Maybe change __size_t definition in <stddef.h>?

I work currently on a better FreeBSD compatibility for Newlib.  For
RTEMS we use the network, USB, SD/MMC stacks from FreeBSD.  It would be
nice to use <sys/types.h>, etc. from Newlib directly to compile FreeBSD
user and kernel space files.

Various FreeBSD source and header files need a typedef __size_t via
<sys/_types.h>.  Unfortunately the GCC provided <stddef.h> uses

    #if (defined (__FreeBSD__) && (__FreeBSD__ >= 5)) \
      || defined(__DragonFly__) \
      || defined(__FreeBSD_kernel__)
    /* __size_t is a typedef on FreeBSD 5, must not trash it. */
    #elif defined (__VMS__)
    /* __size_t is also a typedef on VMS.  */
    #else
    #define __size_t
    #endif

and therefore defines __size_t to nothing on Newlib targets which would
trash a __size_t typedef.  There exists no Newlib builtin define which
we could add here.  As a workaround I include <stddef.h> before
<sys/_types.h> in <sys/types.h> and undefine __size_t in <sys/_types.h>.

This works fine so far, but breaks if someone includes <stddef.h> after
<sys/_types.h> the first time.

This patch would address this issue, but I do not know which other
issues it generates.

gcc/ChangeLog
2016-04-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>

	* ginclude/stddef.h (__size_t): Define to __SIZE_TYPE__.
---
 gcc/ginclude/stddef.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

On Fri, Apr 15, 2016 at 11:44 AM, Sebastian Huber
<sebastian.huber@embedded-brains.de> wrote:
> I work currently on a better FreeBSD compatibility for Newlib.  For
> RTEMS we use the network, USB, SD/MMC stacks from FreeBSD.  It would be
> nice to use <sys/types.h>, etc. from Newlib directly to compile FreeBSD
> user and kernel space files.
>
> Various FreeBSD source and header files need a typedef __size_t via
> <sys/_types.h>.  Unfortunately the GCC provided <stddef.h> uses
>
>     #if (defined (__FreeBSD__) && (__FreeBSD__ >= 5)) \
>       || defined(__DragonFly__) \
>       || defined(__FreeBSD_kernel__)
>     /* __size_t is a typedef on FreeBSD 5, must not trash it. */
>     #elif defined (__VMS__)
>     /* __size_t is also a typedef on VMS.  */
>     #else
>     #define __size_t

This looks odd - maybe it was meant to be a #undef?  OTOH there are
a whole lot of other similar cases there which do not make sense to me...

Richard.

>     #endif
>
> and therefore defines __size_t to nothing on Newlib targets which would
> trash a __size_t typedef.  There exists no Newlib builtin define which
> we could add here.  As a workaround I include <stddef.h> before
> <sys/_types.h> in <sys/types.h> and undefine __size_t in <sys/_types.h>.
>
> This works fine so far, but breaks if someone includes <stddef.h> after
> <sys/_types.h> the first time.
>
> This patch would address this issue, but I do not know which other
> issues it generates.
>
> gcc/ChangeLog
> 2016-04-15  Sebastian Huber  <sebastian.huber@embedded-brains.de>
>
>         * ginclude/stddef.h (__size_t): Define to __SIZE_TYPE__.
> ---
>  gcc/ginclude/stddef.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/gcc/ginclude/stddef.h b/gcc/ginclude/stddef.h
> index d711530..8c2e22e 100644
> --- a/gcc/ginclude/stddef.h
> +++ b/gcc/ginclude/stddef.h
> @@ -207,7 +207,7 @@ typedef __PTRDIFF_TYPE__ ptrdiff_t;
>  #elif defined (__VMS__)
>  /* __size_t is also a typedef on VMS.  */
>  #else
> -#define __size_t
> +#define __size_t __SIZE_TYPE__
>  #endif
>  #ifndef __SIZE_TYPE__
>  #define __SIZE_TYPE__ long unsigned int
> --
> 1.8.4.5
>

On 15/04/16 12:23, Richard Biener wrote:
> On Fri, Apr 15, 2016 at 11:44 AM, Sebastian Huber
> <sebastian.huber@embedded-brains.de>  wrote:
>> >I work currently on a better FreeBSD compatibility for Newlib.  For
>> >RTEMS we use the network, USB, SD/MMC stacks from FreeBSD.  It would be
>> >nice to use <sys/types.h>, etc. from Newlib directly to compile FreeBSD
>> >user and kernel space files.
>> >
>> >Various FreeBSD source and header files need a typedef __size_t via
>> ><sys/_types.h>.  Unfortunately the GCC provided <stddef.h> uses
>> >
>> >     #if (defined (__FreeBSD__) && (__FreeBSD__ >= 5)) \
>> >       || defined(__DragonFly__) \
>> >       || defined(__FreeBSD_kernel__)
>> >     /* __size_t is a typedef on FreeBSD 5, must not trash it. */
>> >     #elif defined (__VMS__)
>> >     /* __size_t is also a typedef on VMS.  */
>> >     #else
>> >     #define __size_t
> This looks odd - maybe it was meant to be a #undef?  OTOH there are
> a whole lot of other similar cases there which do not make sense to me...

No, in case __size_t is defined, then we don't end up in this spot. 
stddef.h tests for various guard defines and if none of them is defined, 
then it typedefs size_t and defines all the guard defines to nothing, 
except for __size_t on some targets.