Merge pull request #2626 from TheBlueMatt/2023-09-revert-2476
[rust-lightning] / lightning / src / chain / chainmonitor.rs
index e7c2b0f18ec001d703eb0249964b90a8a3f1a5ed..cf51dbab72019d9eb87748e0817171c1c0abfb35 100644 (file)
@@ -31,7 +31,7 @@ use crate::chain::{ChannelMonitorUpdateStatus, Filter, WatchedOutput};
 use crate::chain::chaininterface::{BroadcasterInterface, FeeEstimator};
 use crate::chain::channelmonitor::{ChannelMonitor, ChannelMonitorUpdate, Balance, MonitorEvent, TransactionOutputs, LATENCY_GRACE_PERIOD_BLOCKS};
 use crate::chain::transaction::{OutPoint, TransactionData};
-use crate::chain::keysinterface::WriteableEcdsaChannelSigner;
+use crate::sign::WriteableEcdsaChannelSigner;
 use crate::events;
 use crate::events::{Event, EventHandler};
 use crate::util::atomic_counter::AtomicCounter;
@@ -42,11 +42,12 @@ use crate::ln::channelmanager::ChannelDetails;
 
 use crate::prelude::*;
 use crate::sync::{RwLock, RwLockReadGuard, Mutex, MutexGuard};
+use core::iter::FromIterator;
 use core::ops::Deref;
-use core::sync::atomic::{AtomicBool, AtomicUsize, Ordering};
+use core::sync::atomic::{AtomicUsize, Ordering};
 use bitcoin::secp256k1::PublicKey;
 
-#[derive(Clone, Copy, Hash, PartialEq, Eq)]
+#[derive(Debug, Clone, Copy, Hash, PartialEq, Eq)]
 /// A specific update's ID stored in a `MonitorUpdateId`, separated out to make the contents
 /// entirely opaque.
 enum UpdateOrigin {
@@ -60,7 +61,7 @@ enum UpdateOrigin {
 }
 
 /// An opaque identifier describing a specific [`Persist`] method call.
-#[derive(Clone, Copy, Hash, PartialEq, Eq)]
+#[derive(Debug, Clone, Copy, Hash, PartialEq, Eq)]
 pub struct MonitorUpdateId {
        contents: UpdateOrigin,
 }
@@ -77,23 +78,58 @@ impl MonitorUpdateId {
 /// `Persist` defines behavior for persisting channel monitors: this could mean
 /// writing once to disk, and/or uploading to one or more backup services.
 ///
-/// Each method can return three possible values:
-///  * If persistence (including any relevant `fsync()` calls) happens immediately, the
-///    implementation should return [`ChannelMonitorUpdateStatus::Completed`], indicating normal
-///    channel operation should continue.
-///  * If persistence happens asynchronously, implementations should first ensure the
-///    [`ChannelMonitor`] or [`ChannelMonitorUpdate`] are written durably to disk, and then return
-///    [`ChannelMonitorUpdateStatus::InProgress`] while the update continues in the background.
-///    Once the update completes, [`ChainMonitor::channel_monitor_updated`] should be called with
-///    the corresponding [`MonitorUpdateId`].
+/// Persistence can happen in one of two ways - synchronously completing before the trait method
+/// calls return or asynchronously in the background.
 ///
-///    Note that unlike the direct [`chain::Watch`] interface,
-///    [`ChainMonitor::channel_monitor_updated`] must be called once for *each* update which occurs.
+/// # For those implementing synchronous persistence
 ///
-///  * If persistence fails for some reason, implementations should return
-///    [`ChannelMonitorUpdateStatus::PermanentFailure`], in which case the channel will likely be
-///    closed without broadcasting the latest state. See
-///    [`ChannelMonitorUpdateStatus::PermanentFailure`] for more details.
+///  * If persistence completes fully (including any relevant `fsync()` calls), the implementation
+///    should return [`ChannelMonitorUpdateStatus::Completed`], indicating normal channel operation
+///    should continue.
+///
+///  * If persistence fails for some reason, implementations should consider returning
+///    [`ChannelMonitorUpdateStatus::InProgress`] and retry all pending persistence operations in
+///    the background with [`ChainMonitor::list_pending_monitor_updates`] and
+///    [`ChainMonitor::get_monitor`].
+///
+///    Once a full [`ChannelMonitor`] has been persisted, all pending updates for that channel can
+///    be marked as complete via [`ChainMonitor::channel_monitor_updated`].
+///
+///    If at some point no further progress can be made towards persisting the pending updates, the
+///    node should simply shut down.
+///
+///  * If the persistence has failed and cannot be retried further (e.g. because of an outage),
+///    [`ChannelMonitorUpdateStatus::UnrecoverableError`] can be used, though this will result in
+///    an immediate panic and future operations in LDK generally failing.
+///
+/// # For those implementing asynchronous persistence
+///
+///  All calls should generally spawn a background task and immediately return
+///  [`ChannelMonitorUpdateStatus::InProgress`]. Once the update completes,
+///  [`ChainMonitor::channel_monitor_updated`] should be called with the corresponding
+///  [`MonitorUpdateId`].
+///
+///  Note that unlike the direct [`chain::Watch`] interface,
+///  [`ChainMonitor::channel_monitor_updated`] must be called once for *each* update which occurs.
+///
+///  If at some point no further progress can be made towards persisting a pending update, the node
+///  should simply shut down. Until then, the background task should either loop indefinitely, or
+///  persistence should be regularly retried with [`ChainMonitor::list_pending_monitor_updates`]
+///  and [`ChainMonitor::get_monitor`] (note that if a full monitor is persisted all pending
+///  monitor updates may be marked completed).
+///
+/// # Using remote watchtowers
+///
+/// Watchtowers may be updated as a part of an implementation of this trait, utilizing the async
+/// update process described above while the watchtower is being updated. The following methods are
+/// provided for bulding transactions for a watchtower:
+/// [`ChannelMonitor::initial_counterparty_commitment_tx`],
+/// [`ChannelMonitor::counterparty_commitment_txs_from_update`],
+/// [`ChannelMonitor::sign_to_local_justice_tx`], [`TrustedCommitmentTransaction::revokeable_output_index`],
+/// [`TrustedCommitmentTransaction::build_to_local_justice_tx`].
+///
+/// [`TrustedCommitmentTransaction::revokeable_output_index`]: crate::ln::chan_utils::TrustedCommitmentTransaction::revokeable_output_index
+/// [`TrustedCommitmentTransaction::build_to_local_justice_tx`]: crate::ln::chan_utils::TrustedCommitmentTransaction::build_to_local_justice_tx
 pub trait Persist<ChannelSigner: WriteableEcdsaChannelSigner> {
        /// Persist a new channel's data in response to a [`chain::Watch::watch_channel`] call. This is
        /// called by [`ChannelManager`] for new channels, or may be called directly, e.g. on startup.
@@ -168,12 +204,6 @@ struct MonitorHolder<ChannelSigner: WriteableEcdsaChannelSigner> {
        /// the ChannelManager re-adding the same payment entry, before the same block is replayed,
        /// resulting in a duplicate PaymentSent event.
        pending_monitor_updates: Mutex<Vec<MonitorUpdateId>>,
-       /// When the user returns a PermanentFailure error from an update_persisted_channel call during
-       /// block processing, we inform the ChannelManager that the channel should be closed
-       /// asynchronously. In order to ensure no further changes happen before the ChannelManager has
-       /// processed the closure event, we set this to true and return PermanentFailure for any other
-       /// chain::Watch events.
-       channel_perm_failed: AtomicBool,
        /// The last block height at which no [`UpdateOrigin::ChainSync`] monitor updates were present
        /// in `pending_monitor_updates`.
        /// If it's been more than [`LATENCY_GRACE_PERIOD_BLOCKS`] since we started waiting on a chain
@@ -274,7 +304,34 @@ where C::Target: chain::Filter,
        where
                FN: Fn(&ChannelMonitor<ChannelSigner>, &TransactionData) -> Vec<TransactionOutputs>
        {
+               let err_str = "ChannelMonitor[Update] persistence failed unrecoverably. This indicates we cannot continue normal operation and must shut down.";
+               let funding_outpoints: HashSet<OutPoint> = HashSet::from_iter(self.monitors.read().unwrap().keys().cloned());
+               for funding_outpoint in funding_outpoints.iter() {
+                       let monitor_lock = self.monitors.read().unwrap();
+                       if let Some(monitor_state) = monitor_lock.get(funding_outpoint) {
+                               if self.update_monitor_with_chain_data(header, best_height, txdata, &process, funding_outpoint, &monitor_state).is_err() {
+                                       // Take the monitors lock for writing so that we poison it and any future
+                                       // operations going forward fail immediately.
+                                       core::mem::drop(monitor_state);
+                                       core::mem::drop(monitor_lock);
+                                       let _poison = self.monitors.write().unwrap();
+                                       log_error!(self.logger, "{}", err_str);
+                                       panic!("{}", err_str);
+                               }
+                       }
+               }
+
+               // do some followup cleanup if any funding outpoints were added in between iterations
                let monitor_states = self.monitors.write().unwrap();
+               for (funding_outpoint, monitor_state) in monitor_states.iter() {
+                       if !funding_outpoints.contains(funding_outpoint) {
+                               if self.update_monitor_with_chain_data(header, best_height, txdata, &process, funding_outpoint, &monitor_state).is_err() {
+                                       log_error!(self.logger, "{}", err_str);
+                                       panic!("{}", err_str);
+                               }
+                       }
+               }
+
                if let Some(height) = best_height {
                        // If the best block height is being updated, update highest_chain_height under the
                        // monitors write lock.
@@ -284,58 +341,60 @@ where C::Target: chain::Filter,
                                self.highest_chain_height.store(new_height, Ordering::Release);
                        }
                }
+       }
 
-               for (funding_outpoint, monitor_state) in monitor_states.iter() {
-                       let monitor = &monitor_state.monitor;
-                       let mut txn_outputs;
-                       {
-                               txn_outputs = process(monitor, txdata);
-                               let update_id = MonitorUpdateId {
-                                       contents: UpdateOrigin::ChainSync(self.sync_persistence_id.get_increment()),
-                               };
-                               let mut pending_monitor_updates = monitor_state.pending_monitor_updates.lock().unwrap();
-                               if let Some(height) = best_height {
-                                       if !monitor_state.has_pending_chainsync_updates(&pending_monitor_updates) {
-                                               // If there are not ChainSync persists awaiting completion, go ahead and
-                                               // set last_chain_persist_height here - we wouldn't want the first
-                                               // InProgress to always immediately be considered "overly delayed".
-                                               monitor_state.last_chain_persist_height.store(height as usize, Ordering::Release);
-                                       }
+       fn update_monitor_with_chain_data<FN>(
+               &self, header: &BlockHeader, best_height: Option<u32>, txdata: &TransactionData,
+               process: FN, funding_outpoint: &OutPoint, monitor_state: &MonitorHolder<ChannelSigner>
+       ) -> Result<(), ()> where FN: Fn(&ChannelMonitor<ChannelSigner>, &TransactionData) -> Vec<TransactionOutputs> {
+               let monitor = &monitor_state.monitor;
+               let mut txn_outputs;
+               {
+                       txn_outputs = process(monitor, txdata);
+                       let update_id = MonitorUpdateId {
+                               contents: UpdateOrigin::ChainSync(self.sync_persistence_id.get_increment()),
+                       };
+                       let mut pending_monitor_updates = monitor_state.pending_monitor_updates.lock().unwrap();
+                       if let Some(height) = best_height {
+                               if !monitor_state.has_pending_chainsync_updates(&pending_monitor_updates) {
+                                       // If there are not ChainSync persists awaiting completion, go ahead and
+                                       // set last_chain_persist_height here - we wouldn't want the first
+                                       // InProgress to always immediately be considered "overly delayed".
+                                       monitor_state.last_chain_persist_height.store(height as usize, Ordering::Release);
                                }
+                       }
 
-                               log_trace!(self.logger, "Syncing Channel Monitor for channel {}", log_funding_info!(monitor));
-                               match self.persister.update_persisted_channel(*funding_outpoint, None, monitor, update_id) {
-                                       ChannelMonitorUpdateStatus::Completed =>
-                                               log_trace!(self.logger, "Finished syncing Channel Monitor for channel {}", log_funding_info!(monitor)),
-                                       ChannelMonitorUpdateStatus::PermanentFailure => {
-                                               monitor_state.channel_perm_failed.store(true, Ordering::Release);
-                                               self.pending_monitor_events.lock().unwrap().push((*funding_outpoint, vec![MonitorEvent::UpdateFailed(*funding_outpoint)], monitor.get_counterparty_node_id()));
-                                               self.event_notifier.notify();
-                                       },
-                                       ChannelMonitorUpdateStatus::InProgress => {
-                                               log_debug!(self.logger, "Channel Monitor sync for channel {} in progress, holding events until completion!", log_funding_info!(monitor));
-                                               pending_monitor_updates.push(update_id);
-                                       },
-                               }
+                       log_trace!(self.logger, "Syncing Channel Monitor for channel {}", log_funding_info!(monitor));
+                       match self.persister.update_persisted_channel(*funding_outpoint, None, monitor, update_id) {
+                               ChannelMonitorUpdateStatus::Completed =>
+                                       log_trace!(self.logger, "Finished syncing Channel Monitor for channel {}", log_funding_info!(monitor)),
+                               ChannelMonitorUpdateStatus::InProgress => {
+                                       log_debug!(self.logger, "Channel Monitor sync for channel {} in progress, holding events until completion!", log_funding_info!(monitor));
+                                       pending_monitor_updates.push(update_id);
+                               },
+                               ChannelMonitorUpdateStatus::UnrecoverableError => {
+                                       return Err(());
+                               },
                        }
+               }
 
-                       // Register any new outputs with the chain source for filtering, storing any dependent
-                       // transactions from within the block that previously had not been included in txdata.
-                       if let Some(ref chain_source) = self.chain_source {
-                               let block_hash = header.block_hash();
-                               for (txid, mut outputs) in txn_outputs.drain(..) {
-                                       for (idx, output) in outputs.drain(..) {
-                                               // Register any new outputs with the chain source for filtering
-                                               let output = WatchedOutput {
-                                                       block_hash: Some(block_hash),
-                                                       outpoint: OutPoint { txid, index: idx as u16 },
-                                                       script_pubkey: output.script_pubkey,
-                                               };
-                                               chain_source.register_output(output)
-                                       }
+               // Register any new outputs with the chain source for filtering, storing any dependent
+               // transactions from within the block that previously had not been included in txdata.
+               if let Some(ref chain_source) = self.chain_source {
+                       let block_hash = header.block_hash();
+                       for (txid, mut outputs) in txn_outputs.drain(..) {
+                               for (idx, output) in outputs.drain(..) {
+                                       // Register any new outputs with the chain source for filtering
+                                       let output = WatchedOutput {
+                                               block_hash: Some(block_hash),
+                                               outpoint: OutPoint { txid, index: idx as u16 },
+                                               script_pubkey: output.script_pubkey,
+                                       };
+                                       chain_source.register_output(output)
                                }
                        }
                }
+               Ok(())
        }
 
        /// Creates a new `ChainMonitor` used to watch on-chain activity pertaining to channels.
@@ -465,9 +524,8 @@ where C::Target: chain::Filter,
                                // `MonitorEvent`s from the monitor back to the `ChannelManager` until they
                                // complete.
                                let monitor_is_pending_updates = monitor_data.has_pending_offchain_updates(&pending_monitor_updates);
-                               if monitor_is_pending_updates || monitor_data.channel_perm_failed.load(Ordering::Acquire) {
-                                       // If there are still monitor updates pending (or an old monitor update
-                                       // finished after a later one perm-failed), we cannot yet construct an
+                               if monitor_is_pending_updates {
+                                       // If there are still monitor updates pending, we cannot yet construct a
                                        // Completed event.
                                        return Ok(());
                                }
@@ -502,7 +560,7 @@ where C::Target: chain::Filter,
                self.event_notifier.notify();
        }
 
-       #[cfg(any(test, fuzzing, feature = "_test_utils"))]
+       #[cfg(any(test, feature = "_test_utils"))]
        pub fn get_and_clear_pending_events(&self) -> Vec<events::Event> {
                use crate::events::EventsProvider;
                let events = core::cell::RefCell::new(Vec::new());
@@ -520,12 +578,13 @@ where C::Target: chain::Filter,
        pub async fn process_pending_events_async<Future: core::future::Future, H: Fn(Event) -> Future>(
                &self, handler: H
        ) {
-               let mut pending_events = Vec::new();
-               for monitor_state in self.monitors.read().unwrap().values() {
-                       pending_events.append(&mut monitor_state.monitor.get_and_clear_pending_events());
-               }
-               for event in pending_events {
-                       handler(event).await;
+               // Sadly we can't hold the monitors read lock through an async call. Thus we have to do a
+               // crazy dance to process a monitor's events then only remove them once we've done so.
+               let mons_to_process = self.monitors.read().unwrap().keys().cloned().collect::<Vec<_>>();
+               for funding_txo in mons_to_process {
+                       let mut ev;
+                       super::channelmonitor::process_events_body!(
+                               self.monitors.read().unwrap().get(&funding_txo).map(|m| &m.monitor), ev, handler(ev).await);
                }
        }
 
@@ -640,18 +699,12 @@ where C::Target: chain::Filter,
            L::Target: Logger,
            P::Target: Persist<ChannelSigner>,
 {
-       /// Adds the monitor that watches the channel referred to by the given outpoint.
-       ///
-       /// Calls back to [`chain::Filter`] with the funding transaction and outputs to watch.
-       ///
-       /// Note that we persist the given `ChannelMonitor` while holding the `ChainMonitor`
-       /// monitors lock.
-       fn watch_channel(&self, funding_outpoint: OutPoint, monitor: ChannelMonitor<ChannelSigner>) -> ChannelMonitorUpdateStatus {
+       fn watch_channel(&self, funding_outpoint: OutPoint, monitor: ChannelMonitor<ChannelSigner>) -> Result<ChannelMonitorUpdateStatus, ()> {
                let mut monitors = self.monitors.write().unwrap();
                let entry = match monitors.entry(funding_outpoint) {
                        hash_map::Entry::Occupied(_) => {
                                log_error!(self.logger, "Failed to add new channel data: channel monitor for given outpoint is already present");
-                               return ChannelMonitorUpdateStatus::PermanentFailure
+                               return Err(());
                        },
                        hash_map::Entry::Vacant(e) => e,
                };
@@ -664,13 +717,14 @@ where C::Target: chain::Filter,
                                log_info!(self.logger, "Persistence of new ChannelMonitor for channel {} in progress", log_funding_info!(monitor));
                                pending_monitor_updates.push(update_id);
                        },
-                       ChannelMonitorUpdateStatus::PermanentFailure => {
-                               log_error!(self.logger, "Persistence of new ChannelMonitor for channel {} failed", log_funding_info!(monitor));
-                               return persist_res;
-                       },
                        ChannelMonitorUpdateStatus::Completed => {
                                log_info!(self.logger, "Persistence of new ChannelMonitor for channel {} completed", log_funding_info!(monitor));
-                       }
+                       },
+                       ChannelMonitorUpdateStatus::UnrecoverableError => {
+                               let err_str = "ChannelMonitor[Update] persistence failed unrecoverably. This indicates we cannot continue normal operation and must shut down.";
+                               log_error!(self.logger, "{}", err_str);
+                               panic!("{}", err_str);
+                       },
                }
                if let Some(ref chain_source) = self.chain_source {
                        monitor.load_outputs_to_watch(chain_source);
@@ -678,28 +732,25 @@ where C::Target: chain::Filter,
                entry.insert(MonitorHolder {
                        monitor,
                        pending_monitor_updates: Mutex::new(pending_monitor_updates),
-                       channel_perm_failed: AtomicBool::new(false),
                        last_chain_persist_height: AtomicUsize::new(self.highest_chain_height.load(Ordering::Acquire)),
                });
-               persist_res
+               Ok(persist_res)
        }
 
-       /// Note that we persist the given `ChannelMonitor` update while holding the
-       /// `ChainMonitor` monitors lock.
        fn update_channel(&self, funding_txo: OutPoint, update: &ChannelMonitorUpdate) -> ChannelMonitorUpdateStatus {
                // Update the monitor that watches the channel referred to by the given outpoint.
                let monitors = self.monitors.read().unwrap();
-               match monitors.get(&funding_txo) {
+               let ret = match monitors.get(&funding_txo) {
                        None => {
                                log_error!(self.logger, "Failed to update channel monitor: no such monitor registered");
 
                                // We should never ever trigger this from within ChannelManager. Technically a
                                // user could use this object with some proxying in between which makes this
                                // possible, but in tests and fuzzing, this should be a panic.
-                               #[cfg(any(test, fuzzing))]
+                               #[cfg(debug_assertions)]
                                panic!("ChannelManager generated a channel update for a channel that was not yet registered!");
-                               #[cfg(not(any(test, fuzzing)))]
-                               ChannelMonitorUpdateStatus::PermanentFailure
+                               #[cfg(not(debug_assertions))]
+                               ChannelMonitorUpdateStatus::InProgress
                        },
                        Some(monitor_state) => {
                                let monitor = &monitor_state.monitor;
@@ -718,23 +769,28 @@ where C::Target: chain::Filter,
                                                pending_monitor_updates.push(update_id);
                                                log_debug!(self.logger, "Persistence of ChannelMonitorUpdate for channel {} in progress", log_funding_info!(monitor));
                                        },
-                                       ChannelMonitorUpdateStatus::PermanentFailure => {
-                                               monitor_state.channel_perm_failed.store(true, Ordering::Release);
-                                               log_error!(self.logger, "Persistence of ChannelMonitorUpdate for channel {} failed", log_funding_info!(monitor));
-                                       },
                                        ChannelMonitorUpdateStatus::Completed => {
                                                log_debug!(self.logger, "Persistence of ChannelMonitorUpdate for channel {} completed", log_funding_info!(monitor));
                                        },
+                                       ChannelMonitorUpdateStatus::UnrecoverableError => { /* we'll panic in a moment */ },
                                }
                                if update_res.is_err() {
-                                       ChannelMonitorUpdateStatus::PermanentFailure
-                               } else if monitor_state.channel_perm_failed.load(Ordering::Acquire) {
-                                       ChannelMonitorUpdateStatus::PermanentFailure
+                                       ChannelMonitorUpdateStatus::InProgress
                                } else {
                                        persist_res
                                }
                        }
+               };
+               if let ChannelMonitorUpdateStatus::UnrecoverableError = ret {
+                       // Take the monitors lock for writing so that we poison it and any future
+                       // operations going forward fail immediately.
+                       core::mem::drop(monitors);
+                       let _poison = self.monitors.write().unwrap();
+                       let err_str = "ChannelMonitor[Update] persistence failed unrecoverably. This indicates we cannot continue normal operation and must shut down.";
+                       log_error!(self.logger, "{}", err_str);
+                       panic!("{}", err_str);
                }
+               ret
        }
 
        fn release_pending_monitor_events(&self) -> Vec<(OutPoint, Vec<MonitorEvent>, Option<PublicKey>)> {
@@ -745,19 +801,8 @@ where C::Target: chain::Filter,
                                        monitor_state.last_chain_persist_height.load(Ordering::Acquire) + LATENCY_GRACE_PERIOD_BLOCKS as usize
                                                > self.highest_chain_height.load(Ordering::Acquire)
                        {
-                               log_info!(self.logger, "A Channel Monitor sync is still in progress, refusing to provide monitor events!");
+                               log_debug!(self.logger, "A Channel Monitor sync is still in progress, refusing to provide monitor events!");
                        } else {
-                               if monitor_state.channel_perm_failed.load(Ordering::Acquire) {
-                                       // If a `UpdateOrigin::ChainSync` persistence failed with `PermanantFailure`,
-                                       // we don't really know if the latest `ChannelMonitor` state is on disk or not.
-                                       // We're supposed to hold monitor updates until the latest state is on disk to
-                                       // avoid duplicate events, but the user told us persistence is screw-y and may
-                                       // not complete. We can't hold events forever because we may learn some payment
-                                       // preimage, so instead we just log and hope the user complied with the
-                                       // `PermanentFailure` requirements of having at least the local-disk copy
-                                       // updated.
-                                       log_info!(self.logger, "A Channel Monitor sync returned PermanentFailure. Returning monitor events but duplicate events may appear after reload!");
-                               }
                                if is_pending_monitor_update {
                                        log_error!(self.logger, "A ChannelMonitor sync took longer than {} blocks to complete.", LATENCY_GRACE_PERIOD_BLOCKS);
                                        log_error!(self.logger, "   To avoid funds-loss, we are allowing monitor updates to be released.");
@@ -782,30 +827,13 @@ impl<ChannelSigner: WriteableEcdsaChannelSigner, C: Deref, T: Deref, F: Deref, L
              L::Target: Logger,
              P::Target: Persist<ChannelSigner>,
 {
-       #[cfg(not(anchors))]
-       /// Processes [`SpendableOutputs`] events produced from each [`ChannelMonitor`] upon maturity.
-       ///
-       /// An [`EventHandler`] may safely call back to the provider, though this shouldn't be needed in
-       /// order to handle these events.
-       ///
-       /// [`SpendableOutputs`]: events::Event::SpendableOutputs
-       fn process_pending_events<H: Deref>(&self, handler: H) where H::Target: EventHandler {
-               let mut pending_events = Vec::new();
-               for monitor_state in self.monitors.read().unwrap().values() {
-                       pending_events.append(&mut monitor_state.monitor.get_and_clear_pending_events());
-               }
-               for event in pending_events {
-                       handler.handle_event(event);
-               }
-       }
-       #[cfg(anchors)]
        /// Processes [`SpendableOutputs`] events produced from each [`ChannelMonitor`] upon maturity.
        ///
        /// For channels featuring anchor outputs, this method will also process [`BumpTransaction`]
        /// events produced from each [`ChannelMonitor`] while there is a balance to claim onchain
        /// within each channel. As the confirmation of a commitment transaction may be critical to the
-       /// safety of funds, this method must be invoked frequently, ideally once for every chain tip
-       /// update (block connected or disconnected).
+       /// safety of funds, we recommend invoking this every 30 seconds, or lower if running in an
+       /// environment with spotty connections, like on mobile.
        ///
        /// An [`EventHandler`] may safely call back to the provider, though this shouldn't be needed in
        /// order to handle these events.
@@ -813,26 +841,20 @@ impl<ChannelSigner: WriteableEcdsaChannelSigner, C: Deref, T: Deref, F: Deref, L
        /// [`SpendableOutputs`]: events::Event::SpendableOutputs
        /// [`BumpTransaction`]: events::Event::BumpTransaction
        fn process_pending_events<H: Deref>(&self, handler: H) where H::Target: EventHandler {
-               let mut pending_events = Vec::new();
                for monitor_state in self.monitors.read().unwrap().values() {
-                       pending_events.append(&mut monitor_state.monitor.get_and_clear_pending_events());
-               }
-               for event in pending_events {
-                       handler.handle_event(event);
+                       monitor_state.monitor.process_pending_events(&handler);
                }
        }
 }
 
 #[cfg(test)]
 mod tests {
-       use bitcoin::{BlockHeader, TxMerkleNode};
-       use bitcoin::hashes::Hash;
-       use crate::{check_added_monitors, check_closed_broadcast, check_closed_event};
-       use crate::{expect_payment_sent, expect_payment_claimed, expect_payment_sent_without_paths, expect_payment_path_successful, get_event_msg};
+       use crate::check_added_monitors;
+       use crate::{expect_payment_claimed, expect_payment_path_successful, get_event_msg};
        use crate::{get_htlc_update_msgs, get_local_commitment_txn, get_revoke_commit_msgs, get_route_and_payment_hash, unwrap_send_err};
        use crate::chain::{ChannelMonitorUpdateStatus, Confirm, Watch};
        use crate::chain::channelmonitor::LATENCY_GRACE_PERIOD_BLOCKS;
-       use crate::events::{Event, ClosureReason, MessageSendEvent, MessageSendEventsProvider};
+       use crate::events::{Event, MessageSendEvent, MessageSendEventsProvider};
        use crate::ln::channelmanager::{PaymentSendFailure, PaymentId, RecipientOnionFields};
        use crate::ln::functional_test_utils::*;
        use crate::ln::msgs::ChannelMessageHandler;
@@ -850,8 +872,8 @@ mod tests {
                create_announced_chan_between_nodes(&nodes, 0, 1);
 
                // Route two payments to be claimed at the same time.
-               let (payment_preimage_1, payment_hash_1, _) = route_payment(&nodes[0], &[&nodes[1]], 1_000_000);
-               let (payment_preimage_2, payment_hash_2, _) = route_payment(&nodes[0], &[&nodes[1]], 1_000_000);
+               let (payment_preimage_1, payment_hash_1, ..) = route_payment(&nodes[0], &[&nodes[1]], 1_000_000);
+               let (payment_preimage_2, payment_hash_2, ..) = route_payment(&nodes[0], &[&nodes[1]], 1_000_000);
 
                chanmon_cfgs[1].persister.offchain_monitor_updates.lock().unwrap().clear();
                chanmon_cfgs[1].persister.set_update_ret(ChannelMonitorUpdateStatus::InProgress);
@@ -911,7 +933,7 @@ mod tests {
 
                let updates = get_htlc_update_msgs!(nodes[1], nodes[0].node.get_our_node_id());
                nodes[0].node.handle_update_fulfill_htlc(&nodes[1].node.get_our_node_id(), &updates.update_fulfill_htlcs[0]);
-               expect_payment_sent_without_paths!(nodes[0], payment_preimage_1);
+               expect_payment_sent(&nodes[0], payment_preimage_1, None, false, false);
                nodes[0].node.handle_commitment_signed(&nodes[1].node.get_our_node_id(), &updates.commitment_signed);
                check_added_monitors!(nodes[0], 1);
                let (as_first_raa, as_first_update) = get_revoke_commit_msgs!(nodes[0], nodes[1].node.get_our_node_id());
@@ -924,7 +946,7 @@ mod tests {
                let bs_first_raa = get_event_msg!(nodes[1], MessageSendEvent::SendRevokeAndACK, nodes[0].node.get_our_node_id());
 
                nodes[0].node.handle_update_fulfill_htlc(&nodes[1].node.get_our_node_id(), &bs_second_updates.update_fulfill_htlcs[0]);
-               expect_payment_sent_without_paths!(nodes[0], payment_preimage_2);
+               expect_payment_sent(&nodes[0], payment_preimage_2, None, false, false);
                nodes[0].node.handle_commitment_signed(&nodes[1].node.get_our_node_id(), &bs_second_updates.commitment_signed);
                check_added_monitors!(nodes[0], 1);
                nodes[0].node.handle_revoke_and_ack(&nodes[1].node.get_our_node_id(), &bs_first_raa);
@@ -958,7 +980,7 @@ mod tests {
                let (route, second_payment_hash, _, second_payment_secret) = get_route_and_payment_hash!(nodes[0], nodes[1], 100_000);
 
                // First route a payment that we will claim on chain and give the recipient the preimage.
-               let (payment_preimage, payment_hash, _) = route_payment(&nodes[0], &[&nodes[1]], 1_000_000);
+               let (payment_preimage, payment_hash, ..) = route_payment(&nodes[0], &[&nodes[1]], 1_000_000);
                nodes[1].node.claim_funds(payment_preimage);
                expect_payment_claimed!(nodes[1], payment_hash, 1_000_000);
                nodes[1].node.get_and_clear_pending_msg_events();
@@ -972,10 +994,7 @@ mod tests {
 
                // Connect B's commitment transaction, but only to the ChainMonitor/ChannelMonitor. The
                // channel is now closed, but the ChannelManager doesn't know that yet.
-               let new_header = BlockHeader {
-                       version: 2, time: 0, bits: 0, nonce: 0,
-                       prev_blockhash: nodes[0].best_block_info().0,
-                       merkle_root: TxMerkleNode::all_zeros() };
+               let new_header = create_dummy_header(nodes[0].best_block_info().0, 0);
                nodes[0].chain_monitor.chain_monitor.transactions_confirmed(&new_header,
                        &[(0, &remote_txn[0]), (1, &remote_txn[1])], nodes[0].best_block_info().1 + 1);
                assert!(nodes[0].chain_monitor.release_pending_monitor_events().is_empty());
@@ -987,11 +1006,8 @@ mod tests {
                chanmon_cfgs[0].persister.set_update_ret(ChannelMonitorUpdateStatus::Completed);
                unwrap_send_err!(nodes[0].node.send_payment_with_route(&route, second_payment_hash,
                                RecipientOnionFields::secret_only(second_payment_secret), PaymentId(second_payment_hash.0)
-                       ), true, APIError::ChannelUnavailable { ref err },
-                       assert!(err.contains("ChannelMonitor storage failure")));
-               check_added_monitors!(nodes[0], 2); // After the failure we generate a close-channel monitor update
-               check_closed_broadcast!(nodes[0], true);
-               check_closed_event!(nodes[0], 1, ClosureReason::ProcessingError { err: "ChannelMonitor storage failure".to_string() });
+                       ), false, APIError::MonitorUpdateInProgress, {});
+               check_added_monitors!(nodes[0], 1);
 
                // However, as the ChainMonitor is still waiting for the original persistence to complete,
                // it won't yet release the MonitorEvents.
@@ -999,10 +1015,7 @@ mod tests {
 
                if block_timeout {
                        // After three blocks, pending MontiorEvents should be released either way.
-                       let latest_header = BlockHeader {
-                               version: 2, time: 0, bits: 0, nonce: 0,
-                               prev_blockhash: nodes[0].best_block_info().0,
-                               merkle_root: TxMerkleNode::all_zeros() };
+                       let latest_header = create_dummy_header(nodes[0].best_block_info().0, 0);
                        nodes[0].chain_monitor.chain_monitor.best_block_updated(&latest_header, nodes[0].best_block_info().1 + LATENCY_GRACE_PERIOD_BLOCKS);
                } else {
                        let persistences = chanmon_cfgs[0].persister.chain_sync_monitor_persistences.lock().unwrap().clone();
@@ -1013,7 +1026,7 @@ mod tests {
                        }
                }
 
-               expect_payment_sent!(nodes[0], payment_preimage);
+               expect_payment_sent(&nodes[0], payment_preimage, None, true, false);
        }
 
        #[test]
@@ -1023,7 +1036,8 @@ mod tests {
        }
 
        #[test]
-       fn update_during_chainsync_fails_channel() {
+       #[cfg(feature = "std")]
+       fn update_during_chainsync_poisons_channel() {
                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]);
@@ -1031,17 +1045,15 @@ mod tests {
                create_announced_chan_between_nodes(&nodes, 0, 1);
 
                chanmon_cfgs[0].persister.chain_sync_monitor_persistences.lock().unwrap().clear();
-               chanmon_cfgs[0].persister.set_update_ret(ChannelMonitorUpdateStatus::PermanentFailure);
-
-               connect_blocks(&nodes[0], 1);
-               // Before processing events, the ChannelManager will still think the Channel is open and
-               // there won't be any ChannelMonitorUpdates
-               assert_eq!(nodes[0].node.list_channels().len(), 1);
-               check_added_monitors!(nodes[0], 0);
-               // ... however once we get events once, the channel will close, creating a channel-closed
-               // ChannelMonitorUpdate.
-               check_closed_broadcast!(nodes[0], true);
-               check_closed_event!(nodes[0], 1, ClosureReason::ProcessingError { err: "Failed to persist ChannelMonitor update during chain sync".to_string() });
-               check_added_monitors!(nodes[0], 1);
+               chanmon_cfgs[0].persister.set_update_ret(ChannelMonitorUpdateStatus::UnrecoverableError);
+
+               assert!(std::panic::catch_unwind(|| {
+                       // Returning an UnrecoverableError should always panic immediately
+                       connect_blocks(&nodes[0], 1);
+               }).is_err());
+               assert!(std::panic::catch_unwind(|| {
+                       // ...and also poison our locks causing later use to panic as well
+                       core::mem::drop(nodes);
+               }).is_err());
        }
 }