async_req: make async_connect_send() "reentrant"

Allow callers to pass in socket fds that where already passed to an
earlier call of async_connect_send(). Callers expect this behaviour and
it was working until 05d4dbda8357712cb81008e0d611fdb0e7239587 broke it.

The proper fix would be to change callers to close the fd and start from
scratch with a fresh socket.

Bug: https://bugzilla.samba.org/show_bug.cgi?id=12105

Signed-off-by: Ralph Boehme <slow@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>

Autobuild-User(master): Ralph Böhme <slow@samba.org>
Autobuild-Date(master): Thu Aug  4 05:03:21 CEST 2016 on sn-devel-144

1 files changed, 13 insertions, 3 deletions

diff --git a/lib/async_req/async_sock.c b/lib/async_req/async_sock.c
index c14acf3fd62..3af17484a2b 100644
--- a/lib/async_req/async_sock.c
+++ b/lib/async_req/async_sock.c
@@ -128,11 +128,21 @@ struct tevent_req *async_connect_send(
 	}
 
 	/*
-	 * The only errno indicating that the connect is still in
-	 * flight is EINPROGRESS, everything else is an error
+	 * The only errno indicating that an initial connect is still
+	 * in flight is EINPROGRESS.
+	 *
+	 * We get EALREADY when someone calls us a second time for a
+	 * given fd and the connect is still in flight (and returned
+	 * EINPROGRESS the first time).
+	 *
+	 * This allows callers like open_socket_out_send() to reuse
+	 * fds and call us with an fd for which the connect is still
+	 * in flight. The proper thing to do for callers would be
+	 * closing the fd and starting from scratch with a fresh
+	 * socket.
 	 */
 
-	if (errno != EINPROGRESS) {
+	if (errno != EINPROGRESS && errno != EALREADY) {
 		tevent_req_error(req, errno);
 		return tevent_req_post(req, ev);
 	}