rxrpc: Fix use of Don't Fragment flag

rxrpc normally has the Don't Fragment flag set on the UDP packets it
transmits, except when it has decided that DATA packets aren't getting
through - in which case it turns it off just for the DATA transmissions.
This can be a problem, however, for RESPONSE packets that convey
authentication and crypto data from the client to the server as ticket may
be larger than can fit in the MTU.

In such a case, rxrpc gets itself into an infinite loop as the sendmsg
returns an error (EMSGSIZE), which causes rxkad_send_response() to return
-EAGAIN - and the CHALLENGE packet is put back on the Rx queue to retry,
leading to the I/O thread endlessly attempting to perform the transmission.

Fix this by disabling DF on RESPONSE packets for now.  The use of DF and
best data MTU determination needs reconsidering at some point in the
future.

Fixes: 17926a7 ("[AF_RXRPC]: Provide secure RxRPC sockets for use by userspace and kernel both")
Reported-by: Marc Dionne <marc.dionne@auristor.com>
Signed-off-by: David Howells <dhowells@redhat.com>
cc: linux-afs@lists.infradead.org
Acked-by: Paolo Abeni <pabeni@redhat.com>
Link: https://lore.kernel.org/r/1581852.1704813048@warthog.procyon.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Author: dhowells

net/rxrpc/ar-internal.h

@@ -1079,6 +1079,7 @@ void rxrpc_send_version_request(struct rxrpc_local *local,
 /*
  * local_object.c
  */
+void rxrpc_local_dont_fragment(const struct rxrpc_local *local, bool set);
 struct rxrpc_local *rxrpc_lookup_local(struct net *, const struct sockaddr_rxrpc *);
 struct rxrpc_local *rxrpc_get_local(struct rxrpc_local *, enum rxrpc_local_trace);
 struct rxrpc_local *rxrpc_get_local_maybe(struct rxrpc_local *, enum rxrpc_local_trace);

net/rxrpc/local_object.c

@@ -36,6 +36,17 @@ static void rxrpc_encap_err_rcv(struct sock *sk, struct sk_buff *skb, int err,
 		return ipv6_icmp_error(sk, skb, err, port, info, payload);
 }
 
+/*
+ * Set or clear the Don't Fragment flag on a socket.
+ */
+void rxrpc_local_dont_fragment(const struct rxrpc_local *local, bool set)
+{
+	if (set)
+		ip_sock_set_mtu_discover(local->socket->sk, IP_PMTUDISC_DO);
+	else
+		ip_sock_set_mtu_discover(local->socket->sk, IP_PMTUDISC_DONT);
+}
+
 /*
  * Compare a local to an address.  Return -ve, 0 or +ve to indicate less than,
  * same or greater than.
@@ -203,7 +214,7 @@ static int rxrpc_open_socket(struct rxrpc_local *local, struct net *net)
 		ip_sock_set_recverr(usk);
 
 		/* we want to set the don't fragment bit */
-		ip_sock_set_mtu_discover(usk, IP_PMTUDISC_DO);
+		rxrpc_local_dont_fragment(local, true);
 
 		/* We want receive timestamps. */
 		sock_enable_timestamps(usk);

net/rxrpc/output.c

@@ -494,14 +494,12 @@ int rxrpc_send_data_packet(struct rxrpc_call *call, struct rxrpc_txbuf *txb)
 	switch (conn->local->srx.transport.family) {
 	case AF_INET6:
 	case AF_INET:
-		ip_sock_set_mtu_discover(conn->local->socket->sk,
-					 IP_PMTUDISC_DONT);
+		rxrpc_local_dont_fragment(conn->local, false);
 		rxrpc_inc_stat(call->rxnet, stat_tx_data_send_frag);
 		ret = do_udp_sendmsg(conn->local->socket, &msg, len);
 		conn->peer->last_tx_at = ktime_get_seconds();
 
-		ip_sock_set_mtu_discover(conn->local->socket->sk,
-					 IP_PMTUDISC_DO);
+		rxrpc_local_dont_fragment(conn->local, true);
 		break;
 
 	default:

net/rxrpc/rxkad.c

@@ -724,7 +724,9 @@ static int rxkad_send_response(struct rxrpc_connection *conn,
 	serial = atomic_inc_return(&conn->serial);
 	whdr.serial = htonl(serial);
 
+	rxrpc_local_dont_fragment(conn->local, false);
 	ret = kernel_sendmsg(conn->local->socket, &msg, iov, 3, len);
+	rxrpc_local_dont_fragment(conn->local, true);
 	if (ret < 0) {
 		trace_rxrpc_tx_fail(conn->debug_id, serial, ret,
 				    rxrpc_tx_point_rxkad_response);