On Sun, Jan 26, 2025 at 11:00 PM Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> On Sun, 26 Jan 2025 at 11:47, Uros Bizjak <ubizjak@gmail.com> wrote:
> >
> > >   #if __clang_major__ >= 19
> > >   # define CC_HAS_TYPEOF_UNQUAL 1
> > >   #endif
> >
> > It is available in gcc-14.
>
> Ok, let's take that approach, instead of messing with CONFIG_CC_HAS_xyz.

Please find attached an incremental patch that implements the proposed approach.

The detection is put in include/linux/compiler.h where we can
consolidate checks for both compilers:

--cut here--
/*
 * Declare compiler support for __typeof_unqual__() operator.
 *
 * bindgen uses LLVM even if our C compiler is gcc, so we cannot
 * rely on the auto-detected CONFIG_CC_HAS_TYPEOF_UNQUAL.
 *
 * XXX: Remove test for __CHECKER__ once
 * sparse learns about __typeof_unqual__.
 */
#if ((defined(__GNUC__) && __GNUC__ >= 14) ||        \
     (defined(__clang__) && __clang_major__ >= 19)) &&    \
    !defined(__CHECKER__)
# define CC_HAS_TYPEOF_UNQUAL 1
#endif

/*
 * Define TYPEOF_UNQUAL() to use __typeof_unqual__() as typeof
 * operator when available, to return an unqualified type of the exp.
 */
#if defined(CC_HAS_TYPEOF_UNQUAL)
# define TYPEOF_UNQUAL(exp) __typeof_unqual__(exp)
#else
# define TYPEOF_UNQUAL(exp) __typeof__(exp)
#endif
--cut here--

The above also consolidates checks for __CHECKER__, resulting in a
much simpler compile check in arch/x86/include/percpu.h.

I have tested the new patch series with a bunch of compilers, works as
expected, but it is getting a bit late here. I'll resend the series
tomorrow, after some more testing.

Uros.