[Emacs-diffs] trunk r113963: * process.c (allocate_pty) [PTY_OPEN]: Set fd's FD_CLOEXEC flag.

[Paul Eggert]

------------------------------------------------------------
revno: 113963
revision-id: address@hidden
parent: address@hidden
committer: Paul Eggert <address@hidden>
branch nick: trunk
timestamp: Tue 2013-08-20 22:39:51 -0700
message:
  * process.c (allocate_pty) [PTY_OPEN]: Set fd's FD_CLOEXEC flag.
  
  We can't portably rely on PTY_OPEN doing that, even if
  it calls posix_openpt with O_CLOEXEC.
modified:
  src/ChangeLog                  changelog-20091113204419-o5vbwnq5f7feedwu-1438
  src/process.c                  process.c-20091113204419-o5vbwnq5f7feedwu-462

=== modified file 'src/ChangeLog'
--- a/src/ChangeLog     2013-08-20 14:56:03 +0000
+++ b/src/ChangeLog     2013-08-21 05:39:51 +0000
@@ -1,3 +1,9 @@
+2013-08-21  Paul Eggert  <address@hidden>
+
+       * process.c (allocate_pty) [PTY_OPEN]: Set fd's FD_CLOEXEC flag.
+       We can't portably rely on PTY_OPEN doing that, even if
+       it calls posix_openpt with O_CLOEXEC.
+
 2013-08-20  Kenichi Handa  <address@hidden>
 
        * character.c (string_char): Improve commentary.

=== modified file 'src/process.c'
--- a/src/process.c     2013-08-19 05:46:17 +0000
+++ b/src/process.c     2013-08-21 05:39:51 +0000
@@ -687,6 +687,15 @@
 
        if (fd >= 0)
          {
+#ifdef PTY_OPEN
+           /* Set FD's close-on-exec flag.  This is needed even if
+              PT_OPEN calls posix_openpt with O_CLOEXEC, since POSIX
+              doesn't require support for that combination.
+              Multithreaded platforms where posix_openpt ignores
+              O_CLOEXEC (or where PTY_OPEN doesn't call posix_openpt)
+              have a race condition between the PTY_OPEN and here.  */
+           fcntl (fd, F_SETFD, FD_CLOEXEC);
+#endif
            /* check to make certain that both sides are available
               this avoids a nasty yet stupid bug in rlogins */
 #ifdef PTY_TTY_NAME_SPRINTF