From bb7c4d1853fe539de946cad975001593c2f845fa Mon Sep 17 00:00:00 2001 From: Matt Corallo Date: Thu, 17 Aug 2023 22:34:24 +0000 Subject: [PATCH] Work around LND bug 6039 LND hasn't properly handled shutdown messages ever, and force-closes any time we send one while HTLCs are still present. The issue is tracked at https://github.com/lightningnetwork/lnd/issues/6039 and has had multiple patches to fix it but none so far have managed to land upstream. The issue appears to be very low priority for the LND team despite being marked "P1". We're not going to bother handling this in a sensible way, instead simply repeated the Shutdown message on repeat until morale improves. --- lightning/src/ln/channel.rs | 19 +++++++----- lightning/src/ln/channelmanager.rs | 30 ++++++++++++++++++ lightning/src/ln/shutdown_tests.rs | 49 ++++++++++++++++++++++++++++++ 3 files changed, 91 insertions(+), 7 deletions(-) diff --git a/lightning/src/ln/channel.rs b/lightning/src/ln/channel.rs index d5cefc943..9ab152f5d 100644 --- a/lightning/src/ln/channel.rs +++ b/lightning/src/ln/channel.rs @@ -3811,6 +3811,17 @@ impl Channel { } } + /// Gets the `Shutdown` message we should send our peer on reconnect, if any. + pub fn get_outbound_shutdown(&self) -> Option { + if self.context.channel_state & (ChannelState::LocalShutdownSent as u32) != 0 { + assert!(self.context.shutdown_scriptpubkey.is_some()); + Some(msgs::Shutdown { + channel_id: self.context.channel_id, + scriptpubkey: self.get_closing_scriptpubkey(), + }) + } else { None } + } + /// May panic if some calls other than message-handling calls (which will all Err immediately) /// have been called between remove_uncommitted_htlcs_and_mark_paused and this call. /// @@ -3877,13 +3888,7 @@ impl Channel { self.context.channel_state &= !(ChannelState::PeerDisconnected as u32); self.context.sent_message_awaiting_response = None; - let shutdown_msg = if self.context.channel_state & (ChannelState::LocalShutdownSent as u32) != 0 { - assert!(self.context.shutdown_scriptpubkey.is_some()); - Some(msgs::Shutdown { - channel_id: self.context.channel_id, - scriptpubkey: self.get_closing_scriptpubkey(), - }) - } else { None }; + let shutdown_msg = self.get_outbound_shutdown(); let announcement_sigs = self.get_announcement_sigs(node_signer, genesis_block_hash, user_config, best_block.height(), logger); diff --git a/lightning/src/ln/channelmanager.rs b/lightning/src/ln/channelmanager.rs index 45fffd540..9794dc905 100644 --- a/lightning/src/ln/channelmanager.rs +++ b/lightning/src/ln/channelmanager.rs @@ -7489,6 +7489,36 @@ where fn handle_error(&self, counterparty_node_id: &PublicKey, msg: &msgs::ErrorMessage) { let _persistence_guard = PersistenceNotifierGuard::notify_on_drop(self); + match &msg.data as &str { + "cannot co-op close channel w/ active htlcs"| + "link failed to shutdown" => + { + // LND hasn't properly handled shutdown messages ever, and force-closes any time we + // send one while HTLCs are still present. The issue is tracked at + // https://github.com/lightningnetwork/lnd/issues/6039 and has had multiple patches + // to fix it but none so far have managed to land upstream. The issue appears to be + // very low priority for the LND team despite being marked "P1". + // We're not going to bother handling this in a sensible way, instead simply + // repeating the Shutdown message on repeat until morale improves. + if msg.channel_id != [0; 32] { + let per_peer_state = self.per_peer_state.read().unwrap(); + let peer_state_mutex_opt = per_peer_state.get(counterparty_node_id); + if peer_state_mutex_opt.is_none() { return; } + let mut peer_state = peer_state_mutex_opt.unwrap().lock().unwrap(); + if let Some(chan) = peer_state.channel_by_id.get(&msg.channel_id) { + if let Some(msg) = chan.get_outbound_shutdown() { + peer_state.pending_msg_events.push(events::MessageSendEvent::SendShutdown { + node_id: *counterparty_node_id, + msg, + }); + } + } + } + return; + } + _ => {} + } + if msg.channel_id == [0; 32] { let channel_ids: Vec<[u8; 32]> = { let per_peer_state = self.per_peer_state.read().unwrap(); diff --git a/lightning/src/ln/shutdown_tests.rs b/lightning/src/ln/shutdown_tests.rs index bd4df0af0..23d3a30dc 100644 --- a/lightning/src/ln/shutdown_tests.rs +++ b/lightning/src/ln/shutdown_tests.rs @@ -198,6 +198,55 @@ fn expect_channel_shutdown_state_with_htlc() { assert!(nodes[0].node.list_channels().is_empty()); } +#[test] +fn test_lnd_bug_6039() { + // LND sends a nonsense error message any time it gets a shutdown if there are still HTLCs + // pending. We currently swallow that error to work around LND's bug #6039. This test emulates + // the LND nonsense and ensures we at least kinda handle it. + 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 mut nodes = create_network(2, &node_cfgs, &node_chanmgrs); + let chan = create_announced_chan_between_nodes(&nodes, 0, 1); + + let (payment_preimage, payment_hash, _) = route_payment(&nodes[0], &[&nodes[1]], 100_000); + + nodes[0].node.close_channel(&chan.2, &nodes[1].node.get_our_node_id()).unwrap(); + let node_0_shutdown = get_event_msg!(nodes[0], MessageSendEvent::SendShutdown, nodes[1].node.get_our_node_id()); + nodes[1].node.handle_shutdown(&nodes[0].node.get_our_node_id(), &node_0_shutdown); + + // Generate an lnd-like error message and check that we respond by simply screaming louder to + // see if LND will accept our protocol compliance. + let err_msg = msgs::ErrorMessage { channel_id: chan.2, data: "link failed to shutdown".to_string() }; + nodes[0].node.handle_error(&nodes[1].node.get_our_node_id(), &err_msg); + let _node_0_repeated_shutdown = get_event_msg!(nodes[0], MessageSendEvent::SendShutdown, nodes[1].node.get_our_node_id()); + + let node_1_shutdown = get_event_msg!(nodes[1], MessageSendEvent::SendShutdown, nodes[0].node.get_our_node_id()); + + assert!(nodes[0].node.get_and_clear_pending_msg_events().is_empty()); + assert!(nodes[1].node.get_and_clear_pending_msg_events().is_empty()); + + claim_payment(&nodes[0], &[&nodes[1]], payment_preimage); + + // Assume that LND will eventually respond to our Shutdown if we clear all the remaining HTLCs + nodes[0].node.handle_shutdown(&nodes[1].node.get_our_node_id(), &node_1_shutdown); + + // ClosingSignNegotion process + let node_0_closing_signed = get_event_msg!(nodes[0], MessageSendEvent::SendClosingSigned, nodes[1].node.get_our_node_id()); + nodes[1].node.handle_closing_signed(&nodes[0].node.get_our_node_id(), &node_0_closing_signed); + let node_1_closing_signed = get_event_msg!(nodes[1], MessageSendEvent::SendClosingSigned, nodes[0].node.get_our_node_id()); + nodes[0].node.handle_closing_signed(&nodes[1].node.get_our_node_id(), &node_1_closing_signed); + let (_, node_0_2nd_closing_signed) = get_closing_signed_broadcast!(nodes[0].node, nodes[1].node.get_our_node_id()); + nodes[1].node.handle_closing_signed(&nodes[0].node.get_our_node_id(), &node_0_2nd_closing_signed.unwrap()); + let (_, node_1_none) = get_closing_signed_broadcast!(nodes[1].node, nodes[0].node.get_our_node_id()); + assert!(node_1_none.is_none()); + check_closed_event!(nodes[0], 1, ClosureReason::CooperativeClosure, [nodes[1].node.get_our_node_id()], 100000); + check_closed_event!(nodes[1], 1, ClosureReason::CooperativeClosure, [nodes[0].node.get_our_node_id()], 100000); + + // Shutdown basically removes the channelDetails, testing of shutdowncomplete state unnecessary + assert!(nodes[0].node.list_channels().is_empty()); +} + #[test] fn expect_channel_shutdown_state_with_force_closure() { // Test sending a shutdown prior to channel_ready after funding generation -- 2.39.5