Make threading lock-free

The core and the headless tools no longer use mutexes, condition variables,
spinlocks, std::atomic, std::thread or volatile flags. Threads hand off through
libretro-common's retro_atomic and rthreads plus a few small pieces on top of
them in Common/Thread: Thread, MpscQueue, EventCounter, ParkingLot waits and
ReaderGate. Every wait blocks until it's woken; nothing polls on a timer.

State that only one thread touches just loses its lock. The PSP filesystem,
sceIo and the savedata dialog now belong to the emulation thread: async reads
and writes run when they're queued and complete later in emulated time, and
the IO and savedata threads are gone. The friend finder only talks to the
adhoc server and hands what it gets to the emulation thread. GPU init always
runs on the thread that starts the game.

The Vulkan render and compile queues, pipeline and render pass caches, delete
list and SPIR-V cache, the GL render queue, the software renderer's jit caches,
the file loader caches, the web debugger, UPnP, input, savestates, logging and
the GE debugger all hand off through queues and atomics. The IR jit's spinning
profiler thread goes.

A CPU thread with nothing to run now sleeps until work is queued instead of
spinning. Headless with --debugger waits for a client while stopped, and ends
the run once every client that connected has left. Web server shutdown keeps
servicing the CPU queue until the server has finished.

Common/Thread/ThreadUtil.cpp

@@ -18,27 +18,13 @@
 #include <cstring>
 #include <cstdint>
 
+#include <retro_atomic.h>
+#include <rthreads/rthreads.h>
+
 #include "Common/Log.h"
 #include "Common/Thread/ThreadUtil.h"
 #include "Common/Data/Encoding/Utf8.h"
 
-#if (PPSSPP_PLATFORM(ANDROID) || PPSSPP_PLATFORM(LINUX)) && !defined(_GNU_SOURCE)
-#define _GNU_SOURCE
-#endif
-
-#if !PPSSPP_PLATFORM(WINDOWS)
-#include <pthread.h>
-#include <sys/types.h>
-#include <unistd.h>
-#include <sys/syscall.h>
-#endif
-
-#if defined(__DragonFly__) || defined(__FreeBSD__) || defined(__OpenBSD__)
-#include <pthread_np.h>
-#elif defined(__NetBSD__)
-#include <lwp.h>
-#endif
-
 #ifdef TLS_SUPPORTED
 static thread_local const char *curThreadName;
 #endif
@@ -64,25 +50,22 @@ static EXCEPTION_DISPOSITION NTAPI ignore_handler(EXCEPTION_RECORD *rec,
 #if PPSSPP_PLATFORM(WINDOWS) && !PPSSPP_PLATFORM(UWP)
 typedef HRESULT (WINAPI *TSetThreadDescription)(HANDLE, PCWSTR);
 
-static TSetThreadDescription g_pSetThreadDescription = nullptr;
-static bool g_failedToFindSetThreadDescription = false;
+// Resolved once by whichever thread gets there first; any number may race to do it.
+// Holds the function, or the address of g_noSetThreadDescription when it doesn't exist.
+static retro_atomic_ptr_t g_setThreadDescription;
+static char g_noSetThreadDescription;
 
-static void InitializeSetThreadDescription() {
-	if (g_pSetThreadDescription != nullptr || g_failedToFindSetThreadDescription) {
-		return;
-	}
-
-	HMODULE hKernel32 = GetModuleHandle(L"kernelbase.dll");
-	if (hKernel32 == nullptr) {
-		// Failed to find the function. Windows version too old, most likely.
-		g_failedToFindSetThreadDescription = true;
-		return;
-	}
-	g_pSetThreadDescription = reinterpret_cast<TSetThreadDescription>(GetProcAddress(hKernel32, "SetThreadDescription"));
-	if (g_pSetThreadDescription == nullptr) {
-		g_failedToFindSetThreadDescription = true;
-		return;
+static TSetThreadDescription GetSetThreadDescription() {
+	void *fn = retro_atomic_load_acquire_ptr(&g_setThreadDescription);
+	if (!fn) {
+		HMODULE hKernel32 = GetModuleHandle(L"kernelbase.dll");
+		// Missing on Windows versions that are too old.
+		fn = hKernel32 ? (void *)GetProcAddress(hKernel32, "SetThreadDescription") : nullptr;
+		if (!fn)
+			fn = &g_noSetThreadDescription;
+		retro_atomic_store_release_ptr(&g_setThreadDescription, fn);
 	}
+	return fn == &g_noSetThreadDescription ? nullptr : (TSetThreadDescription)fn;
 }
 
 void SetCurrentThreadNameThroughException(const char *threadName);
@@ -98,12 +81,12 @@ const char *GetCurrentThreadName() {
 
 void SetCurrentThreadName(const char *threadName) {
 #if PPSSPP_PLATFORM(WINDOWS) && !PPSSPP_PLATFORM(UWP)
-	InitializeSetThreadDescription();
-	if (g_pSetThreadDescription) {
+	TSetThreadDescription setThreadDescription = GetSetThreadDescription();
+	if (setThreadDescription) {
 		// Use the modern API
 		wchar_t buffer[256];
 		ConvertUTF8ToWString(buffer, ARRAY_SIZE(buffer), threadName);
-		g_pSetThreadDescription(GetCurrentThread(), buffer);
+		setThreadDescription(GetCurrentThread(), buffer);
 	} else {
 		// Use the old exception hack.
 		SetCurrentThreadNameThroughException(threadName);
@@ -112,14 +95,8 @@ void SetCurrentThreadName(const char *threadName) {
 	wchar_t buffer[256];
 	ConvertUTF8ToWString(buffer, ARRAY_SIZE(buffer), threadName);
 	SetThreadDescription(GetCurrentThread(), buffer);
-#elif PPSSPP_PLATFORM(ANDROID) || PPSSPP_PLATFORM(LINUX)
-	pthread_setname_np(pthread_self(), threadName);
-#elif defined(__APPLE__)
-	pthread_setname_np(threadName);
-#elif defined(__DragonFly__) || defined(__FreeBSD__) || defined(__OpenBSD__)
-	pthread_set_name_np(pthread_self(), threadName);
-#elif defined(__NetBSD__)
-	pthread_setname_np(pthread_self(), "%s", (void*)threadName);
+#else
+	sthread_setname(threadName);
 #endif
 
 	// Set the locally known threadname using a thread local variable.
@@ -190,27 +167,5 @@ void AssertCurrentThreadName(const char *threadName) {
 }
 
 int GetCurrentThreadIdForDebug() {
-#if __LIBRETRO__
-	// Not sure why gettid() would not be available, but it isn't.
-	// The return value of this function is only used in unit tests anyway...
-	return 1;
-#elif PPSSPP_PLATFORM(WINDOWS)
-	return (int)GetCurrentThreadId();
-#elif PPSSPP_PLATFORM(MAC) || PPSSPP_PLATFORM(IOS)
-	uint64_t tid = 0;
-	pthread_threadid_np(NULL, &tid);
-	return (int)tid;
-#elif PPSSPP_PLATFORM(ANDROID) || PPSSPP_PLATFORM(LINUX)
-	// See issue 14545
-	return (int)syscall(__NR_gettid);
-	// return (int)gettid();
-#elif defined(__DragonFly__) || defined(__FreeBSD__)
-	return pthread_getthreadid_np();
-#elif defined(__NetBSD__)
-	return _lwp_self();
-#elif defined(__OpenBSD__)
-	return getthrid();
-#else
-	return 1;
-#endif
+	return (int)sthread_get_current_thread_id();
 }