Support atomic partial updates to ChannelConfig
[rust-lightning] / lightning / src / ln / channel.rs
index 7778b92025db044399feba49af7b2ef85ce2fc98..99b3c1691149960f599ab532a98419da03b66c8c 100644 (file)
@@ -35,7 +35,7 @@ use crate::chain::BestBlock;
 use crate::chain::chaininterface::{FeeEstimator, ConfirmationTarget, LowerBoundedFeeEstimator};
 use crate::chain::channelmonitor::{ChannelMonitor, ChannelMonitorUpdate, ChannelMonitorUpdateStep, LATENCY_GRACE_PERIOD_BLOCKS, CLOSED_CHANNEL_UPDATE_ID};
 use crate::chain::transaction::{OutPoint, TransactionData};
-use crate::chain::keysinterface::{WriteableEcdsaChannelSigner, EntropySource, ChannelSigner, SignerProvider, NodeSigner, Recipient};
+use crate::sign::{WriteableEcdsaChannelSigner, EntropySource, ChannelSigner, SignerProvider, NodeSigner, Recipient};
 use crate::events::ClosureReason;
 use crate::routing::gossip::NodeId;
 use crate::util::ser::{Readable, ReadableArgs, Writeable, Writer, VecWriter};
@@ -432,6 +432,12 @@ pub(super) struct ReestablishResponses {
        pub shutdown_msg: Option<msgs::Shutdown>,
 }
 
+/// The return type of `force_shutdown`
+pub(crate) type ShutdownResult = (
+       Option<(PublicKey, OutPoint, ChannelMonitorUpdate)>,
+       Vec<(HTLCSource, PaymentHash, PublicKey, [u8; 32])>
+);
+
 /// If the majority of the channels funds are to the fundee and the initiator holds only just
 /// enough funds to cover their reserve value, channels are at risk of getting "stuck". Because the
 /// initiator controls the feerate, if they then go to increase the channel fee, they may have no
@@ -479,6 +485,28 @@ pub(crate) const MIN_AFFORDABLE_HTLC_COUNT: usize = 4;
 ///   * `EXPIRE_PREV_CONFIG_TICKS` = convergence_delay / tick_interval
 pub(crate) const EXPIRE_PREV_CONFIG_TICKS: usize = 5;
 
+/// The number of ticks that may elapse while we're waiting for a response to a
+/// [`msgs::RevokeAndACK`] or [`msgs::ChannelReestablish`] message before we attempt to disconnect
+/// them.
+///
+/// See [`Channel::sent_message_awaiting_response`] for more information.
+pub(crate) const DISCONNECT_PEER_AWAITING_RESPONSE_TICKS: usize = 2;
+
+struct PendingChannelMonitorUpdate {
+       update: ChannelMonitorUpdate,
+       /// In some cases we need to delay letting the [`ChannelMonitorUpdate`] go until after an
+       /// `Event` is processed by the user. This bool indicates the [`ChannelMonitorUpdate`] is
+       /// blocked on some external event and the [`ChannelManager`] will update us when we're ready.
+       ///
+       /// [`ChannelManager`]: super::channelmanager::ChannelManager
+       blocked: bool,
+}
+
+impl_writeable_tlv_based!(PendingChannelMonitorUpdate, {
+       (0, update, required),
+       (2, blocked, required),
+});
+
 // TODO: We should refactor this to be an Inbound/OutboundChannel until initial setup handshaking
 // has been completed, and then turn into a Channel to get compiler-time enforcement of things like
 // calling channel_id() before we're set up or things like get_outbound_funding_signed on an
@@ -700,6 +728,19 @@ pub(super) struct Channel<Signer: ChannelSigner> {
        /// See-also <https://github.com/lightningnetwork/lnd/issues/4006>
        pub workaround_lnd_bug_4006: Option<msgs::ChannelReady>,
 
+       /// An option set when we wish to track how many ticks have elapsed while waiting for a response
+       /// from our counterparty after sending a message. If the peer has yet to respond after reaching
+       /// `DISCONNECT_PEER_AWAITING_RESPONSE_TICKS`, a reconnection should be attempted to try to
+       /// unblock the state machine.
+       ///
+       /// This behavior is mostly motivated by a lnd bug in which we don't receive a message we expect
+       /// to in a timely manner, which may lead to channels becoming unusable and/or force-closed. An
+       /// example of such can be found at <https://github.com/lightningnetwork/lnd/issues/7682>.
+       ///
+       /// This is currently only used when waiting for a [`msgs::ChannelReestablish`] or
+       /// [`msgs::RevokeAndACK`] message from the counterparty.
+       sent_message_awaiting_response: Option<usize>,
+
        #[cfg(any(test, fuzzing))]
        // When we receive an HTLC fulfill on an outbound path, we may immediately fulfill the
        // corresponding HTLC on the inbound path. If, then, the outbound path channel is
@@ -744,7 +785,7 @@ pub(super) struct Channel<Signer: ChannelSigner> {
        /// If we then persist the [`channelmanager::ChannelManager`] and crash before the persistence
        /// completes we still need to be able to complete the persistence. Thus, we have to keep a
        /// copy of the [`ChannelMonitorUpdate`] here until it is complete.
-       pending_monitor_updates: Vec<ChannelMonitorUpdate>,
+       pending_monitor_updates: Vec<PendingChannelMonitorUpdate>,
 }
 
 #[cfg(any(test, fuzzing))]
@@ -986,7 +1027,10 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                secp_ctx.seeded_randomize(&entropy_source.get_secure_random_bytes());
 
                let shutdown_scriptpubkey = if config.channel_handshake_config.commit_upfront_shutdown_pubkey {
-                       Some(signer_provider.get_shutdown_scriptpubkey())
+                       match signer_provider.get_shutdown_scriptpubkey() {
+                               Ok(scriptpubkey) => Some(scriptpubkey),
+                               Err(_) => return Err(APIError::ChannelUnavailable { err: "Failed to get shutdown scriptpubkey".to_owned()}),
+                       }
                } else { None };
 
                if let Some(shutdown_scriptpubkey) = &shutdown_scriptpubkey {
@@ -995,6 +1039,11 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        }
                }
 
+               let destination_script = match signer_provider.get_destination_script() {
+                       Ok(script) => script,
+                       Err(_) => return Err(APIError::ChannelUnavailable { err: "Failed to get destination script".to_owned()}),
+               };
+
                let temporary_channel_id = entropy_source.get_secure_random_bytes();
 
                Ok(Channel {
@@ -1021,7 +1070,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
 
                        holder_signer,
                        shutdown_scriptpubkey,
-                       destination_script: signer_provider.get_destination_script(),
+                       destination_script,
 
                        cur_holder_commitment_transaction_number: INITIAL_COMMITMENT_NUMBER,
                        cur_counterparty_commitment_transaction_number: INITIAL_COMMITMENT_NUMBER,
@@ -1107,6 +1156,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        next_remote_commitment_tx_fee_info_cached: Mutex::new(None),
 
                        workaround_lnd_bug_4006: None,
+                       sent_message_awaiting_response: None,
 
                        latest_inbound_scid_alias: None,
                        outbound_scid_alias,
@@ -1333,7 +1383,10 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                } else { None };
 
                let shutdown_scriptpubkey = if config.channel_handshake_config.commit_upfront_shutdown_pubkey {
-                       Some(signer_provider.get_shutdown_scriptpubkey())
+                       match signer_provider.get_shutdown_scriptpubkey() {
+                               Ok(scriptpubkey) => Some(scriptpubkey),
+                               Err(_) => return Err(ChannelError::Close("Failed to get upfront shutdown scriptpubkey".to_owned())),
+                       }
                } else { None };
 
                if let Some(shutdown_scriptpubkey) = &shutdown_scriptpubkey {
@@ -1342,6 +1395,11 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        }
                }
 
+               let destination_script = match signer_provider.get_destination_script() {
+                       Ok(script) => script,
+                       Err(_) => return Err(ChannelError::Close("Failed to get destination script".to_owned())),
+               };
+
                let mut secp_ctx = Secp256k1::new();
                secp_ctx.seeded_randomize(&entropy_source.get_secure_random_bytes());
 
@@ -1368,7 +1426,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
 
                        holder_signer,
                        shutdown_scriptpubkey,
-                       destination_script: signer_provider.get_destination_script(),
+                       destination_script,
 
                        cur_holder_commitment_transaction_number: INITIAL_COMMITMENT_NUMBER,
                        cur_counterparty_commitment_transaction_number: INITIAL_COMMITMENT_NUMBER,
@@ -1458,6 +1516,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        next_remote_commitment_tx_fee_info_cached: Mutex::new(None),
 
                        workaround_lnd_bug_4006: None,
+                       sent_message_awaiting_response: None,
 
                        latest_inbound_scid_alias: None,
                        outbound_scid_alias,
@@ -1979,28 +2038,52 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
        }
 
        pub fn get_update_fulfill_htlc_and_commit<L: Deref>(&mut self, htlc_id: u64, payment_preimage: PaymentPreimage, logger: &L) -> UpdateFulfillCommitFetch where L::Target: Logger {
+               let release_cs_monitor = self.pending_monitor_updates.iter().all(|upd| !upd.blocked);
                match self.get_update_fulfill_htlc(htlc_id, payment_preimage, logger) {
-                       UpdateFulfillFetch::NewClaim { mut monitor_update, htlc_value_msat, msg: Some(_) } => {
-                               let mut additional_update = self.build_commitment_no_status_check(logger);
-                               // build_commitment_no_status_check may bump latest_monitor_id but we want them to be
-                               // strictly increasing by one, so decrement it here.
-                               self.latest_monitor_update_id = monitor_update.update_id;
-                               monitor_update.updates.append(&mut additional_update.updates);
-                               self.monitor_updating_paused(false, true, false, Vec::new(), Vec::new(), Vec::new());
-                               self.pending_monitor_updates.push(monitor_update);
+                       UpdateFulfillFetch::NewClaim { mut monitor_update, htlc_value_msat, msg } => {
+                               // Even if we aren't supposed to let new monitor updates with commitment state
+                               // updates run, we still need to push the preimage ChannelMonitorUpdateStep no
+                               // matter what. Sadly, to push a new monitor update which flies before others
+                               // already queued, we have to insert it into the pending queue and update the
+                               // update_ids of all the following monitors.
+                               let unblocked_update_pos = if release_cs_monitor && msg.is_some() {
+                                       let mut additional_update = self.build_commitment_no_status_check(logger);
+                                       // build_commitment_no_status_check may bump latest_monitor_id but we want them
+                                       // to be strictly increasing by one, so decrement it here.
+                                       self.latest_monitor_update_id = monitor_update.update_id;
+                                       monitor_update.updates.append(&mut additional_update.updates);
+                                       self.pending_monitor_updates.push(PendingChannelMonitorUpdate {
+                                               update: monitor_update, blocked: false,
+                                       });
+                                       self.pending_monitor_updates.len() - 1
+                               } else {
+                                       let insert_pos = self.pending_monitor_updates.iter().position(|upd| upd.blocked)
+                                               .unwrap_or(self.pending_monitor_updates.len());
+                                       let new_mon_id = self.pending_monitor_updates.get(insert_pos)
+                                               .map(|upd| upd.update.update_id).unwrap_or(monitor_update.update_id);
+                                       monitor_update.update_id = new_mon_id;
+                                       self.pending_monitor_updates.insert(insert_pos, PendingChannelMonitorUpdate {
+                                               update: monitor_update, blocked: false,
+                                       });
+                                       for held_update in self.pending_monitor_updates.iter_mut().skip(insert_pos + 1) {
+                                               held_update.update.update_id += 1;
+                                       }
+                                       if msg.is_some() {
+                                               debug_assert!(false, "If there is a pending blocked monitor we should have MonitorUpdateInProgress set");
+                                               let update = self.build_commitment_no_status_check(logger);
+                                               self.pending_monitor_updates.push(PendingChannelMonitorUpdate {
+                                                       update, blocked: true,
+                                               });
+                                       }
+                                       insert_pos
+                               };
+                               self.monitor_updating_paused(false, msg.is_some(), false, Vec::new(), Vec::new(), Vec::new());
                                UpdateFulfillCommitFetch::NewClaim {
-                                       monitor_update: self.pending_monitor_updates.last().unwrap(),
+                                       monitor_update: &self.pending_monitor_updates.get(unblocked_update_pos)
+                                               .expect("We just pushed the monitor update").update,
                                        htlc_value_msat,
                                }
                        },
-                       UpdateFulfillFetch::NewClaim { monitor_update, htlc_value_msat, msg: None } => {
-                               self.monitor_updating_paused(false, false, false, Vec::new(), Vec::new(), Vec::new());
-                               self.pending_monitor_updates.push(monitor_update);
-                               UpdateFulfillCommitFetch::NewClaim {
-                                       monitor_update: self.pending_monitor_updates.last().unwrap(),
-                                       htlc_value_msat,
-                               }
-                       }
                        UpdateFulfillFetch::DuplicateClaim {} => UpdateFulfillCommitFetch::DuplicateClaim {},
                }
        }
@@ -3068,7 +3151,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                Ok(())
        }
 
-       pub fn commitment_signed<L: Deref>(&mut self, msg: &msgs::CommitmentSigned, logger: &L) -> Result<&ChannelMonitorUpdate, ChannelError>
+       pub fn commitment_signed<L: Deref>(&mut self, msg: &msgs::CommitmentSigned, logger: &L) -> Result<Option<&ChannelMonitorUpdate>, ChannelError>
                where L::Target: Logger
        {
                if (self.channel_state & (ChannelState::ChannelReady as u32)) != (ChannelState::ChannelReady as u32) {
@@ -3268,8 +3351,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        }
                        log_debug!(logger, "Received valid commitment_signed from peer in channel {}, updated HTLC state but awaiting a monitor update resolution to reply.",
                                log_bytes!(self.channel_id));
-                       self.pending_monitor_updates.push(monitor_update);
-                       return Ok(self.pending_monitor_updates.last().unwrap());
+                       return Ok(self.push_ret_blockable_mon_update(monitor_update));
                }
 
                let need_commitment_signed = if need_commitment && (self.channel_state & (ChannelState::AwaitingRemoteRevoke as u32)) == 0 {
@@ -3286,9 +3368,8 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
 
                log_debug!(logger, "Received valid commitment_signed from peer in channel {}, updating HTLC state and responding with{} a revoke_and_ack.",
                        log_bytes!(self.channel_id()), if need_commitment_signed { " our own commitment_signed and" } else { "" });
-               self.pending_monitor_updates.push(monitor_update);
                self.monitor_updating_paused(true, need_commitment_signed, false, Vec::new(), Vec::new(), Vec::new());
-               return Ok(self.pending_monitor_updates.last().unwrap());
+               return Ok(self.push_ret_blockable_mon_update(monitor_update));
        }
 
        /// Public version of the below, checking relevant preconditions first.
@@ -3403,8 +3484,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                                update_add_htlcs.len(), update_fulfill_htlcs.len(), update_fail_htlcs.len());
 
                        self.monitor_updating_paused(false, true, false, Vec::new(), Vec::new(), Vec::new());
-                       self.pending_monitor_updates.push(monitor_update);
-                       (Some(self.pending_monitor_updates.last().unwrap()), htlcs_to_fail)
+                       (self.push_ret_blockable_mon_update(monitor_update), htlcs_to_fail)
                } else {
                        (None, Vec::new())
                }
@@ -3415,7 +3495,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
        /// waiting on this revoke_and_ack. The generation of this new commitment_signed may also fail,
        /// generating an appropriate error *after* the channel state has been updated based on the
        /// revoke_and_ack message.
-       pub fn revoke_and_ack<L: Deref>(&mut self, msg: &msgs::RevokeAndACK, logger: &L) -> Result<(Vec<(HTLCSource, PaymentHash)>, &ChannelMonitorUpdate), ChannelError>
+       pub fn revoke_and_ack<L: Deref>(&mut self, msg: &msgs::RevokeAndACK, logger: &L) -> Result<(Vec<(HTLCSource, PaymentHash)>, Option<&ChannelMonitorUpdate>), ChannelError>
                where L::Target: Logger,
        {
                if (self.channel_state & (ChannelState::ChannelReady as u32)) != (ChannelState::ChannelReady as u32) {
@@ -3474,6 +3554,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                // OK, we step the channel here and *then* if the new generation fails we can fail the
                // channel based on that, but stepping stuff here should be safe either way.
                self.channel_state &= !(ChannelState::AwaitingRemoteRevoke as u32);
+               self.sent_message_awaiting_response = None;
                self.counterparty_prev_commitment_point = self.counterparty_cur_commitment_point;
                self.counterparty_cur_commitment_point = Some(msg.next_per_commitment_point);
                self.cur_counterparty_commitment_transaction_number -= 1;
@@ -3612,21 +3693,19 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        self.monitor_pending_failures.append(&mut revoked_htlcs);
                        self.monitor_pending_finalized_fulfills.append(&mut finalized_claimed_htlcs);
                        log_debug!(logger, "Received a valid revoke_and_ack for channel {} but awaiting a monitor update resolution to reply.", log_bytes!(self.channel_id()));
-                       self.pending_monitor_updates.push(monitor_update);
-                       return Ok((Vec::new(), self.pending_monitor_updates.last().unwrap()));
+                       return Ok((Vec::new(), self.push_ret_blockable_mon_update(monitor_update)));
                }
 
                match self.free_holding_cell_htlcs(logger) {
                        (Some(_), htlcs_to_fail) => {
-                               let mut additional_update = self.pending_monitor_updates.pop().unwrap();
+                               let mut additional_update = self.pending_monitor_updates.pop().unwrap().update;
                                // free_holding_cell_htlcs may bump latest_monitor_id multiple times but we want them to be
                                // strictly increasing by one, so decrement it here.
                                self.latest_monitor_update_id = monitor_update.update_id;
                                monitor_update.updates.append(&mut additional_update.updates);
 
                                self.monitor_updating_paused(false, true, false, to_forward_infos, revoked_htlcs, finalized_claimed_htlcs);
-                               self.pending_monitor_updates.push(monitor_update);
-                               Ok((htlcs_to_fail, self.pending_monitor_updates.last().unwrap()))
+                               Ok((htlcs_to_fail, self.push_ret_blockable_mon_update(monitor_update)))
                        },
                        (None, htlcs_to_fail) => {
                                if require_commitment {
@@ -3640,13 +3719,11 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                                        log_debug!(logger, "Received a valid revoke_and_ack for channel {}. Responding with a commitment update with {} HTLCs failed.",
                                                log_bytes!(self.channel_id()), update_fail_htlcs.len() + update_fail_malformed_htlcs.len());
                                        self.monitor_updating_paused(false, true, false, to_forward_infos, revoked_htlcs, finalized_claimed_htlcs);
-                                       self.pending_monitor_updates.push(monitor_update);
-                                       Ok((htlcs_to_fail, self.pending_monitor_updates.last().unwrap()))
+                                       Ok((htlcs_to_fail, self.push_ret_blockable_mon_update(monitor_update)))
                                } else {
                                        log_debug!(logger, "Received a valid revoke_and_ack for channel {} with no reply necessary.", log_bytes!(self.channel_id()));
                                        self.monitor_updating_paused(false, false, false, to_forward_infos, revoked_htlcs, finalized_claimed_htlcs);
-                                       self.pending_monitor_updates.push(monitor_update);
-                                       Ok((htlcs_to_fail, self.pending_monitor_updates.last().unwrap()))
+                                       Ok((htlcs_to_fail, self.push_ret_blockable_mon_update(monitor_update)))
                                }
                        }
                }
@@ -3793,6 +3870,8 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        }
                }
 
+               self.sent_message_awaiting_response = None;
+
                self.channel_state |= ChannelState::PeerDisconnected as u32;
                log_trace!(logger, "Peer disconnection resulted in {} remote-announced HTLC drops on channel {}", inbound_drop_count, log_bytes!(self.channel_id()));
        }
@@ -3835,7 +3914,12 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
        {
                assert_eq!(self.channel_state & ChannelState::MonitorUpdateInProgress as u32, ChannelState::MonitorUpdateInProgress as u32);
                self.channel_state &= !(ChannelState::MonitorUpdateInProgress as u32);
-               self.pending_monitor_updates.clear();
+               let mut found_blocked = false;
+               self.pending_monitor_updates.retain(|upd| {
+                       if found_blocked { debug_assert!(upd.blocked, "No mons may be unblocked after a blocked one"); }
+                       if upd.blocked { found_blocked = true; }
+                       upd.blocked
+               });
 
                // If we're past (or at) the FundingSent stage on an outbound channel, try to
                // (re-)broadcast the funding transaction as we may have declined to broadcast it when we
@@ -3890,6 +3974,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        Some(self.get_last_revoke_and_ack())
                } else { None };
                let commitment_update = if self.monitor_pending_commitment_signed {
+                       self.mark_awaiting_response();
                        Some(self.get_last_commitment_update(logger))
                } else { None };
 
@@ -4039,7 +4124,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
 
                if msg.next_local_commitment_number >= INITIAL_COMMITMENT_NUMBER || msg.next_remote_commitment_number >= INITIAL_COMMITMENT_NUMBER ||
                        msg.next_local_commitment_number == 0 {
-                       return Err(ChannelError::Close("Peer sent a garbage channel_reestablish".to_owned()));
+                       return Err(ChannelError::Close("Peer sent a garbage channel_reestablish (usually an lnd node with lost state asking us to force-close for them)".to_owned()));
                }
 
                if msg.next_remote_commitment_number > 0 {
@@ -4079,6 +4164,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                // Go ahead and unmark PeerDisconnected as various calls we may make check for it (and all
                // remaining cases either succeed or ErrorMessage-fail).
                self.channel_state &= !(ChannelState::PeerDisconnected as u32);
+               self.sent_message_awaiting_response = None;
 
                let shutdown_msg = if self.channel_state & (ChannelState::LocalShutdownSent as u32) != 0 {
                        assert!(self.shutdown_scriptpubkey.is_some());
@@ -4139,7 +4225,11 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                // revoke_and_ack, not on sending commitment_signed, so we add one if have
                // AwaitingRemoteRevoke set, which indicates we sent a commitment_signed but haven't gotten
                // the corresponding revoke_and_ack back yet.
-               let next_counterparty_commitment_number = INITIAL_COMMITMENT_NUMBER - self.cur_counterparty_commitment_transaction_number + if (self.channel_state & ChannelState::AwaitingRemoteRevoke as u32) != 0 { 1 } else { 0 };
+               let is_awaiting_remote_revoke = self.channel_state & ChannelState::AwaitingRemoteRevoke as u32 != 0;
+               if is_awaiting_remote_revoke && !self.is_awaiting_monitor_update() {
+                       self.mark_awaiting_response();
+               }
+               let next_counterparty_commitment_number = INITIAL_COMMITMENT_NUMBER - self.cur_counterparty_commitment_transaction_number + if is_awaiting_remote_revoke { 1 } else { 0 };
 
                let channel_ready = if msg.next_local_commitment_number == 1 && INITIAL_COMMITMENT_NUMBER - self.cur_holder_commitment_transaction_number == 1 {
                        // We should never have to worry about MonitorUpdateInProgress resending ChannelReady
@@ -4308,6 +4398,28 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                }), None))
        }
 
+       // Marks a channel as waiting for a response from the counterparty. If it's not received
+       // [`DISCONNECT_PEER_AWAITING_RESPONSE_TICKS`] after sending our own to them, then we'll attempt
+       // a reconnection.
+       fn mark_awaiting_response(&mut self) {
+               self.sent_message_awaiting_response = Some(0);
+       }
+
+       /// Determines whether we should disconnect the counterparty due to not receiving a response
+       /// within our expected timeframe.
+       ///
+       /// This should be called on every [`super::channelmanager::ChannelManager::timer_tick_occurred`].
+       pub fn should_disconnect_peer_awaiting_response(&mut self) -> bool {
+               let ticks_elapsed = if let Some(ticks_elapsed) = self.sent_message_awaiting_response.as_mut() {
+                       ticks_elapsed
+               } else {
+                       // Don't disconnect when we're not waiting on a response.
+                       return false;
+               };
+               *ticks_elapsed += 1;
+               *ticks_elapsed >= DISCONNECT_PEER_AWAITING_RESPONSE_TICKS
+       }
+
        pub fn shutdown<SP: Deref>(
                &mut self, signer_provider: &SP, their_features: &InitFeatures, msg: &msgs::Shutdown
        ) -> Result<(Option<msgs::Shutdown>, Option<&ChannelMonitorUpdate>, Vec<(HTLCSource, PaymentHash)>), ChannelError>
@@ -4350,7 +4462,10 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        Some(_) => false,
                        None => {
                                assert!(send_shutdown);
-                               let shutdown_scriptpubkey = signer_provider.get_shutdown_scriptpubkey();
+                               let shutdown_scriptpubkey = match signer_provider.get_shutdown_scriptpubkey() {
+                                       Ok(scriptpubkey) => scriptpubkey,
+                                       Err(_) => return Err(ChannelError::Close("Failed to get shutdown scriptpubkey".to_owned())),
+                               };
                                if !shutdown_scriptpubkey.is_compatible(their_features) {
                                        return Err(ChannelError::Close(format!("Provided a scriptpubkey format not accepted by peer: {}", shutdown_scriptpubkey)));
                                }
@@ -4373,8 +4488,9 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                                }],
                        };
                        self.monitor_updating_paused(false, false, false, Vec::new(), Vec::new(), Vec::new());
-                       self.pending_monitor_updates.push(monitor_update);
-                       Some(self.pending_monitor_updates.last().unwrap())
+                       if self.push_blockable_mon_update(monitor_update) {
+                               self.pending_monitor_updates.last().map(|upd| &upd.update)
+                       } else { None }
                } else { None };
                let shutdown = if send_shutdown {
                        Some(msgs::Shutdown {
@@ -4946,8 +5062,64 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                (self.channel_state & ChannelState::MonitorUpdateInProgress as u32) != 0
        }
 
-       pub fn get_next_monitor_update(&self) -> Option<&ChannelMonitorUpdate> {
-               self.pending_monitor_updates.first()
+       pub fn get_latest_complete_monitor_update_id(&self) -> u64 {
+               if self.pending_monitor_updates.is_empty() { return self.get_latest_monitor_update_id(); }
+               self.pending_monitor_updates[0].update.update_id - 1
+       }
+
+       /// Returns the next blocked monitor update, if one exists, and a bool which indicates a
+       /// further blocked monitor update exists after the next.
+       pub fn unblock_next_blocked_monitor_update(&mut self) -> Option<(&ChannelMonitorUpdate, bool)> {
+               for i in 0..self.pending_monitor_updates.len() {
+                       if self.pending_monitor_updates[i].blocked {
+                               self.pending_monitor_updates[i].blocked = false;
+                               return Some((&self.pending_monitor_updates[i].update,
+                                       self.pending_monitor_updates.len() > i + 1));
+                       }
+               }
+               None
+       }
+
+       /// Pushes a new monitor update into our monitor update queue, returning whether it should be
+       /// immediately given to the user for persisting or if it should be held as blocked.
+       fn push_blockable_mon_update(&mut self, update: ChannelMonitorUpdate) -> bool {
+               let release_monitor = self.pending_monitor_updates.iter().all(|upd| !upd.blocked);
+               self.pending_monitor_updates.push(PendingChannelMonitorUpdate {
+                       update, blocked: !release_monitor
+               });
+               release_monitor
+       }
+
+       /// Pushes a new monitor update into our monitor update queue, returning a reference to it if
+       /// it should be immediately given to the user for persisting or `None` if it should be held as
+       /// blocked.
+       fn push_ret_blockable_mon_update(&mut self, update: ChannelMonitorUpdate)
+       -> Option<&ChannelMonitorUpdate> {
+               let release_monitor = self.push_blockable_mon_update(update);
+               if release_monitor { self.pending_monitor_updates.last().map(|upd| &upd.update) } else { None }
+       }
+
+       pub fn no_monitor_updates_pending(&self) -> bool {
+               self.pending_monitor_updates.is_empty()
+       }
+
+       pub fn complete_all_mon_updates_through(&mut self, update_id: u64) {
+               self.pending_monitor_updates.retain(|upd| {
+                       if upd.update.update_id <= update_id {
+                               assert!(!upd.blocked, "Completed update must have flown");
+                               false
+                       } else { true }
+               });
+       }
+
+       pub fn complete_one_mon_update(&mut self, update_id: u64) {
+               self.pending_monitor_updates.retain(|upd| upd.update.update_id != update_id);
+       }
+
+       /// Returns an iterator over all unblocked monitor updates which have not yet completed.
+       pub fn uncompleted_unblocked_mon_updates(&self) -> impl Iterator<Item=&ChannelMonitorUpdate> {
+               self.pending_monitor_updates.iter()
+                       .filter_map(|upd| if upd.blocked { None } else { Some(&upd.update) })
        }
 
        /// Returns true if funding_created was sent/received.
@@ -5635,7 +5807,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
 
        /// May panic if called on a channel that wasn't immediately-previously
        /// self.remove_uncommitted_htlcs_and_mark_paused()'d
-       pub fn get_channel_reestablish<L: Deref>(&self, logger: &L) -> msgs::ChannelReestablish where L::Target: Logger {
+       pub fn get_channel_reestablish<L: Deref>(&mut self, logger: &L) -> msgs::ChannelReestablish where L::Target: Logger {
                assert_eq!(self.channel_state & ChannelState::PeerDisconnected as u32, ChannelState::PeerDisconnected as u32);
                assert_ne!(self.cur_counterparty_commitment_transaction_number, INITIAL_COMMITMENT_NUMBER);
                // Prior to static_remotekey, my_current_per_commitment_point was critical to claiming
@@ -5654,6 +5826,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        log_info!(logger, "Sending a data_loss_protect with no previous remote per_commitment_secret for channel {}", log_bytes!(self.channel_id()));
                        [0;32]
                };
+               self.mark_awaiting_response();
                msgs::ChannelReestablish {
                        channel_id: self.channel_id(),
                        // The protocol has two different commitment number concepts - the "commitment
@@ -5676,6 +5849,10 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        next_remote_commitment_number: INITIAL_COMMITMENT_NUMBER - self.cur_counterparty_commitment_transaction_number - 1,
                        your_last_per_commitment_secret: remote_last_secret,
                        my_current_per_commitment_point: dummy_pubkey,
+                       // TODO(dual_funding): If we've sent `commtiment_signed` for an interactive transaction
+                       // construction but have not received `tx_signatures` we MUST set `next_funding_txid` to the
+                       // txid of that interactive transaction, else we MUST NOT set it.
+                       next_funding_txid: None,
                }
        }
 
@@ -5990,8 +6167,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        Some(_) => {
                                let monitor_update = self.build_commitment_no_status_check(logger);
                                self.monitor_updating_paused(false, true, false, Vec::new(), Vec::new(), Vec::new());
-                               self.pending_monitor_updates.push(monitor_update);
-                               Ok(Some(self.pending_monitor_updates.last().unwrap()))
+                               Ok(self.push_ret_blockable_mon_update(monitor_update))
                        },
                        None => Ok(None)
                }
@@ -6021,7 +6197,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
        /// May jump to the channel being fully shutdown (see [`Self::is_shutdown`]) in which case no
        /// [`ChannelMonitorUpdate`] will be returned).
        pub fn get_shutdown<SP: Deref>(&mut self, signer_provider: &SP, their_features: &InitFeatures,
-               target_feerate_sats_per_kw: Option<u32>)
+               target_feerate_sats_per_kw: Option<u32>, override_shutdown_script: Option<ShutdownScript>)
        -> Result<(msgs::Shutdown, Option<&ChannelMonitorUpdate>, Vec<(HTLCSource, PaymentHash)>), APIError>
        where SP::Target: SignerProvider {
                for htlc in self.pending_outbound_htlcs.iter() {
@@ -6037,6 +6213,9 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                                return Err(APIError::ChannelUnavailable{err: "Shutdown already in progress by remote".to_owned()});
                        }
                }
+               if self.shutdown_scriptpubkey.is_some() && override_shutdown_script.is_some() {
+                       return Err(APIError::APIMisuseError{err: "Cannot override shutdown script for a channel with one already set".to_owned()});
+               }
                assert_eq!(self.channel_state & ChannelState::ShutdownComplete as u32, 0);
                if self.channel_state & (ChannelState::PeerDisconnected as u32 | ChannelState::MonitorUpdateInProgress as u32) != 0 {
                        return Err(APIError::ChannelUnavailable{err: "Cannot begin shutdown while peer is disconnected or we're waiting on a monitor update, maybe force-close instead?".to_owned()});
@@ -6052,7 +6231,17 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                let update_shutdown_script = match self.shutdown_scriptpubkey {
                        Some(_) => false,
                        None if !chan_closed => {
-                               let shutdown_scriptpubkey = signer_provider.get_shutdown_scriptpubkey();
+                               // use override shutdown script if provided
+                               let shutdown_scriptpubkey = match override_shutdown_script {
+                                       Some(script) => script,
+                                       None => {
+                                               // otherwise, use the shutdown scriptpubkey provided by the signer
+                                               match signer_provider.get_shutdown_scriptpubkey() {
+                                                       Ok(scriptpubkey) => scriptpubkey,
+                                                       Err(_) => return Err(APIError::ChannelUnavailable{err: "Failed to get shutdown scriptpubkey".to_owned()}),
+                                               }
+                                       },
+                               };
                                if !shutdown_scriptpubkey.is_compatible(their_features) {
                                        return Err(APIError::IncompatibleShutdownScript { script: shutdown_scriptpubkey.clone() });
                                }
@@ -6080,8 +6269,9 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                                }],
                        };
                        self.monitor_updating_paused(false, false, false, Vec::new(), Vec::new(), Vec::new());
-                       self.pending_monitor_updates.push(monitor_update);
-                       Some(self.pending_monitor_updates.last().unwrap())
+                       if self.push_blockable_mon_update(monitor_update) {
+                               self.pending_monitor_updates.last().map(|upd| &upd.update)
+                       } else { None }
                } else { None };
                let shutdown = msgs::Shutdown {
                        channel_id: self.channel_id,
@@ -6113,7 +6303,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
        /// those explicitly stated to be allowed after shutdown completes, eg some simple getters).
        /// Also returns the list of payment_hashes for channels which we can safely fail backwards
        /// immediately (others we will have to allow to time out).
-       pub fn force_shutdown(&mut self, should_broadcast: bool) -> (Option<(OutPoint, ChannelMonitorUpdate)>, Vec<(HTLCSource, PaymentHash, PublicKey, [u8; 32])>) {
+       pub fn force_shutdown(&mut self, should_broadcast: bool) -> ShutdownResult {
                // Note that we MUST only generate a monitor update that indicates force-closure - we're
                // called during initialization prior to the chain_monitor in the encompassing ChannelManager
                // being fully configured in some cases. Thus, its likely any monitor events we generate will
@@ -6142,7 +6332,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Channel<Signer> {
                        // See test_duplicate_chan_id and test_pre_lockin_no_chan_closed_update for more.
                        if self.channel_state & (ChannelState::FundingSent as u32 | ChannelState::ChannelReady as u32 | ChannelState::ShutdownComplete as u32) != 0 {
                                self.latest_monitor_update_id = CLOSED_CHANNEL_UPDATE_ID;
-                               Some((funding_txo, ChannelMonitorUpdate {
+                               Some((self.get_counterparty_node_id(), funding_txo, ChannelMonitorUpdate {
                                        update_id: self.latest_monitor_update_id,
                                        updates: vec![ChannelMonitorUpdateStep::ChannelForceClosed { should_broadcast }],
                                }))
@@ -6518,6 +6708,7 @@ impl<Signer: WriteableEcdsaChannelSigner> Writeable for Channel<Signer> {
                        (28, holder_max_accepted_htlcs, option),
                        (29, self.temporary_channel_id, option),
                        (31, channel_pending_event_emitted, option),
+                       (33, self.pending_monitor_updates, vec_type),
                });
 
                Ok(())
@@ -6794,6 +6985,8 @@ impl<'a, 'b, 'c, ES: Deref, SP: Deref> ReadableArgs<(&'a ES, &'b SP, u32, &'c Ch
                let mut temporary_channel_id: Option<[u8; 32]> = None;
                let mut holder_max_accepted_htlcs: Option<u16> = None;
 
+               let mut pending_monitor_updates = Some(Vec::new());
+
                read_tlv_fields!(reader, {
                        (0, announcement_sigs, option),
                        (1, minimum_depth, option),
@@ -6816,6 +7009,7 @@ impl<'a, 'b, 'c, ES: Deref, SP: Deref> ReadableArgs<(&'a ES, &'b SP, u32, &'c Ch
                        (28, holder_max_accepted_htlcs, option),
                        (29, temporary_channel_id, option),
                        (31, channel_pending_event_emitted, option),
+                       (33, pending_monitor_updates, vec_type),
                });
 
                let (channel_keys_id, holder_signer) = if let Some(channel_keys_id) = channel_keys_id {
@@ -6971,6 +7165,7 @@ impl<'a, 'b, 'c, ES: Deref, SP: Deref> ReadableArgs<(&'a ES, &'b SP, u32, &'c Ch
                        next_remote_commitment_tx_fee_info_cached: Mutex::new(None),
 
                        workaround_lnd_bug_4006: None,
+                       sent_message_awaiting_response: None,
 
                        latest_inbound_scid_alias,
                        // Later in the ChannelManager deserialization phase we scan for channels and assign scid aliases if its missing
@@ -6985,7 +7180,7 @@ impl<'a, 'b, 'c, ES: Deref, SP: Deref> ReadableArgs<(&'a ES, &'b SP, u32, &'c Ch
                        channel_type: channel_type.unwrap(),
                        channel_keys_id,
 
-                       pending_monitor_updates: Vec::new(),
+                       pending_monitor_updates: pending_monitor_updates.unwrap(),
                })
        }
 }
@@ -7012,7 +7207,7 @@ mod tests {
        use crate::ln::chan_utils::{htlc_success_tx_weight, htlc_timeout_tx_weight};
        use crate::chain::BestBlock;
        use crate::chain::chaininterface::{FeeEstimator, LowerBoundedFeeEstimator, ConfirmationTarget};
-       use crate::chain::keysinterface::{ChannelSigner, InMemorySigner, EntropySource, SignerProvider};
+       use crate::sign::{ChannelSigner, InMemorySigner, EntropySource, SignerProvider};
        use crate::chain::transaction::OutPoint;
        use crate::routing::router::Path;
        use crate::util::config::UserConfig;
@@ -7077,17 +7272,17 @@ mod tests {
 
                fn read_chan_signer(&self, _data: &[u8]) -> Result<Self::Signer, DecodeError> { panic!(); }
 
-               fn get_destination_script(&self) -> Script {
+               fn get_destination_script(&self) -> Result<Script, ()> {
                        let secp_ctx = Secp256k1::signing_only();
                        let channel_monitor_claim_key = SecretKey::from_slice(&hex::decode("0fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff").unwrap()[..]).unwrap();
                        let channel_monitor_claim_key_hash = WPubkeyHash::hash(&PublicKey::from_secret_key(&secp_ctx, &channel_monitor_claim_key).serialize());
-                       Builder::new().push_opcode(opcodes::all::OP_PUSHBYTES_0).push_slice(&channel_monitor_claim_key_hash[..]).into_script()
+                       Ok(Builder::new().push_opcode(opcodes::all::OP_PUSHBYTES_0).push_slice(&channel_monitor_claim_key_hash[..]).into_script())
                }
 
-               fn get_shutdown_scriptpubkey(&self) -> ShutdownScript {
+               fn get_shutdown_scriptpubkey(&self) -> Result<ShutdownScript, ()> {
                        let secp_ctx = Secp256k1::signing_only();
                        let channel_close_key = SecretKey::from_slice(&hex::decode("0fffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffff").unwrap()[..]).unwrap();
-                       ShutdownScript::new_p2wpkh_from_pubkey(PublicKey::from_secret_key(&secp_ctx, &channel_close_key))
+                       Ok(ShutdownScript::new_p2wpkh_from_pubkey(PublicKey::from_secret_key(&secp_ctx, &channel_close_key)))
                }
        }
 
@@ -7506,7 +7701,7 @@ mod tests {
                use bitcoin::hashes::hex::FromHex;
                use bitcoin::hash_types::Txid;
                use bitcoin::secp256k1::Message;
-               use crate::chain::keysinterface::EcdsaChannelSigner;
+               use crate::sign::EcdsaChannelSigner;
                use crate::ln::PaymentPreimage;
                use crate::ln::channel::{HTLCOutputInCommitment ,TxCreationKeys};
                use crate::ln::chan_utils::{ChannelPublicKeys, HolderCommitmentTransaction, CounterpartyChannelTransactionParameters};