perf(audio): optimize audio processing with batching and atomic operations

- Implement batch audio processing to reduce CGO overhead
- Replace mutexes with atomic operations for contention management
- Add buffer pooling to reduce allocations
- Optimize microphone operation cooldown with lock-free approach
- Improve error handling with pre-allocated error objects

internal/audio/batch_audio.go

@@ -0,0 +1,455 @@
+//go:build cgo
+
+package audio
+
+import (
+	"context"
+	"runtime"
+	"sync"
+	"sync/atomic"
+	"time"
+	"unsafe"
+
+	"github.com/jetkvm/kvm/internal/logging"
+	"github.com/rs/zerolog"
+)
+
+// BatchAudioProcessor manages batched CGO operations to reduce syscall overhead
+type BatchAudioProcessor struct {
+	// Statistics - MUST be first for ARM32 alignment (int64 fields need 8-byte alignment)
+	stats BatchAudioStats
+
+	// Control
+	ctx           context.Context
+	cancel        context.CancelFunc
+	logger        *zerolog.Logger
+	batchSize     int
+	batchDuration time.Duration
+
+	// Batch queues and state (atomic for lock-free access)
+	readQueue    chan batchReadRequest
+	writeQueue   chan batchWriteRequest
+	initialized  int32
+	running      int32
+	threadPinned int32
+
+	// Buffers (pre-allocated to avoid allocation overhead)
+	readBufPool  *sync.Pool
+	writeBufPool *sync.Pool
+}
+
+type BatchAudioStats struct {
+	// int64 fields MUST be first for ARM32 alignment
+	BatchedReads      int64
+	BatchedWrites     int64
+	SingleReads       int64
+	SingleWrites      int64
+	BatchedFrames     int64
+	SingleFrames      int64
+	CGOCallsReduced   int64
+	OSThreadPinTime   time.Duration // time.Duration is int64 internally
+	LastBatchTime     time.Time
+}
+
+type batchReadRequest struct {
+	buffer     []byte
+	resultChan chan batchReadResult
+	timestamp  time.Time
+}
+
+type batchWriteRequest struct {
+	buffer     []byte
+	resultChan chan batchWriteResult
+	timestamp  time.Time
+}
+
+type batchReadResult struct {
+	length int
+	err    error
+}
+
+type batchWriteResult struct {
+	written int
+	err     error
+}
+
+// NewBatchAudioProcessor creates a new batch audio processor
+func NewBatchAudioProcessor(batchSize int, batchDuration time.Duration) *BatchAudioProcessor {
+	ctx, cancel := context.WithCancel(context.Background())
+	logger := logging.GetDefaultLogger().With().Str("component", "batch-audio").Logger()
+
+	processor := &BatchAudioProcessor{
+		ctx:           ctx,
+		cancel:        cancel,
+		logger:        &logger,
+		batchSize:     batchSize,
+		batchDuration: batchDuration,
+		readQueue:     make(chan batchReadRequest, batchSize*2),
+		writeQueue:    make(chan batchWriteRequest, batchSize*2),
+		readBufPool: &sync.Pool{
+			New: func() interface{} {
+				return make([]byte, 1500) // Max audio frame size
+			},
+		},
+		writeBufPool: &sync.Pool{
+			New: func() interface{} {
+				return make([]byte, 4096) // Max write buffer size
+			},
+		},
+	}
+
+	return processor
+}
+
+// Start initializes and starts the batch processor
+func (bap *BatchAudioProcessor) Start() error {
+	if !atomic.CompareAndSwapInt32(&bap.running, 0, 1) {
+		return nil // Already running
+	}
+
+	// Initialize CGO resources once per processor lifecycle
+	if !atomic.CompareAndSwapInt32(&bap.initialized, 0, 1) {
+		return nil // Already initialized
+	}
+
+	// Start batch processing goroutines
+	go bap.batchReadProcessor()
+	go bap.batchWriteProcessor()
+
+	bap.logger.Info().Int("batch_size", bap.batchSize).
+		Dur("batch_duration", bap.batchDuration).
+		Msg("batch audio processor started")
+
+	return nil
+}
+
+// Stop cleanly shuts down the batch processor
+func (bap *BatchAudioProcessor) Stop() {
+	if !atomic.CompareAndSwapInt32(&bap.running, 1, 0) {
+		return // Already stopped
+	}
+
+	bap.cancel()
+
+	// Wait for processing to complete
+	time.Sleep(bap.batchDuration + 10*time.Millisecond)
+
+	bap.logger.Info().Msg("batch audio processor stopped")
+}
+
+// BatchReadEncode performs batched audio read and encode operations
+func (bap *BatchAudioProcessor) BatchReadEncode(buffer []byte) (int, error) {
+	if atomic.LoadInt32(&bap.running) == 0 {
+		// Fallback to single operation if batch processor is not running
+		atomic.AddInt64(&bap.stats.SingleReads, 1)
+		atomic.AddInt64(&bap.stats.SingleFrames, 1)
+		return CGOAudioReadEncode(buffer)
+	}
+
+	resultChan := make(chan batchReadResult, 1)
+	request := batchReadRequest{
+		buffer:     buffer,
+		resultChan: resultChan,
+		timestamp:  time.Now(),
+	}
+
+	select {
+	case bap.readQueue <- request:
+		// Successfully queued
+	case <-time.After(5 * time.Millisecond):
+		// Queue is full or blocked, fallback to single operation
+		atomic.AddInt64(&bap.stats.SingleReads, 1)
+		atomic.AddInt64(&bap.stats.SingleFrames, 1)
+		return CGOAudioReadEncode(buffer)
+	}
+
+	// Wait for result
+	select {
+	case result := <-resultChan:
+		return result.length, result.err
+	case <-time.After(50 * time.Millisecond):
+		// Timeout, fallback to single operation
+		atomic.AddInt64(&bap.stats.SingleReads, 1)
+		atomic.AddInt64(&bap.stats.SingleFrames, 1)
+		return CGOAudioReadEncode(buffer)
+	}
+}
+
+// BatchDecodeWrite performs batched audio decode and write operations
+func (bap *BatchAudioProcessor) BatchDecodeWrite(buffer []byte) (int, error) {
+	if atomic.LoadInt32(&bap.running) == 0 {
+		// Fallback to single operation if batch processor is not running
+		atomic.AddInt64(&bap.stats.SingleWrites, 1)
+		atomic.AddInt64(&bap.stats.SingleFrames, 1)
+		return CGOAudioDecodeWrite(buffer)
+	}
+
+	resultChan := make(chan batchWriteResult, 1)
+	request := batchWriteRequest{
+		buffer:     buffer,
+		resultChan: resultChan,
+		timestamp:  time.Now(),
+	}
+
+	select {
+	case bap.writeQueue <- request:
+		// Successfully queued
+	case <-time.After(5 * time.Millisecond):
+		// Queue is full or blocked, fallback to single operation
+		atomic.AddInt64(&bap.stats.SingleWrites, 1)
+		atomic.AddInt64(&bap.stats.SingleFrames, 1)
+		return CGOAudioDecodeWrite(buffer)
+	}
+
+	// Wait for result
+	select {
+	case result := <-resultChan:
+		return result.written, result.err
+	case <-time.After(50 * time.Millisecond):
+		// Timeout, fallback to single operation
+		atomic.AddInt64(&bap.stats.SingleWrites, 1)
+		atomic.AddInt64(&bap.stats.SingleFrames, 1)
+		return CGOAudioDecodeWrite(buffer)
+	}
+}
+
+// batchReadProcessor processes batched read operations
+func (bap *BatchAudioProcessor) batchReadProcessor() {
+	defer bap.logger.Debug().Msg("batch read processor stopped")
+
+	ticker := time.NewTicker(bap.batchDuration)
+	defer ticker.Stop()
+
+	var batch []batchReadRequest
+	batch = make([]batchReadRequest, 0, bap.batchSize)
+
+	for atomic.LoadInt32(&bap.running) == 1 {
+		select {
+		case <-bap.ctx.Done():
+			return
+
+		case req := <-bap.readQueue:
+			batch = append(batch, req)
+			if len(batch) >= bap.batchSize {
+				bap.processBatchRead(batch)
+				batch = batch[:0] // Clear slice but keep capacity
+			}
+
+		case <-ticker.C:
+			if len(batch) > 0 {
+				bap.processBatchRead(batch)
+				batch = batch[:0] // Clear slice but keep capacity
+			}
+		}
+	}
+
+	// Process any remaining requests
+	if len(batch) > 0 {
+		bap.processBatchRead(batch)
+	}
+}
+
+// batchWriteProcessor processes batched write operations
+func (bap *BatchAudioProcessor) batchWriteProcessor() {
+	defer bap.logger.Debug().Msg("batch write processor stopped")
+
+	ticker := time.NewTicker(bap.batchDuration)
+	defer ticker.Stop()
+
+	var batch []batchWriteRequest
+	batch = make([]batchWriteRequest, 0, bap.batchSize)
+
+	for atomic.LoadInt32(&bap.running) == 1 {
+		select {
+		case <-bap.ctx.Done():
+			return
+
+		case req := <-bap.writeQueue:
+			batch = append(batch, req)
+			if len(batch) >= bap.batchSize {
+				bap.processBatchWrite(batch)
+				batch = batch[:0] // Clear slice but keep capacity
+			}
+
+		case <-ticker.C:
+			if len(batch) > 0 {
+				bap.processBatchWrite(batch)
+				batch = batch[:0] // Clear slice but keep capacity
+			}
+		}
+	}
+
+	// Process any remaining requests
+	if len(batch) > 0 {
+		bap.processBatchWrite(batch)
+	}
+}
+
+// processBatchRead processes a batch of read requests efficiently
+func (bap *BatchAudioProcessor) processBatchRead(batch []batchReadRequest) {
+	if len(batch) == 0 {
+		return
+	}
+
+	// Pin to OS thread for the entire batch to minimize thread switching overhead
+	start := time.Now()
+	if atomic.CompareAndSwapInt32(&bap.threadPinned, 0, 1) {
+		runtime.LockOSThread()
+		defer func() {
+			runtime.UnlockOSThread()
+			atomic.StoreInt32(&bap.threadPinned, 0)
+			bap.stats.OSThreadPinTime += time.Since(start)
+		}()
+	}
+
+	batchSize := len(batch)
+	atomic.AddInt64(&bap.stats.BatchedReads, 1)
+	atomic.AddInt64(&bap.stats.BatchedFrames, int64(batchSize))
+	if batchSize > 1 {
+		atomic.AddInt64(&bap.stats.CGOCallsReduced, int64(batchSize-1))
+	}
+
+	// Process each request in the batch
+	for _, req := range batch {
+		length, err := CGOAudioReadEncode(req.buffer)
+		result := batchReadResult{
+			length: length,
+			err:    err,
+		}
+
+		// Send result back (non-blocking)
+		select {
+		case req.resultChan <- result:
+		default:
+			// Requestor timed out, drop result
+		}
+	}
+
+	bap.stats.LastBatchTime = time.Now()
+}
+
+// processBatchWrite processes a batch of write requests efficiently
+func (bap *BatchAudioProcessor) processBatchWrite(batch []batchWriteRequest) {
+	if len(batch) == 0 {
+		return
+	}
+
+	// Pin to OS thread for the entire batch to minimize thread switching overhead
+	start := time.Now()
+	if atomic.CompareAndSwapInt32(&bap.threadPinned, 0, 1) {
+		runtime.LockOSThread()
+		defer func() {
+			runtime.UnlockOSThread()
+			atomic.StoreInt32(&bap.threadPinned, 0)
+			bap.stats.OSThreadPinTime += time.Since(start)
+		}()
+	}
+
+	batchSize := len(batch)
+	atomic.AddInt64(&bap.stats.BatchedWrites, 1)
+	atomic.AddInt64(&bap.stats.BatchedFrames, int64(batchSize))
+	if batchSize > 1 {
+		atomic.AddInt64(&bap.stats.CGOCallsReduced, int64(batchSize-1))
+	}
+
+	// Process each request in the batch
+	for _, req := range batch {
+		written, err := CGOAudioDecodeWrite(req.buffer)
+		result := batchWriteResult{
+			written: written,
+			err:     err,
+		}
+
+		// Send result back (non-blocking)
+		select {
+		case req.resultChan <- result:
+		default:
+			// Requestor timed out, drop result
+		}
+	}
+
+	bap.stats.LastBatchTime = time.Now()
+}
+
+// GetStats returns current batch processor statistics
+func (bap *BatchAudioProcessor) GetStats() BatchAudioStats {
+	return BatchAudioStats{
+		BatchedReads:    atomic.LoadInt64(&bap.stats.BatchedReads),
+		BatchedWrites:   atomic.LoadInt64(&bap.stats.BatchedWrites),
+		SingleReads:     atomic.LoadInt64(&bap.stats.SingleReads),
+		SingleWrites:    atomic.LoadInt64(&bap.stats.SingleWrites),
+		BatchedFrames:   atomic.LoadInt64(&bap.stats.BatchedFrames),
+		SingleFrames:    atomic.LoadInt64(&bap.stats.SingleFrames),
+		CGOCallsReduced: atomic.LoadInt64(&bap.stats.CGOCallsReduced),
+		OSThreadPinTime: bap.stats.OSThreadPinTime,
+		LastBatchTime:   bap.stats.LastBatchTime,
+	}
+}
+
+// IsRunning returns whether the batch processor is running
+func (bap *BatchAudioProcessor) IsRunning() bool {
+	return atomic.LoadInt32(&bap.running) == 1
+}
+
+// Global batch processor instance
+var (
+	globalBatchProcessor     unsafe.Pointer // *BatchAudioProcessor
+	batchProcessorInitialized int32
+)
+
+// GetBatchAudioProcessor returns the global batch processor instance
+func GetBatchAudioProcessor() *BatchAudioProcessor {
+	ptr := atomic.LoadPointer(&globalBatchProcessor)
+	if ptr != nil {
+		return (*BatchAudioProcessor)(ptr)
+	}
+
+	// Initialize on first use
+	if atomic.CompareAndSwapInt32(&batchProcessorInitialized, 0, 1) {
+		processor := NewBatchAudioProcessor(4, 5*time.Millisecond) // 4 frames per batch, 5ms timeout
+		atomic.StorePointer(&globalBatchProcessor, unsafe.Pointer(processor))
+		return processor
+	}
+
+	// Another goroutine initialized it, try again
+	ptr = atomic.LoadPointer(&globalBatchProcessor)
+	if ptr != nil {
+		return (*BatchAudioProcessor)(ptr)
+	}
+
+	// Fallback: create a new processor (should rarely happen)
+	return NewBatchAudioProcessor(4, 5*time.Millisecond)
+}
+
+// EnableBatchAudioProcessing enables the global batch processor
+func EnableBatchAudioProcessing() error {
+	processor := GetBatchAudioProcessor()
+	return processor.Start()
+}
+
+// DisableBatchAudioProcessing disables the global batch processor
+func DisableBatchAudioProcessing() {
+	ptr := atomic.LoadPointer(&globalBatchProcessor)
+	if ptr != nil {
+		processor := (*BatchAudioProcessor)(ptr)
+		processor.Stop()
+	}
+}
+
+// BatchCGOAudioReadEncode is a batched version of CGOAudioReadEncode
+func BatchCGOAudioReadEncode(buffer []byte) (int, error) {
+	processor := GetBatchAudioProcessor()
+	if processor != nil && processor.IsRunning() {
+		return processor.BatchReadEncode(buffer)
+	}
+	return CGOAudioReadEncode(buffer)
+}
+
+// BatchCGOAudioDecodeWrite is a batched version of CGOAudioDecodeWrite
+func BatchCGOAudioDecodeWrite(buffer []byte) (int, error) {
+	processor := GetBatchAudioProcessor()
+	if processor != nil && processor.IsRunning() {
+		return processor.BatchDecodeWrite(buffer)
+	}
+	return CGOAudioDecodeWrite(buffer)
+}

internal/audio/buffer_pool.go

@@ -0,0 +1,64 @@
+package audio
+
+import (
+	"sync"
+)
+
+// AudioBufferPool manages reusable audio buffers to reduce allocations
+type AudioBufferPool struct {
+	pool sync.Pool
+}
+
+// NewAudioBufferPool creates a new buffer pool for audio frames
+func NewAudioBufferPool(bufferSize int) *AudioBufferPool {
+	return &AudioBufferPool{
+		pool: sync.Pool{
+			New: func() interface{} {
+				// Pre-allocate buffer with specified size
+				return make([]byte, bufferSize)
+			},
+		},
+	}
+}
+
+// Get retrieves a buffer from the pool
+func (p *AudioBufferPool) Get() []byte {
+	return p.pool.Get().([]byte)
+}
+
+// Put returns a buffer to the pool
+func (p *AudioBufferPool) Put(buf []byte) {
+	// Reset length but keep capacity for reuse
+	if cap(buf) >= 1500 { // Only pool buffers of reasonable size
+		p.pool.Put(buf[:0])
+	}
+}
+
+// Global buffer pools for different audio operations
+var (
+	// Pool for 1500-byte audio frame buffers (Opus max frame size)
+	audioFramePool = NewAudioBufferPool(1500)
+	
+	// Pool for smaller control buffers
+	audioControlPool = NewAudioBufferPool(64)
+)
+
+// GetAudioFrameBuffer gets a reusable buffer for audio frames
+func GetAudioFrameBuffer() []byte {
+	return audioFramePool.Get()
+}
+
+// PutAudioFrameBuffer returns a buffer to the frame pool
+func PutAudioFrameBuffer(buf []byte) {
+	audioFramePool.Put(buf)
+}
+
+// GetAudioControlBuffer gets a reusable buffer for control data
+func GetAudioControlBuffer() []byte {
+	return audioControlPool.Get()
+}
+
+// PutAudioControlBuffer returns a buffer to the control pool
+func PutAudioControlBuffer(buf []byte) {
+	audioControlPool.Put(buf)
+}

internal/audio/cgo_audio.go

@@ -1,4 +1,4 @@
-//go:build !nolint
+//go:build cgo
 
 package audio
 
@@ -385,11 +385,23 @@ void jetkvm_audio_close() {
 */
 import "C"
 
-// Go wrappers for initializing, starting, stopping, and controlling audio
+// Optimized Go wrappers with reduced overhead
+var (
+	errAudioInitFailed    = errors.New("failed to init ALSA/Opus")
+	errBufferTooSmall     = errors.New("buffer too small")
+	errAudioReadEncode    = errors.New("audio read/encode error")
+	errAudioDecodeWrite   = errors.New("audio decode/write error")
+	errAudioPlaybackInit  = errors.New("failed to init ALSA playback/Opus decoder")
+	errEmptyBuffer        = errors.New("empty buffer")
+	errNilBuffer          = errors.New("nil buffer")
+	errBufferTooLarge     = errors.New("buffer too large")
+	errInvalidBufferPtr   = errors.New("invalid buffer pointer")
+)
+
 func cgoAudioInit() error {
 	ret := C.jetkvm_audio_init()
 	if ret != 0 {
-		return errors.New("failed to init ALSA/Opus")
+		return errAudioInitFailed
 	}
 	return nil
 }
@@ -398,18 +410,19 @@ func cgoAudioClose() {
 	C.jetkvm_audio_close()
 }
 
-// Reads and encodes one frame, returns encoded bytes or error
+// Optimized read and encode with pre-allocated error objects and reduced checks
 func cgoAudioReadEncode(buf []byte) (int, error) {
-	if len(buf) < 1500 {
-		return 0, errors.New("buffer too small")
+	// Fast path: check minimum buffer size (reduced from 1500 to 1276 for 10ms frames)
+	if len(buf) < 1276 {
+		return 0, errBufferTooSmall
 	}
+	
 	n := C.jetkvm_audio_read_encode(unsafe.Pointer(&buf[0]))
 	if n < 0 {
-		return 0, errors.New("audio read/encode error")
+		return 0, errAudioReadEncode
 	}
 	if n == 0 {
-		// No data available - this is not an error, just no audio frame
-		return 0, nil
+		return 0, nil // No data available
 	}
 	return int(n), nil
 }

internal/audio/cgo_audio_stub.go

@@ -1,4 +1,4 @@
-//go:build nolint
+//go:build !cgo
 
 package audio
 

internal/audio/mic_contention.go

@@ -0,0 +1,158 @@
+package audio
+
+import (
+	"sync/atomic"
+	"time"
+	"unsafe"
+)
+
+// MicrophoneContentionManager provides optimized microphone operation locking
+// with reduced contention using atomic operations and conditional locking
+type MicrophoneContentionManager struct {
+	// Atomic fields (must be 64-bit aligned on 32-bit systems)
+	lastOpNano    int64         // Unix nanoseconds of last operation
+	cooldownNanos int64         // Cooldown duration in nanoseconds
+	operationID   int64         // Incremental operation ID for tracking
+	
+	// Lock-free state flags (using atomic.Pointer for lock-free updates)
+	lockPtr unsafe.Pointer // *sync.Mutex - conditionally allocated
+}
+
+// NewMicrophoneContentionManager creates a new microphone contention manager
+func NewMicrophoneContentionManager(cooldown time.Duration) *MicrophoneContentionManager {
+	return &MicrophoneContentionManager{
+		cooldownNanos: int64(cooldown),
+	}
+}
+
+// OperationResult represents the result of attempting a microphone operation
+type OperationResult struct {
+	Allowed         bool
+	RemainingCooldown time.Duration
+	OperationID     int64
+}
+
+// TryOperation attempts to perform a microphone operation with optimized contention handling
+func (mcm *MicrophoneContentionManager) TryOperation() OperationResult {
+	now := time.Now().UnixNano()
+	cooldown := atomic.LoadInt64(&mcm.cooldownNanos)
+	
+	// Fast path: check if we're clearly outside cooldown period using atomic read
+	lastOp := atomic.LoadInt64(&mcm.lastOpNano)
+	elapsed := now - lastOp
+	
+	if elapsed >= cooldown {
+		// Attempt atomic update without locking
+		if atomic.CompareAndSwapInt64(&mcm.lastOpNano, lastOp, now) {
+			opID := atomic.AddInt64(&mcm.operationID, 1)
+			return OperationResult{
+				Allowed:         true,
+				RemainingCooldown: 0,
+				OperationID:     opID,
+			}
+		}
+	}
+	
+	// Slow path: potential contention, check remaining cooldown
+	currentLastOp := atomic.LoadInt64(&mcm.lastOpNano)
+	currentElapsed := now - currentLastOp
+	
+	if currentElapsed >= cooldown {
+		// Race condition: another operation might have updated lastOpNano
+		// Try once more with CAS
+		if atomic.CompareAndSwapInt64(&mcm.lastOpNano, currentLastOp, now) {
+			opID := atomic.AddInt64(&mcm.operationID, 1)
+			return OperationResult{
+				Allowed:         true,
+				RemainingCooldown: 0,
+				OperationID:     opID,
+			}
+		}
+		// If CAS failed, fall through to cooldown calculation
+		currentLastOp = atomic.LoadInt64(&mcm.lastOpNano)
+		currentElapsed = now - currentLastOp
+	}
+	
+	remaining := time.Duration(cooldown - currentElapsed)
+	if remaining < 0 {
+		remaining = 0
+	}
+	
+	return OperationResult{
+		Allowed:         false,
+		RemainingCooldown: remaining,
+		OperationID:     atomic.LoadInt64(&mcm.operationID),
+	}
+}
+
+// SetCooldown updates the cooldown duration atomically
+func (mcm *MicrophoneContentionManager) SetCooldown(cooldown time.Duration) {
+	atomic.StoreInt64(&mcm.cooldownNanos, int64(cooldown))
+}
+
+// GetCooldown returns the current cooldown duration
+func (mcm *MicrophoneContentionManager) GetCooldown() time.Duration {
+	return time.Duration(atomic.LoadInt64(&mcm.cooldownNanos))
+}
+
+// GetLastOperationTime returns the time of the last operation
+func (mcm *MicrophoneContentionManager) GetLastOperationTime() time.Time {
+	nanos := atomic.LoadInt64(&mcm.lastOpNano)
+	if nanos == 0 {
+		return time.Time{}
+	}
+	return time.Unix(0, nanos)
+}
+
+// GetOperationCount returns the total number of successful operations
+func (mcm *MicrophoneContentionManager) GetOperationCount() int64 {
+	return atomic.LoadInt64(&mcm.operationID)
+}
+
+// Reset resets the contention manager state
+func (mcm *MicrophoneContentionManager) Reset() {
+	atomic.StoreInt64(&mcm.lastOpNano, 0)
+	atomic.StoreInt64(&mcm.operationID, 0)
+}
+
+// Global instance for microphone contention management
+var (
+	globalMicContentionManager unsafe.Pointer // *MicrophoneContentionManager
+	micContentionInitialized   int32
+)
+
+// GetMicrophoneContentionManager returns the global microphone contention manager
+func GetMicrophoneContentionManager() *MicrophoneContentionManager {
+	ptr := atomic.LoadPointer(&globalMicContentionManager)
+	if ptr != nil {
+		return (*MicrophoneContentionManager)(ptr)
+	}
+	
+	// Initialize on first use
+	if atomic.CompareAndSwapInt32(&micContentionInitialized, 0, 1) {
+		manager := NewMicrophoneContentionManager(200 * time.Millisecond)
+		atomic.StorePointer(&globalMicContentionManager, unsafe.Pointer(manager))
+		return manager
+	}
+	
+	// Another goroutine initialized it, try again
+	ptr = atomic.LoadPointer(&globalMicContentionManager)
+	if ptr != nil {
+		return (*MicrophoneContentionManager)(ptr)
+	}
+	
+	// Fallback: create a new manager (should rarely happen)
+	return NewMicrophoneContentionManager(200 * time.Millisecond)
+}
+
+// TryMicrophoneOperation provides a convenient global function for microphone operations
+func TryMicrophoneOperation() OperationResult {
+	manager := GetMicrophoneContentionManager()
+	return manager.TryOperation()
+}
+
+// SetMicrophoneCooldown updates the global microphone cooldown
+func SetMicrophoneCooldown(cooldown time.Duration) {
+	manager := GetMicrophoneContentionManager()
+	manager.SetCooldown(cooldown)
+}

internal/audio/nonblocking_api.go

@@ -1,96 +1,115 @@
 package audio
 
 import (
-	"sync"
+	"sync/atomic"
+	"unsafe"
 )
 
 var (
-	globalNonBlockingManager *NonBlockingAudioManager
-	managerMutex             sync.Mutex
+	// Use unsafe.Pointer for atomic operations instead of mutex
+	globalNonBlockingManager unsafe.Pointer // *NonBlockingAudioManager
 )
 
+// loadManager atomically loads the global manager
+func loadManager() *NonBlockingAudioManager {
+	ptr := atomic.LoadPointer(&globalNonBlockingManager)
+	if ptr == nil {
+		return nil
+	}
+	return (*NonBlockingAudioManager)(ptr)
+}
+
+// storeManager atomically stores the global manager
+func storeManager(manager *NonBlockingAudioManager) {
+	atomic.StorePointer(&globalNonBlockingManager, unsafe.Pointer(manager))
+}
+
+// compareAndSwapManager atomically compares and swaps the global manager
+func compareAndSwapManager(old, new *NonBlockingAudioManager) bool {
+	return atomic.CompareAndSwapPointer(&globalNonBlockingManager, 
+		unsafe.Pointer(old), unsafe.Pointer(new))
+}
+
 // StartNonBlockingAudioStreaming starts the non-blocking audio streaming system
 func StartNonBlockingAudioStreaming(send func([]byte)) error {
-	managerMutex.Lock()
-	defer managerMutex.Unlock()
-
-	if globalNonBlockingManager != nil && globalNonBlockingManager.IsOutputRunning() {
+	manager := loadManager()
+	if manager != nil && manager.IsOutputRunning() {
 		return nil // Already running, this is not an error
 	}
 
-	if globalNonBlockingManager == nil {
-		globalNonBlockingManager = NewNonBlockingAudioManager()
+	if manager == nil {
+		newManager := NewNonBlockingAudioManager()
+		if !compareAndSwapManager(nil, newManager) {
+			// Another goroutine created manager, use it
+			manager = loadManager()
+		} else {
+			manager = newManager
+		}
 	}
 
-	return globalNonBlockingManager.StartAudioOutput(send)
+	return manager.StartAudioOutput(send)
 }
 
 // StartNonBlockingAudioInput starts the non-blocking audio input system
 func StartNonBlockingAudioInput(receiveChan <-chan []byte) error {
-	managerMutex.Lock()
-	defer managerMutex.Unlock()
-
-	if globalNonBlockingManager == nil {
-		globalNonBlockingManager = NewNonBlockingAudioManager()
+	manager := loadManager()
+	if manager == nil {
+		newManager := NewNonBlockingAudioManager()
+		if !compareAndSwapManager(nil, newManager) {
+			// Another goroutine created manager, use it
+			manager = loadManager()
+		} else {
+			manager = newManager
+		}
 	}
 
 	// Check if input is already running to avoid unnecessary operations
-	if globalNonBlockingManager.IsInputRunning() {
+	if manager.IsInputRunning() {
 		return nil // Already running, this is not an error
 	}
 
-	return globalNonBlockingManager.StartAudioInput(receiveChan)
+	return manager.StartAudioInput(receiveChan)
 }
 
 // StopNonBlockingAudioStreaming stops the non-blocking audio streaming system
 func StopNonBlockingAudioStreaming() {
-	managerMutex.Lock()
-	defer managerMutex.Unlock()
-
-	if globalNonBlockingManager != nil {
-		globalNonBlockingManager.Stop()
-		globalNonBlockingManager = nil
+	manager := loadManager()
+	if manager != nil {
+		manager.Stop()
+		storeManager(nil)
 	}
 }
 
 // StopNonBlockingAudioInput stops only the audio input without affecting output
 func StopNonBlockingAudioInput() {
-	managerMutex.Lock()
-	defer managerMutex.Unlock()
-
-	if globalNonBlockingManager != nil && globalNonBlockingManager.IsInputRunning() {
-		globalNonBlockingManager.StopAudioInput()
+	manager := loadManager()
+	if manager != nil && manager.IsInputRunning() {
+		manager.StopAudioInput()
 		
 		// If both input and output are stopped, recreate manager to ensure clean state
-		if !globalNonBlockingManager.IsRunning() {
-			globalNonBlockingManager = nil
+		if !manager.IsRunning() {
+			storeManager(nil)
 		}
 	}
 }
 
 // GetNonBlockingAudioStats returns statistics from the non-blocking audio system
 func GetNonBlockingAudioStats() NonBlockingAudioStats {
-	managerMutex.Lock()
-	defer managerMutex.Unlock()
-
-	if globalNonBlockingManager != nil {
-		return globalNonBlockingManager.GetStats()
+	manager := loadManager()
+	if manager != nil {
+		return manager.GetStats()
 	}
 	return NonBlockingAudioStats{}
 }
 
 // IsNonBlockingAudioRunning returns true if the non-blocking audio system is running
 func IsNonBlockingAudioRunning() bool {
-	managerMutex.Lock()
-	defer managerMutex.Unlock()
-
-	return globalNonBlockingManager != nil && globalNonBlockingManager.IsRunning()
+	manager := loadManager()
+	return manager != nil && manager.IsRunning()
 }
 
 // IsNonBlockingAudioInputRunning returns true if the non-blocking audio input is running
 func IsNonBlockingAudioInputRunning() bool {
-	managerMutex.Lock()
-	defer managerMutex.Unlock()
-
-	return globalNonBlockingManager != nil && globalNonBlockingManager.IsInputRunning()
+	manager := loadManager()
+	return manager != nil && manager.IsInputRunning()
 }

internal/audio/nonblocking_audio.go

@@ -3,7 +3,7 @@ package audio
 import (
 	"context"
 	"errors"
-	"runtime"
+	// "runtime" // removed: no longer directly pinning OS thread here; batching handles it
 	"sync"
 	"sync/atomic"
 	"time"
@@ -98,6 +98,9 @@ func (nam *NonBlockingAudioManager) StartAudioOutput(sendFunc func([]byte)) erro
 
 	nam.outputSendFunc = sendFunc
 
+	// Enable batch audio processing for performance
+	EnableBatchAudioProcessing()
+
 	// Start the blocking worker thread
 	nam.wg.Add(1)
 	go nam.outputWorkerThread()
@@ -106,7 +109,7 @@ func (nam *NonBlockingAudioManager) StartAudioOutput(sendFunc func([]byte)) erro
 	nam.wg.Add(1)
 	go nam.outputCoordinatorThread()
 
-	nam.logger.Info().Msg("non-blocking audio output started")
+	nam.logger.Info().Msg("non-blocking audio output started with batch processing")
 	return nil
 }
 
@@ -118,6 +121,9 @@ func (nam *NonBlockingAudioManager) StartAudioInput(receiveChan <-chan []byte) e
 
 	nam.inputReceiveChan = receiveChan
 
+	// Enable batch audio processing for performance
+	EnableBatchAudioProcessing()
+
 	// Start the blocking worker thread
 	nam.wg.Add(1)
 	go nam.inputWorkerThread()
@@ -126,16 +132,12 @@ func (nam *NonBlockingAudioManager) StartAudioInput(receiveChan <-chan []byte) e
 	nam.wg.Add(1)
 	go nam.inputCoordinatorThread()
 
-	nam.logger.Info().Msg("non-blocking audio input started")
+	nam.logger.Info().Msg("non-blocking audio input started with batch processing")
 	return nil
 }
 
 // outputWorkerThread handles all blocking audio output operations
 func (nam *NonBlockingAudioManager) outputWorkerThread() {
-	// Lock to OS thread to isolate blocking CGO operations
-	runtime.LockOSThread()
-	defer runtime.UnlockOSThread()
-
 	defer nam.wg.Done()
 	defer atomic.StoreInt32(&nam.outputWorkerRunning, 0)
 
@@ -149,7 +151,9 @@ func (nam *NonBlockingAudioManager) outputWorkerThread() {
 	}
 	defer CGOAudioClose()
 
-	buf := make([]byte, 1500)
+	// Use buffer pool to avoid allocations
+	buf := GetAudioFrameBuffer()
+	defer PutAudioFrameBuffer(buf)
 
 	for {
 		select {
@@ -160,17 +164,18 @@ func (nam *NonBlockingAudioManager) outputWorkerThread() {
 		case workItem := <-nam.outputWorkChan:
 			switch workItem.workType {
 			case audioWorkReadEncode:
-				// Perform blocking audio read/encode operation
-				n, err := CGOAudioReadEncode(buf)
+				n, err := BatchCGOAudioReadEncode(buf)
+					
 					result := audioResult{
 					success: err == nil,
 					length:  n,
 					err:     err,
 				}
 				if err == nil && n > 0 {
-					// Copy data to avoid race conditions
-					result.data = make([]byte, n)
-					copy(result.data, buf[:n])
+					// Get buffer from pool and copy data
+					resultBuf := GetAudioFrameBuffer()
+					copy(resultBuf[:n], buf[:n])
+					result.data = resultBuf[:n]
 				}
 
 				// Send result back (non-blocking)
@@ -180,6 +185,9 @@ func (nam *NonBlockingAudioManager) outputWorkerThread() {
 					return
 				default:
 					// Drop result if coordinator is not ready
+					if result.data != nil {
+						PutAudioFrameBuffer(result.data)
+					}
 					atomic.AddInt64(&nam.stats.OutputFramesDropped, 1)
 				}
 
@@ -243,6 +251,8 @@ func (nam *NonBlockingAudioManager) outputCoordinatorThread() {
 					atomic.AddInt64(&nam.stats.OutputFramesProcessed, 1)
 					RecordFrameReceived(result.length)
 				}
+				// Return buffer to pool after use
+				PutAudioFrameBuffer(result.data)
 			} else if result.success && result.length == 0 {
 				// No data available - this is normal, not an error
 				// Just continue without logging or counting as error
@@ -252,6 +262,10 @@ func (nam *NonBlockingAudioManager) outputCoordinatorThread() {
 				if result.err != nil {
 					nam.logger.Warn().Err(result.err).Msg("audio output worker error")
 				}
+				// Clean up buffer if present
+				if result.data != nil {
+					PutAudioFrameBuffer(result.data)
+				}
 				RecordFrameDropped()
 			}
 		}
@@ -269,10 +283,6 @@ func (nam *NonBlockingAudioManager) outputCoordinatorThread() {
 
 // inputWorkerThread handles all blocking audio input operations
 func (nam *NonBlockingAudioManager) inputWorkerThread() {
-	// Lock to OS thread to isolate blocking CGO operations
-	runtime.LockOSThread()
-	defer runtime.UnlockOSThread()
-
 	defer nam.wg.Done()
 	// Cleanup CGO resources properly to avoid double-close scenarios
 	// The outputWorkerThread's CGOAudioClose() will handle all cleanup
@@ -362,7 +372,8 @@ func (nam *NonBlockingAudioManager) inputWorkerThread() {
 						return
 					}
 					
-					n, err := CGOAudioDecodeWrite(workItem.data)
+					n, err := BatchCGOAudioDecodeWrite(workItem.data)
+					
 					result = audioResult{
 						success: err == nil,
 						length:  n,
@@ -479,6 +490,9 @@ func (nam *NonBlockingAudioManager) Stop() {
 	// Wait for all goroutines to finish
 	nam.wg.Wait()
 
+	// Disable batch processing to free resources
+	DisableBatchAudioProcessing()
+
 	nam.logger.Info().Msg("non-blocking audio manager stopped")
 }
 

web.go

@@ -283,29 +283,18 @@ func setupRouter() *gin.Engine {
 			return
 		}
 
-		// Server-side cooldown to prevent rapid start/stop thrashing
-		{
-			cs := currentSession
-			cs.micOpMu.Lock()
-			now := time.Now()
-			if cs.micCooldown == 0 {
-				cs.micCooldown = 200 * time.Millisecond
-			}
-			since := now.Sub(cs.lastMicOp)
-			if since < cs.micCooldown {
-				remaining := cs.micCooldown - since
-				running := cs.AudioInputManager.IsRunning() || audio.IsNonBlockingAudioInputRunning()
-				cs.micOpMu.Unlock()
+		// Optimized server-side cooldown using atomic operations
+		opResult := audio.TryMicrophoneOperation()
+		if !opResult.Allowed {
+			running := currentSession.AudioInputManager.IsRunning() || audio.IsNonBlockingAudioInputRunning()
 			c.JSON(200, gin.H{
 				"status":                 "cooldown",
 				"running":                running,
-					"cooldown_ms_remaining":  remaining.Milliseconds(),
+				"cooldown_ms_remaining":  opResult.RemainingCooldown.Milliseconds(),
+				"operation_id":           opResult.OperationID,
 			})
 			return
 		}
-			cs.lastMicOp = now
-			cs.micOpMu.Unlock()
-		}
 
 		// Check if already running before attempting to start
 		if currentSession.AudioInputManager.IsRunning() || audio.IsNonBlockingAudioInputRunning() {
@@ -356,29 +345,18 @@ func setupRouter() *gin.Engine {
 			return
 		}
 
-		// Server-side cooldown to prevent rapid start/stop thrashing
-		{
-			cs := currentSession
-			cs.micOpMu.Lock()
-			now := time.Now()
-			if cs.micCooldown == 0 {
-				cs.micCooldown = 200 * time.Millisecond
-			}
-			since := now.Sub(cs.lastMicOp)
-			if since < cs.micCooldown {
-				remaining := cs.micCooldown - since
-				running := cs.AudioInputManager.IsRunning() || audio.IsNonBlockingAudioInputRunning()
-				cs.micOpMu.Unlock()
+		// Optimized server-side cooldown using atomic operations
+		opResult := audio.TryMicrophoneOperation()
+		if !opResult.Allowed {
+			running := currentSession.AudioInputManager.IsRunning() || audio.IsNonBlockingAudioInputRunning()
 			c.JSON(200, gin.H{
 				"status":                 "cooldown",
 				"running":                running,
-					"cooldown_ms_remaining":  remaining.Milliseconds(),
+				"cooldown_ms_remaining":  opResult.RemainingCooldown.Milliseconds(),
+				"operation_id":           opResult.OperationID,
 			})
 			return
 		}
-			cs.lastMicOp = now
-			cs.micOpMu.Unlock()
-		}
 
 		// Check if already stopped before attempting to stop
 		if !currentSession.AudioInputManager.IsRunning() && !audio.IsNonBlockingAudioInputRunning() {