rpcsx/rpcs3/util/atomic.hpp

#pragma once

#include "Utilities/types.h"
#include <functional>
#include <mutex>

#ifdef _MSC_VER
#include <atomic>
#endif

// Wait timeout extension (in nanoseconds)
enum class atomic_wait_timeout : u64
{
	inf = 0xffffffffffffffff,
};

// Various extensions for atomic_t::wait
namespace atomic_wait
{
	// Max number of simultaneous atomic variables to wait on (can be extended if really necessary)
	constexpr uint max_list = 8;

	enum class op : u8
	{
		eq, // Wait while value is bitwise equal to
		slt, // Wait while signed value is less than
		sgt, // Wait while signed value is greater than
		ult, // Wait while unsigned value is less than
		ugt, // Wait while unsigned value is greater than
		alt, // Wait while absolute value is less than
		agt, // Wait while absolute value is greater than
		pop, // Wait while set bit count of the value is less than
		__max
	};

	static_assert(static_cast<u8>(op::__max) == 8);

	enum class op_flag : u8
	{
		inverse = 1 << 4, // Perform inverse operation (negate the result)
		bit_not = 1 << 5, // Perform bitwise NOT on loaded value before operation
		byteswap = 1 << 6, // Perform byteswap on both arguments and masks when applicable
	};

	constexpr op_flag op_be = std::endian::native == std::endian::little ? op_flag::byteswap : op_flag{0};
	constexpr op_flag op_le = std::endian::native == std::endian::little ? op_flag{0} : op_flag::byteswap;

	constexpr op operator |(op_flag lhs, op_flag rhs)
	{
		return op{static_cast<u8>(static_cast<u8>(lhs) | static_cast<u8>(rhs))};
	}

	constexpr op operator |(op_flag lhs, op rhs)
	{
		return op{static_cast<u8>(static_cast<u8>(lhs) | static_cast<u8>(rhs))};
	}

	constexpr op operator |(op lhs, op_flag rhs)
	{
		return op{static_cast<u8>(static_cast<u8>(lhs) | static_cast<u8>(rhs))};
	}

	constexpr op op_ne = op::eq | op_flag::inverse;

	struct info
	{
		const void* data;
		u32 size;
		__m128i old;
		__m128i mask;

		template <typename T>
		static constexpr __m128i get_value(T value = T{})
		{
			static_assert((sizeof(T) & (sizeof(T) - 1)) == 0);
			static_assert(sizeof(T) <= 16);

			if constexpr (sizeof(T) <= 8)
			{
				return _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>, T>(value));
			}
			else if constexpr (sizeof(T) == 16)
			{
				return std::bit_cast<__m128i>(value);
			}
		}

		template <typename T>
		constexpr void set_value(T value = T{})
		{
			old = get_value<T>(value);
		}

		template <typename T>
		constexpr void set_mask(T value)
		{
			static_assert((sizeof(T) & (sizeof(T) - 1)) == 0);
			static_assert(sizeof(T) <= 16);

			if constexpr (sizeof(T) <= 8)
			{
				mask = _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>, T>(value));
			}
			else if constexpr (sizeof(T) == 16)
			{
				mask = std::bit_cast<__m128i>(value);
			}
		}

		template <typename T>
		constexpr void set_mask()
		{
			mask = get_mask<T>();
		}

		template <typename T>
		static constexpr __m128i get_mask()
		{
			if constexpr (sizeof(T) <= 8)
			{
				return _mm_cvtsi64_si128(UINT64_MAX >> ((64 - sizeof(T) * 8) & 63));
			}
			else
			{
				return _mm_set1_epi64x(-1);
			}
		}
	};

	template <uint Max, typename... T>
	class list
	{
		static_assert(Max <= max_list, "Too many elements in the atomic wait list.");

		// Null-terminated list of wait info
		info m_info[Max + 1]{};

	public:
		constexpr list() noexcept = default;

		constexpr list(const list&) noexcept = default;

		constexpr list& operator=(const list&) noexcept = default;

		template <typename... U, std::size_t... Align>
		constexpr list(atomic_t<U, Align>&... vars)
			: m_info{{&vars.raw(), sizeof(U), info::get_value<U>(), info::get_mask<U>()}...}
		{
			static_assert(sizeof...(U) == Max, "Inconsistent amount of atomics.");
		}

		template <typename... U>
		constexpr list& values(U... values)
		{
			static_assert(sizeof...(U) == Max, "Inconsistent amount of values.");

			auto* ptr = m_info;
			((ptr++)->template set_value<T>(values), ...);
			return *this;
		}

		template <typename... U>
		constexpr list& masks(U... masks)
		{
			static_assert(sizeof...(U) <= Max, "Too many masks.");

			auto* ptr = m_info;
			((ptr++)->template set_mask<T>(masks), ...);
			return *this;
		}

		template <uint Index, op Flags = op::eq, typename T2, std::size_t Align, typename U>
		constexpr void set(atomic_t<T2, Align>& var, U value)
		{
			static_assert(Index < Max);

			m_info[Index].data = &var.raw();
			m_info[Index].size = sizeof(T2) | (static_cast<u8>(Flags) << 8);
			m_info[Index].template set_value<T2>(value);
			m_info[Index].template set_mask<T2>();
		}

		template <uint Index, op Flags = op::eq, typename T2, std::size_t Align, typename U, typename V>
		constexpr void set(atomic_t<T2, Align>& var, U value, V mask)
		{
			static_assert(Index < Max);

			m_info[Index].data = &var.raw();
			m_info[Index].size = sizeof(T2) | (static_cast<u8>(Flags) << 8);
			m_info[Index].template set_value<T2>(value);
			m_info[Index].template set_mask<T2>(mask);
		}

		// Timeout is discouraged
		void wait(atomic_wait_timeout timeout = atomic_wait_timeout::inf);

		// Same as wait
		void start()
		{
			wait();
		}
	};

	template <typename... T, std::size_t... Align>
	list(atomic_t<T, Align>&... vars) -> list<sizeof...(T), T...>;

	// RDTSC with adjustment for being unique
	u64 get_unique_tsc();
}

// Helper for waitable atomics (as in C++20 std::atomic)
struct atomic_wait_engine
{
private:
	template <typename T, std::size_t Align>
	friend class atomic_t;

	template <uint Max, typename... T>
	friend class atomic_wait::list;

	static void
#ifdef _WIN32
	__vectorcall
#endif
	wait(const void* data, u32 size, __m128i old128, u64 timeout, __m128i mask128, atomic_wait::info* extension = nullptr);

	static void
#ifdef _WIN32
	__vectorcall
#endif
	notify_one(const void* data, u32 size, __m128i mask128, __m128i val128);

	static void
#ifdef _WIN32
	__vectorcall
#endif
	notify_all(const void* data, u32 size, __m128i mask128);

public:
	static void set_wait_callback(bool(*cb)(const void* data, u64 attempts, u64 stamp0));
	static void set_notify_callback(void(*cb)(const void* data, u64 progress));
	static bool raw_notify(const void* data, u64 thread_id = 0);
};

template <uint Max, typename... T>
void atomic_wait::list<Max, T...>::wait(atomic_wait_timeout timeout)
{
	static_assert(Max, "Cannot initiate atomic wait with empty list.");

	atomic_wait_engine::wait(m_info[0].data, m_info[0].size, m_info[0].old, static_cast<u64>(timeout), m_info[0].mask, m_info + 1);
}

// Helper class, provides access to compiler-specific atomic intrinsics
template <typename T, std::size_t Size = sizeof(T)>
struct atomic_storage
{
	static_assert(sizeof(T) <= 16 && sizeof(T) == alignof(T), "atomic_storage<> error: invalid type");

	/* First part: Non-MSVC intrinsics */

#ifndef _MSC_VER

#if defined(__ATOMIC_HLE_ACQUIRE) && defined(__ATOMIC_HLE_RELEASE)
	static constexpr int s_hle_ack = __ATOMIC_SEQ_CST | __ATOMIC_HLE_ACQUIRE;
	static constexpr int s_hle_rel = __ATOMIC_SEQ_CST | __ATOMIC_HLE_RELEASE;
#else
	static constexpr int s_hle_ack = __ATOMIC_SEQ_CST;
	static constexpr int s_hle_rel = __ATOMIC_SEQ_CST;
#endif

	static inline bool compare_exchange(T& dest, T& comp, T exch)
	{
		return __atomic_compare_exchange(&dest, &comp, &exch, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST);
	}

	static inline bool compare_exchange_hle_acq(T& dest, T& comp, T exch)
	{
		static_assert(sizeof(T) == 4 || sizeof(T) == 8);
		return __atomic_compare_exchange(&dest, &comp, &exch, false, s_hle_ack, s_hle_ack);
	}

	static inline T load(const T& dest)
	{
		T result;
		__atomic_load(&dest, &result, __ATOMIC_SEQ_CST);
		return result;
	}

	static inline void store(T& dest, T value)
	{
		static_cast<void>(exchange(dest, value));
	}

	static inline void release(T& dest, T value)
	{
		__atomic_store(&dest, &value, __ATOMIC_RELEASE);
	}

	static inline T exchange(T& dest, T value)
	{
		T result;
		__atomic_exchange(&dest, &value, &result, __ATOMIC_SEQ_CST);
		return result;
	}

	static inline T fetch_add(T& dest, T value)
	{
		return __atomic_fetch_add(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T fetch_add_hle_rel(T& dest, T value)
	{
		static_assert(sizeof(T) == 4 || sizeof(T) == 8);
		return __atomic_fetch_add(&dest, value, s_hle_rel);
	}

	static inline T add_fetch(T& dest, T value)
	{
		return __atomic_add_fetch(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T fetch_sub(T& dest, T value)
	{
		return __atomic_fetch_sub(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T sub_fetch(T& dest, T value)
	{
		return __atomic_sub_fetch(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T fetch_and(T& dest, T value)
	{
		return __atomic_fetch_and(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T and_fetch(T& dest, T value)
	{
		return __atomic_and_fetch(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T fetch_xor(T& dest, T value)
	{
		return __atomic_fetch_xor(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T xor_fetch(T& dest, T value)
	{
		return __atomic_xor_fetch(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T fetch_or(T& dest, T value)
	{
		return __atomic_fetch_or(&dest, value, __ATOMIC_SEQ_CST);
	}

	static inline T or_fetch(T& dest, T value)
	{
		return __atomic_or_fetch(&dest, value, __ATOMIC_SEQ_CST);
	}
#endif

	/* Second part: MSVC-specific */

#ifdef _MSC_VER
	static inline T add_fetch(T& dest, T value)
	{
		return atomic_storage<T>::fetch_add(dest, value) + value;
	}

	static inline T fetch_sub(T& dest, T value)
	{
		return atomic_storage<T>::fetch_add(dest, 0 - value);
	}

	static inline T sub_fetch(T& dest, T value)
	{
		return atomic_storage<T>::fetch_add(dest, 0 - value) - value;
	}

	static inline T and_fetch(T& dest, T value)
	{
		return atomic_storage<T>::fetch_and(dest, value) & value;
	}

	static inline T or_fetch(T& dest, T value)
	{
		return atomic_storage<T>::fetch_or(dest, value) | value;
	}

	static inline T xor_fetch(T& dest, T value)
	{
		return atomic_storage<T>::fetch_xor(dest, value) ^ value;
	}
#endif

	/* Third part: fallbacks, may be hidden by subsequent atomic_storage<> specializations */

	static inline T fetch_inc(T& dest)
	{
		return atomic_storage<T>::fetch_add(dest, 1);
	}

	static inline T inc_fetch(T& dest)
	{
		return atomic_storage<T>::add_fetch(dest, 1);
	}

	static inline T fetch_dec(T& dest)
	{
		return atomic_storage<T>::fetch_sub(dest, 1);
	}

	static inline T dec_fetch(T& dest)
	{
		return atomic_storage<T>::sub_fetch(dest, 1);
	}

	static inline bool test_and_set(T& dest, T mask)
	{
		return (atomic_storage<T>::fetch_or(dest, mask) & mask) != 0;
	}

	static inline bool test_and_reset(T& dest, T mask)
	{
		return (atomic_storage<T>::fetch_and(dest, ~mask) & mask) != 0;
	}

	static inline bool test_and_complement(T& dest, T mask)
	{
		return (atomic_storage<T>::fetch_xor(dest, mask) & mask) != 0;
	}

	static inline bool bts(T& dest, uint bit)
	{
		return atomic_storage<T>::test_and_set(dest, static_cast<T>(1) << bit);
	}

	static inline bool btr(T& dest, uint bit)
	{
		return atomic_storage<T>::test_and_reset(dest, static_cast<T>(1) << bit);
	}

	static inline bool btc(T& dest, uint bit)
	{
		return atomic_storage<T>::test_and_complement(dest, static_cast<T>(1) << bit);
	}
};

/* The rest: ugly MSVC intrinsics + inline asm implementations */

template <typename T>
struct atomic_storage<T, 1> : atomic_storage<T, 0>
{
#ifdef _MSC_VER
	static inline bool compare_exchange(T& dest, T& comp, T exch)
	{
		const char v = std::bit_cast<char>(comp);
		const char r = _InterlockedCompareExchange8(reinterpret_cast<volatile char*>(&dest), std::bit_cast<char>(exch), v);
		comp = std::bit_cast<T>(r);
		return r == v;
	}

	static inline T load(const T& dest)
	{
		std::atomic_thread_fence(std::memory_order_acquire);
		const char value = *reinterpret_cast<const volatile char*>(&dest);
		std::atomic_thread_fence(std::memory_order_acquire);
		return std::bit_cast<T>(value);
	}

	static inline void release(T& dest, T value)
	{
		std::atomic_thread_fence(std::memory_order_release);
		*reinterpret_cast<volatile char*>(&dest) = std::bit_cast<char>(value);
		std::atomic_thread_fence(std::memory_order_release);
	}

	static inline T exchange(T& dest, T value)
	{
		const char r = _InterlockedExchange8(reinterpret_cast<volatile char*>(&dest), std::bit_cast<char>(value));
		return std::bit_cast<T>(r);
	}

	static inline void store(T& dest, T value)
	{
		exchange(dest, value);
	}

	static inline T fetch_add(T& dest, T value)
	{
		const char r = _InterlockedExchangeAdd8(reinterpret_cast<volatile char*>(&dest), std::bit_cast<char>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_and(T& dest, T value)
	{
		const char r = _InterlockedAnd8(reinterpret_cast<volatile char*>(&dest), std::bit_cast<char>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_or(T& dest, T value)
	{
		const char r = _InterlockedOr8(reinterpret_cast<volatile char*>(&dest), std::bit_cast<char>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_xor(T& dest, T value)
	{
		const char r = _InterlockedXor8(reinterpret_cast<volatile char*>(&dest), std::bit_cast<char>(value));
		return std::bit_cast<T>(r);
	}
#endif
};

template <typename T>
struct atomic_storage<T, 2> : atomic_storage<T, 0>
{
#ifdef _MSC_VER
	static inline bool compare_exchange(T& dest, T& comp, T exch)
	{
		const short v = std::bit_cast<short>(comp);
		const short r = _InterlockedCompareExchange16(reinterpret_cast<volatile short*>(&dest), std::bit_cast<short>(exch), v);
		comp = std::bit_cast<T>(r);
		return r == v;
	}

	static inline T load(const T& dest)
	{
		std::atomic_thread_fence(std::memory_order_acquire);
		const short value = *reinterpret_cast<const volatile short*>(&dest);
		std::atomic_thread_fence(std::memory_order_acquire);
		return std::bit_cast<T>(value);
	}

	static inline void release(T& dest, T value)
	{
		std::atomic_thread_fence(std::memory_order_release);
		*reinterpret_cast<volatile short*>(&dest) = std::bit_cast<short>(value);
		std::atomic_thread_fence(std::memory_order_release);
	}

	static inline T exchange(T& dest, T value)
	{
		const short r = _InterlockedExchange16(reinterpret_cast<volatile short*>(&dest), std::bit_cast<short>(value));
		return std::bit_cast<T>(r);
	}

	static inline void store(T& dest, T value)
	{
		exchange(dest, value);
	}

	static inline T fetch_add(T& dest, T value)
	{
		const short r = _InterlockedExchangeAdd16(reinterpret_cast<volatile short*>(&dest), std::bit_cast<short>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_and(T& dest, T value)
	{
		const short r = _InterlockedAnd16(reinterpret_cast<volatile short*>(&dest), std::bit_cast<short>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_or(T& dest, T value)
	{
		const short r = _InterlockedOr16(reinterpret_cast<volatile short*>(&dest), std::bit_cast<short>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_xor(T& dest, T value)
	{
		const short r = _InterlockedXor16(reinterpret_cast<volatile short*>(&dest), std::bit_cast<short>(value));
		return std::bit_cast<T>(r);
	}

	static inline T inc_fetch(T& dest)
	{
		const short r = _InterlockedIncrement16(reinterpret_cast<volatile short*>(&dest));
		return std::bit_cast<T>(r);
	}

	static inline T dec_fetch(T& dest)
	{
		const short r = _InterlockedDecrement16(reinterpret_cast<volatile short*>(&dest));
		return std::bit_cast<T>(r);
	}
#else
	static inline bool bts(T& dest, uint bit)
	{
		bool result;
		ushort _bit = static_cast<ushort>(bit);
		__asm__("lock btsw %2, %0\n" : "+m" (dest), "=@ccc" (result) : "Ir" (_bit) : "cc");
		return result;
	}

	static inline bool btr(T& dest, uint bit)
	{
		bool result;
		ushort _bit = static_cast<ushort>(bit);
		__asm__("lock btrw %2, %0\n": "+m" (dest), "=@ccc" (result) : "Ir" (_bit) : "cc");
		return result;
	}

	static inline bool btc(T& dest, uint bit)
	{
		bool result;
		ushort _bit = static_cast<ushort>(bit);
		__asm__("lock btcw %2, %0\n": "+m" (dest), "=@ccc" (result) : "Ir" (_bit) : "cc");
		return result;
	}
#endif
};

template <typename T>
struct atomic_storage<T, 4> : atomic_storage<T, 0>
{
#ifdef _MSC_VER
	static inline bool compare_exchange(T& dest, T& comp, T exch)
	{
		const long v = std::bit_cast<long>(comp);
		const long r = _InterlockedCompareExchange(reinterpret_cast<volatile long*>(&dest), std::bit_cast<long>(exch), v);
		comp = std::bit_cast<T>(r);
		return r == v;
	}

	static inline bool compare_exchange_hle_acq(T& dest, T& comp, T exch)
	{
		const long v = std::bit_cast<long>(comp);
		const long r = _InterlockedCompareExchange_HLEAcquire(reinterpret_cast<volatile long*>(&dest), std::bit_cast<long>(exch), v);
		comp = std::bit_cast<T>(r);
		return r == v;
	}

	static inline T load(const T& dest)
	{
		std::atomic_thread_fence(std::memory_order_acquire);
		const long value = *reinterpret_cast<const volatile long*>(&dest);
		std::atomic_thread_fence(std::memory_order_acquire);
		return std::bit_cast<T>(value);
	}

	static inline void release(T& dest, T value)
	{
		std::atomic_thread_fence(std::memory_order_release);
		*reinterpret_cast<volatile long*>(&dest) = std::bit_cast<long>(value);
		std::atomic_thread_fence(std::memory_order_release);
	}

	static inline T exchange(T& dest, T value)
	{
		const long r = _InterlockedExchange(reinterpret_cast<volatile long*>(&dest), std::bit_cast<long>(value));
		return std::bit_cast<T>(r);
	}

	static inline void store(T& dest, T value)
	{
		exchange(dest, value);
	}

	static inline T fetch_add(T& dest, T value)
	{
		const long r = _InterlockedExchangeAdd(reinterpret_cast<volatile long*>(&dest), std::bit_cast<long>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_add_hle_rel(T& dest, T value)
	{
		const long r = _InterlockedExchangeAdd_HLERelease(reinterpret_cast<volatile long*>(&dest), std::bit_cast<long>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_and(T& dest, T value)
	{
		long r = _InterlockedAnd(reinterpret_cast<volatile long*>(&dest), std::bit_cast<long>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_or(T& dest, T value)
	{
		const long r = _InterlockedOr(reinterpret_cast<volatile long*>(&dest), std::bit_cast<long>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_xor(T& dest, T value)
	{
		const long r = _InterlockedXor(reinterpret_cast<volatile long*>(&dest), std::bit_cast<long>(value));
		return std::bit_cast<T>(r);
	}

	static inline T inc_fetch(T& dest)
	{
		const long r = _InterlockedIncrement(reinterpret_cast<volatile long*>(&dest));
		return std::bit_cast<T>(r);
	}

	static inline T dec_fetch(T& dest)
	{
		const long r = _InterlockedDecrement(reinterpret_cast<volatile long*>(&dest));
		return std::bit_cast<T>(r);
	}

	static inline bool bts(T& dest, uint bit)
	{
		return _interlockedbittestandset(reinterpret_cast<volatile long*>(&dest), bit) != 0;
	}

	static inline bool btr(T& dest, uint bit)
	{
		return _interlockedbittestandreset(reinterpret_cast<volatile long*>(&dest), bit) != 0;
	}
#else
	static inline bool bts(T& dest, uint bit)
	{
		bool result;
		__asm__("lock btsl %2, %0\n" : "+m" (dest), "=@ccc" (result) : "Ir" (bit) : "cc");
		return result;
	}

	static inline bool btr(T& dest, uint bit)
	{
		bool result;
		__asm__("lock btrl %2, %0\n" : "+m" (dest), "=@ccc" (result) : "Ir" (bit) : "cc");
		return result;
	}

	static inline bool btc(T& dest, uint bit)
	{
		bool result;
		__asm__("lock btcl %2, %0\n" : "+m" (dest), "=@ccc" (result) : "Ir" (bit) : "cc");
		return result;
	}
#endif
};

template <typename T>
struct atomic_storage<T, 8> : atomic_storage<T, 0>
{
#ifdef _MSC_VER
	static inline bool compare_exchange(T& dest, T& comp, T exch)
	{
		const llong v = std::bit_cast<llong>(comp);
		const llong r = _InterlockedCompareExchange64(reinterpret_cast<volatile llong*>(&dest), std::bit_cast<llong>(exch), v);
		comp = std::bit_cast<T>(r);
		return r == v;
	}

	static inline bool compare_exchange_hle_acq(T& dest, T& comp, T exch)
	{
		const llong v = std::bit_cast<llong>(comp);
		const llong r = _InterlockedCompareExchange64_HLEAcquire(reinterpret_cast<volatile llong*>(&dest), std::bit_cast<llong>(exch), v);
		comp = std::bit_cast<T>(r);
		return r == v;
	}

	static inline T load(const T& dest)
	{
		std::atomic_thread_fence(std::memory_order_acquire);
		const llong value = *reinterpret_cast<const volatile llong*>(&dest);
		std::atomic_thread_fence(std::memory_order_acquire);
		return std::bit_cast<T>(value);
	}

	static inline void release(T& dest, T value)
	{
		std::atomic_thread_fence(std::memory_order_release);
		*reinterpret_cast<volatile llong*>(&dest) = std::bit_cast<llong>(value);
		std::atomic_thread_fence(std::memory_order_release);
	}

	static inline T exchange(T& dest, T value)
	{
		const llong r = _InterlockedExchange64(reinterpret_cast<volatile llong*>(&dest), std::bit_cast<llong>(value));
		return std::bit_cast<T>(r);
	}

	static inline void store(T& dest, T value)
	{
		exchange(dest, value);
	}

	static inline T fetch_add(T& dest, T value)
	{
		const llong r = _InterlockedExchangeAdd64(reinterpret_cast<volatile llong*>(&dest), std::bit_cast<llong>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_add_hle_rel(T& dest, T value)
	{
		const llong r = _InterlockedExchangeAdd64_HLERelease(reinterpret_cast<volatile llong*>(&dest), std::bit_cast<llong>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_and(T& dest, T value)
	{
		const llong r = _InterlockedAnd64(reinterpret_cast<volatile llong*>(&dest), std::bit_cast<llong>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_or(T& dest, T value)
	{
		const llong r = _InterlockedOr64(reinterpret_cast<volatile llong*>(&dest), std::bit_cast<llong>(value));
		return std::bit_cast<T>(r);
	}

	static inline T fetch_xor(T& dest, T value)
	{
		const llong r = _InterlockedXor64(reinterpret_cast<volatile llong*>(&dest), std::bit_cast<llong>(value));
		return std::bit_cast<T>(r);
	}

	static inline T inc_fetch(T& dest)
	{
		const llong r = _InterlockedIncrement64(reinterpret_cast<volatile llong*>(&dest));
		return std::bit_cast<T>(r);
	}

	static inline T dec_fetch(T& dest)
	{
		const llong r = _InterlockedDecrement64(reinterpret_cast<volatile llong*>(&dest));
		return std::bit_cast<T>(r);
	}

	static inline bool bts(T& dest, uint bit)
	{
		return _interlockedbittestandset64(reinterpret_cast<volatile llong*>(&dest), bit) != 0;
	}

	static inline bool btr(T& dest, uint bit)
	{
		return _interlockedbittestandreset64(reinterpret_cast<volatile llong*>(&dest), bit) != 0;
	}
#else
	static inline bool bts(T& dest, uint bit)
	{
		bool result;
		ullong _bit = bit;
		__asm__("lock btsq %2, %0\n" : "+m" (dest), "=@ccc" (result) : "Ir" (_bit) : "cc");
		return result;
	}

	static inline bool btr(T& dest, uint bit)
	{
		bool result;
		ullong _bit = bit;
		__asm__("lock btrq %2, %0\n" : "+m" (dest), "=@ccc" (result) : "Ir" (_bit) : "cc");
		return result;
	}

	static inline bool btc(T& dest, uint bit)
	{
		bool result;
		ullong _bit = bit;
		__asm__("lock btcq %2, %0\n" : "+m" (dest), "=@ccc" (result) : "Ir" (_bit) : "cc");
		return result;
	}
#endif
};

template <typename T>
struct atomic_storage<T, 16> : atomic_storage<T, 0>
{
#ifdef _MSC_VER
	static inline T load(const T& dest)
	{
		std::atomic_thread_fence(std::memory_order_acquire);
		__m128i val = _mm_load_si128(reinterpret_cast<const __m128i*>(&dest));
		std::atomic_thread_fence(std::memory_order_acquire);
		return std::bit_cast<T>(val);
	}

	static inline bool compare_exchange(T& dest, T& comp, T exch)
	{
		struct alignas(16) llong2 { llong ll[2]; };
		const llong2 _exch = std::bit_cast<llong2>(exch);
		return _InterlockedCompareExchange128(reinterpret_cast<volatile llong*>(&dest), _exch.ll[1], _exch.ll[0], reinterpret_cast<llong*>(&comp)) != 0;
	}

	static inline T exchange(T& dest, T value)
	{
		struct alignas(16) llong2 { llong ll[2]; };
		const llong2 _value = std::bit_cast<llong2>(value);

		const auto llptr = reinterpret_cast<volatile llong*>(&dest);
		llong2 cmp{ llptr[0], llptr[1] };
		while (!_InterlockedCompareExchange128(llptr, _value.ll[1], _value.ll[0], cmp.ll));
		return std::bit_cast<T>(cmp);
	}

	static inline void store(T& dest, T value)
	{
		exchange(dest, value);
	}

	static inline void release(T& dest, T value)
	{
		std::atomic_thread_fence(std::memory_order_release);
		_mm_store_si128(reinterpret_cast<__m128i*>(&dest), std::bit_cast<__m128i>(value));
		std::atomic_thread_fence(std::memory_order_release);
	}
#else
	static inline T load(const T& dest)
	{
		__atomic_thread_fence(__ATOMIC_ACQUIRE);
		__m128i val = _mm_load_si128(reinterpret_cast<const __m128i*>(&dest));
		__atomic_thread_fence(__ATOMIC_ACQUIRE);
		return std::bit_cast<T>(val);
	}

	static inline bool compare_exchange(T& dest, T& comp, T exch)
	{
		bool result;
		ullong cmp_lo = 0;
		ullong cmp_hi = 0;
		ullong exc_lo = 0;
		ullong exc_hi = 0;

		if constexpr (std::is_same_v<T, u128> || std::is_same_v<T, s128>)
		{
			cmp_lo = comp;
			cmp_hi = comp >> 64;
			exc_lo = exch;
			exc_hi = exch >> 64;
		}
		else
		{
			std::memcpy(&cmp_lo, reinterpret_cast<char*>(&comp) + 0, 8);
			std::memcpy(&cmp_hi, reinterpret_cast<char*>(&comp) + 8, 8);
			std::memcpy(&exc_lo, reinterpret_cast<char*>(&exch) + 0, 8);
			std::memcpy(&exc_hi, reinterpret_cast<char*>(&exch) + 8, 8);
		}

		__asm__ volatile("lock cmpxchg16b %1;"
			: "=@ccz" (result)
			, "+m" (dest)
			, "+d" (cmp_hi)
			, "+a" (cmp_lo)
			: "c" (exc_hi)
			, "b" (exc_lo)
			: "cc");

		if constexpr (std::is_same_v<T, u128> || std::is_same_v<T, s128>)
		{
			comp = T{cmp_hi} << 64 | cmp_lo;
		}
		else
		{
			std::memcpy(reinterpret_cast<char*>(&comp) + 0, &cmp_lo, 8);
			std::memcpy(reinterpret_cast<char*>(&comp) + 8, &cmp_hi, 8);
		}

		return result;
	}

	static inline T exchange(T& dest, T value)
	{
		return std::bit_cast<T>(__sync_lock_test_and_set(reinterpret_cast<u128*>(&dest), std::bit_cast<u128>(value)));
	}

	static inline void store(T& dest, T value)
	{
		exchange(dest, value);
	}

	static inline void release(T& dest, T value)
	{
		__atomic_thread_fence(__ATOMIC_RELEASE);
		_mm_store_si128(reinterpret_cast<__m128i*>(&dest), std::bit_cast<__m128i>(value));
		__atomic_thread_fence(__ATOMIC_RELEASE);
	}
#endif

	// TODO
};

// Atomic type with lock-free and standard layout guarantees (and appropriate limitations)
template <typename T, std::size_t Align = alignof(T)>
class atomic_t
{
protected:
	using type = typename std::remove_cv<T>::type;

	using ptr_rt = std::conditional_t<std::is_pointer_v<type>, ullong, type>;

	static_assert(alignof(type) == sizeof(type), "atomic_t<> error: unexpected alignment, use alignas() if necessary");

	alignas(Align) type m_data;

public:
	atomic_t() noexcept = default;

	atomic_t(const atomic_t&) = delete;

	atomic_t& operator =(const atomic_t&) = delete;

	// Define simple type
	using simple_type = simple_t<T>;

	constexpr atomic_t(const type& value) noexcept
		: m_data(value)
	{
	}

	// Unsafe direct access
	type& raw()
	{
		return m_data;
	}

	// Unsafe direct access
	const type& raw() const
	{
		return m_data;
	}

	// Atomically compare data with cmp, replace with exch if equal, return previous data value anyway
	type compare_and_swap(const type& cmp, const type& exch)
	{
		type old = cmp;
		atomic_storage<type>::compare_exchange(m_data, old, exch);
		return old;
	}

	// Atomically compare data with cmp, replace with exch if equal, return true if data was replaced
	bool compare_and_swap_test(const type& cmp, const type& exch)
	{
		type old = cmp;
		return atomic_storage<type>::compare_exchange(m_data, old, exch);
	}

	// As in std::atomic
	bool compare_exchange(type& cmp_and_old, const type& exch)
	{
		return atomic_storage<type>::compare_exchange(m_data, cmp_and_old, exch);
	}

	// Atomic operation; returns old value, or pair of old value and return value (cancel op if evaluates to false)
	template <typename F, typename RT = std::invoke_result_t<F, T&>>
	std::conditional_t<std::is_void_v<RT>, type, std::pair<type, RT>> fetch_op(F func)
	{
		type _new, old = atomic_storage<type>::load(m_data);

		while (true)
		{
			_new = old;

			if constexpr (std::is_void_v<RT>)
			{
				std::invoke(func, _new);

				if (atomic_storage<type>::compare_exchange(m_data, old, _new)) [[likely]]
				{
					return old;
				}
			}
			else
			{
				RT ret = std::invoke(func, _new);

				if (!ret || atomic_storage<type>::compare_exchange(m_data, old, _new)) [[likely]]
				{
					return {old, std::move(ret)};
				}
			}
		}
	}

	// Atomic operation; returns function result value, function is the lambda
	template <typename F, typename RT = std::invoke_result_t<F, T&>>
	RT atomic_op(F func)
	{
		type _new, old = atomic_storage<type>::load(m_data);

		while (true)
		{
			_new = old;

			if constexpr (std::is_void_v<RT>)
			{
				std::invoke(func, _new);

				if (atomic_storage<type>::compare_exchange(m_data, old, _new)) [[likely]]
				{
					return;
				}
			}
			else
			{
				RT result = std::invoke(func, _new);

				if (atomic_storage<type>::compare_exchange(m_data, old, _new)) [[likely]]
				{
					return result;
				}
			}
		}
	}

	// Atomically read data
	type load() const
	{
		return atomic_storage<type>::load(m_data);
	}

	// Atomically read data
	operator simple_type() const
	{
		return atomic_storage<type>::load(m_data);
	}

	// Atomically write data
	void store(const type& rhs)
	{
		atomic_storage<type>::store(m_data, rhs);
	}

	type operator =(const type& rhs)
	{
		atomic_storage<type>::store(m_data, rhs);
		return rhs;
	}

	// Atomically write data with release memory order (faster on x86)
	void release(const type& rhs)
	{
		atomic_storage<type>::release(m_data, rhs);
	}

	// Atomically replace data with value, return previous data value
	type exchange(const type& rhs)
	{
		return atomic_storage<type>::exchange(m_data, rhs);
	}

	auto fetch_add(const ptr_rt& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::fetch_add(m_data, rhs);
		}

		return fetch_op([&](T& v)
		{
			v += rhs;
		});
	}

	auto add_fetch(const ptr_rt& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::add_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			v += rhs;
			return v;
		});
	}

	auto operator +=(const ptr_rt& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::add_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			return v += rhs;
		});
	}

	auto fetch_sub(const ptr_rt& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::fetch_sub(m_data, rhs);
		}

		return fetch_op([&](T& v)
		{
			v -= rhs;
		});
	}

	auto sub_fetch(const ptr_rt& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::sub_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			v -= rhs;
			return v;
		});
	}

	auto operator -=(const ptr_rt& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::sub_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			return v -= rhs;
		});
	}

	auto fetch_and(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::fetch_and(m_data, rhs);
		}

		return fetch_op([&](T& v)
		{
			v &= rhs;
		});
	}

	auto and_fetch(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::and_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			v &= rhs;
			return v;
		});
	}

	auto operator &=(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::and_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			return v &= rhs;
		});
	}

	auto fetch_or(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::fetch_or(m_data, rhs);
		}

		return fetch_op([&](T& v)
		{
			v |= rhs;
		});
	}

	auto or_fetch(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::or_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			v |= rhs;
			return v;
		});
	}

	auto operator |=(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::or_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			return v |= rhs;
		});
	}

	auto fetch_xor(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::fetch_xor(m_data, rhs);
		}

		return fetch_op([&](T& v)
		{
			v ^= rhs;
		});
	}

	auto xor_fetch(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::xor_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			v ^= rhs;
			return v;
		});
	}

	auto operator ^=(const type& rhs)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::xor_fetch(m_data, rhs);
		}

		return atomic_op([&](T& v)
		{
			return v ^= rhs;
		});
	}

	auto operator ++()
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::inc_fetch(m_data);
		}

		return atomic_op([](T& v)
		{
			return ++v;
		});
	}

	auto operator --()
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::dec_fetch(m_data);
		}

		return atomic_op([](T& v)
		{
			return --v;
		});
	}

	auto operator ++(int)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::fetch_inc(m_data);
		}

		return atomic_op([](T& v)
		{
			return v++;
		});
	}

	auto operator --(int)
	{
		if constexpr(std::is_integral<type>::value)
		{
			return atomic_storage<type>::fetch_dec(m_data);
		}

		return atomic_op([](T& v)
		{
			return v--;
		});
	}

	// Conditionally decrement
	bool try_dec(simple_type greater_than = std::numeric_limits<simple_type>::min())
	{
		type _new, old = atomic_storage<type>::load(m_data);

		while (true)
		{
			_new = old;

			if (!(_new > greater_than))
			{
				return false;
			}

			_new -= 1;

			if (atomic_storage<type>::compare_exchange(m_data, old, _new)) [[likely]]
			{
				return true;
			}
		}
	}

	// Conditionally increment
	bool try_inc(simple_type less_than = std::numeric_limits<simple_type>::max())
	{
		type _new, old = atomic_storage<type>::load(m_data);

		while (true)
		{
			_new = old;

			if (!(_new < less_than))
			{
				return false;
			}

			_new += 1;

			if (atomic_storage<type>::compare_exchange(m_data, old, _new)) [[likely]]
			{
				return true;
			}
		}
	}

	bool bts(uint bit)
	{
		return atomic_storage<type>::bts(m_data, bit);
	}

	bool btr(uint bit)
	{
		return atomic_storage<type>::btr(m_data, bit);
	}

	// Timeout is discouraged
	template <atomic_wait::op Flags = atomic_wait::op::eq>
	void wait(type old_value, atomic_wait_timeout timeout = atomic_wait_timeout::inf) const noexcept
	{
		if constexpr (sizeof(T) <= 8)
		{
			const __m128i old = _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>>(old_value));
			const __m128i mask = _mm_cvtsi64_si128(UINT64_MAX >> ((64 - sizeof(T) * 8) & 63));
			atomic_wait_engine::wait(&m_data, sizeof(T) | (static_cast<u8>(Flags) << 8), old, static_cast<u64>(timeout), mask);
		}
		else if constexpr (sizeof(T) == 16)
		{
			const __m128i old = std::bit_cast<__m128i>(old_value);
			atomic_wait_engine::wait(&m_data, sizeof(T) | (static_cast<u8>(Flags) << 8), old, static_cast<u64>(timeout), _mm_set1_epi64x(-1));
		}
	}

	// Overload with mask (only selected bits are checked), timeout is discouraged
	template <atomic_wait::op Flags = atomic_wait::op::eq>
	void wait(type old_value, type mask_value, atomic_wait_timeout timeout = atomic_wait_timeout::inf) const noexcept
	{
		if constexpr (sizeof(T) <= 8)
		{
			const __m128i old = _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>>(old_value));
			const __m128i mask = _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>>(mask_value));
			atomic_wait_engine::wait(&m_data, sizeof(T) | (static_cast<u8>(Flags) << 8), old, static_cast<u64>(timeout), mask);
		}
		else if constexpr (sizeof(T) == 16)
		{
			const __m128i old = std::bit_cast<__m128i>(old_value);
			const __m128i mask = std::bit_cast<__m128i>(mask_value);
			atomic_wait_engine::wait(&m_data, sizeof(T) | (static_cast<u8>(Flags) << 8), old, static_cast<u64>(timeout), mask);
		}
	}

	void notify_one() noexcept
	{
		if constexpr (sizeof(T) <= 8)
		{
			atomic_wait_engine::notify_one(&m_data, -1, _mm_cvtsi64_si128(UINT64_MAX >> ((64 - sizeof(T) * 8) & 63)), _mm_setzero_si128());
		}
		else if constexpr (sizeof(T) == 16)
		{
			atomic_wait_engine::notify_one(&m_data, -1, _mm_set1_epi64x(-1), _mm_setzero_si128());
		}
	}

	// Notify with mask, allowing to not wake up thread which doesn't wait on this mask
	void notify_one(type mask_value) noexcept
	{
		if constexpr (sizeof(T) <= 8)
		{
			const __m128i mask = _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>>(mask_value));
			atomic_wait_engine::notify_one(&m_data, -1, mask, _mm_setzero_si128());
		}
		else if constexpr (sizeof(T) == 16)
		{
			const __m128i mask = std::bit_cast<__m128i>(mask_value);
			atomic_wait_engine::notify_one(&m_data, -1, mask, _mm_setzero_si128());
		}
	}

	// Notify with mask and value, allowing to not wake up thread which doesn't wait on them
	[[deprecated("Incomplete")]] void notify_one(type mask_value, type phantom_value) noexcept
	{
		if constexpr (sizeof(T) <= 8)
		{
			const __m128i mask = _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>>(mask_value));
			const __m128i _new = _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>>(phantom_value));
			atomic_wait_engine::notify_one(&m_data, sizeof(T), mask, _new);
		}
		else if constexpr (sizeof(T) == 16)
		{
			const __m128i mask = std::bit_cast<__m128i>(mask_value);
			const __m128i _new = std::bit_cast<__m128i>(phantom_value);
			atomic_wait_engine::notify_one(&m_data, sizeof(T), mask, _new);
		}
	}

	void notify_all() noexcept
	{
		if constexpr (sizeof(T) <= 8)
		{
			atomic_wait_engine::notify_all(&m_data, -1, _mm_cvtsi64_si128(UINT64_MAX >> ((64 - sizeof(T) * 8) & 63)));
		}
		else if constexpr (sizeof(T) == 16)
		{
			atomic_wait_engine::notify_all(&m_data, -1, _mm_set1_epi64x(-1));
		}
	}

	// Notify all threads with mask, allowing to not wake up threads which don't wait on them
	void notify_all(type mask_value) noexcept
	{
		if constexpr (sizeof(T) <= 8)
		{
			const __m128i mask = _mm_cvtsi64_si128(std::bit_cast<get_uint_t<sizeof(T)>>(mask_value));
			atomic_wait_engine::notify_all(&m_data, -1, mask);
		}
		else if constexpr (sizeof(T) == 16)
		{
			const __m128i mask = std::bit_cast<__m128i>(mask_value);
			atomic_wait_engine::notify_all(&m_data, -1, mask);
		}
	}
};