Fixed some libultra thread implementation issues, fixed flash implementation not getting detected by the game

MMRecomp.vcxproj

@@ -167,7 +167,12 @@ XCOPY "$(ProjectDir)lib\SDL2-2.24.0\lib\$(Platform)\SDL2.dll" "$(TargetDir)" /S
     </PostBuildEvent>
   </ItemDefinitionGroup>
   <ItemGroup>
-    <ClCompile Include="portultra\audio.cpp" />
+    <ClCompile Include="portultra\audio.cpp">
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">MaxSpeed</Optimization>
+      <Optimization Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">MaxSpeed</Optimization>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|Win32'">Default</BasicRuntimeChecks>
+      <BasicRuntimeChecks Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">Default</BasicRuntimeChecks>
+    </ClCompile>
     <ClCompile Include="portultra\events.cpp" />
     <ClCompile Include="portultra\mesgqueue.cpp" />
     <ClCompile Include="portultra\misc_ultra.cpp" />

RecompiledFuncs.vcxproj

@@ -97,6 +97,7 @@
       <PrecompiledHeader>NotUsing</PrecompiledHeader>
       <EnableParallelCodeGeneration>true</EnableParallelCodeGeneration>
       <MultiProcessorCompilation>true</MultiProcessorCompilation>
+      <ExceptionHandling>SyncCThrow</ExceptionHandling>
     </ClCompile>
     <Link>
       <SubSystem>
@@ -118,6 +119,7 @@
       <MultiProcessorCompilation>true</MultiProcessorCompilation>
       <WholeProgramOptimization>false</WholeProgramOptimization>
       <PrecompiledHeaderFile />
+      <ExceptionHandling>SyncCThrow</ExceptionHandling>
     </ClCompile>
     <Link>
       <SubSystem>
@@ -136,6 +138,7 @@
       <PrecompiledHeader>NotUsing</PrecompiledHeader>
       <EnableParallelCodeGeneration>true</EnableParallelCodeGeneration>
       <MultiProcessorCompilation>true</MultiProcessorCompilation>
+      <ExceptionHandling>SyncCThrow</ExceptionHandling>
     </ClCompile>
     <Link>
       <SubSystem>
@@ -157,6 +160,7 @@
       <MultiProcessorCompilation>true</MultiProcessorCompilation>
       <WholeProgramOptimization>false</WholeProgramOptimization>
       <PrecompiledHeaderFile />
+      <ExceptionHandling>SyncCThrow</ExceptionHandling>
     </ClCompile>
     <Link>
       <SubSystem>

portultra/scheduler.cpp

@@ -73,7 +73,7 @@ void handle_thread_stopping(thread_queue_t& running_thread_queue) {
     }
 }
 
-void handle_thread_cleanup(thread_queue_t& running_thread_queue) {
+void handle_thread_cleanup(thread_queue_t& running_thread_queue, OSThread*& cur_running_thread) {
     std::lock_guard lock{scheduler_context.mutex};
 
     while (!scheduler_context.to_cleanup.empty()) {
@@ -83,8 +83,20 @@ void handle_thread_cleanup(thread_queue_t& running_thread_queue) {
         
         debug_printf("[Scheduler] Destroying thread %d\n", to_cleanup->id);
         running_thread_queue.remove(to_cleanup);
+        // If the cleaned up thread was the running thread, schedule a new one to run.
+        if (to_cleanup == cur_running_thread) {
+            // If there's a thread queued to run, set it as the new running thread.
+            if (!running_thread_queue.empty()) {
+                cur_running_thread = running_thread_queue.top();
+            }
+            // Otherwise, set the running thread to null so the next thread that can be run gets started.
+            else {
+                cur_running_thread = nullptr;
+            }
+        }
         to_cleanup->context->host_thread.join();
         delete to_cleanup->context;
+        to_cleanup->context = nullptr;
     }
 }
 
@@ -153,7 +165,7 @@ void scheduler_func() {
         handle_thread_stopping(running_thread_queue);
 
         // Handle cleaning up threads
-        handle_thread_cleanup(running_thread_queue);
+        handle_thread_cleanup(running_thread_queue, cur_running_thread);
 
         // Handle queueing threads to run
         handle_thread_queueing(running_thread_queue);

portultra/threads.cpp

@@ -41,6 +41,8 @@ void run_thread_function(uint8_t* rdram, uint64_t addr, uint64_t sp, uint64_t ar
 #define run_thread_function(func, sp, arg) func(arg)
 #endif
 
+struct thread_terminated : std::exception {};
+
 static void _thread_func(RDRAM_ARG PTR(OSThread) self_, PTR(thread_func_t) entrypoint, PTR(void) arg) {
     OSThread *self = TO_PTR(OSThread, self_);
     debug_printf("[Thread] Thread created: %d\n", self->id);
@@ -72,8 +74,12 @@ static void _thread_func(RDRAM_ARG PTR(OSThread) self_, PTR(thread_func_t) entry
 
     debug_printf("[Thread] Thread started: %d\n", self->id);
 
-    // Run the thread's function with the provided argument.
+    try {
-    run_thread_function(PASS_RDRAM entrypoint, self->sp, arg);
+        // Run the thread's function with the provided argument.
+        run_thread_function(PASS_RDRAM entrypoint, self->sp, arg);
+    } catch (thread_terminated& terminated) {
+
+    }
 
     // Dispose of this thread after it completes.
     Multilibultra::cleanup_thread(self);
@@ -110,6 +116,7 @@ extern "C" void osCreateThread(RDRAM_ARG PTR(OSThread) t_, OSId id, PTR(thread_f
     t->id = id;
     t->state = OSThreadState::PAUSED;
     t->sp = sp - 0x10; // Set up the first stack frame
+    t->destroyed = false;
 
     // Spawn a new thread, which will immediately pause itself and wait until it's been started.
     t->context = new UltraThreadContext{};
@@ -124,7 +131,16 @@ extern "C" void osStopThread(RDRAM_ARG PTR(OSThread) t_) {
 }
 
 extern "C" void osDestroyThread(RDRAM_ARG PTR(OSThread) t_) {
-    assert(false);
+    // Check if the thread is destroying itself (arg is null or thread_self)
+    if (t_ == NULLPTR || t_ == thread_self) {
+        throw thread_terminated{};
+    }
+    // Otherwise, mark the target thread as destroyed. Next time it reaches a stopping point,
+    // it'll check this and terminate itself instead of pausing.
+    else {
+        OSThread* t = TO_PTR(OSThread, t_);
+        t->destroyed = true;
+    }
 }
 
 extern "C" void osSetThreadPri(RDRAM_ARG PTR(OSThread) t, OSPri pri) {
@@ -168,8 +184,16 @@ void Multilibultra::set_self_paused(RDRAM_ARG1) {
     TO_PTR(OSThread, thread_self)->context->running.notify_all();
 }
 
+void check_destroyed(OSThread* t) {
+    if (t->destroyed) {
+        throw thread_terminated{};
+    }
+}
+
 void Multilibultra::wait_for_resumed(RDRAM_ARG1) {
+    check_destroyed(TO_PTR(OSThread, thread_self));
     TO_PTR(OSThread, thread_self)->context->running.wait(false);
+    check_destroyed(TO_PTR(OSThread, thread_self));
 }
 
 void Multilibultra::pause_thread_impl(OSThread* t) {

portultra/timer.cpp

@@ -163,11 +163,6 @@ extern "C" OSTime osGetTime() {
 extern "C" int osSetTimer(RDRAM_ARG PTR(OSTimer) t_, OSTime countdown, OSTime interval, PTR(OSMesgQueue) mq, OSMesg msg) {
     OSTimer* t = TO_PTR(OSTimer, t_);
 
-    // HACK: Skip the RCP timeout detection
-    if ((countdown == 140625000 || countdown == 1500000) && (uintptr_t)msg == 666) {
-        return 0;
-    }
-
     // Determine the time when this timer will trigger off
     if (countdown == 0) {
         // Set the timestamp based on the interval

portultra/ultra64.h

@@ -93,6 +93,7 @@ typedef struct OSThread_t {
     int32_t pad3;
     UltraThreadContext* context; // An actual pointer regardless of platform
     int32_t sp;
+    bool destroyed;
 } OSThread;
 
 typedef u32 OSEvent;

src/flash.cpp

@@ -31,8 +31,9 @@ extern "C" void osFlashReadId_recomp(uint8_t * rdram, recomp_context * ctx) {
 	PTR(u32) flash_type = ctx->r4;
 	PTR(u32) flash_maker = ctx->r5;
 
-	MEM_B(0, flash_type) = 0;
+	// Mimic a real flash chip's type and maker, as some games actually check if one is present.
-	MEM_B(0, flash_maker) = 0;
+	MEM_W(0, flash_type) = 0x11118001;
+	MEM_W(0, flash_maker) = 0x00C2001E;
 }
 
 extern "C" void osFlashClearStatus_recomp(uint8_t * rdram, recomp_context * ctx) {