Don't attempt to resume channels if we already exchanged funding

ddf75af introduced the ability to re-exchange our `ChannelOpen`
after a peer disconnects if we didn't complete funding on our end.
It did not implement nor consider what would happen if we
re-connected after we created our own funding transactions, and
currently it panics (and even if it did not it would replay the
`FundingTransactionGenerated` event to users).

While we'd very much like to replay the `open_channel` flow even
if we have already received an `accept_channel` and funded the
channel, we cannot as the peer will likely provide different key
material in their `accept_channel`, causing us to need a different
funding transaction.

Thus, here, we simply close channels which have been funded but not
yet signed when our peer disconnects.

Author: TheBlueMatt

lightning/src/ln/channel.rs

@@ -7520,6 +7520,12 @@ impl<SP: Deref> OutboundV1Channel<SP> where SP::Target: SignerProvider {
 		Ok(self.get_open_channel(chain_hash))
 	}
 
+	/// Returns true if we can resume the channel by sending the [`msgs::OpenChannel`] again.
+	pub fn is_resumable(&self) -> bool {
+		!self.context.have_received_message() &&
+			self.context.cur_holder_commitment_transaction_number == INITIAL_COMMITMENT_NUMBER
+	}
+
 	pub fn get_open_channel(&self, chain_hash: ChainHash) -> msgs::OpenChannel {
 		if !self.context.is_outbound() {
 			panic!("Tried to open a channel for an inbound channel?");

lightning/src/ln/channelmanager.rs

@@ -9954,11 +9954,14 @@ where
 							}
 							&mut chan.context
 						},
-						// We retain UnfundedOutboundV1 channel for some time in case
-						// peer unexpectedly disconnects, and intends to reconnect again.
-						ChannelPhase::UnfundedOutboundV1(_) => {
-							return true;
-						},
+						// If we get disconnected and haven't yet committed to a funding
+						// transaction, we can replay the `open_channel` on reconnection, so don't
+						// bother dropping the channel here. However, if we already committed to
+						// the funding transaction we don't yet support replaying the funding
+						// handshake (and bailing if the peer rejects it), so we force-close in
+						// that case.
+						ChannelPhase::UnfundedOutboundV1(chan) if chan.is_resumable() => return true,
+						ChannelPhase::UnfundedOutboundV1(chan) => &mut chan.context,
 						// Unfunded inbound channels will always be removed.
 						ChannelPhase::UnfundedInboundV1(chan) => {
 							&mut chan.context

lightning/src/ln/functional_tests.rs

@@ -61,6 +61,38 @@ use crate::ln::chan_utils::CommitmentTransaction;
 
 use super::channel::UNFUNDED_CHANNEL_AGE_LIMIT_TICKS;
 
+#[test]
+fn test_channel_resumption_fail_post_funding() {
+	// If we fail to exchange funding with a peer prior to it disconnecting we'll resume the
+	// channel open on reconnect, however if we do exchange funding we do not currently support
+	// replaying it and here test that the channel closes.
+	let chanmon_cfgs = create_chanmon_cfgs(2);
+	let node_cfgs = create_node_cfgs(2, &chanmon_cfgs);
+	let node_chanmgrs = create_node_chanmgrs(2, &node_cfgs, &[None, None]);
+	let nodes = create_network(2, &node_cfgs, &node_chanmgrs);
+
+	nodes[0].node.create_channel(nodes[1].node.get_our_node_id(), 1_000_000, 0, 42, None, None).unwrap();
+	let open_chan = get_event_msg!(nodes[0], MessageSendEvent::SendOpenChannel, nodes[1].node.get_our_node_id());
+	nodes[1].node.handle_open_channel(&nodes[0].node.get_our_node_id(), &open_chan);
+	let accept_chan = get_event_msg!(nodes[1], MessageSendEvent::SendAcceptChannel, nodes[0].node.get_our_node_id());
+	nodes[0].node.handle_accept_channel(&nodes[1].node.get_our_node_id(), &accept_chan);
+
+	let (temp_chan_id, tx, funding_output) =
+		create_funding_transaction(&nodes[0], &nodes[1].node.get_our_node_id(), 1_000_000, 42);
+	let new_chan_id = ChannelId::v1_from_funding_outpoint(funding_output);
+	nodes[0].node.funding_transaction_generated(&temp_chan_id, &nodes[1].node.get_our_node_id(), tx).unwrap();
+
+	nodes[0].node.peer_disconnected(&nodes[1].node.get_our_node_id());
+	check_closed_events(&nodes[0], &[ExpectedCloseEvent::from_id_reason(new_chan_id, true, ClosureReason::DisconnectedPeer)]);
+
+	// After ddf75afd16 we'd panic on reconnection if we exchanged funding info, so test that
+	// explicitly here.
+	nodes[0].node.peer_connected(&nodes[1].node.get_our_node_id(), &msgs::Init {
+		features: nodes[1].node.init_features(), networks: None, remote_network_address: None
+	}, true).unwrap();
+	assert_eq!(nodes[0].node.get_and_clear_pending_msg_events(), Vec::new());
+}
+
 #[test]
 fn test_insane_channel_opens() {
 	// Stand up a network of 2 nodes
@@ -3738,10 +3770,10 @@ fn test_peer_disconnected_before_funding_broadcasted() {
 		nodes[0].node.timer_tick_occurred();
 	}
 
-	// Ensure that the channel is closed with `ClosureReason::HolderForceClosed`
-	// when the peers are disconnected and do not reconnect before the funding
-	// transaction is broadcasted.
-	check_closed_event!(&nodes[0], 2, ClosureReason::HolderForceClosed, true
+	// Ensure that the channel is closed with `ClosureReason::DisconnectedPeer` and a
+	// `DiscardFunding` event when the peers are disconnected and do not reconnect before the
+	// funding transaction is broadcasted.
+	check_closed_event!(&nodes[0], 2, ClosureReason::DisconnectedPeer, true
 		, [nodes[1].node.get_our_node_id()], 1000000);
 	check_closed_event!(&nodes[1], 1, ClosureReason::DisconnectedPeer, false
 		, [nodes[0].node.get_our_node_id()], 1000000);