On Fri, Dec 13, 2019 at 01:56:18PM +0100, Peter Zijlstra wrote:
>
> Excellent! I had to change it to something like:
>
> #define unqual_typeof(x) typeof(({_Atomic typeof(x) ___x __maybe_unused;
> ___x; }))
>
> but that does indeed work!
>
> Now I suppose we should wrap that in a symbol that indicates our
> compiler does indeed support _Atomic, otherwise things will come apart.
>
> That is, my gcc-4.6 doesn't seem to have it, while gcc-4.8 does, which
> is exactly the range that needs the daft READ_ONCE() construct, how
> convenient :/
>
> Something a little like this perhaps?
Yes, this looks good to me.
Just a small nit here below.
> ---
>
> diff --git a/arch/arm64/include/asm/barrier.h
> b/arch/arm64/include/asm/barrier.h
> index 7d9cc5ec4971..c389af602da8 100644
> --- a/arch/arm64/include/asm/barrier.h
> +++ b/arch/arm64/include/asm/barrier.h
> @@ -75,9 +75,9 @@ static inline unsigned long
> array_index_mask_nospec(unsigned long idx,
>
> #define __smp_store_release(p, v) \
> do { \
> - typeof(p) __p = (p); \
> - union { typeof(*p) __val; char __c[1]; } __u = \
> - { .__val = (__force typeof(*p)) (v) }; \
> + unqual_typeof(p) __p = (p); \
> + union { unqual_typeof(*p) __val; char __c[1]; } __u = \
> + { .__val = (__force unqual_typeof(*p)) (v) }; \
The 2 two trailing backslashes are now off by one tab.
-- Luc
