Complete managed macOS platform bring-up

core/src/audio/audio_engine.cpp

@@ -566,13 +566,19 @@ void AudioEngine::init_recv_stream(uint32_t ssrc, const codec::OpusParams& p,
 }
 
 void AudioEngine::set_pcm_sink(PcmSink cb, void* user) {
-    pcm_sink_user_.store(user, std::memory_order_relaxed);
-    pcm_sink_.store(cb, std::memory_order_release);
+    std::lock_guard setter_lk(pcm_sink_set_mu_);
+    pcm_sink_.store(nullptr);
+    while (pcm_sink_active_.load() != 0) std::this_thread::yield();
+    pcm_sink_user_.store(user);
+    pcm_sink_.store(cb);
 }
 
 void AudioEngine::set_mixed_output_sink(MixedSink cb, void* user) {
-    mixed_sink_user_.store(user, std::memory_order_relaxed);
-    mixed_sink_.store(cb, std::memory_order_release);
+    std::lock_guard setter_lk(mixed_sink_set_mu_);
+    mixed_sink_.store(nullptr);
+    while (mixed_sink_active_.load() != 0) std::this_thread::yield();
+    mixed_sink_user_.store(user);
+    mixed_sink_.store(cb);
 }
 
 void AudioEngine::capture_data_cb(ma_device* dev, void* /*out*/,
@@ -734,17 +740,19 @@ void AudioEngine::on_playback(int16_t* out, ma_uint32 frames) {
             }
 
             // PCM sink: deliver decoded per-stream audio to external consumer (bots,
-            // transcription, recording) before it enters the hardware mix. Atomic relaxed-
-            // load is safe on the RT thread — the fn-ptr and user-ptr are independent
-            // pointer-sized values written together by set_pcm_sink (release store).
-            if (auto sink = pcm_sink_.load(std::memory_order_relaxed)) {
-                sink(pcm_sink_user_.load(std::memory_order_relaxed),
+            // transcription, recording) before it enters the hardware mix. The reader count
+            // makes disabling the callback a quiescence barrier while keeping this RT path
+            // lock-free and non-blocking.
+            pcm_sink_active_.fetch_add(1);
+            if (auto sink = pcm_sink_.load()) {
+                sink(pcm_sink_user_.load(),
                      stream.user_id, stream.stream_id,
                      stream.decode_scratch.data(),
                      static_cast<size_t>(n),
                      static_cast<uint32_t>(dec_channels),
                      params_.sample_rate);
             }
+            pcm_sink_active_.fetch_sub(1);
 
             stream.push_ring(stream.decode_scratch.data(), static_cast<size_t>(n));
             stream.playout_ts += static_cast<uint32_t>(n);
@@ -797,10 +805,12 @@ void AudioEngine::run_mixer_timer() {
     auto next = clock::now();
     while (!mixer_timer_stop_.load(std::memory_order_acquire)) {
         on_playback(mixer_scratch_.data(), spc);
-        if (auto sink = mixed_sink_.load(std::memory_order_relaxed)) {
-            sink(mixed_sink_user_.load(std::memory_order_relaxed), mixer_scratch_.data(), spc, ch,
+        mixed_sink_active_.fetch_add(1);
+        if (auto sink = mixed_sink_.load()) {
+            sink(mixed_sink_user_.load(), mixer_scratch_.data(), spc, ch,
                  params_.sample_rate);
         }
+        mixed_sink_active_.fetch_sub(1);
         next += period;
         // If we fell badly behind (e.g. the thread was descheduled), reset the deadline rather
         // than spin to catch up — the VPIO renderer rides its own clock + jitter ring.

core/src/audio/audio_engine.h

@@ -514,16 +514,22 @@ class AudioEngine {
 
     int frame_samples_ = 960;  // 20 ms @48 kHz
 
-    // External PCM tap: atomic fn-ptr + user-ptr pair. Written by set_pcm_sink (any thread);
-    // read by on_playback (RT thread) via relaxed load — safe for pointer-sized atomics.
+    // External PCM tap. The in-flight counter lets set_pcm_sink(nullptr, nullptr) wait for an
+    // already-entered callback without making the RT thread lock or block. Sequentially
+    // consistent operations are intentional here: once the setter observes zero readers, a
+    // later reader must observe the disabled callback before it can dereference the old user.
     std::atomic<PcmSink> pcm_sink_{nullptr};
     std::atomic<void*>   pcm_sink_user_{nullptr};
+    std::atomic<uint32_t> pcm_sink_active_{0};
+    std::mutex            pcm_sink_set_mu_;
 
     // External mixed-output sink + mode flag (iOS VPIO). mixed_sink_ is written by
     // set_mixed_output_sink (any thread); read by run_mixer_timer via relaxed load.
     // external_playback_ is read in start() to gate hardware-playback-device creation.
     std::atomic<MixedSink> mixed_sink_{nullptr};
     std::atomic<void*>     mixed_sink_user_{nullptr};
+    std::atomic<uint32_t>  mixed_sink_active_{0};
+    std::mutex             mixed_sink_set_mu_;
     bool                   external_playback_ = false;
 
     ::OpusDREDDecoder* dred_dec_ = nullptr;  // shared DRED decoder; null if unsupported