On Fri, Jan 10, 2020 at 08:42:37PM +0100, Arnd Bergmann wrote:
> On Fri, Jan 10, 2020 at 5:57 PM Will Deacon <will@xxxxxxxxxx> wrote:
>
> > @@ -128,10 +128,10 @@ do { \
> > #ifndef __smp_load_acquire
> > #define __smp_load_acquire(p) \
> > ({ \
> > - typeof(*p) ___p1 = READ_ONCE(*p); \
> > + __unqual_scalar_typeof(*p) ___p1 = READ_ONCE(*p); \
> > compiletime_assert_atomic_type(*p); \
> > __smp_mb(); \
> > - ___p1; \
> > + (typeof(*p))___p1; \
> > })
>
> Doesn't that last (typeof(*p))___p1 mean you put the potential
> 'volatile' back on the assignment after you went through the
> effort of taking it out?
Yes, but that's ok wrt codegen since the local variable isn't volatile,
and I definitely ran into issues with 'const' if I returned the unqualified
type here.
Will
