tcp_splice: Don't use flow_trace() before setting flow type

In tcp_splice_conn_from_sock() we can call flow_trace() if there's an
error setting TCP_QUICKACK.  However, we do so before we've set the
flow type in the flow entry.  That means that flow_trace() will print
nonsense when it tries to print the flow type.

There's no reason the setsockopt() has to happen before initialising
the flow entry, so just move it after.

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

1 files changed, 3 insertions, 3 deletions

diff --git a/tcp_splice.c b/tcp_splice.c
index 8187220..49585f2 100644
--- a/tcp_splice.c
+++ b/tcp_splice.c
@@ -453,9 +453,6 @@ bool tcp_splice_conn_from_sock(const struct ctx *c,
 	if (!inany_is_loopback(&aany))
 		return false;
 
-	if (setsockopt(s, SOL_TCP, TCP_QUICKACK, &((int){ 1 }), sizeof(int)))
-		flow_trace(conn, "failed to set TCP_QUICKACK on %i", s);
-
 	conn->f.type = FLOW_TCP_SPLICE;
 	conn->flags = inany_v4(&aany) ? 0 : SPLICE_V6;
 	conn->s[0] = s;
@@ -463,6 +460,9 @@ bool tcp_splice_conn_from_sock(const struct ctx *c,
 	conn->pipe[0][0] = conn->pipe[0][1] = -1;
 	conn->pipe[1][0] = conn->pipe[1][1] = -1;
 
+	if (setsockopt(s, SOL_TCP, TCP_QUICKACK, &((int){ 1 }), sizeof(int)))
+		flow_trace(conn, "failed to set TCP_QUICKACK on %i", s);
+
 	if (tcp_splice_new(c, conn, ref.port, ref.pif))
 		conn_flag(c, conn, CLOSING);