Fix music never playing - game speed nearly doubled (task #49)

Channel::isPlaying answered FALSE for a channel System::playSound had just
handed out with a success code. The guest polled it, concluded the track
had finished, and started the next one - forever. Each cycle reopened the
OBB asset filesystem ON THE GLThread, stalling frames directly.

Found from a user observation ("треки меняются очень быстро в меню паузы"),
then measured: a different track roughly every 1.5s, picked
non-sequentially (DEAD SARA, DEAD SARA, GREEN DAY, SKRILLEX, GREEN DAY,
CROSSES, ICONA POP, JOY FORMIDABLE), so not a walk through playlists.sb but
a fresh selection each time.

Now tracks the channels playSound hands out and reports those as playing
until Channel::stop drops them, so normal transitions still work. A
one-time log states plainly that there is no audio backend and playback is
simulated, instead of leaving that only in a comment.

Measured on the Pixel 6a with the race's own clock, same method both times:

    game time vs wall clock     7.69x  ->  3.99x
    useAssetsFileSystem       1.00/s   ->  0.00/s   (in-race, fully stopped)
    faults                         0   ->  0

27s of wall time advanced the race timer 18.82 -> 25.58 (6.76s); the two
sub-intervals gave 4.18x and 3.81x.

Third instance today of one failure shape: a shim reporting SUCCESS while
handing back an empty or negative value. See AndroidBitmap_lockPixels
(task #41) and the synthetic-swap premise (task #47).

Co-Authored-By: Claude <noreply@anthropic.com>

mpcore/src/main/cpp/emu/fmod_shims.cpp

@@ -1,5 +1,7 @@
 #include "fmod_shims.h"
 #include <cstring>
+#include <mutex>
+#include <set>
 
 namespace {
 
@@ -256,9 +258,37 @@ uint32_t Shim_System_getSoftwareFormat(GuestEngine& eng, uint32_t, uint32_t out1
 }
 uint32_t Shim_System_setSoftwareFormat(GuestEngine&, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t) { return kFmodOk; }
 uint32_t Shim_System_set3DSpeakerPosition(GuestEngine&, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t) { return kFmodOk; }
+// FIX (2026-09-19, task #49). Channels handed out by System::playSound and
+// not yet stopped. Without this the music never played: playSound reported
+// success and returned a fake channel, then Channel::isPlaying answered
+// FALSE for it, so the game concluded the track had finished and immediately
+// started the next one - forever. Measured in the pause menu: a different
+// track roughly every 1.5s, picked non-sequentially (DEAD SARA, DEAD SARA,
+// GREEN DAY, SKRILLEX, GREEN DAY, CROSSES, ICONA POP, JOY FORMIDABLE), so
+// not a walk through playlists.sb but a fresh selection each time.
+//
+// Third instance today of the same failure shape - a shim reporting SUCCESS
+// while handing back an empty/negative value (see AndroidBitmap_lockPixels
+// in gles_shim.cpp, task #41). Hence the log below: this engine has no audio
+// backend, so playback here is simulated, and that fact belongs in the log
+// rather than only in a comment.
+std::set<uint32_t> g_playingChannels;
+std::mutex g_playingChannelsMutex;
+
 uint32_t Shim_System_playSound(GuestEngine& eng, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t sp) {
     uint32_t out = eng.ReadIncomingArg(4, 0, 0, 0, 0, sp); // 5th total arg (this,channelid,sound,paused,+channel**)
-    OutPtr(eng, out, FakeHandle(eng));
+    GuestAddr channel = FakeHandle(eng);
+    OutPtr(eng, out, channel);
+    {
+        std::lock_guard<std::mutex> lock(g_playingChannelsMutex);
+        g_playingChannels.insert(channel);
+    }
+    static std::once_flag once;
+    std::call_once(once, [] {
+        Log("fmod_shims: no audio backend - System::playSound will report channels as PLAYING until "
+            "the guest stops them. Sound is silent by design; without this the guest sees every track "
+            "finish instantly and thrashes through the playlist (task #49).");
+    });
     return kFmodOk;
 }
 uint32_t Shim_System_setOutput(GuestEngine&, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t) { return kFmodOk; }
@@ -266,14 +296,29 @@ uint32_t Shim_System_setOutput(GuestEngine&, uint32_t, uint32_t, uint32_t, uint3
 // ---- Channel ----
 uint32_t Shim_Channel_setCallback(GuestEngine&, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t) { return kFmodOk; }
 uint32_t Shim_Channel_setPriority(GuestEngine&, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t) { return kFmodOk; }
-uint32_t Shim_Channel_stop(GuestEngine&, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t) { return kFmodOk; }
+uint32_t Shim_Channel_stop(GuestEngine&, uint32_t r0, uint32_t, uint32_t, uint32_t, uint32_t) {
+    // r0 is `this`. Dropping it here is what lets the guest end a track on its
+    // own terms - the next isPlaying() then honestly answers false, so normal
+    // transitions (race ends, menu change) still work.
+    std::lock_guard<std::mutex> lock(g_playingChannelsMutex);
+    g_playingChannels.erase(r0);
+    return kFmodOk;
+}
 uint32_t Shim_Channel_setMute(GuestEngine&, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t) { return kFmodOk; }
 uint32_t Shim_Channel_getPaused(GuestEngine& eng, uint32_t, uint32_t out, uint32_t, uint32_t, uint32_t) {
     OutBool(eng, out, false);
     return kFmodOk;
 }
-uint32_t Shim_Channel_isPlaying(GuestEngine& eng, uint32_t, uint32_t out, uint32_t, uint32_t, uint32_t) {
-    OutBool(eng, out, false);
+uint32_t Shim_Channel_isPlaying(GuestEngine& eng, uint32_t r0, uint32_t out, uint32_t, uint32_t, uint32_t) {
+    // See g_playingChannels' comment. Answering "playing" only for channels we
+    // actually handed out - rather than a blanket true - keeps a stale or
+    // never-started channel honest.
+    bool playing;
+    {
+        std::lock_guard<std::mutex> lock(g_playingChannelsMutex);
+        playing = g_playingChannels.count(r0) != 0;
+    }
+    OutBool(eng, out, playing);
     return kFmodOk;
 }
 uint32_t Shim_Channel_setPaused(GuestEngine&, uint32_t, uint32_t, uint32_t, uint32_t, uint32_t) { return kFmodOk; }