Applied upstream patch (in conjunction with
5113963):
Port close-on-exec pty creation to FreeBSD 9.1-RELEASE.
* configure.ac (PTY_OPEN): If posix_openpt with O_CLOEXEC fails
and reports EINVAL, try it again without O_CLOEXEC. This should
port PTY_OPEN to FreeBSD 9, which stupidly rejects O_CLOEXEC.
What were they thinking?
Origin: upstream, commit:
5113964,
5c41cec3858f1828303b67be634b8c39937c8580
Added-by: Rob Browning <rlb@defaultvalue.org>
Provided-By: Paul Eggert <eggert@cs.ucla.edu>
Bug: http://debbugs.gnu.org/cgi/bugreport.cgi?bug=15129
+2013-08-21 Paul Eggert <eggert@cs.ucla.edu>
+
+ Port close-on-exec pty creation to FreeBSD 9.1-RELEASE (Bug#15129).
+ * configure.ac (PTY_OPEN): If posix_openpt with O_CLOEXEC fails
+ and reports EINVAL, try it again without O_CLOEXEC. This should
+ port PTY_OPEN to FreeBSD 9, which stupidly rejects O_CLOEXEC.
+ What were they thinking?
+
2013-03-11 Glenn Morris <rgm@gnu.org>
* Version 24.3 released.
AC_DEFINE(PTY_TTY_NAME_SPRINTF, [{ char *ptyname = 0; sigset_t blocked; sigemptyset (&blocked); sigaddset (&blocked, SIGCHLD); pthread_sigmask (SIG_BLOCK, &blocked, 0); if (grantpt (fd) != -1 && unlockpt (fd) != -1) ptyname = ptsname(fd); pthread_sigmask (SIG_UNBLOCK, &blocked, 0); if (!ptyname) { close (fd); return -1; } snprintf (pty_name, sizeof pty_name, "%s", ptyname); }])
dnl if HAVE_POSIX_OPENPT
if test "x$ac_cv_func_posix_openpt" = xyes; then
- AC_DEFINE(PTY_OPEN, [fd = posix_openpt (O_RDWR | O_NOCTTY)])
+ AC_DEFINE(PTY_OPEN, [do { fd = posix_openpt (O_RDWR | O_CLOEXEC | O_NOCTTY); if (fd < 0 && errno == EINVAL) fd = posix_openpt (O_RDWR | O_NOCTTY); } while (0)])
AC_DEFINE(PTY_NAME_SPRINTF, [])
dnl if HAVE_GETPT
elif test "x$ac_cv_func_getpt" = xyes; then