For the shortlog, it's not the "static call" that's incorrectly NULL, it's the
kvm_x86_ops.func that's wrong. E.g. yank out static calls and KVM could still
keep the WARN.
On Mon, Feb 14, 2022, Paolo Bonzini wrote:
> Use the newly corrected KVM_X86_OP annotations to warn about possible
> NULL pointer dereferences as soon as the vendor module is loaded.
>
> Signed-off-by: Paolo Bonzini <pbonzini@xxxxxxxxxx>
> ---
> arch/x86/include/asm/kvm_host.h | 5 +++--
> 1 file changed, 3 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h
> index e3f7d958c150..5dce6fbd9ab6 100644
> --- a/arch/x86/include/asm/kvm_host.h
> +++ b/arch/x86/include/asm/kvm_host.h
> @@ -1546,9 +1546,10 @@ extern struct kvm_x86_ops kvm_x86_ops;
>
> static inline void kvm_ops_static_call_update(void)
> {
> -#define KVM_X86_OP(func) \
> +#define KVM_X86_OP_OPTIONAL(func) \
> static_call_update(kvm_x86_##func, kvm_x86_ops.func);
> -#define KVM_X86_OP_OPTIONAL KVM_X86_OP
> +#define KVM_X86_OP(func) \
> + WARN_ON(!kvm_x86_ops.func); KVM_X86_OP_OPTIONAL(func)
As before, I'd prefer that we not have a KVM_X86_OP => KVM_X86_OP_OPTIONAL
dependency and instead have a common __KVM_X86_OP for this one case.
> #include <asm/kvm-x86-ops.h>
> }
>
> --
> 2.31.1
>
>
