Define C23 header version macros
Adhemerval Zanella Netto
adhemerval.zanella@linaro.org
Fri Nov 21 14:14:58 GMT 2025
On 20/11/25 18:48, Joseph Myers wrote:
> C23 defines library macros __STDC_VERSION_<header>_H__ to indicate
> that a header has support for new / changed features from C23. Now
> that all the required library features are implemented in glibc,
> define these macros. I'm not sure this is sufficiently much of a
> user-visible feature to be worth a mention in NEWS.
>
> Tested for x86_64.
>
> There are various optional C23 features we don't yet have, of which I
> might look at the Annex H ones (floating-point encoding conversion
> functions and _Float16 functions) next.
>
> * Optional time bases TIME_MONOTONIC, TIME_ACTIVE, TIME_THREAD_ACTIVE.
> See
> <https://sourceware.org/pipermail/libc-alpha/2023-June/149264.html>
> - we need to review / update that patch. (I think patch 2/2,
> inventing new names for all the nonstandard CLOCK_* supported by the
> Linux kernel, is rather more dubious.)
I will take a look at this. I had it on my backlog and it seems simple
enough for next release.
>
> * Updating conform/ tests for C23.
This a more comprehensible but I think feasible work for next release.
I will check if I have bandwidth.
>
> * Defining the rounding mode macro FE_TONEARESTFROMZERO for RISC-V (as
> far as I know, the only architecture supported by glibc that has
> hardware support for this rounding mode for binary floating point)
> and supporting it throughout glibc and its tests (especially the
> string/numeric conversions in both directions that explicitly handle
> each possible rounding mode, and various tests that do likewise).
I think this would require not only adjustments in tests, but also on
some printf/strtox/etc to handle the rounding mode. This is a more
extensive change and most likely we will need to adjust for further
releases.
>
> * Annex H floating-point encoding conversion functions. (It's not
> entirely clear which are optional even given support for Annex H;
> there's some wording applied inconsistently about only being
> required when non-arithmetic interchange formats are supported; see
> the comments I raised on the WG14 reflector on 23 Oct 2025.)
I don't have access to the WG14 reflector, what are the comments you
have raised? And what kind of additional work do you think we still
need to add on glibc?
>
> * _Float16 functions (and other header and testcase support for this
> type).
>
The _Float16 supported generated some interest from my GNU Cauldron
presentation. I think this can be a possible project to futures
releases.
> * Decimal floating-point support.
>
This comes from time to time, but I don't see much traction for enablement
and the reaction of my GNU Cauldron when I proposed it as future subproject
was also tepid.
I am also not sure about libdfp status, and if we can readily integrate
its implementation on glibc.
It also a *lot* of work, as you already laid it out in a libc-alpha
thread.
> * Fully supporting __int128 and unsigned __int128 as integer types
> wider than intmax_t, as permitted by C23. Would need doing in
> coordination with GCC, see GCC bug 113887 for more discussion of
> what's involved.
>
> diff --git a/assert/assert.h b/assert/assert.h
> index 839406ddf7..373a1142d8 100644
> --- a/assert/assert.h
> +++ b/assert/assert.h
> @@ -34,6 +34,12 @@
> #define _ASSERT_H 1
> #include <features.h>
>
> +#if __GLIBC_USE (ISOC23)
> +# ifndef __STDC_VERSION_ASSERT_H__
> +# define __STDC_VERSION_ASSERT_H__ 202311L
> +# endif
> +#endif
> +
> #if defined __cplusplus && __GNUC_PREREQ (2,95)
> # define __ASSERT_VOID_CAST static_cast<void>
> #else
> diff --git a/include/limits.h b/include/limits.h
> index 6370ccd84d..baf79df609 100644
> --- a/include/limits.h
> +++ b/include/limits.h
> @@ -180,7 +180,7 @@
>
> /* The macros for _Bool are not defined by GCC's <limits.h> before GCC
> 11, or if _GNU_SOURCE is defined rather than enabling C23 support
> - with -std. */
> + with -std; likewise for the version macro before GCC 13. */
> #if __GLIBC_USE (ISOC23)
> # ifndef BOOL_MAX
> # define BOOL_MAX 1
> @@ -188,6 +188,9 @@
> # ifndef BOOL_WIDTH
> # define BOOL_WIDTH 1
> # endif
> +# ifndef __STDC_VERSION_LIMITS_H__
> +# define __STDC_VERSION_LIMITS_H__ 202311L
> +# endif
> #endif
>
> #ifdef __USE_POSIX
> diff --git a/libio/stdio.h b/libio/stdio.h
> index e0e70945fa..d5a44edfa6 100644
> --- a/libio/stdio.h
> +++ b/libio/stdio.h
> @@ -29,6 +29,10 @@
>
> __BEGIN_DECLS
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_STDIO_H__ 202311L
> +#endif
> +
> #define __need_size_t
> #define __need_NULL
> #include <stddef.h>
> diff --git a/math/complex.h b/math/complex.h
> index c925c5c0a3..2c84953429 100644
> --- a/math/complex.h
> +++ b/math/complex.h
> @@ -33,6 +33,10 @@
>
> __BEGIN_DECLS
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_COMPLEX_H__ 202311L
> +#endif
> +
> /* We might need to add support for more compilers here. But since ISO
> C99 is out hopefully all maintained compilers will soon provide the data
> types `float complex' and `double complex'. */
> diff --git a/math/fenv.h b/math/fenv.h
> index 07bbbf61df..8db719f687 100644
> --- a/math/fenv.h
> +++ b/math/fenv.h
> @@ -65,6 +65,10 @@
>
> __BEGIN_DECLS
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_FENV_H__ 202311L
> +#endif
> +
> /* Floating-point exception handling. */
>
> /* Clear the supported exceptions represented by EXCEPTS. */
> diff --git a/math/math.h b/math/math.h
> index 5f1933a2f9..26cea186fc 100644
> --- a/math/math.h
> +++ b/math/math.h
> @@ -39,6 +39,10 @@ __BEGIN_DECLS
> /* Gather machine dependent type support. */
> #include <bits/floatn.h>
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_MATH_H__ 202311L
> +#endif
> +
> /* Value returned on overflow. With IEEE 754 floating point, this is
> +Infinity, otherwise the largest representable positive value. */
> #if __GNUC_PREREQ (3, 3)
> diff --git a/math/tgmath.h b/math/tgmath.h
> index 27a87c599d..89ee94e1f7 100644
> --- a/math/tgmath.h
> +++ b/math/tgmath.h
> @@ -30,6 +30,10 @@
> #include <math.h>
> #include <complex.h>
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_TGMATH_H__ 202311L
> +#endif
> +
>
> /* There are two variant implementations of type-generic macros in
> this file: one for GCC 8 and later, using __builtin_tgmath and
> diff --git a/setjmp/setjmp.h b/setjmp/setjmp.h
> index 7a4760e291..f232ddcd6b 100644
> --- a/setjmp/setjmp.h
> +++ b/setjmp/setjmp.h
> @@ -26,6 +26,10 @@
>
> __BEGIN_DECLS
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_SETJMP_H__ 202311L
> +#endif
> +
> #include <bits/setjmp.h> /* Get `__jmp_buf'. */
> #include <bits/types/struct___jmp_buf_tag.h>
>
> diff --git a/stdlib/inttypes.h b/stdlib/inttypes.h
> index 6b64917b54..13cb7fae36 100644
> --- a/stdlib/inttypes.h
> +++ b/stdlib/inttypes.h
> @@ -23,6 +23,11 @@
> #define _INTTYPES_H 1
>
> #include <features.h>
> +
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_INTTYPES_H__ 202311L
> +#endif
> +
> /* Get the type definitions. */
> #include <stdint.h>
>
> diff --git a/stdlib/stdint.h b/stdlib/stdint.h
> index 302230914c..814718a366 100644
> --- a/stdlib/stdint.h
> +++ b/stdlib/stdint.h
> @@ -28,6 +28,10 @@
> #include <bits/wchar.h>
> #include <bits/wordsize.h>
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_STDINT_H__ 202311L
> +#endif
> +
> /* Exact integral types. */
>
> /* Signed. */
> diff --git a/stdlib/stdlib.h b/stdlib/stdlib.h
> index 4a79c612de..a1ea003658 100644
> --- a/stdlib/stdlib.h
> +++ b/stdlib/stdlib.h
> @@ -35,6 +35,10 @@ __BEGIN_DECLS
>
> #define _STDLIB_H 1
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_STDLIB_H__ 202311L
> +#endif
> +
> #if (defined __USE_XOPEN || defined __USE_XOPEN2K8) && !defined _SYS_WAIT_H
> /* XPG requires a few symbols from <sys/wait.h> being defined. */
> # include <bits/waitflags.h>
> diff --git a/string/string.h b/string/string.h
> index 1dc5995c17..b8747506db 100644
> --- a/string/string.h
> +++ b/string/string.h
> @@ -27,6 +27,10 @@
>
> __BEGIN_DECLS
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_STRING_H__ 202311L
> +#endif
> +
> /* Get size_t and NULL from <stddef.h>. */
> #define __need_size_t
> #define __need_NULL
> diff --git a/time/time.h b/time/time.h
> index 4913508d86..ca4f2828de 100644
> --- a/time/time.h
> +++ b/time/time.h
> @@ -28,6 +28,10 @@
> #define __need_NULL
> #include <stddef.h>
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_TIME_H__ 202311L
> +#endif
> +
> /* This defines CLOCKS_PER_SEC, which is the number of processor clock
> ticks per second, and possibly a number of other constants. */
> #include <bits/time.h>
> diff --git a/wcsmbs/uchar.h b/wcsmbs/uchar.h
> index 335bfa343d..c774b1cd34 100644
> --- a/wcsmbs/uchar.h
> +++ b/wcsmbs/uchar.h
> @@ -31,6 +31,10 @@
> #include <bits/types.h>
> #include <bits/types/mbstate_t.h>
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_UCHAR_H__ 202311L
> +#endif
> +
> /* Declare the C23 char8_t typedef in C23 modes, but only if the C++
> __cpp_char8_t feature test macro is not defined. */
> #if __GLIBC_USE (ISOC23) && !defined __cpp_char8_t
> diff --git a/wcsmbs/wchar.h b/wcsmbs/wchar.h
> index 19f71f90cf..e7170d2173 100644
> --- a/wcsmbs/wchar.h
> +++ b/wcsmbs/wchar.h
> @@ -60,6 +60,10 @@ typedef __gnuc_va_list va_list;
> # include <bits/types/locale_t.h>
> #endif
>
> +#if __GLIBC_USE (ISOC23)
> +# define __STDC_VERSION_WCHAR_H__ 202311L
> +#endif
> +
> /* Tell the caller that we provide correct C++ prototypes. */
> #if defined __cplusplus && __GNUC_PREREQ (4, 4)
> # define __CORRECT_ISO_CPP_WCHAR_H_PROTO
>
More information about the Libc-alpha
mailing list