[PATCH 2/3] arm64: Optimize __READ_ONCE() with CONFIG_LTO=y

[Marco Elver]

On Mon, 26 Jan 2026 at 23:24, Arnd Bergmann <arnd at arndb.de> wrote:
>
> On Mon, Jan 26, 2026, at 20:54, Marco Elver wrote:
> > On Mon, Jan 26, 2026 at 08:56AM +0100, Arnd Bergmann wrote:
> >> On Mon, Jan 26, 2026, at 01:25, Marco Elver wrote:
> >>
> >> How does this work with CC_HAS_TYPEOF_UNQUAL=false?
> >>
> >> As far as I can tell, TYPEOF_UNQUAL() falls back to __typeof__
> >> on gcc-13, clang-18 and earlier, and not strip out qualifiers.
> >
> > I think we only need to worry about Clang for LTO builds. But yeah, our
> > minimum supported Clang is 15, so between 15-18 it'd be broken.
>
> Right, I missed the #ifdef CONFIG_LTO check, so indeed gcc is
> fine here.
>
> >> With fd69b2f7d5f4 ("compiler: Use __typeof_unqual__() for
> >> __unqual_scalar_typeof()"), I would expect __unqual_scalar_typeof()
> >> to do the right thing already.
> >
> > It'd still be broken for Clang 15-18, so it won't help much. We need
> > this to work for more than "scalar", so even though it'll work for Clang
> > 19+ given the redefinition to __typeof_unqual__, we should deprecate the
> > _Generic-based __unqual_scalar_typeof() sooner than later.
> >
> > I was able to make this work for older compilers:
> >
> ...
> >  #define __READ_ONCE(x)                                                       \
> >  ({                                                                   \
> >       auto __x = &(x);                                                \
> > -     auto __ret = (TYPEOF_UNQUAL(*__x) *)__x, *__retp = &__ret;      \
> > -     union { TYPEOF_UNQUAL(*__x) __val; char __c[1]; } __u;          \
> > +     auto __ret = (__read_once_typeof(*__x) *)__x, *__retp = &__ret; \
> > +     union { __read_once_typeof(*__x) __val; char __c[1]; } __u;     \
> >       *__retp = &__u.__val;                                           \
>
> >
> > Thoughts?
>
> Looks better than __unqual_scalar_typeof() to me. Would it make
> sense to do the same __read_once_typeof() in the asm-generic
> version of __READ_ONCE()? I don't remember if we discussed it
> in the thread leading up to dee081bf8f82 ("READ_ONCE: Drop
> pointer qualifiers when reading from scalar types").
> We probably didn't have __auto_type back then.

I don't see the point for the asm-generic __READ_ONCE(): it's not
wrong to cast to 'const volatile volatile T*' nor 'const volatile
const T*' etc., which is dereferenced directly and not stored in any
temporary variable when used in READ_ONCE(). I actually don't know why
__unqual_scalar_typeof() is used in the asm-generic __READ_ONCE(),
which just adds 'const volatile' right back.

And __READ_ONCE_SCALAR() appears to be gone, where the
qualifier-stripping resulted in better code-gen. Then there's alpha,
but I think nobody cares sufficiently about it.
Did I miss some variant?

For this arm64 LTO variant we do need the qualifier stripping because
of the temporary __val.

And I'm hoping that once we update minimum compiler versions, we can
delete all these typeof-hacks and just use typeof_unqual().