From: Ondřej Surý <ondrej@sury.org>
Date: Mon, 5 Oct 2020 11:14:04 +0000 (+0200)
Subject: Split reusing the addr/port and load-balancing socket options
X-Git-Tag: v9.17.6~14^2
X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=fd975a551dc23b18c2a63105c0b2bb9cf0015021;p=thirdparty%2Fbind9.git

Split reusing the addr/port and load-balancing socket options

The SO_REUSEADDR, SO_REUSEPORT and SO_REUSEPORT_LB has different meaning
on different platform. In this commit, we split the function to set the
reuse of address/port and setting the load-balancing into separate
functions.

The libuv library already have multiplatform support for setting
SO_REUSEADDR and SO_REUSEPORT that allows binding to the same address
and port, but unfortunately, when used after the load-balancing socket
options have been already set, it overrides the previous setting, so we
need our own helper function to enable the SO_REUSEADDR/SO_REUSEPORT
first and then enable the load-balancing socket option.
---

diff --git a/lib/isc/netmgr/netmgr-int.h b/lib/isc/netmgr/netmgr-int.h
index 1ea281e251f..ceeb5cbb276 100644
--- a/lib/isc/netmgr/netmgr-int.h
+++ b/lib/isc/netmgr/netmgr-int.h
@@ -840,9 +840,15 @@ isc__nm_socket_freebind(uv_os_sock_t fd, sa_family_t sa_family);
  */
 
 isc_result_t
-isc__nm_socket_reuseport(uv_os_sock_t fd);
+isc__nm_socket_reuse(uv_os_sock_t fd);
 /*%<
- * Set the SO_REUSEPORT (or equivalent) socket option on the fd
+ * Set the SO_REUSEADDR or SO_REUSEPORT (or equivalent) socket option on the fd
+ */
+
+isc_result_t
+isc__nm_socket_reuse_lb(uv_os_sock_t fd);
+/*%<
+ * Set the SO_REUSEPORT_LB (or equivalent) socket option on the fd
  */
 
 isc_result_t
diff --git a/lib/isc/netmgr/netmgr.c b/lib/isc/netmgr/netmgr.c
index 7567ee30ffc..28dc6f22e11 100644
--- a/lib/isc/netmgr/netmgr.c
+++ b/lib/isc/netmgr/netmgr.c
@@ -1625,46 +1625,66 @@ isc__nm_socket_freebind(uv_os_sock_t fd, sa_family_t sa_family) {
 }
 
 isc_result_t
-isc__nm_socket_reuseport(uv_os_sock_t fd) {
+isc__nm_socket_reuse(uv_os_sock_t fd) {
 	/*
-	 * This is SO_REUSE**** hell:
-	 *
 	 * Generally, the SO_REUSEADDR socket option allows reuse of
-	 * local addresses.  On Windows, it also allows a socket to
-	 * forcibly bind to a port in use by another socket.
+	 * local addresses.
 	 *
-	 * On Linux, SO_REUSEPORT socket option allows sockets to be
-	 * bound to an identical socket address. For UDP sockets, the
-	 * use of this option can provide better distribution of
-	 * incoming datagrams to multiple processes (or threads) as
-	 * compared to the traditional technique of having multiple
-	 * processes compete to receive datagrams on the same socket.
+	 * On the BSDs, SO_REUSEPORT implies SO_REUSEADDR but with some
+	 * additional refinements for programs that use multicast.
 	 *
-	 * On FreeBSD 12+, the same thing is achieved with SO_REUSEPORT_LB.
+	 * On Linux, SO_REUSEPORT has different semantics: it _shares_ the port
+	 * rather than steal it from the current listener, so we don't use it
+	 * here, but rather in isc__nm_socket_reuse_lb().
 	 *
+	 * On Windows, it also allows a socket to forcibly bind to a port in use
+	 * by another socket.
 	 */
-	isc_result_t result = ISC_R_NOTIMPLEMENTED;
-#if defined(SO_REUSEADDR)
+
+#if defined(SO_REUSEPORT) && !defined(__linux__)
+	if (setsockopt_on(fd, SOL_SOCKET, SO_REUSEPORT) == -1) {
+		return (ISC_R_FAILURE);
+	}
+	return (ISC_R_SUCCESS);
+#elif defined(SO_REUSEADDR)
 	if (setsockopt_on(fd, SOL_SOCKET, SO_REUSEADDR) == -1) {
 		return (ISC_R_FAILURE);
-	} else {
-		result = ISC_R_SUCCESS;
 	}
+	return (ISC_R_SUCCESS);
+#else
+	UNUSED(fd);
+	return (ISC_R_NOTIMPLEMENTED);
 #endif
+}
+
+isc_result_t
+isc__nm_socket_reuse_lb(uv_os_sock_t fd) {
+	/*
+	 * On FreeBSD 12+, SO_REUSEPORT_LB socket option allows sockets to be
+	 * bound to an identical socket address. For UDP sockets, the use of
+	 * this option can provide better distribution of incoming datagrams to
+	 * multiple processes (or threads) as compared to the traditional
+	 * technique of having multiple processes compete to receive datagrams
+	 * on the same socket.
+	 *
+	 * On Linux, the same thing is achieved simply with SO_REUSEPORT.
+	 */
 #if defined(SO_REUSEPORT_LB)
 	if (setsockopt_on(fd, SOL_SOCKET, SO_REUSEPORT_LB) == -1) {
 		return (ISC_R_FAILURE);
 	} else {
-		result = ISC_R_SUCCESS;
+		return (ISC_R_SUCCESS);
 	}
-#elif defined(SO_REUSEPORT)
+#elif defined(SO_REUSEPORT) && defined(__linux__)
 	if (setsockopt_on(fd, SOL_SOCKET, SO_REUSEPORT) == -1) {
 		return (ISC_R_FAILURE);
 	} else {
-		result = ISC_R_SUCCESS;
+		return (ISC_R_SUCCESS);
 	}
+#else
+	UNUSED(fd);
+	return (ISC_R_NOTIMPLEMENTED);
 #endif
-	return (result);
 }
 
 isc_result_t
diff --git a/lib/isc/netmgr/udp.c b/lib/isc/netmgr/udp.c
index e531857e1c6..7056a29bf27 100644
--- a/lib/isc/netmgr/udp.c
+++ b/lib/isc/netmgr/udp.c
@@ -82,7 +82,11 @@ isc_nm_listenudp(isc_nm_t *mgr, isc_nmiface_t *iface, isc_nm_recv_cb_t cb,
 		csock->fd = socket(family, SOCK_DGRAM, 0);
 		RUNTIME_CHECK(csock->fd >= 0);
 
-		result = isc__nm_socket_reuseport(csock->fd);
+		result = isc__nm_socket_reuse(csock->fd);
+		RUNTIME_CHECK(result == ISC_R_SUCCESS ||
+			      result == ISC_R_NOTIMPLEMENTED);
+
+		result = isc__nm_socket_reuse_lb(csock->fd);
 		RUNTIME_CHECK(result == ISC_R_SUCCESS ||
 			      result == ISC_R_NOTIMPLEMENTED);