`west spdx` does not produce valid SBOMs for `native_sim`

[kartben]

Describe the bug

It would appear that native_sim is maybe a bit special in that it doesn't has a zephyr_final CMake target, which prevent the SPDX tooling to generate a valid SBOM for it does not know what the "main" target is (it's expecting it to be zephyr_final) and there does not include the quite important "Relationship: SPDXRef-DOCUMENT DESCRIBES SPDXRef-zephyr-final" entry in the resulting SBOM.

Flagging as "annoyance" given it's reasonably easy to build with another emulated target if one is only interested in a general evaluation of SPDX support in Zephyr with no particular ties to a given hardware target (and the dependencies it may be pulling).

Regression

• This is a regression.

Steps to reproduce

1. west spdx --init -d BUILD_DIR
2. west build -d BUILD_DIR samples/hello_world -b native_sim -- -DCONFIG_BUILD_OUTPUT_META=y
3. west spdx -d BUILD_DIR --spdx-version 2.2

Notice how while generation is not failing, the build.spdx file is missing an Relationship: SPDXRef-DOCUMENT DESCRIBES SPDXRef-zephyr-final entry.

Relevant log output

Impact

Annoyance – Minor irritation; no significant impact on usability or functionality.

Environment

macOS
SDK 0.17.1
main @d4758f02b18b9261c0b98df244ff0bad7e824551

Additional Context

zephyr/CMakeLists.txt

Lines 53 to 54 in 8ab712a

	# When only a single linking stage is required, the `zephyr_pre0` will be mapped
	# into the `zephyr_final` target.

seems to indicate zephyr_final should exist even for native_sim, but this doesn't seem to be the case. zephyr_pre0 is there though.

[pdgendt]

@tejlmand could you provide some feedback here?

[aescolar]

The SPDX part is very much out of my turf, and about targets probably @tejlmand can comment much better than me,
but is the problem that the west spdx command assumes there is a zephyr_final target (which is not always the case)?

(The native simulator final executable assembly pigybacks on whatever target ${logical_target_for_zephyr_elf} is, and I tend to recall that was what we were meant to base any later target on(?))

Maybe this is silly, but should the west spdx build use some other target than zephyr_final?
add_custom_target(spdx_target DEPENDS ${logical_target_for_zephyr_elf} )?

In any case, I have the impression that the fact that this presents itself in native_sim is not so much about native_sim but about the assumption that zephyr_final exists and it not existing always/for native_sim. So I will assign this to @tejlmand hoping he can provide better input or guidance.

[kartben]

but is the problem that the west spdx command assumes there is a zephyr_final target (which is not always the case)?

Yep -- and according to the comment in CMakeLists.txt I quoted, it was my assumption that there should always be one (even if in some case it might be equivalent to pre0)

Maybe this is silly, but should the west spdx build use some other target than zephyr_final?
add_custom_target(spdx_target DEPENDS ${logical_target_for_zephyr_elf} )?

Possibly, yes :) Thanks for the hint!