svn commit: trunk/uClibc/libc: misc/ctype misc/wchar sysdeps/linux/comm etc...

[Bernhard Reutner-Fischer]

On Fri, Nov 21, 2008 at 12:15:19AM -0800, vda at uclibc.org wrote:
>Author: vda
>Date: 2008-11-21 00:15:19 -0800 (Fri, 21 Nov 2008)
>New Revision: 24108

>Modified: trunk/uClibc/libc/sysdeps/linux/common/mremap.c
>===================================================================
>--- trunk/uClibc/libc/sysdeps/linux/common/mremap.c	2008-11-21 05:22:12 UTC (rev 24107)
>+++ trunk/uClibc/libc/sysdeps/linux/common/mremap.c	2008-11-21 08:15:19 UTC (rev 24108)
>@@ -12,7 +12,7 @@
> 
> #ifdef __NR_mremap
> 
>-//vda:TODO: add comment and explain why??
>+/* Why do we do this?! */
> 
> #define mremap _hidemremap
> #include <sys/mman.h>

IIRC this was an attempt to bypass http://gcc.gnu.org/PR36849
but as you can see i gave up on it (it's compiled separately to please C and/or
GCC now). The rename is still needed, though. Ugly, isn't it.


And there is a second category of problems, weakrefs:
gcc.gnu.org/PR31537 has a wrong bypass in the BE (but it's a FE/ME bug
so my hack is just that)
gcc.gnu.org/PR35034

So, if somebody knows how weakrefs were supposed to work then please
take a look. The initial proposal by lxo that i found is a bit unclear
to me in this respect. Perhaps you could ask your colleague Alexandre
to have a look.