Re: [XEN PATCH 1/3] x86/emul: define pseudo keyword fallthrough

[Jan Beulich <jbeulich@xxxxxxxx>]

On 06.11.2024 10:04, Federico Serafini wrote:
> The pseudo keyword fallthrough shall be used to make explicit the
> fallthrough intention at the end of a case statement (doing this
> through comments is deprecated).
> 
> Signed-off-by: Federico Serafini <federico.serafini@xxxxxxxxxxx>
> ---
>  xen/arch/x86/x86_emulate/x86_emulate.h | 10 ++++++++++
>  1 file changed, 10 insertions(+)

When you had asked my privately on Matrix, I specifically said: "Adding
the pseudo-keyword to x86-emulate.h (not x86_emulate.h) is probably best,
unless problems with that approach turn up." Even if identical re-
definitions are deemed fine, I for one consider such bad practice. Yet
by playing with this file (and outside of any relevant #ifdef) means
there will be such a re-definition when building Xen itself.

In fact the patch subject should also already clarify that the auxiliary
definition is only needed for the test and fuzzing harnesses.

> --- a/xen/arch/x86/x86_emulate/x86_emulate.h
> +++ b/xen/arch/x86/x86_emulate/x86_emulate.h
> @@ -23,6 +23,16 @@
>  # error Unknown compilation width
>  #endif
>  
> +/*
> + * Pseudo keyword 'fallthrough' to make explicit the fallthrough intention at
> + * the end of a case statement.
> + */
> +#if (!defined(__clang__) && (__GNUC__ >= 7))

I realize xen/compiler.h has it like that, but may I ask that you omit
the meaningless outer pair of parentheses?

Jan