tcp: avoid infinite loop in tcp_splice_read()

commit ccf7abb upstream.

Splicing from TCP socket is vulnerable when a packet with URG flag is
received and stored into receive queue.

__tcp_splice_read() returns 0, and sk_wait_data() immediately
returns since there is the problematic skb in queue.

This is a nice way to burn cpu (aka infinite loop) and trigger
soft lockups.

Again, this gem was found by syzkaller tool.

Fixes: 9c55e01 ("[TCP]: Splice receive support.")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Dmitry Vyukov  <dvyukov@google.com>
Cc: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
[bwh: Backported to 3.16: adjust context]
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>

net/ipv4/tcp.c

@@ -765,6 +765,12 @@ ssize_t tcp_splice_read(struct socket *sock, loff_t *ppos,
 				ret = -EAGAIN;
 				break;
 			}
+			/* if __tcp_splice_read() got nothing while we have
+			 * an skb in receive queue, we do not want to loop.
+			 * This might happen with URG data.
+			 */
+			if (!skb_queue_empty(&sk->sk_receive_queue))
+				break;
 			sk_wait_data(sk, &timeo);
 			if (signal_pending(current)) {
 				ret = sock_intr_errno(timeo);