tcp_splice: Improve logic deciding when to splice

This makes several tweaks to improve the logic which decides whether
we're able to use the splice method for a new connection.

 * Rather than only calling tcp_splice_conn_from_sock() in pasta mode, we
   check for pasta mode within it, better localising the checks.
 * Previously if we got a connection from a non-loopback address we'd
   always fall back to the "tap" path, even if the  connection was on a
   socket in the namespace.  If we did get a non-loopback address on a
   namespace socket, something has gone wrong and the "tap" path certainly
   won't be able to handle it.  Report the error and close, rather than
   passing it along to tap.

Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Stefano Brivio <sbrivio@redhat.com>

3 files changed, 35 insertions, 17 deletions

diff --git a/inany.c b/inany.c
index c11e2aa..1c165b1 100644
--- a/inany.c
+++ b/inany.c
@@ -39,7 +39,6 @@ const union inany_addr inany_any4 = {
  *
  * Return: On success, a non-null pointer to @dst, NULL on failure
  */
-/* cppcheck-suppress unusedFunction */
 const char *inany_ntop(const union inany_addr *src, char *dst, socklen_t size)
 {
 	const struct in_addr *v4 = inany_v4(src);
diff --git a/tcp.c b/tcp.c
index 91163b8..be62a31 100644
--- a/tcp.c
+++ b/tcp.c
@@ -2737,8 +2737,7 @@ void tcp_listen_handler(struct ctx *c, union epoll_ref ref,
 	if (s < 0)
 		goto cancel;
 
-	if (c->mode == MODE_PASTA &&
-	    tcp_splice_conn_from_sock(c, ref.tcp_listen, flow, s, &sa))
+	if (tcp_splice_conn_from_sock(c, ref.tcp_listen, flow, s, &sa))
 		return;
 
 	tcp_tap_conn_from_sock(c, ref.tcp_listen, flow, s, &sa, now);
diff --git a/tcp_splice.c b/tcp_splice.c
index a202715..45b9b29 100644
--- a/tcp_splice.c
+++ b/tcp_splice.c
@@ -431,14 +431,44 @@ bool tcp_splice_conn_from_sock(const struct ctx *c,
 	sa_family_t af;
 	uint8_t pif1;
 
-	ASSERT(c->mode == MODE_PASTA);
-
-	inany_from_sockaddr(&src, &srcport, sa);
-	if (!inany_is_loopback(&src))
+	if (c->mode != MODE_PASTA)
 		return false;
 
+	inany_from_sockaddr(&src, &srcport, sa);
 	af = inany_v4(&src) ? AF_INET : AF_INET6;
 
+	switch (ref.pif) {
+	case PIF_SPLICE:
+		if (!inany_is_loopback(&src)) {
+			char str[INANY_ADDRSTRLEN];
+
+			/* We can't use flow_err() etc. because we haven't set
+			 * the flow type yet
+			 */
+			warn("Bad source address %s for splice, closing",
+			     inany_ntop(&src, str, sizeof(str)));
+
+			/* We *don't* want to fall back to tap */
+			flow_alloc_cancel(flow);
+			return true;
+		}
+
+		pif1 = PIF_HOST;
+		dstport += c->tcp.fwd_out.delta[dstport];
+		break;
+
+	case PIF_HOST:
+		if (!inany_is_loopback(&src))
+			return false;
+
+		pif1 = PIF_SPLICE;
+		dstport += c->tcp.fwd_in.delta[dstport];
+		break;
+
+	default:
+		return false;
+	}
+
 	conn = FLOW_START(flow, FLOW_TCP_SPLICE, tcp_splice, 0);
 
 	conn->flags = af == AF_INET ? 0 : SPLICE_V6;
@@ -450,16 +480,6 @@ bool tcp_splice_conn_from_sock(const struct ctx *c,
 	if (setsockopt(s0, SOL_TCP, TCP_QUICKACK, &((int){ 1 }), sizeof(int)))
 		flow_trace(conn, "failed to set TCP_QUICKACK on %i", s0);
 
-	if (ref.pif == PIF_SPLICE) {
-		pif1 = PIF_HOST;
-		dstport += c->tcp.fwd_out.delta[dstport];
-	} else {
-		ASSERT(ref.pif == PIF_HOST);
-
-		pif1 = PIF_SPLICE;
-		dstport += c->tcp.fwd_in.delta[dstport];
-	}
-
 	if (tcp_splice_connect(c, conn, af, pif1, dstport))
 		conn_flag(c, conn, CLOSING);