Le 05/02/2012 16:44, Andrew Benton a écrit :
> On Sun, 05 Feb 2012 12:46:24 +0100
> Pierre Labastie<[email protected]> wrote:
>
>>> Maybe try (supposing the build tree has not been removed):
>>>
>>> echo '#include<cstdio>' |/mnt/lfs/sources/gcc-build/gcc/xgcc -shared-libgcc
>>> \
>>> -B/mnt/lfs/sources/gcc-build/gcc -nostdinc++ \
>>> -B/tools/x86_64-unknown-linux-gnu/bin/ \
>>> -B/tools/x86_64-unknown-linux-gnu/lib/ \
>>> -isystem /tools/x86_64-unknown-linux-gnu/include \
>>> -isystem /tools/x86_64-unknown-linux-gnu/sys-include \
>>> -I/mnt/lfs/sources/gcc-4.6.2/gcc \
>>> -I/mnt/lfs/sources/gcc-build/x86_64-unknown-linux-gnu/libstdc++-v3/include/x86_64-unknown-linux-gnu
>>> \
>>> -I/mnt/lfs/sources/gcc-build/x86_64-unknown-linux-gnu/libstdc++-v3/include \
>>> -I/mnt/lfs/sources/gcc-4.6.2/libstdc++-v3/libsupc++ \
>>> -D_GNU_SOURCE \
>>> -E -Wp,-dU -
>>>
>>> Regards
>>> pierre
>>>
>>>
>> Sorry, change the last line to:
>>
>> -E -Wp,-dU -xc++-header -
> To save spamming the list I've put that up here:
> http://www.linuxfromscratch.org/~andy/dump-1.txt
>
> Andy
I run the same command after building gcc-pass2 as of current LFS,
and diffed the output respective to Andy's.
There are quite a few differences in line numbers, so I do not put
the resulting diff. Relevant parts (+ is Andy's, - is mine):
----------------------------------------------------
@@ -60,6 +72,7 @@
#undef _POSIX_C_SOURCE
#undef _POSIX_SOURCE
#undef _BSD_SOURCE
+#define _ISOC11_SOURCE 1
#define _SVID_SOURCE 1
#define _BSD_SOURCE 1
#define _XOPEN_SOURCE 700
@@ -67,7 +80,6 @@
#define _POSIX_SOURCE 1
#define _ISOC99_SOURCE 1
#undef __STDC_VERSION__
-#undef _ISOC9X_SOURCE
#undef __STRICT_ANSI__
#define _LARGEFILE_SOURCE 1
#define _LARGEFILE64_SOURCE 1
@@ -82,28 +94,29 @@
@@ -567,17 +579,16 @@
#undef __GNUC_VA_LIST
# 40 "/mnt/lfs/sources/gcc-build/gcc/include/stdarg.h" 3 4
typedef __builtin_va_list __gnuc_va_list;
-# 54 "/tools/include/libio.h" 2 3
+# 55 "/tools/include/libio.h" 2 3
#define __GNUC_VA_LIST
#define __P(args) args
-#undef _PARAMS
#define _G_HAVE_PRINTF_FP 1
#undef EOF
#define NULL __null
@@ -690,7 +701,7 @@
-typedef __ssize_t __io_write_fn (void *__cookie, __const char *__buf,
+typedef __ssize_t __io_write_fn (void *__cookie, const char *__buf,
size_t __n);
(there are a lot of other declarations differing by __const -> const
Do not think it is usefull to put them all)
@@ -1141,18 +1152,13 @@
extern char *fgets (char *__restrict __s, int __n, FILE *__restrict
__stream)
;
+#define __USE_ISOC11 1
+# 646 "/tools/include/stdio.h" 3
-
-
-
-
-
-extern char *gets (char *__s) ;
-
-# 642 "/tools/include/stdio.h" 3
+# 655 "/tools/include/stdio.h" 3
extern char *fgets_unlocked (char *__restrict __s, int __n,
FILE *__restrict __stream) ;
-# 658 "/tools/include/stdio.h" 3
---------------------------------------------------------
Glearly, gets is not declared, because __USE_ISOC11 1 is defined. According
to the patch mentionned earlier in the thread, there should also be a
check of whether __cplusplus
is defined and if it is less than 201103L (actually it is defined and
the value is 1,
which makes the comparison look like nonsense, but 1 <=201103L is true).
Maybe be look at /tools/include/stdio.h around the 630th line and see if
the code
#if !defined __USE_ISOC11 \
|| (defined __cplusplus && __cplusplus <= 201103L)
is there...
If not, your glibc has not been patched. If yes, it looks like there is
something wrong in the preprocessor of gcc.
Maybe Bryan has a better idea...
Regards
Pierre
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
