[PATCH 02/10] compiler.h: add is_const() as a replacement of __is_constexpr()

[David Laight]

From: Linus Torvalds
> Sent: 06 December 2024 19:15
> On Fri, 6 Dec 2024 at 11:07, David Laight <David.Laight at aculab.com> wrote:
> >
> > I'm missing the compiler version and options to generate the error.
> 
> Just -Wall with most recent gcc versions seems to do it. At least I
> can repro it with gcc-14.2.1 and something silly like this:

I may have just missed golbolt returning a warning.
...
> > Does a '+ 0' help?  "(var << 2) + 0 ? 0 : 0"
> 
> Yeah, that actually works.
> 
> And "+0" is nice in that it should work in any context.

Unless it falls foul of the clang test for arthmetic on NULL pointers.
(I'm sure that is only a problem if NULL isn't the all-zero bit pattern.
And pretty much no C code has ever been 'that portable'.)

Adding 0 can also help when compliers are being picky about enums.
Since none have (yet) made them more like a 32bit pointer to a one
byte structure (or the Pascal enum).
In case the relevant people are reading this, maybe only do any
non-integer warnings for named enums?

> > #define const_NULL(x) _Generic(0 ? (x) : (char *)0, char *: 1, void *: 0)
> > #define const_true(x) const_NULL((x) ? NULL : (void *)1L))
> > #define const_expr(x) const_NULL((x) ? NULL : NULL))
> > I send this morning.
> > Needs 's/char/struct kjkjkjkjui/' applied.
> 
> Oh Christ. You really are taking this whole ugly to another level.

I sort of liked that version in a perverse sort of way.
It does give you a simple test for NULL (unless you've used 'struct kjkjkjkjui').

	David

-
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK
Registration No: 1397386 (Wales)