From 971b992e785a69b6b12a740f9906741e022a6554 Mon Sep 17 00:00:00 2001
From: Liu Yuan <namei.unix@gmail.com>
Date: Sat, 4 Apr 2026 11:07:29 +0800
Subject: [PATCH] fix(seahorse): add mutex for assembler lazy initialization

Issue #5 from PR review: The check-then-create pattern for e.assembler
was a data race when multiple goroutines called Assemble() concurrently:
    if e.assembler == nil {
        e.assembler = &Assembler{...}
    }

Changes:
- Add assemblerMu sync.Mutex to Engine struct
- Add initAssemblerOnce() using double-checked locking (same pattern as initCompactionOnce)
- Add TestAssemblerLazyInitRace to verify thread-safety
---
 .gitignore                        |  2 ++
 pkg/seahorse/short_engine.go      | 16 ++++++++---
 pkg/seahorse/short_engine_test.go | 44 +++++++++++++++++++++++++++++++
 3 files changed, 59 insertions(+), 3 deletions(-)

diff --git a/.gitignore b/.gitignore
index b869ecc33..135867842 100644
--- a/.gitignore
+++ b/.gitignore
@@ -67,3 +67,5 @@ web/backend/dist/*
 .claude/
 
 docker/data
+
+.omc/
diff --git a/pkg/seahorse/short_engine.go b/pkg/seahorse/short_engine.go
index 11cbe01d7..4cd4d3887 100644
--- a/pkg/seahorse/short_engine.go
+++ b/pkg/seahorse/short_engine.go
@@ -58,6 +58,7 @@ type Engine struct {
 	compaction        *CompactionEngine
 	compactionMu      sync.Mutex
 	assembler         *Assembler
+	assemblerMu       sync.Mutex
 	retrieval         *RetrievalEngine
 	config            Config
 	complete          CompleteFn
@@ -307,9 +308,7 @@ func (e *Engine) Assemble(ctx context.Context, sessionKey string, input Assemble
 		return nil, fmt.Errorf("get conversation: %w", err)
 	}
 
-	if e.assembler == nil {
-		e.assembler = &Assembler{store: e.store, config: e.config}
-	}
+	e.initAssemblerOnce()
 	return e.assembler.Assemble(ctx, conv.ConversationID, input)
 }
 
@@ -362,6 +361,17 @@ func (e *Engine) initCompactionOnce() {
 	}
 }
 
+// initAssemblerOnce lazily initializes the assembler.
+func (e *Engine) initAssemblerOnce() {
+	if e.assembler == nil {
+		e.assemblerMu.Lock()
+		defer e.assemblerMu.Unlock()
+		if e.assembler == nil {
+			e.assembler = &Assembler{store: e.store, config: e.config}
+		}
+	}
+}
+
 // IngestMessages is an alias for Ingest.
 func (e *Engine) IngestMessages(ctx context.Context, sessionKey string, messages []Message) (*IngestResult, error) {
 	return e.Ingest(ctx, sessionKey, messages)
diff --git a/pkg/seahorse/short_engine_test.go b/pkg/seahorse/short_engine_test.go
index 401b6ca39..d2f108efd 100644
--- a/pkg/seahorse/short_engine_test.go
+++ b/pkg/seahorse/short_engine_test.go
@@ -1329,3 +1329,47 @@ func TestBootstrapAnchorWithDuplicateContent_Simple(t *testing.T) {
 		t.Errorf("last message content = %q, want 'D'", lastMsg.Content)
 	}
 }
+
+// --- Assembler lazy init race detection ---
+
+func TestAssemblerLazyInitRace(t *testing.T) {
+	// This test verifies that Assemble() lazy initialization of e.assembler
+	// is thread-safe. The original code has a data race:
+	//   if e.assembler == nil {
+	//       e.assembler = &Assembler{...}
+	//   }
+
+	// Run multiple iterations to increase chance of catching race
+	for i := 0; i < 30; i++ {
+		// Create fresh engine with nil assembler
+		e := newTestEngine(t)
+
+		ctx := context.Background()
+		sessionKey := fmt.Sprintf("race-test-%d", i)
+
+		// Add message first (avoid SQLite concurrency issues)
+		_, err := e.Ingest(ctx, sessionKey, []Message{
+			{Role: "user", Content: "hello", TokenCount: 5},
+		})
+		if err != nil {
+			t.Fatalf("Ingest: %v", err)
+		}
+
+		// Use a barrier to ensure all goroutines start at the same time
+		start := make(chan struct{})
+		var wg sync.WaitGroup
+
+		for j := 0; j < 20; j++ {
+			wg.Add(1)
+			go func() {
+				defer wg.Done()
+				<-start // Wait for all goroutines to be ready
+				e.Assemble(ctx, sessionKey, AssembleInput{Budget: 1000})
+			}()
+		}
+
+		// Start all goroutines simultaneously
+		close(start)
+		wg.Wait()
+	}
+}