fix: Summit cold start from checkpoint deadlock

.gitignore

@@ -2,3 +2,4 @@
 .idea/
 testnet/
 .vscode/*
+.nvim.lua

finalizer/src/actor.rs

@@ -220,6 +220,36 @@ impl<
             }))
             .await;
 
+        // Send initial forkchoice to the execution client so it knows the chain
+        // head and can start P2P sync. Then wait for sync to complete before
+        // replaying any blocks. Without this, catch-up blocks fail because the
+        // execution client doesn't have them yet.
+        {
+            let forkchoice = self.canonical_state.forkchoice;
+            if !forkchoice.head_block_hash.is_zero() {
+                info!(
+                    head = %forkchoice.head_block_hash,
+                    "sending initial forkchoice update to execution client, waiting for sync..."
+                );
+                loop {
+                    let status = self.engine_client.commit_hash_with_status(forkchoice).await;
+                    if status.is_valid() {
+                        info!("execution client synced to checkpoint head, ready to replay blocks");
+                        break;
+                    } else if status.is_syncing() {
+                        info!("execution client still syncing, waiting 5s...");
+                        self.context.sleep(std::time::Duration::from_secs(5)).await;
+                    } else {
+                        warn!(
+                            ?status,
+                            "unexpected response to initial forkchoice update, proceeding anyway"
+                        );
+                        break;
+                    }
+                }
+            }
+        }
+
         loop {
             if self.validator_exit
                 && is_first_block_of_epoch(

finalizer/src/tests/mocks.rs

@@ -3,7 +3,8 @@
 use alloy_primitives::{Address, FixedBytes, U256};
 use alloy_rpc_types_engine::{
     ExecutionPayloadEnvelopeV3, ExecutionPayloadEnvelopeV4, ExecutionPayloadV1, ExecutionPayloadV2,
-    ExecutionPayloadV3, ForkchoiceState, PayloadId, PayloadStatus, PayloadStatusEnum,
+    ExecutionPayloadV3, ForkchoiceState, ForkchoiceUpdated, PayloadId, PayloadStatus,
+    PayloadStatusEnum,
 };
 use commonware_consensus::simplex::scheme::bls12381_multisig;
 use commonware_consensus::simplex::types::{Finalization, Finalize, Proposal};
@@ -136,6 +137,19 @@ impl EngineClient for MockEngineClient {
     }
 
     async fn commit_hash(&mut self, _fork_choice_state: ForkchoiceState) {}
+
+    async fn commit_hash_with_status(
+        &mut self,
+        _fork_choice_state: ForkchoiceState,
+    ) -> ForkchoiceUpdated {
+        ForkchoiceUpdated {
+            payload_status: PayloadStatus {
+                status: PayloadStatusEnum::Valid,
+                latest_valid_hash: Some([0u8; 32].into()),
+            },
+            payload_id: None,
+        }
+    }
 }
 
 /// Minimal mock NetworkOracle

node/src/engine.rs

@@ -64,7 +64,7 @@ pub const BLOCKS_PER_EPOCH: u64 = 50;
 #[cfg(debug_assertions)]
 pub const BLOCKS_PER_EPOCH: u64 = 10;
 #[cfg(all(not(debug_assertions), not(feature = "e2e")))]
-const BLOCKS_PER_EPOCH: u64 = 10000;
+const BLOCKS_PER_EPOCH: u64 = 50;
 const VALIDATOR_MAX_WITHDRAWALS_PER_BLOCK: usize = 16;
 //
 

node/src/test_harness/mock_engine_client.rs

@@ -4,8 +4,8 @@ use alloy_primitives::hex;
 use alloy_primitives::{Address, B256, Bloom, Bytes, FixedBytes, U256};
 use alloy_rpc_types_engine::{
     BlobsBundleV1, ExecutionPayloadEnvelopeV3, ExecutionPayloadEnvelopeV4, ExecutionPayloadV1,
-    ExecutionPayloadV2, ExecutionPayloadV3, ForkchoiceState, PayloadId, PayloadStatus,
-    PayloadStatusEnum,
+    ExecutionPayloadV2, ExecutionPayloadV3, ForkchoiceState, ForkchoiceUpdated, PayloadId,
+    PayloadStatus, PayloadStatusEnum,
 };
 use rand::RngCore;
 use std::collections::HashMap;
@@ -469,6 +469,17 @@ impl EngineClient for MockEngineClient {
             }
         }
     }
+
+    async fn commit_hash_with_status(
+        &mut self,
+        fork_choice_state: ForkchoiceState,
+    ) -> ForkchoiceUpdated {
+        self.commit_hash(fork_choice_state).await;
+        ForkchoiceUpdated {
+            payload_status: PayloadStatus::new(PayloadStatusEnum::Valid, Some(fork_choice_state.head_block_hash)),
+            payload_id: None,
+        }
+    }
 }
 
 pub struct MockEngineNetworkBuilder {

types/src/engine_client.rs

@@ -21,7 +21,8 @@ use alloy_eips::eip4895::Withdrawal;
 use alloy_primitives::{Address, FixedBytes};
 use alloy_provider::{ProviderBuilder, RootProvider, ext::EngineApi};
 use alloy_rpc_types_engine::{
-    ExecutionPayloadEnvelopeV4, ForkchoiceState, PayloadAttributes, PayloadId, PayloadStatus,
+    ExecutionPayloadEnvelopeV4, ForkchoiceState, ForkchoiceUpdated, PayloadAttributes, PayloadId,
+    PayloadStatus,
 };
 use tracing::{error, warn};
 
@@ -51,6 +52,11 @@ pub trait EngineClient: Clone + Send + Sync + 'static {
         &mut self,
         fork_choice_state: ForkchoiceState,
     ) -> impl Future<Output = ()> + Send;
+
+    fn commit_hash_with_status(
+        &mut self,
+        fork_choice_state: ForkchoiceState,
+    ) -> impl Future<Output = ForkchoiceUpdated> + Send;
 }
 
 #[derive(Clone)]
@@ -192,6 +198,27 @@ impl EngineClient for RethEngineClient {
             Err(_) => panic!("Unable to get a response"),
         };
     }
+
+    async fn commit_hash_with_status(
+        &mut self,
+        fork_choice_state: ForkchoiceState,
+    ) -> ForkchoiceUpdated {
+        match self
+            .provider
+            .fork_choice_updated_v3(fork_choice_state, None)
+            .await
+        {
+            Ok(res) => res,
+            Err(e) if e.is_transport_error() => {
+                self.wait_until_reconnect_available().await;
+                self.provider
+                    .fork_choice_updated_v3(fork_choice_state, None)
+                    .await
+                    .expect("Failed to commit hash after reconnect")
+            }
+            Err(_) => panic!("Unable to get a response"),
+        }
+    }
 }
 
 #[cfg(feature = "bad-blocks")]
@@ -345,6 +372,27 @@ impl EngineClient for BadBlockEngineClient {
             Err(_) => panic!("Unable to get a response"),
         };
     }
+
+    async fn commit_hash_with_status(
+        &mut self,
+        fork_choice_state: ForkchoiceState,
+    ) -> ForkchoiceUpdated {
+        match self
+            .provider
+            .fork_choice_updated_v3(fork_choice_state, None)
+            .await
+        {
+            Ok(res) => res,
+            Err(e) if e.is_transport_error() => {
+                self.wait_until_reconnect_available().await;
+                self.provider
+                    .fork_choice_updated_v3(fork_choice_state, None)
+                    .await
+                    .expect("Failed to commit hash after reconnect")
+            }
+            Err(_) => panic!("Unable to get a response"),
+        }
+    }
 }
 
 #[cfg(feature = "bench")]
@@ -357,7 +405,7 @@ pub mod benchmarking {
     use alloy_provider::{ProviderBuilder, RootProvider, ext::EngineApi};
     use alloy_rpc_types_engine::{
         ExecutionPayloadEnvelopeV3, ExecutionPayloadEnvelopeV4, ExecutionPayloadV3,
-        ForkchoiceState, PayloadId, PayloadStatus,
+        ForkchoiceState, ForkchoiceUpdated, PayloadId, PayloadStatus,
     };
     use alloy_transport_ipc::IpcConnect;
     use serde::{Deserialize, Serialize};
@@ -441,6 +489,16 @@ pub mod benchmarking {
                 .await
                 .unwrap();
         }
+
+        async fn commit_hash_with_status(
+            &mut self,
+            fork_choice_state: ForkchoiceState,
+        ) -> ForkchoiceUpdated {
+            self.provider
+                .fork_choice_updated_v3(fork_choice_state, None)
+                .await
+                .unwrap()
+        }
     }
 
     #[derive(Debug, Serialize, Deserialize)]