cpu/stm32: disable MPU for cortex-m0+ (stm32g0 and stm32l052t8)

[aabadie]

Contribution description

This PR remove the MPU feature from the list of features provided by the stm32g0 family and stm32l052t8 model since MPU is broken on cortex-m0+ and this results in nearly usable state for stm32g0 family and the related stm32l0 model.

Testing procedure

All applications are now working on nucleo-g070rb and i-nucleo-lrwan1

Issues/PRs references

This was reported in #14822 (Let's keep it open until a real fix is proposed and the feature can be re-enabled)

[bergzand]

Ack, similar to #14806

[bergzand]

fixes #14822

[aabadie]

fixes #14822

can we keep it open until this is really fixed ?

[bergzand]

can we keep it open until this is really fixed ?

Only if you change it to "MPU doesn't work on Cortex-m0+" :)

[aabadie]

Only if you change it to "MPU doesn't work on Cortex-m0+" :)

Done

[benpicco]

Only if you change it to "MPU doesn't work on Cortex-m0+" :)

We have a lot more Cortex-m0+ than sam32g0 - should we disable the MPU there too?

[bergzand]

We have a lot more Cortex-m0+ than stm32g0 - should we disable the MPU there too?

I think so yes, they are probably as broken as the stm32g0

[aabadie]

There's only one stm32l0 cpu model left for cortex-m0+. There's also saml1x but it's cortex-m23 and I don't know the state for those ones.
I can extend this PR for the stm32l0, if you want.

[bergzand]

I can extend this PR for the stm32l0, if you want.

Fine with me!

Only stm32l0 cpu model for cortex-m0+. There's also saml1x but it's cortex-m23 and I don't know the state for those ones.

The saml1x is already disabled with #14806. (and yeah, cortex-m23 is also broken.)

[aabadie]

I tried i-nucleo-lrwan1 (the stm32l0) and it's broken the same.

[aabadie]

Wait, this time I forgot to update Kconfig ;)