http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53238
--- Comment #16 from Daniel Richard G. <skunk at iskunk dot org> 2012-07-19
21:10:28 UTC ---
Okay, spoke too soon; I encountered the same error later in the game, when
building the "pthread" variant of libstdc++-v3.
I'm going to try the build with a new patch, revisiting the code snippet in
comment #2:
--- libstdc++-v3/acinclude.m4 (revision 189681)
+++ libstdc++-v3/acinclude.m4 (working copy)
@@ -3282,13 +3282,18 @@
AC_MSG_CHECKING([whether it can be safely assumed that mutex_timedlock is
available])
- AC_TRY_COMPILE([#include <unistd.h>],
+ AC_TRY_COMPILE(
+ [
+ #include <stdlib.h>
+ #include <unistd.h>
+ ],
[
// In case of POSIX threads check _POSIX_TIMEOUTS.
#if (defined(_PTHREADS) \
&& (!defined(_POSIX_TIMEOUTS) || _POSIX_TIMEOUTS <= 0))
#error
#endif
+ return pthread_mutex_timedlock(NULL, NULL)
], [ac_gthread_use_mutex_timedlock=1], [ac_gthread_use_mutex_timedlock=0])
AC_DEFINE_UNQUOTED(_GTHREAD_USE_MUTEX_TIMEDLOCK,
$ac_gthread_use_mutex_timedlock,
Since the code is being compiled as C++, linking is not necessary to get an
error in case the function is missing. Will this work any better as a solution
than my suggestion in comment #4? (Might an "#ifndef WIN32" in there be
helpful?)
