native: Use WNOHANG before signaling

It turns out that on linux, and possibly other platforms, child processes will
continue to accept signals until they have been *reaped*. This means that once
the child has exited, it will succeed to receive signals until waitpid() has
been invoked on it.

This is unfortunate behavior, and differs from what is seen on OSX and windows.
This commit changes the behavior of Process::signal() to be the same across
platforms, and updates the documentation of Process::kill() to note that when
signaling a foreign process it may accept signals until reaped.

Implementation-wise, this invokes waitpid() with WNOHANG before each signal to
the child to ensure that if the child has exited that we will reap it. Other
possibilities include installing a SIGCHLD signal handler, but at this time I
believe that that's too complicated.

Closes #13124

src/libstd/io/process.rs

@@ -331,7 +331,9 @@ impl Process {
     /// signals (SIGTERM/SIGKILL/SIGINT) are translated to `TerminateProcess`.
     ///
     /// Additionally, a signal number of 0 can check for existence of the target
-    /// process.
+    /// process. Note, though, that on some platforms signals will continue to
+    /// be successfully delivered if the child has exited, but not yet been
+    /// reaped.
     pub fn kill(id: libc::pid_t, signal: int) -> IoResult<()> {
         LocalIo::maybe_raise(|io| io.kill(id, signal))
     }
@@ -342,8 +344,16 @@ impl Process {
     /// Sends the specified signal to the child process, returning whether the
     /// signal could be delivered or not.
     ///
-    /// Note that this is purely a wrapper around libuv's `uv_process_kill`
-    /// function.
+    /// Note that signal 0 is interpreted as a poll to check whether the child
+    /// process is still alive or not. If an error is returned, then the child
+    /// process has exited.
+    ///
+    /// On some unix platforms signals will continue to be received after a
+    /// child has exited but not yet been reaped. In order to report the status
+    /// of signal delivery correctly, unix implementations may invoke
+    /// `waitpid()` with `WNOHANG` in order to reap the child as necessary.
+    ///
+    /// # Errors
     ///
     /// If the signal delivery fails, the corresponding error is returned.
     pub fn signal(&mut self, signal: int) -> IoResult<()> {
@@ -833,4 +843,17 @@ mod tests {
         p.signal_kill().unwrap();
         assert!(!p.wait().success());
     })
+
+    iotest!(fn test_zero() {
+        let mut p = sleeper();
+        p.signal_kill().unwrap();
+        for _ in range(0, 20) {
+            if p.signal(0).is_err() {
+                assert!(!p.wait().success());
+                return
+            }
+            timer::sleep(100);
+        }
+        fail!("never saw the child go away");
+    })
 }