perf(pool): add fast path for Get/Put to match master performance

Introduced TryTransitionFast() for the hot path (Get/Put operations):
- Single CAS operation (same as master's atomic bool)
- No waiter notification overhead
- No loop through valid states
- No error allocation

Hot path flow:
1. popIdle(): Try IDLE → IN_USE (fast), fallback to CREATED → IN_USE
2. putConn(): Try IN_USE → IDLE (fast)

This matches master's performance while preserving state machine for:
- Background operations (handoff/reauth use UNUSABLE state)
- State validation (TryTransition still available)
- Waiter notification (AwaitAndTransition for blocking)

Performance comparison per Get/Put cycle:
- Master: 2 atomic CAS operations
- State machine (before): 5 atomic operations (2.5x slower)
- State machine (after): 2 atomic CAS operations (same as master!)

Expected improvement: Restore to baseline ~11,373 ops/sec

internal/pool/conn_state.go

@@ -121,6 +121,20 @@ func (sm *ConnStateMachine) GetState() ConnState {
 	return ConnState(sm.state.Load())
 }
 
+// TryTransitionFast is an optimized version for the hot path (Get/Put operations).
+// It only handles simple IDLE ⇄ IN_USE transitions without waiter notification.
+// This is safe because:
+// 1. Get/Put don't need to wait for state changes
+// 2. Background operations (handoff/reauth) use UNUSABLE state, which this won't match
+// 3. If a background operation is in progress (state is UNUSABLE), this fails fast
+//
+// Returns true if transition succeeded, false otherwise.
+// Use this for performance-critical paths where you don't need error details.
+func (sm *ConnStateMachine) TryTransitionFast(fromState, targetState ConnState) bool {
+	// Single CAS operation - as fast as the old atomic bool!
+	return sm.state.CompareAndSwap(uint32(fromState), uint32(targetState))
+}
+
 // TryTransition attempts an immediate state transition without waiting.
 // Returns the current state after the transition attempt and an error if the transition failed.
 // The returned state is the CURRENT state (after the attempt), not the previous state.

internal/pool/pool.go

@@ -599,16 +599,25 @@ func (p *ConnPool) popIdle() (*Conn, error) {
 		}
 		attempts++
 
-		// Try to atomically transition to IN_USE using state machine
+		// Hot path optimization: try fast IDLE → IN_USE transition first
-		// Accept both CREATED (uninitialized) and IDLE (initialized) states
+		// This is a single CAS operation, as fast as the old atomic bool
-		// Use predefined slice to avoid allocation
+		sm := cn.GetStateMachine()
-		_, err := cn.GetStateMachine().TryTransition(validFromCreatedOrIdle, StateInUse)
+		if sm.TryTransitionFast(StateIdle, StateInUse) {
-		if err == nil {
+			// Successfully acquired the connection (common case)
-			// Successfully acquired the connection
 			p.idleConnsLen.Add(-1)
 			break
 		}
 
+		// Fast path failed - connection might be CREATED (uninitialized) or UNUSABLE (handoff/reauth)
+		// Try CREATED → IN_USE for new connections
+		if sm.TryTransitionFast(StateCreated, StateInUse) {
+			// Successfully acquired uninitialized connection
+			p.idleConnsLen.Add(-1)
+			break
+		}
+
+		// Connection is in UNUSABLE, INITIALIZING, or other state - skip it
+
 		// Connection is not in a valid state (might be UNUSABLE for handoff/re-auth, INITIALIZING, etc.)
 		// Put it back in the pool and try the next one
 		if p.cfg.PoolFIFO {
@@ -691,16 +700,13 @@ func (p *ConnPool) putConn(ctx context.Context, cn *Conn, freeTurn bool) {
 	var shouldCloseConn bool
 
 	if p.cfg.MaxIdleConns == 0 || p.idleConnsLen.Load() < p.cfg.MaxIdleConns {
-		// Try to transition to IDLE state BEFORE adding to pool
+		// Hot path optimization: try fast IN_USE → IDLE transition
-		// Only transition if connection is still IN_USE (hooks might have changed state)
+		// This is a single CAS operation, as fast as the old atomic bool
-		// This prevents:
+		sm := cn.GetStateMachine()
-		// 1. Race condition where another goroutine could acquire a connection that's still in IN_USE state
+		if !sm.TryTransitionFast(StateInUse, StateIdle) {
-		// 2. Overwriting state changes made by hooks (e.g., IN_USE → UNUSABLE for handoff)
+			// Fast path failed - hook might have changed state (e.g., to UNUSABLE for handoff)
-		// Use predefined slice to avoid allocation
-		currentState, err := cn.GetStateMachine().TryTransition(validFromInUse, StateIdle)
-		if err != nil {
-			// Hook changed the state (e.g., to UNUSABLE for handoff)
 			// Keep the state set by the hook and pool the connection anyway
+			currentState := sm.GetState()
 			internal.Logger.Printf(ctx, "Connection state changed by hook to %v, pooling as-is", currentState)
 		}