From b45d9518aebfcd64e8359b9dce99067b00a33841 Mon Sep 17 00:00:00 2001 From: Ken Raeburn Date: Tue, 16 May 2006 02:56:59 +0000 Subject: Don't look for pthread_mutexattr_setrobust_np Apparently Red Hat's Fedora Core 5 defines it but doesn't declare it, so we'd have to declare it before testing the address. While it was once useful for checking whether the pthread code had been loaded, I think the other tests done now are more effective and this isn't needed any more. git-svn-id: svn://anonsvn.mit.edu/krb5/trunk@18012 dc483132-0cff-0310-8789-dd5450dbe970 --- src/include/k5-thread.h | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'src/include') diff --git a/src/include/k5-thread.h b/src/include/k5-thread.h index 5373f836a..13c2391fb 100644 --- a/src/include/k5-thread.h +++ b/src/include/k5-thread.h @@ -388,9 +388,13 @@ typedef k5_os_nothread_mutex k5_os_mutex; Linux: Stub mutex routines exist, but pthread_once does not. - Solaris: In libc there's a pthread_once that doesn't seem - to do anything. Bleah. But pthread_mutexattr_setrobust_np - is defined only in libpthread. + Solaris: In libc there's a pthread_once that doesn't seem to do + anything. Bleah. But pthread_mutexattr_setrobust_np is defined + only in libpthread. However, some version of GNU libc (Red Hat's + Fedora Core 5, reportedly) seems to have that function, but no + declaration, so we'd have to declare it in order to test for its + address. We now have tests to see if pthread_once actually works, + so stick with that for now. IRIX 6.5 stub pthread support in libc is really annoying. The pthread_mutex_lock function returns ENOSYS for a program not linked @@ -414,9 +418,6 @@ typedef k5_os_nothread_mutex k5_os_mutex; # pragma weak pthread_mutex_init # pragma weak pthread_self # pragma weak pthread_equal -# ifdef HAVE_PTHREAD_MUTEXATTR_SETROBUST_NP_IN_THREAD_LIB -# pragma weak pthread_mutexattr_setrobust_np -# endif extern int krb5int_pthread_loaded(void); # define K5_PTHREADS_LOADED (krb5int_pthread_loaded()) #else -- cgit