms by default
fetch modules or drivers that are using multithreading primitives and
because of that fails to compile when multithreading is off.

Limit the test to only qemu platforms since test is arch specific.

Fixes #34739.

Signed-off-by: Krzysztof Chruscinski krzysztof.chruscinski@nordicsemi.no

[ioannisg]

This needs to run on hardware, we cannot do it like this.

[galak]

This needs to run on hardware, we cannot do it like this.

Do you expect a fix soon? If not, I'd like to merge this and than we deal with expanding this to run on other platforms as a follow up PR.

[ioannisg]

This needs to run on hardware, we cannot do it like this.

Do you expect a fix soon? If not, I'd like to merge this and than we deal with expanding this to run on other platforms as a follow up PR.

So let me ask this - we merged changes in the support of CONFIG_MULTITHREADING=n feature, (probably here: #34279) that broke this test, which has been working for quite a long time, i.e. we had support for CONFIG_MULTITHREADING=n on real ARM hardware, and we consider that the fix is to run the test only for QEMU targets?

Looking closely on #34279; it appears to me that this PR has introduced additional tests for the CONFIG_MULTITHREADING=n use-case, but all of them build and run only on QEMU targets, why have we agreed to go this way? Nothing in the description of the PR suggests that from now on, the CONFIG_MULTITHREADING=n feature is only supported on qemu targets, not hardware.

[stephanosio]

Some platforms by default
fetch modules or drivers that are using multithreading primitives and
because of that fails to compile when multithreading is off.

In #34739, we are seeing messages like:

libzephyr.a(rtt_console.c.obj): in function `k_mutex_lock':
libzephyr.a(rtt_console.c.obj): in function `k_mutex_unlock':
libzephyr.a(rtt_console.c.obj): in function `k_sleep':

Were these functions undefined/unsupported prior to #34279 when CONFIG_MULTITHREADING=n?

[nordic-krch]

Sorry guys, i'm on vacation with limited access to PC (back Thursday) #34279 changed kernel CMakeLists.txt to not include multithreading files when CONFIG_MULTITHREADING=n. Previously, this test was compile and was actually fetching multithreading.

Not sure how to filter ok platforms. I checked that test compiles for nrf boards except boards like Thingy (#34468) which has sensors and uses i2c and spi.

@ioannisg @galak

This needs to run on hardware, we cannot do it like this.

I can add some nrf boards? Others i want be able to test.

[ioannisg]

This is not the right thing to do.
CONFIG_MULTITHREADING=n was de-deprecated, so it needs to run in some set of hardware.
I'll file an issue about this.