Set socket TCP keepalive option on incoming connections

If a connected host disappears without our knowledge, as can happen over
wireless or a hibernating machine, we continue to hold the port open waiting
for messages. Because we never try to send anything down this now-broken
pipe, the connection will sit idle taking up a slot in our allowed incoming
connections list.

If enough of these happen, an unintended Denial of Service takes place,
where all connection slots are filled with now-broken, never ending
connections. Setting the TCP keepalive option at least allows these to time
out after the default two hours, which is sufficient in the non-malicious
case.

Signed-off-by: Dan McGee <dan@archlinux.org>

1 files changed, 6 insertions, 2 deletions

diff --git a/src/server_socket.c b/src/server_socket.c
index 82ad81f12..fc8dbd17a 100644
--- a/src/server_socket.c
+++ b/src/server_socket.c
@@ -160,12 +160,16 @@ server_socket_in_event(G_GNUC_UNUSED GIOChannel *source,
 	size_t address_length = sizeof(address);
 	int fd = accept_cloexec_nonblock(s->fd, (struct sockaddr*)&address,
 					 &address_length);
-	if (fd >= 0)
+	if (fd >= 0) {
+		if (socket_keepalive(fd))
+			g_warning("Could not set TCP keepalive option: %s",
+				  g_strerror(errno));
 		s->parent->callback(fd, (const struct sockaddr*)&address,
 				    address_length, get_remote_uid(fd),
 				    s->parent->callback_ctx);
-	else
+	} else {
 		g_warning("accept() failed: %s", g_strerror(errno));
+	}
 
 	return true;
 }