Re: [PATCH v3 3/8] KVM: selftests: Move setting a vCPU's entry point to a dedicated API

[Oliver Upton <oliver.upton@linux.dev>]

+cc Raghavendra

Hey,

On Wed, Feb 28, 2024 at 01:19:48PM -0800, Sean Christopherson wrote:
> but due to a different issue that is fixed in the kvm-arm tree[*], but not in mine,
> I built without -Werror and didn't see the new warn in the sea of GUEST_PRINTF
> warnings.
> 
> Ugh, and I still can't enable -Werror, because there are unused functions in
> aarch64/vpmu_counter_access.c
> 
>   aarch64/vpmu_counter_access.c:96:20: error: unused function 'enable_counter' [-Werror,-Wunused-function]
>   static inline void enable_counter(int idx)
>                    ^
>   aarch64/vpmu_counter_access.c:104:20: error: unused function 'disable_counter' [-Werror,-Wunused-function]
>   static inline void disable_counter(int idx)
>                    ^
>   2 errors generated.
>   make: *** [Makefile:278: /usr/local/google/home/seanjc/go/src/kernel.org/nox/tools/testing/selftests/kvm/aarch64/vpmu_counter_access.o] Error 1
>   make: *** Waiting for unfinished jobs....
> 
>   Commit 49f31cff9c533d264659356b90445023b04e10fb failed to build with 'make-clang make-arm make -j128'.
> 
> Oliver/Marc, any thoughts on how you want to fix the unused function warnings?
> As evidenced by this goof, being able to compile with -Werror is super helpful.

Are these the only remaining warnings we have in the arm64 selftests
build?

Faster than me paging this test back in: Raghu, are we missing any test
cases upstream that these helpers were intended for? If no, mind sending
a patch to get rid of them?

> And another question: is there any reason to not force -Werror for selftests?

Nothing comes to mind. We need to bite the bullet and make the switch.
There might be breakage, but we can certainly handle that.

-- 
Thanks,
Oliver