fix: block assembly reset skips intermediate block finalization and handleReorg returns nil after fallback reset

[freemans13]

Summary

Fix two bugs in block assembly's reset logic that cause intermediate blocks to be improperly processed during reorgs/resets:

• Bug 1: SubtreeProcessor.reset() only called finalizeBlockProcessing() (and thus SetBlockProcessedAt) for the last moveForward block. Intermediate blocks never got processed_at set, meaning they were not recognized as fully processed.
• Bug 2: handleReorg() returned nil after fallback reset (triggered by invalid block or failed Reorg), allowing processNewBlockAnnouncement to overwrite the reset's setBestBlockHeader with a potentially stale value captured before the reset ran.

Changes

Bug 1 Fix (SubtreeProcessor.go)

• Non-legacy path: Added finalizeBlockProcessing(ctx, block) inside the per-block moveForward loop so each block gets SetBlockProcessedAt called individually.
• Legacy sync path: Replaced single currentBlockHeader.Store() + updatePrecomputedMiningData() with per-block finalizeBlockProcessing() loop after concurrent coinbase processing.
• Changed post-loop conditional to only handle the moveBack-only case (no moveForward blocks).

Bug 2 Fix (BlockAssembler.go)

• After fallback reset() succeeds in handleReorg(), return errors.NewBlockAssemblyResetError(...) instead of nil. This matches the large-reorg path which already correctly returns ErrBlockAssemblyReset, preventing processNewBlockAnnouncement from overwriting the reset's state.

Tests

• Unit tests (reset_bug_test.go): Two tests that prove both bugs exist and pass after fixes.
• Integration tests (blockassembly_system_test.go): Two tests using full blockchain daemon (gRPC, real stores) that exercise the reset and reorg-with-invalid-block paths end-to-end.

Testing

All tests pass with no regressions:

ok  github.com/bsv-blockchain/teranode/services/blockassembly         18.460s
ok  github.com/bsv-blockchain/teranode/services/blockassembly/mining   3.961s
ok  github.com/bsv-blockchain/teranode/services/blockassembly/subtreeprocessor  210.212s

Related issue: https://github.com/orgs/bitcoin-sv/projects/5/views/6?pane=issue&itemId=161301562&issue=bitcoin-sv%7Cteranode%7C4507

[github-actions]

🤖 Claude Code Review

Status: Complete

Two critical bug fixes in block assembly reset logic, both properly addressed with comprehensive tests:

Bug 1 Fix (SubtreeProcessor.go): Fixed intermediate block finalization during reset

• Previously only the last moveForward block got SetBlockProcessedAt called
• Now each block is finalized individually in both legacy and non-legacy paths
• Critical for block processing state consistency

Bug 2 Fix (BlockAssembler.go): Fixed handleReorg return value after fallback reset

• Now returns ErrBlockAssemblyReset instead of nil after fallback reset
• Prevents processNewBlockAnnouncement from overwriting reset state at line 609
• Matches the large-reorg path behavior (line 1163)

Test Coverage: Excellent

• Unit tests prove both bugs exist and pass after fixes
• Integration tests exercise full service stack with real stores
• Tests cover both reset-ahead and reorg-with-invalid-block scenarios

No issues found. The fixes are well-reasoned, properly implemented, and thoroughly tested.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
88.9% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud