optee-os_git.bb is overridden by a defective version in meta-linaro/meta-optee

[selbstereg]

---

SHORT VERSION:

Why does meta-renesas (branch krogoth) provide a recipe optee-os_git.bb, if it depends on meta-linaro/meta-optee (branch krogoth), which overrides this recipe (with a version, that seems defective if used together with meta-renesas)?

---

LONG VERSION:

I'm trying to update my RCAR-H3 image build from fido to krogoth.
Now the recipe optee-os_git.bb depends on python-wand and python-pycrypto recipes. To get these I cloned meta-linaro and checked out it's krogoth-branch.
The python-* recipes are provided in meta-linaro/meta-optee which also contains another optee-os_git.bb. This version of the recipe is chosen for my build because of the higher layer priority of meta-linaro/meta-optee. It contains the lines

OPTEEMACHINE ?= "${MACHINE}"
OPTEEOUTPUTMACHINE ?= "${MACHINE}"

where MACHINE ??= "h3ulcb" is set in my local.conf. This setup results in a failure of the build. Taking a look at the meta-renesas version of optee-os_git.bb and the content of the optee-source folder, I've come to the conclusion, that the correct values for OPTEEMACHINE/OPTEEOUTPUTMACHINE should be "rcar" (as can be seen in the meta-renesas version of optee-os_git.bb). I can fix my build by setting

OPTEEMACHINE = "rcar"
OPTEEOUTPUTMACHINE = "rcar"

in local.conf but still the whole matter seems rather queer to me and I wonder, whether there may be problems arising from the usage of the meta-linaro/meta-optee version of optee-os_git.bb.

Thank you in advance for any comments and help.

[ghost]

This way you are using the recipe from meta-linaro layer, you will probably miss the tee-h3ulcb.srec .bin and .elf in your DEPLOYDIR.
Refer to the doc for instance: meta-rcar-gen3/docs/sample/conf/m3ulcb/poky-gcc/bsp/local.conf renesas seems to suggest to MASK the recipes in the local.conf: BBMASK = "meta-linaro/meta-optee/recipes-security/optee"