[PATCH 0/2] Synchronize <sys/cdefs.h> with FreeBSD

[PATCH 0/2] Synchronize <sys/cdefs.h> with FreeBSD

[Sebastian Huber]

John Baldwin (1):
  <sys/cdefs.h>: Decay expression passed to fallback version of
    __generic()

Sebastian Huber (1):
  sys/cdefs.h: fix for use __restrict in C++

 newlib/libc/include/sys/cdefs.h | 21 +++++++++++----------
 1 file changed, 11 insertions(+), 10 deletions(-)

-- 
2.35.3

[PATCH 1/2] <sys/cdefs.h>: Decay expression passed to fallback version of __generic()

[Sebastian Huber]

From: John Baldwin <jhb@FreeBSD.org>

This ensures that __generic() more closely matches _Generic() when
using the fallback version when _Generic() is not available (such as
GCC).

Co-authored by:	jrtc27
Reviewed by:	jrtc27
Differential Revision:	https://reviews.freebsd.org/D38215
---
 newlib/libc/include/sys/cdefs.h | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/newlib/libc/include/sys/cdefs.h b/newlib/libc/include/sys/cdefs.h
index 720e700920..808498c50b 100644
--- a/newlib/libc/include/sys/cdefs.h
+++ b/newlib/libc/include/sys/cdefs.h
@@ -328,6 +328,9 @@
  * __generic().  Unlike _Generic(), this macro can only distinguish
  * between a single type, so it requires nested invocations to
  * distinguish multiple cases.
+ *
+ * Note that the comma operator is used to force expr to decay in
+ * order to match _Generic().
  */
 
 #if (defined(__STDC_VERSION__) && __STDC_VERSION__ >= 201112L) || \
@@ -337,7 +340,7 @@
 #elif __GNUC_PREREQ__(3, 1) && !defined(__cplusplus)
 #define	__generic(expr, t, yes, no)					\
 	__builtin_choose_expr(						\
-	    __builtin_types_compatible_p(__typeof(expr), t), yes, no)
+	    __builtin_types_compatible_p(__typeof((0, (expr))), t), yes, no)
 #endif
 
 /*
-- 
2.35.3

[PATCH 2/2] sys/cdefs.h: fix for use __restrict in C++

[Sebastian Huber]

Newlib shares large parts of <sys/cdefs.h> with FreeBSD and received
this bug report:

https://sourceware.org/pipermail/newlib/2023/020400.html

As an extension, GCC and clang offer C99-style restricted pointers in
C++ mode:
https://gcc.gnu.org/onlinedocs/gcc/Restricted-Pointers.html

We notice that this extension is broken when including newlib headers:
restricted pointers are treated as ordinary pointers.

We traced this to the following section of
newlib/libc/include/sys/cdefs.h:

  /*
   * GCC 2.95 provides `__restrict' as an extension to C90 to support the
   * C99-specific `restrict' type qualifier.  We happen to use `__restrict' as
   * a way to define the `restrict' type qualifier without disturbing older
   * software that is unaware of C99 keywords.
   */
  #if !(__GNUC__ == 2 && __GNUC_MINOR__ == 95)
  #if !defined(__STDC_VERSION__) || __STDC_VERSION__ < 199901
  #define	__restrict
  #else
  #define	__restrict	restrict
  #endif
  #endif

While the GCC __restrict extension was indeed introduced in GCC 2.95, it
is not limited to this version; the extension is also not limited to
C90:
https://gcc.gnu.org/gcc-2.95/c++features.html

Rewrite the logic in the header so that __restrict is kept alone when
available.

PR:	272723
MFC after:	1 week
---
 newlib/libc/include/sys/cdefs.h | 16 +++++++---------
 1 file changed, 7 insertions(+), 9 deletions(-)

diff --git a/newlib/libc/include/sys/cdefs.h b/newlib/libc/include/sys/cdefs.h
index 808498c50b..cc1a8a1ccb 100644
--- a/newlib/libc/include/sys/cdefs.h
+++ b/newlib/libc/include/sys/cdefs.h
@@ -412,17 +412,15 @@
 #endif
 
 /*
- * GCC 2.95 provides `__restrict' as an extension to C90 to support the
- * C99-specific `restrict' type qualifier.  We happen to use `__restrict' as
- * a way to define the `restrict' type qualifier without disturbing older
- * software that is unaware of C99 keywords.
+ * We use `__restrict' as a way to define the `restrict' type qualifier
+ * without disturbing older software that is unaware of C99 keywords.
+ * GCC also provides `__restrict' as an extension to support C99-style
+ * restricted pointers in other language modes.
  */
-#if !(__GNUC__ == 2 && __GNUC_MINOR__ == 95)
-#if !defined(__STDC_VERSION__) || __STDC_VERSION__ < 199901
-#define	__restrict
-#else
+#if defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901
 #define	__restrict	restrict
-#endif
+#elif !__GNUC_PREREQ__(2, 95)
+#define	__restrict
 #endif
 
 /*
-- 
2.35.3

Re: [PATCH 0/2] Synchronize <sys/cdefs.h> with FreeBSD

[Corinna Vinschen]

On Aug  7 11:28, Sebastian Huber wrote:
> John Baldwin (1):
>   <sys/cdefs.h>: Decay expression passed to fallback version of
>     __generic()
> 
> Sebastian Huber (1):
>   sys/cdefs.h: fix for use __restrict in C++
> 
>  newlib/libc/include/sys/cdefs.h | 21 +++++++++++----------
>  1 file changed, 11 insertions(+), 10 deletions(-)
> 
> -- 
> 2.35.3

Okidoki!


Thanks,
Corinna

Re: [PATCH 2/2] sys/cdefs.h: fix for use __restrict in C++

[Corinna Vinschen]

Hi Sebastian,

On Aug  7 11:28, Sebastian Huber wrote:
> diff --git a/newlib/libc/include/sys/cdefs.h b/newlib/libc/include/sys/cdefs.h
> index 808498c50b..cc1a8a1ccb 100644
> --- a/newlib/libc/include/sys/cdefs.h
> +++ b/newlib/libc/include/sys/cdefs.h
> @@ -412,17 +412,15 @@
>  #endif
>  
>  /*
> - * GCC 2.95 provides `__restrict' as an extension to C90 to support the
> - * C99-specific `restrict' type qualifier.  We happen to use `__restrict' as
> - * a way to define the `restrict' type qualifier without disturbing older
> - * software that is unaware of C99 keywords.
> + * We use `__restrict' as a way to define the `restrict' type qualifier
> + * without disturbing older software that is unaware of C99 keywords.
> + * GCC also provides `__restrict' as an extension to support C99-style
> + * restricted pointers in other language modes.
>   */
> -#if !(__GNUC__ == 2 && __GNUC_MINOR__ == 95)
> -#if !defined(__STDC_VERSION__) || __STDC_VERSION__ < 199901
> -#define	__restrict
> -#else
> +#if defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901
>  #define	__restrict	restrict
> -#endif
> +#elif !__GNUC_PREREQ__(2, 95)
> +#define	__restrict
>  #endif
>  
>  /*

Turns out, this leads to a build failuer in Cygwin.

We have a definition in aio.h:

  int lio_listio (int, struct aiocb *__restrict const [__restrict], int,
		  struct sigevent *__restrict);

and a matching one in aio.cc:

  int 
  lio_listio (int mode, struct aiocb *__restrict const aiolist[__restrict],
              int nent, struct sigevent *__restrict sig)

The problem is the bracket expression.  The restrict keyword is allowed
in C90, but not in C++.

GLibc has a special definition __restrict_arr in cdefs.h, which is used
in the brackets of the above definition:

  /* ISO C99 also allows to declare arrays as non-overlapping.  The syntax is
       array_name[restrict]
     GCC 3.1 and clang support this.
     This syntax is not usable in C++ mode.  */
  #if (__GNUC_PREREQ (3,1) || __clang_major__ >= 3) && !defined __cplusplus
  # define __restrict_arr __restrict
  #else
  # ifdef __GNUC__
  #  define __restrict_arr        /* Not supported in old GCC.  */
  # else
  #  if defined __STDC_VERSION__ && __STDC_VERSION__ >= 199901L
  #   define __restrict_arr       restrict
  #  else 
  /* Some other non-C99 compiler.  */
  #   define __restrict_arr       /* Not supported.  */
  #  endif
  # endif
  #endif  

This doesn't exist in FreeBSD, though.

Do you think it's ok to add it to our cdefs.h, nevertheless?


Thanks,
Corinna

Re: [PATCH 2/2] sys/cdefs.h: fix for use __restrict in C++

[Sebastian Huber]

On 07.08.23 15:15, Corinna Vinschen wrote:
> This doesn't exist in FreeBSD, though. Do you think it's ok to add it to 
> our cdefs.h, nevertheless?

It is good to have glibc and FreeBSD compatibility from my point of 
view, so we should also add this __restrict_arr.

-- 
embedded brains GmbH
Herr Sebastian HUBER
Dornierstr. 4
82178 Puchheim
Germany
email: sebastian.huber@embedded-brains.de
phone: +49-89-18 94 741 - 16
fax:   +49-89-18 94 741 - 08

Registergericht: Amtsgericht München
Registernummer: HRB 157899
Vertretungsberechtigte Geschäftsführer: Peter Rasmussen, Thomas Dörfler
Unsere Datenschutzerklärung finden Sie hier:
https://embedded-brains.de/datenschutzerklaerung/

Re: [PATCH 2/2] sys/cdefs.h: fix for use __restrict in C++

[Corinna Vinschen]

On Aug  7 15:35, Sebastian Huber wrote:
> On 07.08.23 15:15, Corinna Vinschen wrote:
> > This doesn't exist in FreeBSD, though. Do you think it's ok to add it to
> > our cdefs.h, nevertheless?
> 
> It is good to have glibc and FreeBSD compatibility from my point of view, so
> we should also add this __restrict_arr.

Thanks, I hope my just pused version is ok.


Corinna