Merge pull request #489 from TheBlueMatt/2020-02-chan-updates

Move to a Monitor-Update return from copying around ChannelMonitors

Author: TheBlueMatt

fuzz/src/chanmon_consistency.rs

@@ -73,52 +73,55 @@ impl Writer for VecWriter {
 	}
 }
 
-static mut IN_RESTORE: bool = false;
 pub struct TestChannelMonitor {
+	pub logger: Arc<dyn Logger>,
 	pub simple_monitor: Arc<channelmonitor::SimpleManyChannelMonitor<OutPoint, EnforcingChannelKeys, Arc<chaininterface::BroadcasterInterface>>>,
 	pub update_ret: Mutex<Result<(), channelmonitor::ChannelMonitorUpdateErr>>,
-	pub latest_good_update: Mutex<HashMap<OutPoint, Vec<u8>>>,
-	pub latest_update_good: Mutex<HashMap<OutPoint, bool>>,
-	pub latest_updates_good_at_last_ser: Mutex<HashMap<OutPoint, bool>>,
+	// If we reload a node with an old copy of ChannelMonitors, the ChannelManager deserialization
+	// logic will automatically force-close our channels for us (as we don't have an up-to-date
+	// monitor implying we are not able to punish misbehaving counterparties). Because this test
+	// "fails" if we ever force-close a channel, we avoid doing so, always saving the latest
+	// fully-serialized monitor state here, as well as the corresponding update_id.
+	pub latest_monitors: Mutex<HashMap<OutPoint, (u64, Vec<u8>)>>,
 	pub should_update_manager: atomic::AtomicBool,
 }
 impl TestChannelMonitor {
 	pub fn new(chain_monitor: Arc<dyn chaininterface::ChainWatchInterface>, broadcaster: Arc<dyn chaininterface::BroadcasterInterface>, logger: Arc<dyn Logger>, feeest: Arc<dyn chaininterface::FeeEstimator>) -> Self {
 		Self {
-			simple_monitor: Arc::new(channelmonitor::SimpleManyChannelMonitor::new(chain_monitor, broadcaster, logger, feeest)),
+			simple_monitor: Arc::new(channelmonitor::SimpleManyChannelMonitor::new(chain_monitor, broadcaster, logger.clone(), feeest)),
+			logger,
 			update_ret: Mutex::new(Ok(())),
-			latest_good_update: Mutex::new(HashMap::new()),
-			latest_update_good: Mutex::new(HashMap::new()),
-			latest_updates_good_at_last_ser: Mutex::new(HashMap::new()),
+			latest_monitors: Mutex::new(HashMap::new()),
 			should_update_manager: atomic::AtomicBool::new(false),
 		}
 	}
 }
 impl channelmonitor::ManyChannelMonitor<EnforcingChannelKeys> for TestChannelMonitor {
-	fn add_update_monitor(&self, funding_txo: OutPoint, monitor: channelmonitor::ChannelMonitor<EnforcingChannelKeys>) -> Result<(), channelmonitor::ChannelMonitorUpdateErr> {
-		let ret = self.update_ret.lock().unwrap().clone();
-		if let Ok(()) = ret {
-			let mut ser = VecWriter(Vec::new());
-			monitor.write_for_disk(&mut ser).unwrap();
-			self.latest_good_update.lock().unwrap().insert(funding_txo, ser.0);
-			match self.latest_update_good.lock().unwrap().entry(funding_txo) {
-				hash_map::Entry::Vacant(e) => { e.insert(true); },
-				hash_map::Entry::Occupied(mut e) => {
-					if !e.get() && unsafe { IN_RESTORE } {
-						// Technically we can't consider an update to be "good" unless we're doing
-						// it in response to a test_restore_channel_monitor as the channel may
-						// still be waiting on such a call, so only set us to good if we're in the
-						// middle of a restore call.
-						e.insert(true);
-					}
-				},
-			}
-			self.should_update_manager.store(true, atomic::Ordering::Relaxed);
-		} else {
-			self.latest_update_good.lock().unwrap().insert(funding_txo, false);
+	fn add_monitor(&self, funding_txo: OutPoint, monitor: channelmonitor::ChannelMonitor<EnforcingChannelKeys>) -> Result<(), channelmonitor::ChannelMonitorUpdateErr> {
+		let mut ser = VecWriter(Vec::new());
+		monitor.write_for_disk(&mut ser).unwrap();
+		if let Some(_) = self.latest_monitors.lock().unwrap().insert(funding_txo, (monitor.get_latest_update_id(), ser.0)) {
+			panic!("Already had monitor pre-add_monitor");
 		}
-		assert!(self.simple_monitor.add_update_monitor(funding_txo, monitor).is_ok());
-		ret
+		self.should_update_manager.store(true, atomic::Ordering::Relaxed);
+		assert!(self.simple_monitor.add_monitor(funding_txo, monitor).is_ok());
+		self.update_ret.lock().unwrap().clone()
+	}
+
+	fn update_monitor(&self, funding_txo: OutPoint, update: channelmonitor::ChannelMonitorUpdate) -> Result<(), channelmonitor::ChannelMonitorUpdateErr> {
+		let mut map_lock = self.latest_monitors.lock().unwrap();
+		let mut map_entry = match map_lock.entry(funding_txo) {
+			hash_map::Entry::Occupied(entry) => entry,
+			hash_map::Entry::Vacant(_) => panic!("Didn't have monitor on update call"),
+		};
+		let mut deserialized_monitor = <(Sha256d, channelmonitor::ChannelMonitor<EnforcingChannelKeys>)>::
+			read(&mut Cursor::new(&map_entry.get().1), Arc::clone(&self.logger)).unwrap().1;
+		deserialized_monitor.update_monitor(update.clone()).unwrap();
+		let mut ser = VecWriter(Vec::new());
+		deserialized_monitor.write_for_disk(&mut ser).unwrap();
+		map_entry.insert((update.update_id, ser.0));
+		self.should_update_manager.store(true, atomic::Ordering::Relaxed);
+		self.update_ret.lock().unwrap().clone()
 	}
 
 	fn get_and_clear_pending_htlcs_updated(&self) -> Vec<HTLCUpdate> {
@@ -210,10 +213,10 @@ pub fn do_test(data: &[u8]) {
 			config.peer_channel_config_limits.min_dust_limit_satoshis = 0;
 
 			let mut monitors = HashMap::new();
-			let mut old_monitors = $old_monitors.latest_good_update.lock().unwrap();
-			for (outpoint, monitor_ser) in old_monitors.drain() {
+			let mut old_monitors = $old_monitors.latest_monitors.lock().unwrap();
+			for (outpoint, (update_id, monitor_ser)) in old_monitors.drain() {
 				monitors.insert(outpoint, <(Sha256d, ChannelMonitor<EnforcingChannelKeys>)>::read(&mut Cursor::new(&monitor_ser), Arc::clone(&logger)).expect("Failed to read monitor").1);
-				monitor.latest_good_update.lock().unwrap().insert(outpoint, monitor_ser);
+				monitor.latest_monitors.lock().unwrap().insert(outpoint, (update_id, monitor_ser));
 			}
 			let mut monitor_refs = HashMap::new();
 			for (outpoint, monitor) in monitors.iter_mut() {
@@ -230,17 +233,7 @@ pub fn do_test(data: &[u8]) {
 				channel_monitors: &mut monitor_refs,
 			};
 
-			let res = (<(Sha256d, ChannelManager<EnforcingChannelKeys, Arc<TestChannelMonitor>, Arc<TestBroadcaster>>)>::read(&mut Cursor::new(&$ser.0), read_args).expect("Failed to read manager").1, monitor);
-			for (_, was_good) in $old_monitors.latest_updates_good_at_last_ser.lock().unwrap().iter() {
-				if !was_good {
-					// If the last time we updated a monitor we didn't successfully update (and we
-					// have sense updated our serialized copy of the ChannelManager) we may
-					// force-close the channel on our counterparty cause we know we're missing
-					// something. Thus, we just return here since we can't continue to test.
-					return;
-				}
-			}
-			res
+			(<(Sha256d, ChannelManager<EnforcingChannelKeys, Arc<TestChannelMonitor>, Arc<TestBroadcaster>>)>::read(&mut Cursor::new(&$ser.0), read_args).expect("Failed to read manager").1, monitor)
 		} }
 	}
 
@@ -266,14 +259,15 @@ pub fn do_test(data: &[u8]) {
 			};
 
 			$source.handle_accept_channel(&$dest.get_our_node_id(), InitFeatures::supported(), &accept_channel);
+			let funding_output;
 			{
 				let events = $source.get_and_clear_pending_events();
 				assert_eq!(events.len(), 1);
 				if let events::Event::FundingGenerationReady { ref temporary_channel_id, ref channel_value_satoshis, ref output_script, .. } = events[0] {
 					let tx = Transaction { version: $chan_id, lock_time: 0, input: Vec::new(), output: vec![TxOut {
 						value: *channel_value_satoshis, script_pubkey: output_script.clone(),
 					}]};
-					let funding_output = OutPoint::new(tx.txid(), 0);
+					funding_output = OutPoint::new(tx.txid(), 0);
 					$source.funding_transaction_generated(&temporary_channel_id, funding_output);
 					channel_txn.push(tx);
 				} else { panic!("Wrong event type"); }
@@ -303,6 +297,7 @@ pub fn do_test(data: &[u8]) {
 				if let events::Event::FundingBroadcastSafe { .. } = events[0] {
 				} else { panic!("Wrong event type"); }
 			}
+			funding_output
 		} }
 	}
 
@@ -359,8 +354,8 @@ pub fn do_test(data: &[u8]) {
 
 	let mut nodes = [node_a, node_b, node_c];
 
-	make_channel!(nodes[0], nodes[1], 0);
-	make_channel!(nodes[1], nodes[2], 1);
+	let chan_1_funding = make_channel!(nodes[0], nodes[1], 0);
+	let chan_2_funding = make_channel!(nodes[1], nodes[2], 1);
 
 	for node in nodes.iter() {
 		confirm_txn!(node);
@@ -631,9 +626,26 @@ pub fn do_test(data: &[u8]) {
 			0x03 => *monitor_a.update_ret.lock().unwrap() = Ok(()),
 			0x04 => *monitor_b.update_ret.lock().unwrap() = Ok(()),
 			0x05 => *monitor_c.update_ret.lock().unwrap() = Ok(()),
-			0x06 => { unsafe { IN_RESTORE = true }; nodes[0].test_restore_channel_monitor(); unsafe { IN_RESTORE = false }; },
-			0x07 => { unsafe { IN_RESTORE = true }; nodes[1].test_restore_channel_monitor(); unsafe { IN_RESTORE = false }; },
-			0x08 => { unsafe { IN_RESTORE = true }; nodes[2].test_restore_channel_monitor(); unsafe { IN_RESTORE = false }; },
+			0x06 => {
+				if let Some((id, _)) = monitor_a.latest_monitors.lock().unwrap().get(&chan_1_funding) {
+					nodes[0].channel_monitor_updated(&chan_1_funding, *id);
+				}
+			},
+			0x07 => {
+				if let Some((id, _)) = monitor_b.latest_monitors.lock().unwrap().get(&chan_1_funding) {
+					nodes[1].channel_monitor_updated(&chan_1_funding, *id);
+				}
+			},
+			0x24 => {
+				if let Some((id, _)) = monitor_b.latest_monitors.lock().unwrap().get(&chan_2_funding) {
+					nodes[1].channel_monitor_updated(&chan_2_funding, *id);
+				}
+			},
+			0x08 => {
+				if let Some((id, _)) = monitor_c.latest_monitors.lock().unwrap().get(&chan_2_funding) {
+					nodes[2].channel_monitor_updated(&chan_2_funding, *id);
+				}
+			},
 			0x09 => send_payment!(nodes[0], (&nodes[1], chan_a)),
 			0x0a => send_payment!(nodes[1], (&nodes[0], chan_a)),
 			0x0b => send_payment!(nodes[1], (&nodes[2], chan_b)),
@@ -722,27 +734,19 @@ pub fn do_test(data: &[u8]) {
 				nodes[2] = node_c.clone();
 				monitor_c = new_monitor_c;
 			},
+			// 0x24 defined above
 			_ => test_return!(),
 		}
 
-		if monitor_a.should_update_manager.load(atomic::Ordering::Relaxed) {
-			node_a_ser.0.clear();
-			nodes[0].write(&mut node_a_ser).unwrap();
-			monitor_a.should_update_manager.store(false, atomic::Ordering::Relaxed);
-			*monitor_a.latest_updates_good_at_last_ser.lock().unwrap() = monitor_a.latest_update_good.lock().unwrap().clone();
-		}
-		if monitor_b.should_update_manager.load(atomic::Ordering::Relaxed) {
-			node_b_ser.0.clear();
-			nodes[1].write(&mut node_b_ser).unwrap();
-			monitor_b.should_update_manager.store(false, atomic::Ordering::Relaxed);
-			*monitor_b.latest_updates_good_at_last_ser.lock().unwrap() = monitor_b.latest_update_good.lock().unwrap().clone();
-		}
-		if monitor_c.should_update_manager.load(atomic::Ordering::Relaxed) {
-			node_c_ser.0.clear();
-			nodes[2].write(&mut node_c_ser).unwrap();
-			monitor_c.should_update_manager.store(false, atomic::Ordering::Relaxed);
-			*monitor_c.latest_updates_good_at_last_ser.lock().unwrap() = monitor_c.latest_update_good.lock().unwrap().clone();
-		}
+		node_a_ser.0.clear();
+		nodes[0].write(&mut node_a_ser).unwrap();
+		monitor_a.should_update_manager.store(false, atomic::Ordering::Relaxed);
+		node_b_ser.0.clear();
+		nodes[1].write(&mut node_b_ser).unwrap();
+		monitor_b.should_update_manager.store(false, atomic::Ordering::Relaxed);
+		node_c_ser.0.clear();
+		nodes[2].write(&mut node_c_ser).unwrap();
+		monitor_c.should_update_manager.store(false, atomic::Ordering::Relaxed);
 	}
 }
 

lightning/src/chain/keysinterface.rs

@@ -135,7 +135,8 @@ pub trait KeysInterface: Send + Sync {
 /// (TODO: We shouldn't require that, and should have an API to get them at deser time, due mostly
 /// to the possibility of reentrancy issues by calling the user's code during our deserialization
 /// routine).
-/// TODO: remove Clone once we start returning ChannelUpdate objects instead of copying ChannelMonitor
+/// TODO: We should remove Clone by instead requesting a new ChannelKeys copy when we create
+/// ChannelMonitors instead of expecting to clone the one out of the Channel into the monitors.
 pub trait ChannelKeys : Send+Clone {
 	/// Gets the private key for the anchor tx
 	fn funding_key<'a>(&'a self) -> &'a SecretKey;

lightning/src/chain/transaction.rs

@@ -39,6 +39,8 @@ impl OutPoint {
 	}
 }
 
+impl_writeable!(OutPoint, 0, { txid, index });
+
 #[cfg(test)]
 mod tests {
 	use chain::transaction::OutPoint;