klibc - Mar 2005 - [PATCH] avoid size_t redefinition

This patch protects against redefinitions of size_t.
There are currently at least two different definitions
provided with klibc:

	unistd.h -> stddef.h -> bits32/bitsize/stddef.h
	sys/times.h -> linux/times.h -> linux/types.h

both define size_t, causing gcc to complain.

I suspect ptrdiff_t has a similar problem; not covered by
this patch.

Regards,
Erik

diff -urN klibc-0.202-pristine/include/bits32/bitsize/stddef.h
klibc-0.202/include/bits32/bitsize/stddef.h
--- klibc-0.202-pristine/include/bits32/bitsize/stddef.h	2004-07-07
02:41:04.000000000 +0200
+++ klibc-0.202/include/bits32/bitsize/stddef.h	2005-03-02 20:41:19.000000000
+0100
@@ -5,12 +5,14 @@
 #ifndef _BITSIZE_STDDEF_H
 #define _BITSIZE_STDDEF_H
 
+#ifndef _SIZE_T
 #define _SIZE_T
 #if defined(__s390__) || defined(__hppa__) || defined(__cris__)
 typedef unsigned long size_t;
 #else
 typedef unsigned int size_t;
 #endif
+#endif
 
 #define _PTRDIFF_T
 typedef signed int   ptrdiff_t;

Erik van Konijnenburg wrote:
> This patch protects against redefinitions of size_t.
> There are currently at least two different definitions
> provided with klibc:
> 
> 	unistd.h -> stddef.h -> bits32/bitsize/stddef.h
> 	sys/times.h -> linux/times.h -> linux/types.h
> 
> both define size_t, causing gcc to complain.
> 
> I suspect ptrdiff_t has a similar problem; not covered by
> this patch.
The problem is <linux/types.h> being invoked without <sys/types.h> 
having been included first.  <linux/types.h> is poison.  I'll add 
#include <sys/types.h> into sys/times.h.

	-hpa