diff options
Diffstat (limited to 'src/common')
30 files changed, 2224 insertions, 962 deletions
diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt index 2c2bd2ee8..788516ded 100644 --- a/src/common/CMakeLists.txt +++ b/src/common/CMakeLists.txt @@ -98,7 +98,6 @@ add_library(common STATIC algorithm.h alignment.h assert.h - atomic_ops.cpp atomic_ops.h detached_tasks.cpp detached_tasks.h @@ -108,7 +107,6 @@ add_library(common STATIC bit_util.h cityhash.cpp cityhash.h - color.h common_funcs.h common_paths.h common_types.h @@ -123,6 +121,7 @@ add_library(common STATIC hash.h hex_util.cpp hex_util.h + intrusive_red_black_tree.h logging/backend.cpp logging/backend.h logging/filter.cpp @@ -139,10 +138,13 @@ add_library(common STATIC microprofile.h microprofileui.h misc.cpp + nvidia_flags.cpp + nvidia_flags.h page_table.cpp page_table.h param_package.cpp param_package.h + parent_of_member.h quaternion.h ring_buffer.h scm_rev.cpp @@ -165,9 +167,8 @@ add_library(common STATIC threadsafe_queue.h time_zone.cpp time_zone.h - timer.cpp - timer.h - uint128.cpp + tiny_mt.h + tree.h uint128.h uuid.cpp uuid.h @@ -205,6 +206,8 @@ if (MSVC) else() target_compile_options(common PRIVATE -Werror + + $<$<CXX_COMPILER_ID:Clang>:-fsized-deallocation> ) endif() diff --git a/src/common/alignment.h b/src/common/alignment.h index 5040043de..32d796ffa 100644 --- a/src/common/alignment.h +++ b/src/common/alignment.h @@ -9,50 +9,50 @@ namespace Common { template <typename T> -[[nodiscard]] constexpr T AlignUp(T value, std::size_t size) { - static_assert(std::is_unsigned_v<T>, "T must be an unsigned value."); +requires std::is_unsigned_v<T>[[nodiscard]] constexpr T AlignUp(T value, size_t size) { auto mod{static_cast<T>(value % size)}; value -= mod; return static_cast<T>(mod == T{0} ? value : value + size); } template <typename T> -[[nodiscard]] constexpr T AlignDown(T value, std::size_t size) { - static_assert(std::is_unsigned_v<T>, "T must be an unsigned value."); - return static_cast<T>(value - value % size); +requires std::is_unsigned_v<T>[[nodiscard]] constexpr T AlignUpLog2(T value, size_t align_log2) { + return static_cast<T>((value + ((1ULL << align_log2) - 1)) >> align_log2 << align_log2); } template <typename T> -[[nodiscard]] constexpr T AlignBits(T value, std::size_t align) { - static_assert(std::is_unsigned_v<T>, "T must be an unsigned value."); - return static_cast<T>((value + ((1ULL << align) - 1)) >> align << align); +requires std::is_unsigned_v<T>[[nodiscard]] constexpr T AlignDown(T value, size_t size) { + return static_cast<T>(value - value % size); } template <typename T> -[[nodiscard]] constexpr bool Is4KBAligned(T value) { - static_assert(std::is_unsigned_v<T>, "T must be an unsigned value."); +requires std::is_unsigned_v<T>[[nodiscard]] constexpr bool Is4KBAligned(T value) { return (value & 0xFFF) == 0; } template <typename T> -[[nodiscard]] constexpr bool IsWordAligned(T value) { - static_assert(std::is_unsigned_v<T>, "T must be an unsigned value."); +requires std::is_unsigned_v<T>[[nodiscard]] constexpr bool IsWordAligned(T value) { return (value & 0b11) == 0; } template <typename T> -[[nodiscard]] constexpr bool IsAligned(T value, std::size_t alignment) { - using U = typename std::make_unsigned<T>::type; +requires std::is_integral_v<T>[[nodiscard]] constexpr bool IsAligned(T value, size_t alignment) { + using U = typename std::make_unsigned_t<T>; const U mask = static_cast<U>(alignment - 1); return (value & mask) == 0; } -template <typename T, std::size_t Align = 16> +template <typename T, typename U> +requires std::is_integral_v<T>[[nodiscard]] constexpr T DivideUp(T x, U y) { + return (x + (y - 1)) / y; +} + +template <typename T, size_t Align = 16> class AlignmentAllocator { public: using value_type = T; - using size_type = std::size_t; - using difference_type = std::ptrdiff_t; + using size_type = size_t; + using difference_type = ptrdiff_t; using propagate_on_container_copy_assignment = std::true_type; using propagate_on_container_move_assignment = std::true_type; diff --git a/src/common/atomic_ops.cpp b/src/common/atomic_ops.cpp deleted file mode 100644 index 1612d0e67..000000000 --- a/src/common/atomic_ops.cpp +++ /dev/null @@ -1,75 +0,0 @@ -// Copyright 2020 yuzu Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#include <cstring> - -#include "common/atomic_ops.h" - -#if _MSC_VER -#include <intrin.h> -#endif - -namespace Common { - -#if _MSC_VER - -bool AtomicCompareAndSwap(volatile u8* pointer, u8 value, u8 expected) { - const u8 result = - _InterlockedCompareExchange8(reinterpret_cast<volatile char*>(pointer), value, expected); - return result == expected; -} - -bool AtomicCompareAndSwap(volatile u16* pointer, u16 value, u16 expected) { - const u16 result = - _InterlockedCompareExchange16(reinterpret_cast<volatile short*>(pointer), value, expected); - return result == expected; -} - -bool AtomicCompareAndSwap(volatile u32* pointer, u32 value, u32 expected) { - const u32 result = - _InterlockedCompareExchange(reinterpret_cast<volatile long*>(pointer), value, expected); - return result == expected; -} - -bool AtomicCompareAndSwap(volatile u64* pointer, u64 value, u64 expected) { - const u64 result = _InterlockedCompareExchange64(reinterpret_cast<volatile __int64*>(pointer), - value, expected); - return result == expected; -} - -bool AtomicCompareAndSwap(volatile u64* pointer, u128 value, u128 expected) { - return _InterlockedCompareExchange128(reinterpret_cast<volatile __int64*>(pointer), value[1], - value[0], - reinterpret_cast<__int64*>(expected.data())) != 0; -} - -#else - -bool AtomicCompareAndSwap(volatile u8* pointer, u8 value, u8 expected) { - return __sync_bool_compare_and_swap(pointer, expected, value); -} - -bool AtomicCompareAndSwap(volatile u16* pointer, u16 value, u16 expected) { - return __sync_bool_compare_and_swap(pointer, expected, value); -} - -bool AtomicCompareAndSwap(volatile u32* pointer, u32 value, u32 expected) { - return __sync_bool_compare_and_swap(pointer, expected, value); -} - -bool AtomicCompareAndSwap(volatile u64* pointer, u64 value, u64 expected) { - return __sync_bool_compare_and_swap(pointer, expected, value); -} - -bool AtomicCompareAndSwap(volatile u64* pointer, u128 value, u128 expected) { - unsigned __int128 value_a; - unsigned __int128 expected_a; - std::memcpy(&value_a, value.data(), sizeof(u128)); - std::memcpy(&expected_a, expected.data(), sizeof(u128)); - return __sync_bool_compare_and_swap((unsigned __int128*)pointer, expected_a, value_a); -} - -#endif - -} // namespace Common diff --git a/src/common/atomic_ops.h b/src/common/atomic_ops.h index b46888589..2b1f515e8 100644 --- a/src/common/atomic_ops.h +++ b/src/common/atomic_ops.h @@ -4,14 +4,75 @@ #pragma once +#include <cstring> +#include <memory> + #include "common/common_types.h" +#if _MSC_VER +#include <intrin.h> +#endif + namespace Common { -[[nodiscard]] bool AtomicCompareAndSwap(volatile u8* pointer, u8 value, u8 expected); -[[nodiscard]] bool AtomicCompareAndSwap(volatile u16* pointer, u16 value, u16 expected); -[[nodiscard]] bool AtomicCompareAndSwap(volatile u32* pointer, u32 value, u32 expected); -[[nodiscard]] bool AtomicCompareAndSwap(volatile u64* pointer, u64 value, u64 expected); -[[nodiscard]] bool AtomicCompareAndSwap(volatile u64* pointer, u128 value, u128 expected); +#if _MSC_VER + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u8* pointer, u8 value, u8 expected) { + const u8 result = + _InterlockedCompareExchange8(reinterpret_cast<volatile char*>(pointer), value, expected); + return result == expected; +} + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u16* pointer, u16 value, u16 expected) { + const u16 result = + _InterlockedCompareExchange16(reinterpret_cast<volatile short*>(pointer), value, expected); + return result == expected; +} + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u32* pointer, u32 value, u32 expected) { + const u32 result = + _InterlockedCompareExchange(reinterpret_cast<volatile long*>(pointer), value, expected); + return result == expected; +} + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u64* pointer, u64 value, u64 expected) { + const u64 result = _InterlockedCompareExchange64(reinterpret_cast<volatile __int64*>(pointer), + value, expected); + return result == expected; +} + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u64* pointer, u128 value, u128 expected) { + return _InterlockedCompareExchange128(reinterpret_cast<volatile __int64*>(pointer), value[1], + value[0], + reinterpret_cast<__int64*>(expected.data())) != 0; +} + +#else + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u8* pointer, u8 value, u8 expected) { + return __sync_bool_compare_and_swap(pointer, expected, value); +} + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u16* pointer, u16 value, u16 expected) { + return __sync_bool_compare_and_swap(pointer, expected, value); +} + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u32* pointer, u32 value, u32 expected) { + return __sync_bool_compare_and_swap(pointer, expected, value); +} + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u64* pointer, u64 value, u64 expected) { + return __sync_bool_compare_and_swap(pointer, expected, value); +} + +[[nodiscard]] inline bool AtomicCompareAndSwap(volatile u64* pointer, u128 value, u128 expected) { + unsigned __int128 value_a; + unsigned __int128 expected_a; + std::memcpy(&value_a, value.data(), sizeof(u128)); + std::memcpy(&expected_a, expected.data(), sizeof(u128)); + return __sync_bool_compare_and_swap((unsigned __int128*)pointer, expected_a, value_a); +} + +#endif } // namespace Common diff --git a/src/common/bit_util.h b/src/common/bit_util.h index 29f59a9a3..64520ca4e 100644 --- a/src/common/bit_util.h +++ b/src/common/bit_util.h @@ -4,13 +4,10 @@ #pragma once +#include <bit> #include <climits> #include <cstddef> -#ifdef _MSC_VER -#include <intrin.h> -#endif - #include "common/common_types.h" namespace Common { @@ -21,124 +18,30 @@ template <typename T> return sizeof(T) * CHAR_BIT; } -#ifdef _MSC_VER -[[nodiscard]] inline u32 CountLeadingZeroes32(u32 value) { - unsigned long leading_zero = 0; - - if (_BitScanReverse(&leading_zero, value) != 0) { - return 31 - leading_zero; - } - - return 32; -} - -[[nodiscard]] inline u32 CountLeadingZeroes64(u64 value) { - unsigned long leading_zero = 0; - - if (_BitScanReverse64(&leading_zero, value) != 0) { - return 63 - leading_zero; - } - - return 64; -} -#else -[[nodiscard]] inline u32 CountLeadingZeroes32(u32 value) { - if (value == 0) { - return 32; - } - - return static_cast<u32>(__builtin_clz(value)); -} - -[[nodiscard]] inline u32 CountLeadingZeroes64(u64 value) { - if (value == 0) { - return 64; - } - - return static_cast<u32>(__builtin_clzll(value)); -} -#endif - -#ifdef _MSC_VER -[[nodiscard]] inline u32 CountTrailingZeroes32(u32 value) { - unsigned long trailing_zero = 0; - - if (_BitScanForward(&trailing_zero, value) != 0) { - return trailing_zero; - } - - return 32; -} - -[[nodiscard]] inline u32 CountTrailingZeroes64(u64 value) { - unsigned long trailing_zero = 0; - - if (_BitScanForward64(&trailing_zero, value) != 0) { - return trailing_zero; - } - - return 64; -} -#else -[[nodiscard]] inline u32 CountTrailingZeroes32(u32 value) { - if (value == 0) { - return 32; - } - - return static_cast<u32>(__builtin_ctz(value)); -} - -[[nodiscard]] inline u32 CountTrailingZeroes64(u64 value) { - if (value == 0) { - return 64; - } - - return static_cast<u32>(__builtin_ctzll(value)); +[[nodiscard]] constexpr u32 MostSignificantBit32(const u32 value) { + return 31U - static_cast<u32>(std::countl_zero(value)); } -#endif - -#ifdef _MSC_VER -[[nodiscard]] inline u32 MostSignificantBit32(const u32 value) { - unsigned long result; - _BitScanReverse(&result, value); - return static_cast<u32>(result); +[[nodiscard]] constexpr u32 MostSignificantBit64(const u64 value) { + return 63U - static_cast<u32>(std::countl_zero(value)); } -[[nodiscard]] inline u32 MostSignificantBit64(const u64 value) { - unsigned long result; - _BitScanReverse64(&result, value); - return static_cast<u32>(result); -} - -#else - -[[nodiscard]] inline u32 MostSignificantBit32(const u32 value) { - return 31U - static_cast<u32>(__builtin_clz(value)); -} - -[[nodiscard]] inline u32 MostSignificantBit64(const u64 value) { - return 63U - static_cast<u32>(__builtin_clzll(value)); -} - -#endif - -[[nodiscard]] inline u32 Log2Floor32(const u32 value) { +[[nodiscard]] constexpr u32 Log2Floor32(const u32 value) { return MostSignificantBit32(value); } -[[nodiscard]] inline u32 Log2Ceil32(const u32 value) { - const u32 log2_f = Log2Floor32(value); - return log2_f + ((value ^ (1U << log2_f)) != 0U); +[[nodiscard]] constexpr u32 Log2Floor64(const u64 value) { + return MostSignificantBit64(value); } -[[nodiscard]] inline u32 Log2Floor64(const u64 value) { - return MostSignificantBit64(value); +[[nodiscard]] constexpr u32 Log2Ceil32(const u32 value) { + const u32 log2_f = Log2Floor32(value); + return log2_f + static_cast<u32>((value ^ (1U << log2_f)) != 0U); } -[[nodiscard]] inline u32 Log2Ceil64(const u64 value) { - const u64 log2_f = static_cast<u64>(Log2Floor64(value)); - return static_cast<u32>(log2_f + ((value ^ (1ULL << log2_f)) != 0ULL)); +[[nodiscard]] constexpr u32 Log2Ceil64(const u64 value) { + const u64 log2_f = Log2Floor64(value); + return static_cast<u32>(log2_f + static_cast<u64>((value ^ (1ULL << log2_f)) != 0ULL)); } } // namespace Common diff --git a/src/common/cityhash.cpp b/src/common/cityhash.cpp index 4e1d874b5..66218fc21 100644 --- a/src/common/cityhash.cpp +++ b/src/common/cityhash.cpp @@ -28,8 +28,10 @@ // compromising on hash quality. #include <algorithm> -#include <string.h> // for memcpy and memset -#include "cityhash.h" +#include <cstring> +#include <utility> + +#include "common/cityhash.h" #include "common/swap.h" // #include "config.h" @@ -42,21 +44,17 @@ using namespace std; -typedef uint8_t uint8; -typedef uint32_t uint32; -typedef uint64_t uint64; - namespace Common { -static uint64 UNALIGNED_LOAD64(const char* p) { - uint64 result; - memcpy(&result, p, sizeof(result)); +static u64 unaligned_load64(const char* p) { + u64 result; + std::memcpy(&result, p, sizeof(result)); return result; } -static uint32 UNALIGNED_LOAD32(const char* p) { - uint32 result; - memcpy(&result, p, sizeof(result)); +static u32 unaligned_load32(const char* p) { + u32 result; + std::memcpy(&result, p, sizeof(result)); return result; } @@ -76,64 +74,64 @@ static uint32 UNALIGNED_LOAD32(const char* p) { #endif #endif -static uint64 Fetch64(const char* p) { - return uint64_in_expected_order(UNALIGNED_LOAD64(p)); +static u64 Fetch64(const char* p) { + return uint64_in_expected_order(unaligned_load64(p)); } -static uint32 Fetch32(const char* p) { - return uint32_in_expected_order(UNALIGNED_LOAD32(p)); +static u32 Fetch32(const char* p) { + return uint32_in_expected_order(unaligned_load32(p)); } // Some primes between 2^63 and 2^64 for various uses. -static const uint64 k0 = 0xc3a5c85c97cb3127ULL; -static const uint64 k1 = 0xb492b66fbe98f273ULL; -static const uint64 k2 = 0x9ae16a3b2f90404fULL; +static constexpr u64 k0 = 0xc3a5c85c97cb3127ULL; +static constexpr u64 k1 = 0xb492b66fbe98f273ULL; +static constexpr u64 k2 = 0x9ae16a3b2f90404fULL; // Bitwise right rotate. Normally this will compile to a single // instruction, especially if the shift is a manifest constant. -static uint64 Rotate(uint64 val, int shift) { +static u64 Rotate(u64 val, int shift) { // Avoid shifting by 64: doing so yields an undefined result. return shift == 0 ? val : ((val >> shift) | (val << (64 - shift))); } -static uint64 ShiftMix(uint64 val) { +static u64 ShiftMix(u64 val) { return val ^ (val >> 47); } -static uint64 HashLen16(uint64 u, uint64 v) { - return Hash128to64(uint128(u, v)); +static u64 HashLen16(u64 u, u64 v) { + return Hash128to64(u128{u, v}); } -static uint64 HashLen16(uint64 u, uint64 v, uint64 mul) { +static u64 HashLen16(u64 u, u64 v, u64 mul) { // Murmur-inspired hashing. - uint64 a = (u ^ v) * mul; + u64 a = (u ^ v) * mul; a ^= (a >> 47); - uint64 b = (v ^ a) * mul; + u64 b = (v ^ a) * mul; b ^= (b >> 47); b *= mul; return b; } -static uint64 HashLen0to16(const char* s, std::size_t len) { +static u64 HashLen0to16(const char* s, size_t len) { if (len >= 8) { - uint64 mul = k2 + len * 2; - uint64 a = Fetch64(s) + k2; - uint64 b = Fetch64(s + len - 8); - uint64 c = Rotate(b, 37) * mul + a; - uint64 d = (Rotate(a, 25) + b) * mul; + u64 mul = k2 + len * 2; + u64 a = Fetch64(s) + k2; + u64 b = Fetch64(s + len - 8); + u64 c = Rotate(b, 37) * mul + a; + u64 d = (Rotate(a, 25) + b) * mul; return HashLen16(c, d, mul); } if (len >= 4) { - uint64 mul = k2 + len * 2; - uint64 a = Fetch32(s); + u64 mul = k2 + len * 2; + u64 a = Fetch32(s); return HashLen16(len + (a << 3), Fetch32(s + len - 4), mul); } if (len > 0) { - uint8 a = s[0]; - uint8 b = s[len >> 1]; - uint8 c = s[len - 1]; - uint32 y = static_cast<uint32>(a) + (static_cast<uint32>(b) << 8); - uint32 z = static_cast<uint32>(len) + (static_cast<uint32>(c) << 2); + u8 a = s[0]; + u8 b = s[len >> 1]; + u8 c = s[len - 1]; + u32 y = static_cast<u32>(a) + (static_cast<u32>(b) << 8); + u32 z = static_cast<u32>(len) + (static_cast<u32>(c) << 2); return ShiftMix(y * k2 ^ z * k0) * k2; } return k2; @@ -141,22 +139,21 @@ static uint64 HashLen0to16(const char* s, std::size_t len) { // This probably works well for 16-byte strings as well, but it may be overkill // in that case. -static uint64 HashLen17to32(const char* s, std::size_t len) { - uint64 mul = k2 + len * 2; - uint64 a = Fetch64(s) * k1; - uint64 b = Fetch64(s + 8); - uint64 c = Fetch64(s + len - 8) * mul; - uint64 d = Fetch64(s + len - 16) * k2; +static u64 HashLen17to32(const char* s, size_t len) { + u64 mul = k2 + len * 2; + u64 a = Fetch64(s) * k1; + u64 b = Fetch64(s + 8); + u64 c = Fetch64(s + len - 8) * mul; + u64 d = Fetch64(s + len - 16) * k2; return HashLen16(Rotate(a + b, 43) + Rotate(c, 30) + d, a + Rotate(b + k2, 18) + c, mul); } // Return a 16-byte hash for 48 bytes. Quick and dirty. // Callers do best to use "random-looking" values for a and b. -static pair<uint64, uint64> WeakHashLen32WithSeeds(uint64 w, uint64 x, uint64 y, uint64 z, uint64 a, - uint64 b) { +static pair<u64, u64> WeakHashLen32WithSeeds(u64 w, u64 x, u64 y, u64 z, u64 a, u64 b) { a += w; b = Rotate(b + a + z, 21); - uint64 c = a; + u64 c = a; a += x; a += y; b += Rotate(a, 44); @@ -164,34 +161,34 @@ static pair<uint64, uint64> WeakHashLen32WithSeeds(uint64 w, uint64 x, uint64 y, } // Return a 16-byte hash for s[0] ... s[31], a, and b. Quick and dirty. -static pair<uint64, uint64> WeakHashLen32WithSeeds(const char* s, uint64 a, uint64 b) { +static pair<u64, u64> WeakHashLen32WithSeeds(const char* s, u64 a, u64 b) { return WeakHashLen32WithSeeds(Fetch64(s), Fetch64(s + 8), Fetch64(s + 16), Fetch64(s + 24), a, b); } // Return an 8-byte hash for 33 to 64 bytes. -static uint64 HashLen33to64(const char* s, std::size_t len) { - uint64 mul = k2 + len * 2; - uint64 a = Fetch64(s) * k2; - uint64 b = Fetch64(s + 8); - uint64 c = Fetch64(s + len - 24); - uint64 d = Fetch64(s + len - 32); - uint64 e = Fetch64(s + 16) * k2; - uint64 f = Fetch64(s + 24) * 9; - uint64 g = Fetch64(s + len - 8); - uint64 h = Fetch64(s + len - 16) * mul; - uint64 u = Rotate(a + g, 43) + (Rotate(b, 30) + c) * 9; - uint64 v = ((a + g) ^ d) + f + 1; - uint64 w = swap64((u + v) * mul) + h; - uint64 x = Rotate(e + f, 42) + c; - uint64 y = (swap64((v + w) * mul) + g) * mul; - uint64 z = e + f + c; +static u64 HashLen33to64(const char* s, size_t len) { + u64 mul = k2 + len * 2; + u64 a = Fetch64(s) * k2; + u64 b = Fetch64(s + 8); + u64 c = Fetch64(s + len - 24); + u64 d = Fetch64(s + len - 32); + u64 e = Fetch64(s + 16) * k2; + u64 f = Fetch64(s + 24) * 9; + u64 g = Fetch64(s + len - 8); + u64 h = Fetch64(s + len - 16) * mul; + u64 u = Rotate(a + g, 43) + (Rotate(b, 30) + c) * 9; + u64 v = ((a + g) ^ d) + f + 1; + u64 w = swap64((u + v) * mul) + h; + u64 x = Rotate(e + f, 42) + c; + u64 y = (swap64((v + w) * mul) + g) * mul; + u64 z = e + f + c; a = swap64((x + z) * mul + y) + b; b = ShiftMix((z + a) * mul + d + h) * mul; return b + x; } -uint64 CityHash64(const char* s, std::size_t len) { +u64 CityHash64(const char* s, size_t len) { if (len <= 32) { if (len <= 16) { return HashLen0to16(s, len); @@ -204,15 +201,15 @@ uint64 CityHash64(const char* s, std::size_t len) { // For strings over 64 bytes we hash the end first, and then as we // loop we keep 56 bytes of state: v, w, x, y, and z. - uint64 x = Fetch64(s + len - 40); - uint64 y = Fetch64(s + len - 16) + Fetch64(s + len - 56); - uint64 z = HashLen16(Fetch64(s + len - 48) + len, Fetch64(s + len - 24)); - pair<uint64, uint64> v = WeakHashLen32WithSeeds(s + len - 64, len, z); - pair<uint64, uint64> w = WeakHashLen32WithSeeds(s + len - 32, y + k1, x); + u64 x = Fetch64(s + len - 40); + u64 y = Fetch64(s + len - 16) + Fetch64(s + len - 56); + u64 z = HashLen16(Fetch64(s + len - 48) + len, Fetch64(s + len - 24)); + pair<u64, u64> v = WeakHashLen32WithSeeds(s + len - 64, len, z); + pair<u64, u64> w = WeakHashLen32WithSeeds(s + len - 32, y + k1, x); x = x * k1 + Fetch64(s); // Decrease len to the nearest multiple of 64, and operate on 64-byte chunks. - len = (len - 1) & ~static_cast<std::size_t>(63); + len = (len - 1) & ~static_cast<size_t>(63); do { x = Rotate(x + y + v.first + Fetch64(s + 8), 37) * k1; y = Rotate(y + v.second + Fetch64(s + 48), 42) * k1; @@ -229,21 +226,21 @@ uint64 CityHash64(const char* s, std::size_t len) { HashLen16(v.second, w.second) + x); } -uint64 CityHash64WithSeed(const char* s, std::size_t len, uint64 seed) { +u64 CityHash64WithSeed(const char* s, size_t len, u64 seed) { return CityHash64WithSeeds(s, len, k2, seed); } -uint64 CityHash64WithSeeds(const char* s, std::size_t len, uint64 seed0, uint64 seed1) { +u64 CityHash64WithSeeds(const char* s, size_t len, u64 seed0, u64 seed1) { return HashLen16(CityHash64(s, len) - seed0, seed1); } // A subroutine for CityHash128(). Returns a decent 128-bit hash for strings // of any length representable in signed long. Based on City and Murmur. -static uint128 CityMurmur(const char* s, std::size_t len, uint128 seed) { - uint64 a = Uint128Low64(seed); - uint64 b = Uint128High64(seed); - uint64 c = 0; - uint64 d = 0; +static u128 CityMurmur(const char* s, size_t len, u128 seed) { + u64 a = seed[0]; + u64 b = seed[1]; + u64 c = 0; + u64 d = 0; signed long l = static_cast<long>(len) - 16; if (l <= 0) { // len <= 16 a = ShiftMix(a * k1) * k1; @@ -266,20 +263,20 @@ static uint128 CityMurmur(const char* s, std::size_t len, uint128 seed) { } a = HashLen16(a, c); b = HashLen16(d, b); - return uint128(a ^ b, HashLen16(b, a)); + return u128{a ^ b, HashLen16(b, a)}; } -uint128 CityHash128WithSeed(const char* s, std::size_t len, uint128 seed) { +u128 CityHash128WithSeed(const char* s, size_t len, u128 seed) { if (len < 128) { return CityMurmur(s, len, seed); } // We expect len >= 128 to be the common case. Keep 56 bytes of state: // v, w, x, y, and z. - pair<uint64, uint64> v, w; - uint64 x = Uint128Low64(seed); - uint64 y = Uint128High64(seed); - uint64 z = len * k1; + pair<u64, u64> v, w; + u64 x = seed[0]; + u64 y = seed[1]; + u64 z = len * k1; v.first = Rotate(y ^ k1, 49) * k1 + Fetch64(s); v.second = Rotate(v.first, 42) * k1 + Fetch64(s + 8); w.first = Rotate(y + z, 35) * k1 + x; @@ -313,7 +310,7 @@ uint128 CityHash128WithSeed(const char* s, std::size_t len, uint128 seed) { w.first *= 9; v.first *= k0; // If 0 < len < 128, hash up to 4 chunks of 32 bytes each from the end of s. - for (std::size_t tail_done = 0; tail_done < len;) { + for (size_t tail_done = 0; tail_done < len;) { tail_done += 32; y = Rotate(x + y, 42) * k0 + v.second; w.first += Fetch64(s + len - tail_done + 16); @@ -328,13 +325,12 @@ uint128 CityHash128WithSeed(const char* s, std::size_t len, uint128 seed) { // different 56-byte-to-8-byte hashes to get a 16-byte final result. x = HashLen16(x, v.first); y = HashLen16(y + z, w.first); - return uint128(HashLen16(x + v.second, w.second) + y, HashLen16(x + w.second, y + v.second)); + return u128{HashLen16(x + v.second, w.second) + y, HashLen16(x + w.second, y + v.second)}; } -uint128 CityHash128(const char* s, std::size_t len) { - return len >= 16 - ? CityHash128WithSeed(s + 16, len - 16, uint128(Fetch64(s), Fetch64(s + 8) + k0)) - : CityHash128WithSeed(s, len, uint128(k0, k1)); +u128 CityHash128(const char* s, size_t len) { + return len >= 16 ? CityHash128WithSeed(s + 16, len - 16, u128{Fetch64(s), Fetch64(s + 8) + k0}) + : CityHash128WithSeed(s, len, u128{k0, k1}); } } // namespace Common diff --git a/src/common/cityhash.h b/src/common/cityhash.h index a00804e01..d74fc7639 100644 --- a/src/common/cityhash.h +++ b/src/common/cityhash.h @@ -62,49 +62,38 @@ #pragma once #include <cstddef> -#include <cstdint> -#include <utility> +#include "common/common_types.h" namespace Common { -using uint128 = std::pair<uint64_t, uint64_t>; - -[[nodiscard]] inline uint64_t Uint128Low64(const uint128& x) { - return x.first; -} -[[nodiscard]] inline uint64_t Uint128High64(const uint128& x) { - return x.second; -} - // Hash function for a byte array. -[[nodiscard]] uint64_t CityHash64(const char* buf, std::size_t len); +[[nodiscard]] u64 CityHash64(const char* buf, size_t len); // Hash function for a byte array. For convenience, a 64-bit seed is also // hashed into the result. -[[nodiscard]] uint64_t CityHash64WithSeed(const char* buf, std::size_t len, uint64_t seed); +[[nodiscard]] u64 CityHash64WithSeed(const char* buf, size_t len, u64 seed); // Hash function for a byte array. For convenience, two seeds are also // hashed into the result. -[[nodiscard]] uint64_t CityHash64WithSeeds(const char* buf, std::size_t len, uint64_t seed0, - uint64_t seed1); +[[nodiscard]] u64 CityHash64WithSeeds(const char* buf, size_t len, u64 seed0, u64 seed1); // Hash function for a byte array. -[[nodiscard]] uint128 CityHash128(const char* s, std::size_t len); +[[nodiscard]] u128 CityHash128(const char* s, size_t len); // Hash function for a byte array. For convenience, a 128-bit seed is also // hashed into the result. -[[nodiscard]] uint128 CityHash128WithSeed(const char* s, std::size_t len, uint128 seed); +[[nodiscard]] u128 CityHash128WithSeed(const char* s, size_t len, u128 seed); // Hash 128 input bits down to 64 bits of output. // This is intended to be a reasonably good hash function. -[[nodiscard]] inline uint64_t Hash128to64(const uint128& x) { +[[nodiscard]] inline u64 Hash128to64(const u128& x) { // Murmur-inspired hashing. - const uint64_t kMul = 0x9ddfea08eb382d69ULL; - uint64_t a = (Uint128Low64(x) ^ Uint128High64(x)) * kMul; + const u64 mul = 0x9ddfea08eb382d69ULL; + u64 a = (x[0] ^ x[1]) * mul; a ^= (a >> 47); - uint64_t b = (Uint128High64(x) ^ a) * kMul; + u64 b = (x[1] ^ a) * mul; b ^= (b >> 47); - b *= kMul; + b *= mul; return b; } diff --git a/src/common/color.h b/src/common/color.h deleted file mode 100644 index bbcac858e..000000000 --- a/src/common/color.h +++ /dev/null @@ -1,271 +0,0 @@ -// Copyright 2014 Citra Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#pragma once - -#include <cstring> - -#include "common/common_types.h" -#include "common/swap.h" -#include "common/vector_math.h" - -namespace Common::Color { - -/// Convert a 1-bit color component to 8 bit -[[nodiscard]] constexpr u8 Convert1To8(u8 value) { - return value * 255; -} - -/// Convert a 4-bit color component to 8 bit -[[nodiscard]] constexpr u8 Convert4To8(u8 value) { - return (value << 4) | value; -} - -/// Convert a 5-bit color component to 8 bit -[[nodiscard]] constexpr u8 Convert5To8(u8 value) { - return (value << 3) | (value >> 2); -} - -/// Convert a 6-bit color component to 8 bit -[[nodiscard]] constexpr u8 Convert6To8(u8 value) { - return (value << 2) | (value >> 4); -} - -/// Convert a 8-bit color component to 1 bit -[[nodiscard]] constexpr u8 Convert8To1(u8 value) { - return value >> 7; -} - -/// Convert a 8-bit color component to 4 bit -[[nodiscard]] constexpr u8 Convert8To4(u8 value) { - return value >> 4; -} - -/// Convert a 8-bit color component to 5 bit -[[nodiscard]] constexpr u8 Convert8To5(u8 value) { - return value >> 3; -} - -/// Convert a 8-bit color component to 6 bit -[[nodiscard]] constexpr u8 Convert8To6(u8 value) { - return value >> 2; -} - -/** - * Decode a color stored in RGBA8 format - * @param bytes Pointer to encoded source color - * @return Result color decoded as Common::Vec4<u8> - */ -[[nodiscard]] inline Common::Vec4<u8> DecodeRGBA8(const u8* bytes) { - return {bytes[3], bytes[2], bytes[1], bytes[0]}; -} - -/** - * Decode a color stored in RGB8 format - * @param bytes Pointer to encoded source color - * @return Result color decoded as Common::Vec4<u8> - */ -[[nodiscard]] inline Common::Vec4<u8> DecodeRGB8(const u8* bytes) { - return {bytes[2], bytes[1], bytes[0], 255}; -} - -/** - * Decode a color stored in RG8 (aka HILO8) format - * @param bytes Pointer to encoded source color - * @return Result color decoded as Common::Vec4<u8> - */ -[[nodiscard]] inline Common::Vec4<u8> DecodeRG8(const u8* bytes) { - return {bytes[1], bytes[0], 0, 255}; -} - -/** - * Decode a color stored in RGB565 format - * @param bytes Pointer to encoded source color - * @return Result color decoded as Common::Vec4<u8> - */ -[[nodiscard]] inline Common::Vec4<u8> DecodeRGB565(const u8* bytes) { - u16_le pixel; - std::memcpy(&pixel, bytes, sizeof(pixel)); - return {Convert5To8((pixel >> 11) & 0x1F), Convert6To8((pixel >> 5) & 0x3F), - Convert5To8(pixel & 0x1F), 255}; -} - -/** - * Decode a color stored in RGB5A1 format - * @param bytes Pointer to encoded source color - * @return Result color decoded as Common::Vec4<u8> - */ -[[nodiscard]] inline Common::Vec4<u8> DecodeRGB5A1(const u8* bytes) { - u16_le pixel; - std::memcpy(&pixel, bytes, sizeof(pixel)); - return {Convert5To8((pixel >> 11) & 0x1F), Convert5To8((pixel >> 6) & 0x1F), - Convert5To8((pixel >> 1) & 0x1F), Convert1To8(pixel & 0x1)}; -} - -/** - * Decode a color stored in RGBA4 format - * @param bytes Pointer to encoded source color - * @return Result color decoded as Common::Vec4<u8> - */ -[[nodiscard]] inline Common::Vec4<u8> DecodeRGBA4(const u8* bytes) { - u16_le pixel; - std::memcpy(&pixel, bytes, sizeof(pixel)); - return {Convert4To8((pixel >> 12) & 0xF), Convert4To8((pixel >> 8) & 0xF), - Convert4To8((pixel >> 4) & 0xF), Convert4To8(pixel & 0xF)}; -} - -/** - * Decode a depth value stored in D16 format - * @param bytes Pointer to encoded source value - * @return Depth value as an u32 - */ -[[nodiscard]] inline u32 DecodeD16(const u8* bytes) { - u16_le data; - std::memcpy(&data, bytes, sizeof(data)); - return data; -} - -/** - * Decode a depth value stored in D24 format - * @param bytes Pointer to encoded source value - * @return Depth value as an u32 - */ -[[nodiscard]] inline u32 DecodeD24(const u8* bytes) { - return (bytes[2] << 16) | (bytes[1] << 8) | bytes[0]; -} - -/** - * Decode a depth value and a stencil value stored in D24S8 format - * @param bytes Pointer to encoded source values - * @return Resulting values stored as a Common::Vec2 - */ -[[nodiscard]] inline Common::Vec2<u32> DecodeD24S8(const u8* bytes) { - return {static_cast<u32>((bytes[2] << 16) | (bytes[1] << 8) | bytes[0]), bytes[3]}; -} - -/** - * Encode a color as RGBA8 format - * @param color Source color to encode - * @param bytes Destination pointer to store encoded color - */ -inline void EncodeRGBA8(const Common::Vec4<u8>& color, u8* bytes) { - bytes[3] = color.r(); - bytes[2] = color.g(); - bytes[1] = color.b(); - bytes[0] = color.a(); -} - -/** - * Encode a color as RGB8 format - * @param color Source color to encode - * @param bytes Destination pointer to store encoded color - */ -inline void EncodeRGB8(const Common::Vec4<u8>& color, u8* bytes) { - bytes[2] = color.r(); - bytes[1] = color.g(); - bytes[0] = color.b(); -} - -/** - * Encode a color as RG8 (aka HILO8) format - * @param color Source color to encode - * @param bytes Destination pointer to store encoded color - */ -inline void EncodeRG8(const Common::Vec4<u8>& color, u8* bytes) { - bytes[1] = color.r(); - bytes[0] = color.g(); -} -/** - * Encode a color as RGB565 format - * @param color Source color to encode - * @param bytes Destination pointer to store encoded color - */ -inline void EncodeRGB565(const Common::Vec4<u8>& color, u8* bytes) { - const u16_le data = - (Convert8To5(color.r()) << 11) | (Convert8To6(color.g()) << 5) | Convert8To5(color.b()); - - std::memcpy(bytes, &data, sizeof(data)); -} - -/** - * Encode a color as RGB5A1 format - * @param color Source color to encode - * @param bytes Destination pointer to store encoded color - */ -inline void EncodeRGB5A1(const Common::Vec4<u8>& color, u8* bytes) { - const u16_le data = (Convert8To5(color.r()) << 11) | (Convert8To5(color.g()) << 6) | - (Convert8To5(color.b()) << 1) | Convert8To1(color.a()); - - std::memcpy(bytes, &data, sizeof(data)); -} - -/** - * Encode a color as RGBA4 format - * @param color Source color to encode - * @param bytes Destination pointer to store encoded color - */ -inline void EncodeRGBA4(const Common::Vec4<u8>& color, u8* bytes) { - const u16 data = (Convert8To4(color.r()) << 12) | (Convert8To4(color.g()) << 8) | - (Convert8To4(color.b()) << 4) | Convert8To4(color.a()); - - std::memcpy(bytes, &data, sizeof(data)); -} - -/** - * Encode a 16 bit depth value as D16 format - * @param value 16 bit source depth value to encode - * @param bytes Pointer where to store the encoded value - */ -inline void EncodeD16(u32 value, u8* bytes) { - const u16_le data = static_cast<u16>(value); - std::memcpy(bytes, &data, sizeof(data)); -} - -/** - * Encode a 24 bit depth value as D24 format - * @param value 24 bit source depth value to encode - * @param bytes Pointer where to store the encoded value - */ -inline void EncodeD24(u32 value, u8* bytes) { - bytes[0] = value & 0xFF; - bytes[1] = (value >> 8) & 0xFF; - bytes[2] = (value >> 16) & 0xFF; -} - -/** - * Encode a 24 bit depth and 8 bit stencil values as D24S8 format - * @param depth 24 bit source depth value to encode - * @param stencil 8 bit source stencil value to encode - * @param bytes Pointer where to store the encoded value - */ -inline void EncodeD24S8(u32 depth, u8 stencil, u8* bytes) { - bytes[0] = depth & 0xFF; - bytes[1] = (depth >> 8) & 0xFF; - bytes[2] = (depth >> 16) & 0xFF; - bytes[3] = stencil; -} - -/** - * Encode a 24 bit depth value as D24X8 format (32 bits per pixel with 8 bits unused) - * @param depth 24 bit source depth value to encode - * @param bytes Pointer where to store the encoded value - * @note unused bits will not be modified - */ -inline void EncodeD24X8(u32 depth, u8* bytes) { - bytes[0] = depth & 0xFF; - bytes[1] = (depth >> 8) & 0xFF; - bytes[2] = (depth >> 16) & 0xFF; -} - -/** - * Encode an 8 bit stencil value as X24S8 format (32 bits per pixel with 24 bits unused) - * @param stencil 8 bit source stencil value to encode - * @param bytes Pointer where to store the encoded value - * @note unused bits will not be modified - */ -inline void EncodeX24S8(u8 stencil, u8* bytes) { - bytes[3] = stencil; -} - -} // namespace Common::Color diff --git a/src/common/common_funcs.h b/src/common/common_funcs.h index 367b6bf6e..4ace2cd33 100644 --- a/src/common/common_funcs.h +++ b/src/common/common_funcs.h @@ -24,10 +24,10 @@ #define INSERT_PADDING_WORDS(num_words) \ std::array<u32, num_words> CONCAT2(pad, __LINE__) {} -/// These are similar to the INSERT_PADDING_* macros, but are needed for padding unions. This is -/// because unions can only be initialized by one member. -#define INSERT_UNION_PADDING_BYTES(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__) -#define INSERT_UNION_PADDING_WORDS(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__) +/// These are similar to the INSERT_PADDING_* macros but do not zero-initialize the contents. +/// This keeps the structure trivial to construct. +#define INSERT_PADDING_BYTES_NOINIT(num_bytes) std::array<u8, num_bytes> CONCAT2(pad, __LINE__) +#define INSERT_PADDING_WORDS_NOINIT(num_words) std::array<u32, num_words> CONCAT2(pad, __LINE__) #ifndef _MSC_VER @@ -52,9 +52,13 @@ __declspec(dllimport) void __stdcall DebugBreak(void); // Generic function to get last error message. // Call directly after the command or use the error num. // This function might change the error code. -// Defined in Misc.cpp. +// Defined in misc.cpp. [[nodiscard]] std::string GetLastErrorMsg(); +// Like GetLastErrorMsg(), but passing an explicit error code. +// Defined in misc.cpp. +[[nodiscard]] std::string NativeErrorToString(int e); + #define DECLARE_ENUM_FLAG_OPERATORS(type) \ [[nodiscard]] constexpr type operator|(type a, type b) noexcept { \ using T = std::underlying_type_t<type>; \ @@ -93,6 +97,31 @@ __declspec(dllimport) void __stdcall DebugBreak(void); return static_cast<T>(key) == 0; \ } +/// Evaluates a boolean expression, and returns a result unless that expression is true. +#define R_UNLESS(expr, res) \ + { \ + if (!(expr)) { \ + if (res.IsError()) { \ + LOG_ERROR(Kernel, "Failed with result: {}", res.raw); \ + } \ + return res; \ + } \ + } + +#define R_SUCCEEDED(res) (res.IsSuccess()) + +/// Evaluates an expression that returns a result, and returns the result if it would fail. +#define R_TRY(res_expr) \ + { \ + const auto _tmp_r_try_rc = (res_expr); \ + if (_tmp_r_try_rc.IsError()) { \ + return _tmp_r_try_rc; \ + } \ + } + +/// Evaluates a boolean expression, and succeeds if that expression is true. +#define R_SUCCEED_IF(expr) R_UNLESS(!(expr), RESULT_SUCCESS) + namespace Common { [[nodiscard]] constexpr u32 MakeMagic(char a, char b, char c, char d) { diff --git a/src/common/fiber.cpp b/src/common/fiber.cpp index 3c1eefcb7..39532ff58 100644 --- a/src/common/fiber.cpp +++ b/src/common/fiber.cpp @@ -116,16 +116,19 @@ void Fiber::Rewind() { boost::context::detail::jump_fcontext(impl->rewind_context, this); } -void Fiber::YieldTo(std::shared_ptr<Fiber> from, std::shared_ptr<Fiber> to) { - ASSERT_MSG(from != nullptr, "Yielding fiber is null!"); - ASSERT_MSG(to != nullptr, "Next fiber is null!"); - to->impl->guard.lock(); - to->impl->previous_fiber = from; - auto transfer = boost::context::detail::jump_fcontext(to->impl->context, to.get()); - ASSERT(from->impl->previous_fiber != nullptr); - from->impl->previous_fiber->impl->context = transfer.fctx; - from->impl->previous_fiber->impl->guard.unlock(); - from->impl->previous_fiber.reset(); +void Fiber::YieldTo(std::weak_ptr<Fiber> weak_from, Fiber& to) { + to.impl->guard.lock(); + to.impl->previous_fiber = weak_from.lock(); + + auto transfer = boost::context::detail::jump_fcontext(to.impl->context, &to); + + // "from" might no longer be valid if the thread was killed + if (auto from = weak_from.lock()) { + ASSERT(from->impl->previous_fiber != nullptr); + from->impl->previous_fiber->impl->context = transfer.fctx; + from->impl->previous_fiber->impl->guard.unlock(); + from->impl->previous_fiber.reset(); + } } std::shared_ptr<Fiber> Fiber::ThreadToFiber() { diff --git a/src/common/fiber.h b/src/common/fiber.h index f7f587f8c..f2a8ff29a 100644 --- a/src/common/fiber.h +++ b/src/common/fiber.h @@ -41,7 +41,7 @@ public: /// Yields control from Fiber 'from' to Fiber 'to' /// Fiber 'from' must be the currently running fiber. - static void YieldTo(std::shared_ptr<Fiber> from, std::shared_ptr<Fiber> to); + static void YieldTo(std::weak_ptr<Fiber> weak_from, Fiber& to); [[nodiscard]] static std::shared_ptr<Fiber> ThreadToFiber(); void SetRewindPoint(std::function<void(void*)>&& rewind_func, void* rewind_param); diff --git a/src/common/intrusive_red_black_tree.h b/src/common/intrusive_red_black_tree.h new file mode 100644 index 000000000..c0bbcd457 --- /dev/null +++ b/src/common/intrusive_red_black_tree.h @@ -0,0 +1,602 @@ +// Copyright 2021 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include "common/parent_of_member.h" +#include "common/tree.h" + +namespace Common { + +namespace impl { + +class IntrusiveRedBlackTreeImpl; + +} + +struct IntrusiveRedBlackTreeNode { +public: + using EntryType = RBEntry<IntrusiveRedBlackTreeNode>; + + constexpr IntrusiveRedBlackTreeNode() = default; + + void SetEntry(const EntryType& new_entry) { + entry = new_entry; + } + + [[nodiscard]] EntryType& GetEntry() { + return entry; + } + + [[nodiscard]] const EntryType& GetEntry() const { + return entry; + } + +private: + EntryType entry{}; + + friend class impl::IntrusiveRedBlackTreeImpl; + + template <class, class, class> + friend class IntrusiveRedBlackTree; +}; + +template <class T, class Traits, class Comparator> +class IntrusiveRedBlackTree; + +namespace impl { + +class IntrusiveRedBlackTreeImpl { +private: + template <class, class, class> + friend class ::Common::IntrusiveRedBlackTree; + + using RootType = RBHead<IntrusiveRedBlackTreeNode>; + RootType root; + +public: + template <bool Const> + class Iterator; + + using value_type = IntrusiveRedBlackTreeNode; + using size_type = size_t; + using difference_type = ptrdiff_t; + using pointer = value_type*; + using const_pointer = const value_type*; + using reference = value_type&; + using const_reference = const value_type&; + using iterator = Iterator<false>; + using const_iterator = Iterator<true>; + + template <bool Const> + class Iterator { + public: + using iterator_category = std::bidirectional_iterator_tag; + using value_type = typename IntrusiveRedBlackTreeImpl::value_type; + using difference_type = typename IntrusiveRedBlackTreeImpl::difference_type; + using pointer = std::conditional_t<Const, IntrusiveRedBlackTreeImpl::const_pointer, + IntrusiveRedBlackTreeImpl::pointer>; + using reference = std::conditional_t<Const, IntrusiveRedBlackTreeImpl::const_reference, + IntrusiveRedBlackTreeImpl::reference>; + + private: + pointer node; + + public: + explicit Iterator(pointer n) : node(n) {} + + bool operator==(const Iterator& rhs) const { + return this->node == rhs.node; + } + + bool operator!=(const Iterator& rhs) const { + return !(*this == rhs); + } + + pointer operator->() const { + return this->node; + } + + reference operator*() const { + return *this->node; + } + + Iterator& operator++() { + this->node = GetNext(this->node); + return *this; + } + + Iterator& operator--() { + this->node = GetPrev(this->node); + return *this; + } + + Iterator operator++(int) { + const Iterator it{*this}; + ++(*this); + return it; + } + + Iterator operator--(int) { + const Iterator it{*this}; + --(*this); + return it; + } + + operator Iterator<true>() const { + return Iterator<true>(this->node); + } + }; + +private: + // Define accessors using RB_* functions. + bool EmptyImpl() const { + return root.IsEmpty(); + } + + IntrusiveRedBlackTreeNode* GetMinImpl() const { + return RB_MIN(const_cast<RootType*>(&root)); + } + + IntrusiveRedBlackTreeNode* GetMaxImpl() const { + return RB_MAX(const_cast<RootType*>(&root)); + } + + IntrusiveRedBlackTreeNode* RemoveImpl(IntrusiveRedBlackTreeNode* node) { + return RB_REMOVE(&root, node); + } + +public: + static IntrusiveRedBlackTreeNode* GetNext(IntrusiveRedBlackTreeNode* node) { + return RB_NEXT(node); + } + + static IntrusiveRedBlackTreeNode* GetPrev(IntrusiveRedBlackTreeNode* node) { + return RB_PREV(node); + } + + static const IntrusiveRedBlackTreeNode* GetNext(const IntrusiveRedBlackTreeNode* node) { + return static_cast<const IntrusiveRedBlackTreeNode*>( + GetNext(const_cast<IntrusiveRedBlackTreeNode*>(node))); + } + + static const IntrusiveRedBlackTreeNode* GetPrev(const IntrusiveRedBlackTreeNode* node) { + return static_cast<const IntrusiveRedBlackTreeNode*>( + GetPrev(const_cast<IntrusiveRedBlackTreeNode*>(node))); + } + +public: + constexpr IntrusiveRedBlackTreeImpl() {} + + // Iterator accessors. + iterator begin() { + return iterator(this->GetMinImpl()); + } + + const_iterator begin() const { + return const_iterator(this->GetMinImpl()); + } + + iterator end() { + return iterator(static_cast<IntrusiveRedBlackTreeNode*>(nullptr)); + } + + const_iterator end() const { + return const_iterator(static_cast<const IntrusiveRedBlackTreeNode*>(nullptr)); + } + + const_iterator cbegin() const { + return this->begin(); + } + + const_iterator cend() const { + return this->end(); + } + + iterator iterator_to(reference ref) { + return iterator(&ref); + } + + const_iterator iterator_to(const_reference ref) const { + return const_iterator(&ref); + } + + // Content management. + bool empty() const { + return this->EmptyImpl(); + } + + reference back() { + return *this->GetMaxImpl(); + } + + const_reference back() const { + return *this->GetMaxImpl(); + } + + reference front() { + return *this->GetMinImpl(); + } + + const_reference front() const { + return *this->GetMinImpl(); + } + + iterator erase(iterator it) { + auto cur = std::addressof(*it); + auto next = GetNext(cur); + this->RemoveImpl(cur); + return iterator(next); + } +}; + +} // namespace impl + +template <typename T> +concept HasLightCompareType = requires { + { std::is_same<typename T::LightCompareType, void>::value } + ->std::convertible_to<bool>; +}; + +namespace impl { + +template <typename T, typename Default> +consteval auto* GetLightCompareType() { + if constexpr (HasLightCompareType<T>) { + return static_cast<typename T::LightCompareType*>(nullptr); + } else { + return static_cast<Default*>(nullptr); + } +} + +} // namespace impl + +template <typename T, typename Default> +using LightCompareType = std::remove_pointer_t<decltype(impl::GetLightCompareType<T, Default>())>; + +template <class T, class Traits, class Comparator> +class IntrusiveRedBlackTree { + +public: + using ImplType = impl::IntrusiveRedBlackTreeImpl; + +private: + ImplType impl{}; + +public: + template <bool Const> + class Iterator; + + using value_type = T; + using size_type = size_t; + using difference_type = ptrdiff_t; + using pointer = T*; + using const_pointer = const T*; + using reference = T&; + using const_reference = const T&; + using iterator = Iterator<false>; + using const_iterator = Iterator<true>; + + using light_value_type = LightCompareType<Comparator, value_type>; + using const_light_pointer = const light_value_type*; + using const_light_reference = const light_value_type&; + + template <bool Const> + class Iterator { + public: + friend class IntrusiveRedBlackTree<T, Traits, Comparator>; + + using ImplIterator = + std::conditional_t<Const, ImplType::const_iterator, ImplType::iterator>; + + using iterator_category = std::bidirectional_iterator_tag; + using value_type = typename IntrusiveRedBlackTree::value_type; + using difference_type = typename IntrusiveRedBlackTree::difference_type; + using pointer = std::conditional_t<Const, IntrusiveRedBlackTree::const_pointer, + IntrusiveRedBlackTree::pointer>; + using reference = std::conditional_t<Const, IntrusiveRedBlackTree::const_reference, + IntrusiveRedBlackTree::reference>; + + private: + ImplIterator iterator; + + private: + explicit Iterator(ImplIterator it) : iterator(it) {} + + explicit Iterator(typename std::conditional<Const, ImplType::const_iterator, + ImplType::iterator>::type::pointer ptr) + : iterator(ptr) {} + + ImplIterator GetImplIterator() const { + return this->iterator; + } + + public: + bool operator==(const Iterator& rhs) const { + return this->iterator == rhs.iterator; + } + + bool operator!=(const Iterator& rhs) const { + return !(*this == rhs); + } + + pointer operator->() const { + return Traits::GetParent(std::addressof(*this->iterator)); + } + + reference operator*() const { + return *Traits::GetParent(std::addressof(*this->iterator)); + } + + Iterator& operator++() { + ++this->iterator; + return *this; + } + + Iterator& operator--() { + --this->iterator; + return *this; + } + + Iterator operator++(int) { + const Iterator it{*this}; + ++this->iterator; + return it; + } + + Iterator operator--(int) { + const Iterator it{*this}; + --this->iterator; + return it; + } + + operator Iterator<true>() const { + return Iterator<true>(this->iterator); + } + }; + +private: + static int CompareImpl(const IntrusiveRedBlackTreeNode* lhs, + const IntrusiveRedBlackTreeNode* rhs) { + return Comparator::Compare(*Traits::GetParent(lhs), *Traits::GetParent(rhs)); + } + + static int LightCompareImpl(const void* elm, const IntrusiveRedBlackTreeNode* rhs) { + return Comparator::Compare(*static_cast<const_light_pointer>(elm), *Traits::GetParent(rhs)); + } + + // Define accessors using RB_* functions. + IntrusiveRedBlackTreeNode* InsertImpl(IntrusiveRedBlackTreeNode* node) { + return RB_INSERT(&impl.root, node, CompareImpl); + } + + IntrusiveRedBlackTreeNode* FindImpl(const IntrusiveRedBlackTreeNode* node) const { + return RB_FIND(const_cast<ImplType::RootType*>(&impl.root), + const_cast<IntrusiveRedBlackTreeNode*>(node), CompareImpl); + } + + IntrusiveRedBlackTreeNode* NFindImpl(const IntrusiveRedBlackTreeNode* node) const { + return RB_NFIND(const_cast<ImplType::RootType*>(&impl.root), + const_cast<IntrusiveRedBlackTreeNode*>(node), CompareImpl); + } + + IntrusiveRedBlackTreeNode* FindLightImpl(const_light_pointer lelm) const { + return RB_FIND_LIGHT(const_cast<ImplType::RootType*>(&impl.root), + static_cast<const void*>(lelm), LightCompareImpl); + } + + IntrusiveRedBlackTreeNode* NFindLightImpl(const_light_pointer lelm) const { + return RB_NFIND_LIGHT(const_cast<ImplType::RootType*>(&impl.root), + static_cast<const void*>(lelm), LightCompareImpl); + } + +public: + constexpr IntrusiveRedBlackTree() = default; + + // Iterator accessors. + iterator begin() { + return iterator(this->impl.begin()); + } + + const_iterator begin() const { + return const_iterator(this->impl.begin()); + } + + iterator end() { + return iterator(this->impl.end()); + } + + const_iterator end() const { + return const_iterator(this->impl.end()); + } + + const_iterator cbegin() const { + return this->begin(); + } + + const_iterator cend() const { + return this->end(); + } + + iterator iterator_to(reference ref) { + return iterator(this->impl.iterator_to(*Traits::GetNode(std::addressof(ref)))); + } + + const_iterator iterator_to(const_reference ref) const { + return const_iterator(this->impl.iterator_to(*Traits::GetNode(std::addressof(ref)))); + } + + // Content management. + bool empty() const { + return this->impl.empty(); + } + + reference back() { + return *Traits::GetParent(std::addressof(this->impl.back())); + } + + const_reference back() const { + return *Traits::GetParent(std::addressof(this->impl.back())); + } + + reference front() { + return *Traits::GetParent(std::addressof(this->impl.front())); + } + + const_reference front() const { + return *Traits::GetParent(std::addressof(this->impl.front())); + } + + iterator erase(iterator it) { + return iterator(this->impl.erase(it.GetImplIterator())); + } + + iterator insert(reference ref) { + ImplType::pointer node = Traits::GetNode(std::addressof(ref)); + this->InsertImpl(node); + return iterator(node); + } + + iterator find(const_reference ref) const { + return iterator(this->FindImpl(Traits::GetNode(std::addressof(ref)))); + } + + iterator nfind(const_reference ref) const { + return iterator(this->NFindImpl(Traits::GetNode(std::addressof(ref)))); + } + + iterator find_light(const_light_reference ref) const { + return iterator(this->FindLightImpl(std::addressof(ref))); + } + + iterator nfind_light(const_light_reference ref) const { + return iterator(this->NFindLightImpl(std::addressof(ref))); + } +}; + +template <auto T, class Derived = impl::GetParentType<T>> +class IntrusiveRedBlackTreeMemberTraits; + +template <class Parent, IntrusiveRedBlackTreeNode Parent::*Member, class Derived> +class IntrusiveRedBlackTreeMemberTraits<Member, Derived> { +public: + template <class Comparator> + using TreeType = IntrusiveRedBlackTree<Derived, IntrusiveRedBlackTreeMemberTraits, Comparator>; + using TreeTypeImpl = impl::IntrusiveRedBlackTreeImpl; + +private: + template <class, class, class> + friend class IntrusiveRedBlackTree; + + friend class impl::IntrusiveRedBlackTreeImpl; + + static constexpr IntrusiveRedBlackTreeNode* GetNode(Derived* parent) { + return std::addressof(parent->*Member); + } + + static constexpr IntrusiveRedBlackTreeNode const* GetNode(Derived const* parent) { + return std::addressof(parent->*Member); + } + + static constexpr Derived* GetParent(IntrusiveRedBlackTreeNode* node) { + return GetParentPointer<Member, Derived>(node); + } + + static constexpr Derived const* GetParent(const IntrusiveRedBlackTreeNode* node) { + return GetParentPointer<Member, Derived>(node); + } + +private: + static constexpr TypedStorage<Derived> DerivedStorage = {}; + static_assert(GetParent(GetNode(GetPointer(DerivedStorage))) == GetPointer(DerivedStorage)); +}; + +template <auto T, class Derived = impl::GetParentType<T>> +class IntrusiveRedBlackTreeMemberTraitsDeferredAssert; + +template <class Parent, IntrusiveRedBlackTreeNode Parent::*Member, class Derived> +class IntrusiveRedBlackTreeMemberTraitsDeferredAssert<Member, Derived> { +public: + template <class Comparator> + using TreeType = + IntrusiveRedBlackTree<Derived, IntrusiveRedBlackTreeMemberTraitsDeferredAssert, Comparator>; + using TreeTypeImpl = impl::IntrusiveRedBlackTreeImpl; + + static constexpr bool IsValid() { + TypedStorage<Derived> DerivedStorage = {}; + return GetParent(GetNode(GetPointer(DerivedStorage))) == GetPointer(DerivedStorage); + } + +private: + template <class, class, class> + friend class IntrusiveRedBlackTree; + + friend class impl::IntrusiveRedBlackTreeImpl; + + static constexpr IntrusiveRedBlackTreeNode* GetNode(Derived* parent) { + return std::addressof(parent->*Member); + } + + static constexpr IntrusiveRedBlackTreeNode const* GetNode(Derived const* parent) { + return std::addressof(parent->*Member); + } + + static constexpr Derived* GetParent(IntrusiveRedBlackTreeNode* node) { + return GetParentPointer<Member, Derived>(node); + } + + static constexpr Derived const* GetParent(const IntrusiveRedBlackTreeNode* node) { + return GetParentPointer<Member, Derived>(node); + } +}; + +template <class Derived> +class IntrusiveRedBlackTreeBaseNode : public IntrusiveRedBlackTreeNode { +public: + constexpr Derived* GetPrev() { + return static_cast<Derived*>(impl::IntrusiveRedBlackTreeImpl::GetPrev(this)); + } + constexpr const Derived* GetPrev() const { + return static_cast<const Derived*>(impl::IntrusiveRedBlackTreeImpl::GetPrev(this)); + } + + constexpr Derived* GetNext() { + return static_cast<Derived*>(impl::IntrusiveRedBlackTreeImpl::GetNext(this)); + } + constexpr const Derived* GetNext() const { + return static_cast<const Derived*>(impl::IntrusiveRedBlackTreeImpl::GetNext(this)); + } +}; + +template <class Derived> +class IntrusiveRedBlackTreeBaseTraits { +public: + template <class Comparator> + using TreeType = IntrusiveRedBlackTree<Derived, IntrusiveRedBlackTreeBaseTraits, Comparator>; + using TreeTypeImpl = impl::IntrusiveRedBlackTreeImpl; + +private: + template <class, class, class> + friend class IntrusiveRedBlackTree; + + friend class impl::IntrusiveRedBlackTreeImpl; + + static constexpr IntrusiveRedBlackTreeNode* GetNode(Derived* parent) { + return static_cast<IntrusiveRedBlackTreeNode*>(parent); + } + + static constexpr IntrusiveRedBlackTreeNode const* GetNode(Derived const* parent) { + return static_cast<const IntrusiveRedBlackTreeNode*>(parent); + } + + static constexpr Derived* GetParent(IntrusiveRedBlackTreeNode* node) { + return static_cast<Derived*>(node); + } + + static constexpr Derived const* GetParent(const IntrusiveRedBlackTreeNode* node) { + return static_cast<const Derived*>(node); + } +}; + +} // namespace Common diff --git a/src/common/logging/backend.cpp b/src/common/logging/backend.cpp index 631f64d05..2d4d2e9e7 100644 --- a/src/common/logging/backend.cpp +++ b/src/common/logging/backend.cpp @@ -145,10 +145,18 @@ void ColorConsoleBackend::Write(const Entry& entry) { PrintColoredMessage(entry); } -// _SH_DENYWR allows read only access to the file for other programs. -// It is #defined to 0 on other platforms -FileBackend::FileBackend(const std::string& filename) - : file(filename, "w", _SH_DENYWR), bytes_written(0) {} +FileBackend::FileBackend(const std::string& filename) : bytes_written(0) { + if (Common::FS::Exists(filename + ".old.txt")) { + Common::FS::Delete(filename + ".old.txt"); + } + if (Common::FS::Exists(filename)) { + Common::FS::Rename(filename, filename + ".old.txt"); + } + + // _SH_DENYWR allows read only access to the file for other programs. + // It is #defined to 0 on other platforms + file = Common::FS::IOFile(filename, "w", _SH_DENYWR); +} void FileBackend::Write(const Entry& entry) { // prevent logs from going over the maximum size (in case its spamming and the user doesn't diff --git a/src/common/misc.cpp b/src/common/misc.cpp index 1d5393597..495385b9e 100644 --- a/src/common/misc.cpp +++ b/src/common/misc.cpp @@ -12,27 +12,41 @@ #include "common/common_funcs.h" -// Generic function to get last error message. -// Call directly after the command or use the error num. -// This function might change the error code. -std::string GetLastErrorMsg() { - static constexpr std::size_t buff_size = 255; - char err_str[buff_size]; - +std::string NativeErrorToString(int e) { #ifdef _WIN32 - FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM, nullptr, GetLastError(), - MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), err_str, buff_size, nullptr); - return std::string(err_str, buff_size); -#elif defined(__GLIBC__) && (_GNU_SOURCE || (_POSIX_C_SOURCE < 200112L && _XOPEN_SOURCE < 600)) + LPSTR err_str; + + DWORD res = FormatMessageA(FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_ALLOCATE_BUFFER | + FORMAT_MESSAGE_IGNORE_INSERTS, + nullptr, e, MAKELANGID(LANG_NEUTRAL, SUBLANG_DEFAULT), + reinterpret_cast<LPSTR>(&err_str), 1, nullptr); + if (!res) { + return "(FormatMessageA failed to format error)"; + } + std::string ret(err_str); + LocalFree(err_str); + return ret; +#else + char err_str[255]; +#if defined(__GLIBC__) && (_GNU_SOURCE || (_POSIX_C_SOURCE < 200112L && _XOPEN_SOURCE < 600)) // Thread safe (GNU-specific) - const char* str = strerror_r(errno, err_str, buff_size); + const char* str = strerror_r(e, err_str, sizeof(err_str)); return std::string(str); #else // Thread safe (XSI-compliant) - const int success = strerror_r(errno, err_str, buff_size); - if (success != 0) { - return {}; + int second_err = strerror_r(e, err_str, sizeof(err_str)); + if (second_err != 0) { + return "(strerror_r failed to format error)"; } return std::string(err_str); +#endif // GLIBC etc. +#endif // _WIN32 +} + +std::string GetLastErrorMsg() { +#ifdef _WIN32 + return NativeErrorToString(GetLastError()); +#else + return NativeErrorToString(errno); #endif } diff --git a/src/common/nvidia_flags.cpp b/src/common/nvidia_flags.cpp new file mode 100644 index 000000000..d537517db --- /dev/null +++ b/src/common/nvidia_flags.cpp @@ -0,0 +1,27 @@ +// Copyright 2021 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#include <filesystem> +#include <stdlib.h> + +#include <fmt/format.h> + +#include "common/file_util.h" +#include "common/nvidia_flags.h" + +namespace Common { + +void ConfigureNvidiaEnvironmentFlags() { +#ifdef _WIN32 + const std::string shader_path = Common::FS::SanitizePath( + fmt::format("{}/nvidia/", Common::FS::GetUserPath(Common::FS::UserPath::ShaderDir))); + const std::string windows_path = + Common::FS::SanitizePath(shader_path, Common::FS::DirectorySeparator::BackwardSlash); + void(Common::FS::CreateFullPath(shader_path + '/')); + void(_putenv(fmt::format("__GL_SHADER_DISK_CACHE_PATH={}", windows_path).c_str())); + void(_putenv("__GL_SHADER_DISK_CACHE_SKIP_CLEANUP=1")); +#endif +} + +} // namespace Common diff --git a/src/common/nvidia_flags.h b/src/common/nvidia_flags.h new file mode 100644 index 000000000..75a0233ac --- /dev/null +++ b/src/common/nvidia_flags.h @@ -0,0 +1,10 @@ +// Copyright 2021 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +namespace Common { + +/// Configure platform specific flags for Nvidia's driver +void ConfigureNvidiaEnvironmentFlags(); + +} // namespace Common diff --git a/src/common/parent_of_member.h b/src/common/parent_of_member.h new file mode 100644 index 000000000..d9a14529d --- /dev/null +++ b/src/common/parent_of_member.h @@ -0,0 +1,191 @@ +// Copyright 2021 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <type_traits> + +#include "common/assert.h" +#include "common/common_types.h" + +namespace Common { +namespace detail { +template <typename T, size_t Size, size_t Align> +struct TypedStorageImpl { + std::aligned_storage_t<Size, Align> storage_; +}; +} // namespace detail + +template <typename T> +using TypedStorage = detail::TypedStorageImpl<T, sizeof(T), alignof(T)>; + +template <typename T> +static constexpr T* GetPointer(TypedStorage<T>& ts) { + return static_cast<T*>(static_cast<void*>(std::addressof(ts.storage_))); +} + +template <typename T> +static constexpr const T* GetPointer(const TypedStorage<T>& ts) { + return static_cast<const T*>(static_cast<const void*>(std::addressof(ts.storage_))); +} + +namespace impl { + +template <size_t MaxDepth> +struct OffsetOfUnionHolder { + template <typename ParentType, typename MemberType, size_t Offset> + union UnionImpl { + using PaddingMember = char; + static constexpr size_t GetOffset() { + return Offset; + } + +#pragma pack(push, 1) + struct { + PaddingMember padding[Offset]; + MemberType members[(sizeof(ParentType) / sizeof(MemberType)) + 1]; + } data; +#pragma pack(pop) + UnionImpl<ParentType, MemberType, Offset + 1> next_union; + }; + + template <typename ParentType, typename MemberType> + union UnionImpl<ParentType, MemberType, 0> { + static constexpr size_t GetOffset() { + return 0; + } + + struct { + MemberType members[(sizeof(ParentType) / sizeof(MemberType)) + 1]; + } data; + UnionImpl<ParentType, MemberType, 1> next_union; + }; + + template <typename ParentType, typename MemberType> + union UnionImpl<ParentType, MemberType, MaxDepth> {}; +}; + +template <typename ParentType, typename MemberType> +struct OffsetOfCalculator { + using UnionHolder = + typename OffsetOfUnionHolder<sizeof(MemberType)>::template UnionImpl<ParentType, MemberType, + 0>; + union Union { + char c{}; + UnionHolder first_union; + TypedStorage<ParentType> parent; + + constexpr Union() : c() {} + }; + static constexpr Union U = {}; + + static constexpr const MemberType* GetNextAddress(const MemberType* start, + const MemberType* target) { + while (start < target) { + start++; + } + return start; + } + + static constexpr std::ptrdiff_t GetDifference(const MemberType* start, + const MemberType* target) { + return (target - start) * sizeof(MemberType); + } + + template <typename CurUnion> + static constexpr std::ptrdiff_t OffsetOfImpl(MemberType ParentType::*member, + CurUnion& cur_union) { + constexpr size_t Offset = CurUnion::GetOffset(); + const auto target = std::addressof(GetPointer(U.parent)->*member); + const auto start = std::addressof(cur_union.data.members[0]); + const auto next = GetNextAddress(start, target); + + if (next != target) { + if constexpr (Offset < sizeof(MemberType) - 1) { + return OffsetOfImpl(member, cur_union.next_union); + } else { + UNREACHABLE(); + } + } + + return (next - start) * sizeof(MemberType) + Offset; + } + + static constexpr std::ptrdiff_t OffsetOf(MemberType ParentType::*member) { + return OffsetOfImpl(member, U.first_union); + } +}; + +template <typename T> +struct GetMemberPointerTraits; + +template <typename P, typename M> +struct GetMemberPointerTraits<M P::*> { + using Parent = P; + using Member = M; +}; + +template <auto MemberPtr> +using GetParentType = typename GetMemberPointerTraits<decltype(MemberPtr)>::Parent; + +template <auto MemberPtr> +using GetMemberType = typename GetMemberPointerTraits<decltype(MemberPtr)>::Member; + +template <auto MemberPtr, typename RealParentType = GetParentType<MemberPtr>> +static inline std::ptrdiff_t OffsetOf = [] { + using DeducedParentType = GetParentType<MemberPtr>; + using MemberType = GetMemberType<MemberPtr>; + static_assert(std::is_base_of<DeducedParentType, RealParentType>::value || + std::is_same<RealParentType, DeducedParentType>::value); + + return OffsetOfCalculator<RealParentType, MemberType>::OffsetOf(MemberPtr); +}(); + +} // namespace impl + +template <auto MemberPtr, typename RealParentType = impl::GetParentType<MemberPtr>> +constexpr RealParentType& GetParentReference(impl::GetMemberType<MemberPtr>* member) { + std::ptrdiff_t Offset = impl::OffsetOf<MemberPtr, RealParentType>; + return *static_cast<RealParentType*>( + static_cast<void*>(static_cast<uint8_t*>(static_cast<void*>(member)) - Offset)); +} + +template <auto MemberPtr, typename RealParentType = impl::GetParentType<MemberPtr>> +constexpr RealParentType const& GetParentReference(impl::GetMemberType<MemberPtr> const* member) { + std::ptrdiff_t Offset = impl::OffsetOf<MemberPtr, RealParentType>; + return *static_cast<const RealParentType*>(static_cast<const void*>( + static_cast<const uint8_t*>(static_cast<const void*>(member)) - Offset)); +} + +template <auto MemberPtr, typename RealParentType = impl::GetParentType<MemberPtr>> +constexpr RealParentType* GetParentPointer(impl::GetMemberType<MemberPtr>* member) { + return std::addressof(GetParentReference<MemberPtr, RealParentType>(member)); +} + +template <auto MemberPtr, typename RealParentType = impl::GetParentType<MemberPtr>> +constexpr RealParentType const* GetParentPointer(impl::GetMemberType<MemberPtr> const* member) { + return std::addressof(GetParentReference<MemberPtr, RealParentType>(member)); +} + +template <auto MemberPtr, typename RealParentType = impl::GetParentType<MemberPtr>> +constexpr RealParentType& GetParentReference(impl::GetMemberType<MemberPtr>& member) { + return GetParentReference<MemberPtr, RealParentType>(std::addressof(member)); +} + +template <auto MemberPtr, typename RealParentType = impl::GetParentType<MemberPtr>> +constexpr RealParentType const& GetParentReference(impl::GetMemberType<MemberPtr> const& member) { + return GetParentReference<MemberPtr, RealParentType>(std::addressof(member)); +} + +template <auto MemberPtr, typename RealParentType = impl::GetParentType<MemberPtr>> +constexpr RealParentType* GetParentPointer(impl::GetMemberType<MemberPtr>& member) { + return std::addressof(GetParentReference<MemberPtr, RealParentType>(member)); +} + +template <auto MemberPtr, typename RealParentType = impl::GetParentType<MemberPtr>> +constexpr RealParentType const* GetParentPointer(impl::GetMemberType<MemberPtr> const& member) { + return std::addressof(GetParentReference<MemberPtr, RealParentType>(member)); +} + +} // namespace Common diff --git a/src/common/ring_buffer.h b/src/common/ring_buffer.h index 138fa0131..4a8d09806 100644 --- a/src/common/ring_buffer.h +++ b/src/common/ring_buffer.h @@ -19,15 +19,14 @@ namespace Common { /// SPSC ring buffer /// @tparam T Element type /// @tparam capacity Number of slots in ring buffer -/// @tparam granularity Slot size in terms of number of elements -template <typename T, std::size_t capacity, std::size_t granularity = 1> +template <typename T, std::size_t capacity> class RingBuffer { - /// A "slot" is made of `granularity` elements of `T`. - static constexpr std::size_t slot_size = granularity * sizeof(T); + /// A "slot" is made of a single `T`. + static constexpr std::size_t slot_size = sizeof(T); // T must be safely memcpy-able and have a trivial default constructor. static_assert(std::is_trivial_v<T>); // Ensure capacity is sensible. - static_assert(capacity < std::numeric_limits<std::size_t>::max() / 2 / granularity); + static_assert(capacity < std::numeric_limits<std::size_t>::max() / 2); static_assert((capacity & (capacity - 1)) == 0, "capacity must be a power of two"); // Ensure lock-free. static_assert(std::atomic_size_t::is_always_lock_free); @@ -47,7 +46,7 @@ public: const std::size_t second_copy = push_count - first_copy; const char* in = static_cast<const char*>(new_slots); - std::memcpy(m_data.data() + pos * granularity, in, first_copy * slot_size); + std::memcpy(m_data.data() + pos, in, first_copy * slot_size); in += first_copy * slot_size; std::memcpy(m_data.data(), in, second_copy * slot_size); @@ -74,7 +73,7 @@ public: const std::size_t second_copy = pop_count - first_copy; char* out = static_cast<char*>(output); - std::memcpy(out, m_data.data() + pos * granularity, first_copy * slot_size); + std::memcpy(out, m_data.data() + pos, first_copy * slot_size); out += first_copy * slot_size; std::memcpy(out, m_data.data(), second_copy * slot_size); @@ -84,9 +83,9 @@ public: } std::vector<T> Pop(std::size_t max_slots = ~std::size_t(0)) { - std::vector<T> out(std::min(max_slots, capacity) * granularity); - const std::size_t count = Pop(out.data(), out.size() / granularity); - out.resize(count * granularity); + std::vector<T> out(std::min(max_slots, capacity)); + const std::size_t count = Pop(out.data(), out.size()); + out.resize(count); return out; } @@ -113,7 +112,7 @@ private: alignas(128) std::atomic_size_t m_write_index{0}; #endif - std::array<T, granularity * capacity> m_data; + std::array<T, capacity> m_data; }; } // namespace Common diff --git a/src/common/scope_exit.h b/src/common/scope_exit.h index fa46cb394..35dac3a8f 100644 --- a/src/common/scope_exit.h +++ b/src/common/scope_exit.h @@ -49,3 +49,9 @@ ScopeExitHelper<Func> ScopeExit(Func&& func) { * \endcode */ #define SCOPE_EXIT(body) auto CONCAT2(scope_exit_helper_, __LINE__) = detail::ScopeExit([&]() body) + +/** + * This macro is similar to SCOPE_EXIT, except the object is caller managed. This is intended to be + * used when the caller might want to cancel the ScopeExit. + */ +#define SCOPE_GUARD(body) detail::ScopeExit([&]() body) diff --git a/src/common/string_util.cpp b/src/common/string_util.cpp index 4cba2aaa4..7b614ad89 100644 --- a/src/common/string_util.cpp +++ b/src/common/string_util.cpp @@ -141,27 +141,13 @@ std::string ReplaceAll(std::string result, const std::string& src, const std::st } std::string UTF16ToUTF8(const std::u16string& input) { -#ifdef _MSC_VER - // Workaround for missing char16_t/char32_t instantiations in MSVC2017 - std::wstring_convert<std::codecvt_utf8_utf16<__int16>, __int16> convert; - std::basic_string<__int16> tmp_buffer(input.cbegin(), input.cend()); - return convert.to_bytes(tmp_buffer); -#else std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> convert; return convert.to_bytes(input); -#endif } std::u16string UTF8ToUTF16(const std::string& input) { -#ifdef _MSC_VER - // Workaround for missing char16_t/char32_t instantiations in MSVC2017 - std::wstring_convert<std::codecvt_utf8_utf16<__int16>, __int16> convert; - auto tmp_buffer = convert.from_bytes(input); - return std::u16string(tmp_buffer.cbegin(), tmp_buffer.cend()); -#else std::wstring_convert<std::codecvt_utf8_utf16<char16_t>, char16_t> convert; return convert.from_bytes(input); -#endif } #ifdef _WIN32 diff --git a/src/common/timer.cpp b/src/common/timer.cpp deleted file mode 100644 index d17dc2a50..000000000 --- a/src/common/timer.cpp +++ /dev/null @@ -1,159 +0,0 @@ -// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#include <ctime> -#include <fmt/format.h> -#include "common/common_types.h" -#include "common/string_util.h" -#include "common/timer.h" - -namespace Common { - -std::chrono::milliseconds Timer::GetTimeMs() { - return std::chrono::duration_cast<std::chrono::milliseconds>( - std::chrono::system_clock::now().time_since_epoch()); -} - -// -------------------------------------------- -// Initiate, Start, Stop, and Update the time -// -------------------------------------------- - -// Set initial values for the class -Timer::Timer() : m_LastTime(0), m_StartTime(0), m_Running(false) { - Update(); -} - -// Write the starting time -void Timer::Start() { - m_StartTime = GetTimeMs(); - m_Running = true; -} - -// Stop the timer -void Timer::Stop() { - // Write the final time - m_LastTime = GetTimeMs(); - m_Running = false; -} - -// Update the last time variable -void Timer::Update() { - m_LastTime = GetTimeMs(); - // TODO(ector) - QPF -} - -// ------------------------------------- -// Get time difference and elapsed time -// ------------------------------------- - -// Get the number of milliseconds since the last Update() -std::chrono::milliseconds Timer::GetTimeDifference() { - return GetTimeMs() - m_LastTime; -} - -// Add the time difference since the last Update() to the starting time. -// This is used to compensate for a paused game. -void Timer::AddTimeDifference() { - m_StartTime += GetTimeDifference(); -} - -// Get the time elapsed since the Start() -std::chrono::milliseconds Timer::GetTimeElapsed() { - // If we have not started yet, return 1 (because then I don't - // have to change the FPS calculation in CoreRerecording.cpp . - if (m_StartTime.count() == 0) - return std::chrono::milliseconds(1); - - // Return the final timer time if the timer is stopped - if (!m_Running) - return (m_LastTime - m_StartTime); - - return (GetTimeMs() - m_StartTime); -} - -// Get the formatted time elapsed since the Start() -std::string Timer::GetTimeElapsedFormatted() const { - // If we have not started yet, return zero - if (m_StartTime.count() == 0) - return "00:00:00:000"; - - // The number of milliseconds since the start. - // Use a different value if the timer is stopped. - std::chrono::milliseconds Milliseconds; - if (m_Running) - Milliseconds = GetTimeMs() - m_StartTime; - else - Milliseconds = m_LastTime - m_StartTime; - // Seconds - std::chrono::seconds Seconds = std::chrono::duration_cast<std::chrono::seconds>(Milliseconds); - // Minutes - std::chrono::minutes Minutes = std::chrono::duration_cast<std::chrono::minutes>(Milliseconds); - // Hours - std::chrono::hours Hours = std::chrono::duration_cast<std::chrono::hours>(Milliseconds); - - std::string TmpStr = fmt::format("{:02}:{:02}:{:02}:{:03}", Hours.count(), Minutes.count() % 60, - Seconds.count() % 60, Milliseconds.count() % 1000); - return TmpStr; -} - -// Get the number of seconds since January 1 1970 -std::chrono::seconds Timer::GetTimeSinceJan1970() { - return std::chrono::duration_cast<std::chrono::seconds>(GetTimeMs()); -} - -std::chrono::seconds Timer::GetLocalTimeSinceJan1970() { - time_t sysTime, tzDiff, tzDST; - struct tm* gmTime; - - time(&sysTime); - - // Account for DST where needed - gmTime = localtime(&sysTime); - if (gmTime->tm_isdst == 1) - tzDST = 3600; - else - tzDST = 0; - - // Lazy way to get local time in sec - gmTime = gmtime(&sysTime); - tzDiff = sysTime - mktime(gmTime); - - return std::chrono::seconds(sysTime + tzDiff + tzDST); -} - -// Return the current time formatted as Minutes:Seconds:Milliseconds -// in the form 00:00:000. -std::string Timer::GetTimeFormatted() { - time_t sysTime; - struct tm* gmTime; - char tmp[13]; - - time(&sysTime); - gmTime = localtime(&sysTime); - - strftime(tmp, 6, "%M:%S", gmTime); - - u64 milliseconds = static_cast<u64>(GetTimeMs().count()) % 1000; - return fmt::format("{}:{:03}", tmp, milliseconds); -} - -// Returns a timestamp with decimals for precise time comparisons -// ---------------- -double Timer::GetDoubleTime() { - // Get continuous timestamp - auto tmp_seconds = static_cast<u64>(GetTimeSinceJan1970().count()); - const auto ms = static_cast<double>(static_cast<u64>(GetTimeMs().count()) % 1000); - - // Remove a few years. We only really want enough seconds to make - // sure that we are detecting actual actions, perhaps 60 seconds is - // enough really, but I leave a year of seconds anyway, in case the - // user's clock is incorrect or something like that. - tmp_seconds = tmp_seconds - (38 * 365 * 24 * 60 * 60); - - // Make a smaller integer that fits in the double - const auto seconds = static_cast<u32>(tmp_seconds); - return seconds + ms; -} - -} // Namespace Common diff --git a/src/common/timer.h b/src/common/timer.h deleted file mode 100644 index 8894a143d..000000000 --- a/src/common/timer.h +++ /dev/null @@ -1,41 +0,0 @@ -// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#pragma once - -#include <chrono> -#include <string> -#include "common/common_types.h" - -namespace Common { -class Timer { -public: - Timer(); - - void Start(); - void Stop(); - void Update(); - - // The time difference is always returned in milliseconds, regardless of alternative internal - // representation - [[nodiscard]] std::chrono::milliseconds GetTimeDifference(); - void AddTimeDifference(); - - [[nodiscard]] static std::chrono::seconds GetTimeSinceJan1970(); - [[nodiscard]] static std::chrono::seconds GetLocalTimeSinceJan1970(); - [[nodiscard]] static double GetDoubleTime(); - - [[nodiscard]] static std::string GetTimeFormatted(); - [[nodiscard]] std::string GetTimeElapsedFormatted() const; - [[nodiscard]] std::chrono::milliseconds GetTimeElapsed(); - - [[nodiscard]] static std::chrono::milliseconds GetTimeMs(); - -private: - std::chrono::milliseconds m_LastTime; - std::chrono::milliseconds m_StartTime; - bool m_Running; -}; - -} // Namespace Common diff --git a/src/common/tiny_mt.h b/src/common/tiny_mt.h new file mode 100644 index 000000000..19ae5b7d6 --- /dev/null +++ b/src/common/tiny_mt.h @@ -0,0 +1,250 @@ +// Copyright 2021 yuzu Emulator Project +// Licensed under GPLv2 or any later version +// Refer to the license.txt file included. + +#pragma once + +#include <array> + +#include "common/alignment.h" +#include "common/common_types.h" + +namespace Common { + +// Implementation of TinyMT (mersenne twister RNG). +// Like Nintendo, we will use the sample parameters. +class TinyMT { +public: + static constexpr std::size_t NumStateWords = 4; + + struct State { + std::array<u32, NumStateWords> data{}; + }; + +private: + static constexpr u32 ParamMat1 = 0x8F7011EE; + static constexpr u32 ParamMat2 = 0xFC78FF1F; + static constexpr u32 ParamTmat = 0x3793FDFF; + + static constexpr u32 ParamMult = 0x6C078965; + static constexpr u32 ParamPlus = 0x0019660D; + static constexpr u32 ParamXor = 0x5D588B65; + + static constexpr u32 TopBitmask = 0x7FFFFFFF; + + static constexpr int MinimumInitIterations = 8; + static constexpr int NumDiscardedInitOutputs = 8; + + static constexpr u32 XorByShifted27(u32 value) { + return value ^ (value >> 27); + } + + static constexpr u32 XorByShifted30(u32 value) { + return value ^ (value >> 30); + } + +private: + State state{}; + +private: + // Internal API. + void FinalizeInitialization() { + const u32 state0 = this->state.data[0] & TopBitmask; + const u32 state1 = this->state.data[1]; + const u32 state2 = this->state.data[2]; + const u32 state3 = this->state.data[3]; + + if (state0 == 0 && state1 == 0 && state2 == 0 && state3 == 0) { + this->state.data[0] = 'T'; + this->state.data[1] = 'I'; + this->state.data[2] = 'N'; + this->state.data[3] = 'Y'; + } + + for (int i = 0; i < NumDiscardedInitOutputs; i++) { + this->GenerateRandomU32(); + } + } + + u32 GenerateRandomU24() { + return (this->GenerateRandomU32() >> 8); + } + + static void GenerateInitialValuePlus(TinyMT::State* state, int index, u32 value) { + u32& state0 = state->data[(index + 0) % NumStateWords]; + u32& state1 = state->data[(index + 1) % NumStateWords]; + u32& state2 = state->data[(index + 2) % NumStateWords]; + u32& state3 = state->data[(index + 3) % NumStateWords]; + + const u32 x = XorByShifted27(state0 ^ state1 ^ state3) * ParamPlus; + const u32 y = x + index + value; + + state0 = y; + state1 += x; + state2 += y; + } + + static void GenerateInitialValueXor(TinyMT::State* state, int index) { + u32& state0 = state->data[(index + 0) % NumStateWords]; + u32& state1 = state->data[(index + 1) % NumStateWords]; + u32& state2 = state->data[(index + 2) % NumStateWords]; + u32& state3 = state->data[(index + 3) % NumStateWords]; + + const u32 x = XorByShifted27(state0 + state1 + state3) * ParamXor; + const u32 y = x - index; + + state0 = y; + state1 ^= x; + state2 ^= y; + } + +public: + constexpr TinyMT() = default; + + // Public API. + + // Initialization. + void Initialize(u32 seed) { + this->state.data[0] = seed; + this->state.data[1] = ParamMat1; + this->state.data[2] = ParamMat2; + this->state.data[3] = ParamTmat; + + for (int i = 1; i < MinimumInitIterations; i++) { + const u32 mixed = XorByShifted30(this->state.data[(i - 1) % NumStateWords]); + this->state.data[i % NumStateWords] ^= mixed * ParamMult + i; + } + + this->FinalizeInitialization(); + } + + void Initialize(const u32* seed, int seed_count) { + this->state.data[0] = 0; + this->state.data[1] = ParamMat1; + this->state.data[2] = ParamMat2; + this->state.data[3] = ParamTmat; + + { + const int num_init_iterations = std::max(seed_count + 1, MinimumInitIterations) - 1; + + GenerateInitialValuePlus(&this->state, 0, seed_count); + + for (int i = 0; i < num_init_iterations; i++) { + GenerateInitialValuePlus(&this->state, (i + 1) % NumStateWords, + (i < seed_count) ? seed[i] : 0); + } + + for (int i = 0; i < static_cast<int>(NumStateWords); i++) { + GenerateInitialValueXor(&this->state, + (i + 1 + num_init_iterations) % NumStateWords); + } + } + + this->FinalizeInitialization(); + } + + // State management. + void GetState(TinyMT::State& out) const { + out.data = this->state.data; + } + + void SetState(const TinyMT::State& state_) { + this->state.data = state_.data; + } + + // Random generation. + void GenerateRandomBytes(void* dst, std::size_t size) { + const uintptr_t start = reinterpret_cast<uintptr_t>(dst); + const uintptr_t end = start + size; + const uintptr_t aligned_start = Common::AlignUp(start, 4); + const uintptr_t aligned_end = Common::AlignDown(end, 4); + + // Make sure we're aligned. + if (start < aligned_start) { + const u32 rnd = this->GenerateRandomU32(); + std::memcpy(dst, &rnd, aligned_start - start); + } + + // Write as many aligned u32s as we can. + { + u32* cur_dst = reinterpret_cast<u32*>(aligned_start); + u32* const end_dst = reinterpret_cast<u32*>(aligned_end); + + while (cur_dst < end_dst) { + *(cur_dst++) = this->GenerateRandomU32(); + } + } + + // Handle any leftover unaligned data. + if (aligned_end < end) { + const u32 rnd = this->GenerateRandomU32(); + std::memcpy(reinterpret_cast<void*>(aligned_end), &rnd, end - aligned_end); + } + } + + u32 GenerateRandomU32() { + // Advance state. + const u32 x0 = + (this->state.data[0] & TopBitmask) ^ this->state.data[1] ^ this->state.data[2]; + const u32 y0 = this->state.data[3]; + const u32 x1 = x0 ^ (x0 << 1); + const u32 y1 = y0 ^ (y0 >> 1) ^ x1; + + const u32 state0 = this->state.data[1]; + u32 state1 = this->state.data[2]; + u32 state2 = x1 ^ (y1 << 10); + const u32 state3 = y1; + + if ((y1 & 1) != 0) { + state1 ^= ParamMat1; + state2 ^= ParamMat2; + } + + this->state.data[0] = state0; + this->state.data[1] = state1; + this->state.data[2] = state2; + this->state.data[3] = state3; + + // Temper. + const u32 t1 = state0 + (state2 >> 8); + u32 t0 = state3 ^ t1; + + if ((t1 & 1) != 0) { + t0 ^= ParamTmat; + } + + return t0; + } + + u64 GenerateRandomU64() { + const u32 lo = this->GenerateRandomU32(); + const u32 hi = this->GenerateRandomU32(); + return (u64{hi} << 32) | u64{lo}; + } + + float GenerateRandomF32() { + // Floats have 24 bits of mantissa. + constexpr u32 MantissaBits = 24; + return static_cast<float>(GenerateRandomU24()) * (1.0f / (1U << MantissaBits)); + } + + double GenerateRandomF64() { + // Doubles have 53 bits of mantissa. + // The smart way to generate 53 bits of random would be to use 32 bits + // from the first rnd32() call, and then 21 from the second. + // Nintendo does not. They use (32 - 5) = 27 bits from the first rnd32() + // call, and (32 - 6) bits from the second. We'll do what they do, but + // There's not a clear reason why. + constexpr u32 MantissaBits = 53; + constexpr u32 Shift1st = (64 - MantissaBits) / 2; + constexpr u32 Shift2nd = (64 - MantissaBits) - Shift1st; + + const u32 first = (this->GenerateRandomU32() >> Shift1st); + const u32 second = (this->GenerateRandomU32() >> Shift2nd); + + return (1.0 * first * (u64{1} << (32 - Shift2nd)) + second) * + (1.0 / (u64{1} << MantissaBits)); + } +}; + +} // namespace Common diff --git a/src/common/tree.h b/src/common/tree.h new file mode 100644 index 000000000..3da49e422 --- /dev/null +++ b/src/common/tree.h @@ -0,0 +1,674 @@ +/* $NetBSD: tree.h,v 1.8 2004/03/28 19:38:30 provos Exp $ */ +/* $OpenBSD: tree.h,v 1.7 2002/10/17 21:51:54 art Exp $ */ +/* $FreeBSD$ */ + +/*- + * Copyright 2002 Niels Provos <provos@citi.umich.edu> + * All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions + * are met: + * 1. Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * 2. Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in the + * documentation and/or other materials provided with the distribution. + * + * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR + * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES + * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. + * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, + * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT + * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, + * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY + * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT + * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF + * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. + */ + +#pragma once + +/* + * This file defines data structures for red-black trees. + * + * A red-black tree is a binary search tree with the node color as an + * extra attribute. It fulfills a set of conditions: + * - every search path from the root to a leaf consists of the + * same number of black nodes, + * - each red node (except for the root) has a black parent, + * - each leaf node is black. + * + * Every operation on a red-black tree is bounded as O(lg n). + * The maximum height of a red-black tree is 2lg (n+1). + */ + +namespace Common { +template <typename T> +class RBHead { +public: + [[nodiscard]] T* Root() { + return rbh_root; + } + + [[nodiscard]] const T* Root() const { + return rbh_root; + } + + void SetRoot(T* root) { + rbh_root = root; + } + + [[nodiscard]] bool IsEmpty() const { + return Root() == nullptr; + } + +private: + T* rbh_root = nullptr; +}; + +enum class EntryColor { + Black, + Red, +}; + +template <typename T> +class RBEntry { +public: + [[nodiscard]] T* Left() { + return rbe_left; + } + + [[nodiscard]] const T* Left() const { + return rbe_left; + } + + void SetLeft(T* left) { + rbe_left = left; + } + + [[nodiscard]] T* Right() { + return rbe_right; + } + + [[nodiscard]] const T* Right() const { + return rbe_right; + } + + void SetRight(T* right) { + rbe_right = right; + } + + [[nodiscard]] T* Parent() { + return rbe_parent; + } + + [[nodiscard]] const T* Parent() const { + return rbe_parent; + } + + void SetParent(T* parent) { + rbe_parent = parent; + } + + [[nodiscard]] bool IsBlack() const { + return rbe_color == EntryColor::Black; + } + + [[nodiscard]] bool IsRed() const { + return rbe_color == EntryColor::Red; + } + + [[nodiscard]] EntryColor Color() const { + return rbe_color; + } + + void SetColor(EntryColor color) { + rbe_color = color; + } + +private: + T* rbe_left = nullptr; + T* rbe_right = nullptr; + T* rbe_parent = nullptr; + EntryColor rbe_color{}; +}; + +template <typename Node> +[[nodiscard]] RBEntry<Node>& RB_ENTRY(Node* node) { + return node->GetEntry(); +} + +template <typename Node> +[[nodiscard]] const RBEntry<Node>& RB_ENTRY(const Node* node) { + return node->GetEntry(); +} + +template <typename Node> +[[nodiscard]] Node* RB_PARENT(Node* node) { + return RB_ENTRY(node).Parent(); +} + +template <typename Node> +[[nodiscard]] const Node* RB_PARENT(const Node* node) { + return RB_ENTRY(node).Parent(); +} + +template <typename Node> +void RB_SET_PARENT(Node* node, Node* parent) { + return RB_ENTRY(node).SetParent(parent); +} + +template <typename Node> +[[nodiscard]] Node* RB_LEFT(Node* node) { + return RB_ENTRY(node).Left(); +} + +template <typename Node> +[[nodiscard]] const Node* RB_LEFT(const Node* node) { + return RB_ENTRY(node).Left(); +} + +template <typename Node> +void RB_SET_LEFT(Node* node, Node* left) { + return RB_ENTRY(node).SetLeft(left); +} + +template <typename Node> +[[nodiscard]] Node* RB_RIGHT(Node* node) { + return RB_ENTRY(node).Right(); +} + +template <typename Node> +[[nodiscard]] const Node* RB_RIGHT(const Node* node) { + return RB_ENTRY(node).Right(); +} + +template <typename Node> +void RB_SET_RIGHT(Node* node, Node* right) { + return RB_ENTRY(node).SetRight(right); +} + +template <typename Node> +[[nodiscard]] bool RB_IS_BLACK(const Node* node) { + return RB_ENTRY(node).IsBlack(); +} + +template <typename Node> +[[nodiscard]] bool RB_IS_RED(const Node* node) { + return RB_ENTRY(node).IsRed(); +} + +template <typename Node> +[[nodiscard]] EntryColor RB_COLOR(const Node* node) { + return RB_ENTRY(node).Color(); +} + +template <typename Node> +void RB_SET_COLOR(Node* node, EntryColor color) { + return RB_ENTRY(node).SetColor(color); +} + +template <typename Node> +void RB_SET(Node* node, Node* parent) { + auto& entry = RB_ENTRY(node); + entry.SetParent(parent); + entry.SetLeft(nullptr); + entry.SetRight(nullptr); + entry.SetColor(EntryColor::Red); +} + +template <typename Node> +void RB_SET_BLACKRED(Node* black, Node* red) { + RB_SET_COLOR(black, EntryColor::Black); + RB_SET_COLOR(red, EntryColor::Red); +} + +template <typename Node> +void RB_ROTATE_LEFT(RBHead<Node>* head, Node* elm, Node*& tmp) { + tmp = RB_RIGHT(elm); + RB_SET_RIGHT(elm, RB_LEFT(tmp)); + if (RB_RIGHT(elm) != nullptr) { + RB_SET_PARENT(RB_LEFT(tmp), elm); + } + + RB_SET_PARENT(tmp, RB_PARENT(elm)); + if (RB_PARENT(tmp) != nullptr) { + if (elm == RB_LEFT(RB_PARENT(elm))) { + RB_SET_LEFT(RB_PARENT(elm), tmp); + } else { + RB_SET_RIGHT(RB_PARENT(elm), tmp); + } + } else { + head->SetRoot(tmp); + } + + RB_SET_LEFT(tmp, elm); + RB_SET_PARENT(elm, tmp); +} + +template <typename Node> +void RB_ROTATE_RIGHT(RBHead<Node>* head, Node* elm, Node*& tmp) { + tmp = RB_LEFT(elm); + RB_SET_LEFT(elm, RB_RIGHT(tmp)); + if (RB_LEFT(elm) != nullptr) { + RB_SET_PARENT(RB_RIGHT(tmp), elm); + } + + RB_SET_PARENT(tmp, RB_PARENT(elm)); + if (RB_PARENT(tmp) != nullptr) { + if (elm == RB_LEFT(RB_PARENT(elm))) { + RB_SET_LEFT(RB_PARENT(elm), tmp); + } else { + RB_SET_RIGHT(RB_PARENT(elm), tmp); + } + } else { + head->SetRoot(tmp); + } + + RB_SET_RIGHT(tmp, elm); + RB_SET_PARENT(elm, tmp); +} + +template <typename Node> +void RB_INSERT_COLOR(RBHead<Node>* head, Node* elm) { + Node* parent = nullptr; + Node* tmp = nullptr; + + while ((parent = RB_PARENT(elm)) != nullptr && RB_IS_RED(parent)) { + Node* gparent = RB_PARENT(parent); + if (parent == RB_LEFT(gparent)) { + tmp = RB_RIGHT(gparent); + if (tmp && RB_IS_RED(tmp)) { + RB_SET_COLOR(tmp, EntryColor::Black); + RB_SET_BLACKRED(parent, gparent); + elm = gparent; + continue; + } + + if (RB_RIGHT(parent) == elm) { + RB_ROTATE_LEFT(head, parent, tmp); + tmp = parent; + parent = elm; + elm = tmp; + } + + RB_SET_BLACKRED(parent, gparent); + RB_ROTATE_RIGHT(head, gparent, tmp); + } else { + tmp = RB_LEFT(gparent); + if (tmp && RB_IS_RED(tmp)) { + RB_SET_COLOR(tmp, EntryColor::Black); + RB_SET_BLACKRED(parent, gparent); + elm = gparent; + continue; + } + + if (RB_LEFT(parent) == elm) { + RB_ROTATE_RIGHT(head, parent, tmp); + tmp = parent; + parent = elm; + elm = tmp; + } + + RB_SET_BLACKRED(parent, gparent); + RB_ROTATE_LEFT(head, gparent, tmp); + } + } + + RB_SET_COLOR(head->Root(), EntryColor::Black); +} + +template <typename Node> +void RB_REMOVE_COLOR(RBHead<Node>* head, Node* parent, Node* elm) { + Node* tmp; + while ((elm == nullptr || RB_IS_BLACK(elm)) && elm != head->Root()) { + if (RB_LEFT(parent) == elm) { + tmp = RB_RIGHT(parent); + if (RB_IS_RED(tmp)) { + RB_SET_BLACKRED(tmp, parent); + RB_ROTATE_LEFT(head, parent, tmp); + tmp = RB_RIGHT(parent); + } + + if ((RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) && + (RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp)))) { + RB_SET_COLOR(tmp, EntryColor::Red); + elm = parent; + parent = RB_PARENT(elm); + } else { + if (RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp))) { + Node* oleft; + if ((oleft = RB_LEFT(tmp)) != nullptr) { + RB_SET_COLOR(oleft, EntryColor::Black); + } + + RB_SET_COLOR(tmp, EntryColor::Red); + RB_ROTATE_RIGHT(head, tmp, oleft); + tmp = RB_RIGHT(parent); + } + + RB_SET_COLOR(tmp, RB_COLOR(parent)); + RB_SET_COLOR(parent, EntryColor::Black); + if (RB_RIGHT(tmp)) { + RB_SET_COLOR(RB_RIGHT(tmp), EntryColor::Black); + } + + RB_ROTATE_LEFT(head, parent, tmp); + elm = head->Root(); + break; + } + } else { + tmp = RB_LEFT(parent); + if (RB_IS_RED(tmp)) { + RB_SET_BLACKRED(tmp, parent); + RB_ROTATE_RIGHT(head, parent, tmp); + tmp = RB_LEFT(parent); + } + + if ((RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) && + (RB_RIGHT(tmp) == nullptr || RB_IS_BLACK(RB_RIGHT(tmp)))) { + RB_SET_COLOR(tmp, EntryColor::Red); + elm = parent; + parent = RB_PARENT(elm); + } else { + if (RB_LEFT(tmp) == nullptr || RB_IS_BLACK(RB_LEFT(tmp))) { + Node* oright; + if ((oright = RB_RIGHT(tmp)) != nullptr) { + RB_SET_COLOR(oright, EntryColor::Black); + } + + RB_SET_COLOR(tmp, EntryColor::Red); + RB_ROTATE_LEFT(head, tmp, oright); + tmp = RB_LEFT(parent); + } + + RB_SET_COLOR(tmp, RB_COLOR(parent)); + RB_SET_COLOR(parent, EntryColor::Black); + + if (RB_LEFT(tmp)) { + RB_SET_COLOR(RB_LEFT(tmp), EntryColor::Black); + } + + RB_ROTATE_RIGHT(head, parent, tmp); + elm = head->Root(); + break; + } + } + } + + if (elm) { + RB_SET_COLOR(elm, EntryColor::Black); + } +} + +template <typename Node> +Node* RB_REMOVE(RBHead<Node>* head, Node* elm) { + Node* child = nullptr; + Node* parent = nullptr; + Node* old = elm; + EntryColor color{}; + + const auto finalize = [&] { + if (color == EntryColor::Black) { + RB_REMOVE_COLOR(head, parent, child); + } + + return old; + }; + + if (RB_LEFT(elm) == nullptr) { + child = RB_RIGHT(elm); + } else if (RB_RIGHT(elm) == nullptr) { + child = RB_LEFT(elm); + } else { + Node* left; + elm = RB_RIGHT(elm); + while ((left = RB_LEFT(elm)) != nullptr) { + elm = left; + } + + child = RB_RIGHT(elm); + parent = RB_PARENT(elm); + color = RB_COLOR(elm); + + if (child) { + RB_SET_PARENT(child, parent); + } + if (parent) { + if (RB_LEFT(parent) == elm) { + RB_SET_LEFT(parent, child); + } else { + RB_SET_RIGHT(parent, child); + } + } else { + head->SetRoot(child); + } + + if (RB_PARENT(elm) == old) { + parent = elm; + } + + elm->SetEntry(old->GetEntry()); + + if (RB_PARENT(old)) { + if (RB_LEFT(RB_PARENT(old)) == old) { + RB_SET_LEFT(RB_PARENT(old), elm); + } else { + RB_SET_RIGHT(RB_PARENT(old), elm); + } + } else { + head->SetRoot(elm); + } + RB_SET_PARENT(RB_LEFT(old), elm); + if (RB_RIGHT(old)) { + RB_SET_PARENT(RB_RIGHT(old), elm); + } + if (parent) { + left = parent; + } + + return finalize(); + } + + parent = RB_PARENT(elm); + color = RB_COLOR(elm); + + if (child) { + RB_SET_PARENT(child, parent); + } + if (parent) { + if (RB_LEFT(parent) == elm) { + RB_SET_LEFT(parent, child); + } else { + RB_SET_RIGHT(parent, child); + } + } else { + head->SetRoot(child); + } + + return finalize(); +} + +// Inserts a node into the RB tree +template <typename Node, typename CompareFunction> +Node* RB_INSERT(RBHead<Node>* head, Node* elm, CompareFunction cmp) { + Node* parent = nullptr; + Node* tmp = head->Root(); + int comp = 0; + + while (tmp) { + parent = tmp; + comp = cmp(elm, parent); + if (comp < 0) { + tmp = RB_LEFT(tmp); + } else if (comp > 0) { + tmp = RB_RIGHT(tmp); + } else { + return tmp; + } + } + + RB_SET(elm, parent); + + if (parent != nullptr) { + if (comp < 0) { + RB_SET_LEFT(parent, elm); + } else { + RB_SET_RIGHT(parent, elm); + } + } else { + head->SetRoot(elm); + } + + RB_INSERT_COLOR(head, elm); + return nullptr; +} + +// Finds the node with the same key as elm +template <typename Node, typename CompareFunction> +Node* RB_FIND(RBHead<Node>* head, Node* elm, CompareFunction cmp) { + Node* tmp = head->Root(); + + while (tmp) { + const int comp = cmp(elm, tmp); + if (comp < 0) { + tmp = RB_LEFT(tmp); + } else if (comp > 0) { + tmp = RB_RIGHT(tmp); + } else { + return tmp; + } + } + + return nullptr; +} + +// Finds the first node greater than or equal to the search key +template <typename Node, typename CompareFunction> +Node* RB_NFIND(RBHead<Node>* head, Node* elm, CompareFunction cmp) { + Node* tmp = head->Root(); + Node* res = nullptr; + + while (tmp) { + const int comp = cmp(elm, tmp); + if (comp < 0) { + res = tmp; + tmp = RB_LEFT(tmp); + } else if (comp > 0) { + tmp = RB_RIGHT(tmp); + } else { + return tmp; + } + } + + return res; +} + +// Finds the node with the same key as lelm +template <typename Node, typename CompareFunction> +Node* RB_FIND_LIGHT(RBHead<Node>* head, const void* lelm, CompareFunction lcmp) { + Node* tmp = head->Root(); + + while (tmp) { + const int comp = lcmp(lelm, tmp); + if (comp < 0) { + tmp = RB_LEFT(tmp); + } else if (comp > 0) { + tmp = RB_RIGHT(tmp); + } else { + return tmp; + } + } + + return nullptr; +} + +// Finds the first node greater than or equal to the search key +template <typename Node, typename CompareFunction> +Node* RB_NFIND_LIGHT(RBHead<Node>* head, const void* lelm, CompareFunction lcmp) { + Node* tmp = head->Root(); + Node* res = nullptr; + + while (tmp) { + const int comp = lcmp(lelm, tmp); + if (comp < 0) { + res = tmp; + tmp = RB_LEFT(tmp); + } else if (comp > 0) { + tmp = RB_RIGHT(tmp); + } else { + return tmp; + } + } + + return res; +} + +template <typename Node> +Node* RB_NEXT(Node* elm) { + if (RB_RIGHT(elm)) { + elm = RB_RIGHT(elm); + while (RB_LEFT(elm)) { + elm = RB_LEFT(elm); + } + } else { + if (RB_PARENT(elm) && (elm == RB_LEFT(RB_PARENT(elm)))) { + elm = RB_PARENT(elm); + } else { + while (RB_PARENT(elm) && (elm == RB_RIGHT(RB_PARENT(elm)))) { + elm = RB_PARENT(elm); + } + elm = RB_PARENT(elm); + } + } + return elm; +} + +template <typename Node> +Node* RB_PREV(Node* elm) { + if (RB_LEFT(elm)) { + elm = RB_LEFT(elm); + while (RB_RIGHT(elm)) { + elm = RB_RIGHT(elm); + } + } else { + if (RB_PARENT(elm) && (elm == RB_RIGHT(RB_PARENT(elm)))) { + elm = RB_PARENT(elm); + } else { + while (RB_PARENT(elm) && (elm == RB_LEFT(RB_PARENT(elm)))) { + elm = RB_PARENT(elm); + } + elm = RB_PARENT(elm); + } + } + return elm; +} + +template <typename Node> +Node* RB_MINMAX(RBHead<Node>* head, bool is_min) { + Node* tmp = head->Root(); + Node* parent = nullptr; + + while (tmp) { + parent = tmp; + if (is_min) { + tmp = RB_LEFT(tmp); + } else { + tmp = RB_RIGHT(tmp); + } + } + + return parent; +} + +template <typename Node> +Node* RB_MIN(RBHead<Node>* head) { + return RB_MINMAX(head, true); +} + +template <typename Node> +Node* RB_MAX(RBHead<Node>* head) { + return RB_MINMAX(head, false); +} +} // namespace Common diff --git a/src/common/uint128.cpp b/src/common/uint128.cpp deleted file mode 100644 index 16bf7c828..000000000 --- a/src/common/uint128.cpp +++ /dev/null @@ -1,71 +0,0 @@ -// Copyright 2019 yuzu Emulator Project -// Licensed under GPLv2 or any later version -// Refer to the license.txt file included. - -#ifdef _MSC_VER -#include <intrin.h> - -#pragma intrinsic(_umul128) -#pragma intrinsic(_udiv128) -#endif -#include <cstring> -#include "common/uint128.h" - -namespace Common { - -#ifdef _MSC_VER - -u64 MultiplyAndDivide64(u64 a, u64 b, u64 d) { - u128 r{}; - r[0] = _umul128(a, b, &r[1]); - u64 remainder; -#if _MSC_VER < 1923 - return udiv128(r[1], r[0], d, &remainder); -#else - return _udiv128(r[1], r[0], d, &remainder); -#endif -} - -#else - -u64 MultiplyAndDivide64(u64 a, u64 b, u64 d) { - const u64 diva = a / d; - const u64 moda = a % d; - const u64 divb = b / d; - const u64 modb = b % d; - return diva * b + moda * divb + moda * modb / d; -} - -#endif - -u128 Multiply64Into128(u64 a, u64 b) { - u128 result; -#ifdef _MSC_VER - result[0] = _umul128(a, b, &result[1]); -#else - unsigned __int128 tmp = a; - tmp *= b; - std::memcpy(&result, &tmp, sizeof(u128)); -#endif - return result; -} - -std::pair<u64, u64> Divide128On32(u128 dividend, u32 divisor) { - u64 remainder = dividend[0] % divisor; - u64 accum = dividend[0] / divisor; - if (dividend[1] == 0) - return {accum, remainder}; - // We ignore dividend[1] / divisor as that overflows - const u64 first_segment = (dividend[1] % divisor) << 32; - accum += (first_segment / divisor) << 32; - const u64 second_segment = (first_segment % divisor) << 32; - accum += (second_segment / divisor); - remainder += second_segment % divisor; - if (remainder >= divisor) { - accum++; - remainder -= divisor; - } - return {accum, remainder}; -} - -} // namespace Common diff --git a/src/common/uint128.h b/src/common/uint128.h index 969259ab6..4780b2f9d 100644 --- a/src/common/uint128.h +++ b/src/common/uint128.h @@ -4,19 +4,118 @@ #pragma once +#include <cstring> #include <utility> + +#ifdef _MSC_VER +#include <intrin.h> +#pragma intrinsic(__umulh) +#pragma intrinsic(_umul128) +#pragma intrinsic(_udiv128) +#else +#include <x86intrin.h> +#endif + #include "common/common_types.h" namespace Common { // This function multiplies 2 u64 values and divides it by a u64 value. -[[nodiscard]] u64 MultiplyAndDivide64(u64 a, u64 b, u64 d); +[[nodiscard]] static inline u64 MultiplyAndDivide64(u64 a, u64 b, u64 d) { +#ifdef _MSC_VER + u128 r{}; + r[0] = _umul128(a, b, &r[1]); + u64 remainder; +#if _MSC_VER < 1923 + return udiv128(r[1], r[0], d, &remainder); +#else + return _udiv128(r[1], r[0], d, &remainder); +#endif +#else + const u64 diva = a / d; + const u64 moda = a % d; + const u64 divb = b / d; + const u64 modb = b % d; + return diva * b + moda * divb + moda * modb / d; +#endif +} // This function multiplies 2 u64 values and produces a u128 value; -[[nodiscard]] u128 Multiply64Into128(u64 a, u64 b); +[[nodiscard]] static inline u128 Multiply64Into128(u64 a, u64 b) { + u128 result; +#ifdef _MSC_VER + result[0] = _umul128(a, b, &result[1]); +#else + unsigned __int128 tmp = a; + tmp *= b; + std::memcpy(&result, &tmp, sizeof(u128)); +#endif + return result; +} + +[[nodiscard]] static inline u64 GetFixedPoint64Factor(u64 numerator, u64 divisor) { +#ifdef __SIZEOF_INT128__ + const auto base = static_cast<unsigned __int128>(numerator) << 64ULL; + return static_cast<u64>(base / divisor); +#elif defined(_M_X64) || defined(_M_ARM64) + std::array<u64, 2> r = {0, numerator}; + u64 remainder; +#if _MSC_VER < 1923 + return udiv128(r[1], r[0], divisor, &remainder); +#else + return _udiv128(r[1], r[0], divisor, &remainder); +#endif +#else + // This one is bit more inaccurate. + return MultiplyAndDivide64(std::numeric_limits<u64>::max(), numerator, divisor); +#endif +} + +[[nodiscard]] static inline u64 MultiplyHigh(u64 a, u64 b) { +#ifdef __SIZEOF_INT128__ + return (static_cast<unsigned __int128>(a) * static_cast<unsigned __int128>(b)) >> 64; +#elif defined(_M_X64) || defined(_M_ARM64) + return __umulh(a, b); // MSVC +#else + // Generic fallback + const u64 a_lo = u32(a); + const u64 a_hi = a >> 32; + const u64 b_lo = u32(b); + const u64 b_hi = b >> 32; + + const u64 a_x_b_hi = a_hi * b_hi; + const u64 a_x_b_mid = a_hi * b_lo; + const u64 b_x_a_mid = b_hi * a_lo; + const u64 a_x_b_lo = a_lo * b_lo; + + const u64 carry_bit = (static_cast<u64>(static_cast<u32>(a_x_b_mid)) + + static_cast<u64>(static_cast<u32>(b_x_a_mid)) + (a_x_b_lo >> 32)) >> + 32; + + const u64 multhi = a_x_b_hi + (a_x_b_mid >> 32) + (b_x_a_mid >> 32) + carry_bit; + + return multhi; +#endif +} // This function divides a u128 by a u32 value and produces two u64 values: // the result of division and the remainder -[[nodiscard]] std::pair<u64, u64> Divide128On32(u128 dividend, u32 divisor); +[[nodiscard]] static inline std::pair<u64, u64> Divide128On32(u128 dividend, u32 divisor) { + u64 remainder = dividend[0] % divisor; + u64 accum = dividend[0] / divisor; + if (dividend[1] == 0) + return {accum, remainder}; + // We ignore dividend[1] / divisor as that overflows + const u64 first_segment = (dividend[1] % divisor) << 32; + accum += (first_segment / divisor) << 32; + const u64 second_segment = (first_segment % divisor) << 32; + accum += (second_segment / divisor); + remainder += second_segment % divisor; + if (remainder >= divisor) { + accum++; + remainder -= divisor; + } + return {accum, remainder}; +} } // namespace Common diff --git a/src/common/uuid.h b/src/common/uuid.h index 4ab9a25f0..2e7a18405 100644 --- a/src/common/uuid.h +++ b/src/common/uuid.h @@ -14,8 +14,8 @@ constexpr u128 INVALID_UUID{{0, 0}}; struct UUID { // UUIDs which are 0 are considered invalid! - u128 uuid = INVALID_UUID; - constexpr UUID() = default; + u128 uuid; + UUID() = default; constexpr explicit UUID(const u128& id) : uuid{id} {} constexpr explicit UUID(const u64 lo, const u64 hi) : uuid{{lo, hi}} {} diff --git a/src/common/wall_clock.cpp b/src/common/wall_clock.cpp index a8c143f85..49830b8ab 100644 --- a/src/common/wall_clock.cpp +++ b/src/common/wall_clock.cpp @@ -2,6 +2,8 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <cstdint> + #include "common/uint128.h" #include "common/wall_clock.h" diff --git a/src/common/x64/native_clock.cpp b/src/common/x64/native_clock.cpp index eb8a7782f..87de40624 100644 --- a/src/common/x64/native_clock.cpp +++ b/src/common/x64/native_clock.cpp @@ -2,16 +2,13 @@ // Licensed under GPLv2 or any later version // Refer to the license.txt file included. +#include <array> #include <chrono> +#include <limits> #include <mutex> #include <thread> -#ifdef _MSC_VER -#include <intrin.h> -#else -#include <x86intrin.h> -#endif - +#include "common/atomic_ops.h" #include "common/uint128.h" #include "common/x64/native_clock.h" @@ -48,54 +45,71 @@ NativeClock::NativeClock(u64 emulated_cpu_frequency_, u64 emulated_clock_frequen : WallClock(emulated_cpu_frequency_, emulated_clock_frequency_, true), rtsc_frequency{ rtsc_frequency_} { _mm_mfence(); - last_measure = __rdtsc(); - accumulated_ticks = 0U; + time_point.inner.last_measure = __rdtsc(); + time_point.inner.accumulated_ticks = 0U; + ns_rtsc_factor = GetFixedPoint64Factor(1000000000, rtsc_frequency); + us_rtsc_factor = GetFixedPoint64Factor(1000000, rtsc_frequency); + ms_rtsc_factor = GetFixedPoint64Factor(1000, rtsc_frequency); + clock_rtsc_factor = GetFixedPoint64Factor(emulated_clock_frequency, rtsc_frequency); + cpu_rtsc_factor = GetFixedPoint64Factor(emulated_cpu_frequency, rtsc_frequency); } u64 NativeClock::GetRTSC() { - std::scoped_lock scope{rtsc_serialize}; - _mm_mfence(); - const u64 current_measure = __rdtsc(); - u64 diff = current_measure - last_measure; - diff = diff & ~static_cast<u64>(static_cast<s64>(diff) >> 63); // max(diff, 0) - if (current_measure > last_measure) { - last_measure = current_measure; - } - accumulated_ticks += diff; + TimePoint new_time_point{}; + TimePoint current_time_point{}; + do { + current_time_point.pack = time_point.pack; + _mm_mfence(); + const u64 current_measure = __rdtsc(); + u64 diff = current_measure - current_time_point.inner.last_measure; + diff = diff & ~static_cast<u64>(static_cast<s64>(diff) >> 63); // max(diff, 0) + new_time_point.inner.last_measure = current_measure > current_time_point.inner.last_measure + ? current_measure + : current_time_point.inner.last_measure; + new_time_point.inner.accumulated_ticks = current_time_point.inner.accumulated_ticks + diff; + } while (!Common::AtomicCompareAndSwap(time_point.pack.data(), new_time_point.pack, + current_time_point.pack)); /// The clock cannot be more precise than the guest timer, remove the lower bits - return accumulated_ticks & inaccuracy_mask; + return new_time_point.inner.accumulated_ticks & inaccuracy_mask; } void NativeClock::Pause(bool is_paused) { if (!is_paused) { - _mm_mfence(); - last_measure = __rdtsc(); + TimePoint current_time_point{}; + TimePoint new_time_point{}; + do { + current_time_point.pack = time_point.pack; + new_time_point.pack = current_time_point.pack; + _mm_mfence(); + new_time_point.inner.last_measure = __rdtsc(); + } while (!Common::AtomicCompareAndSwap(time_point.pack.data(), new_time_point.pack, + current_time_point.pack)); } } std::chrono::nanoseconds NativeClock::GetTimeNS() { const u64 rtsc_value = GetRTSC(); - return std::chrono::nanoseconds{MultiplyAndDivide64(rtsc_value, 1000000000, rtsc_frequency)}; + return std::chrono::nanoseconds{MultiplyHigh(rtsc_value, ns_rtsc_factor)}; } std::chrono::microseconds NativeClock::GetTimeUS() { const u64 rtsc_value = GetRTSC(); - return std::chrono::microseconds{MultiplyAndDivide64(rtsc_value, 1000000, rtsc_frequency)}; + return std::chrono::microseconds{MultiplyHigh(rtsc_value, us_rtsc_factor)}; } std::chrono::milliseconds NativeClock::GetTimeMS() { const u64 rtsc_value = GetRTSC(); - return std::chrono::milliseconds{MultiplyAndDivide64(rtsc_value, 1000, rtsc_frequency)}; + return std::chrono::milliseconds{MultiplyHigh(rtsc_value, ms_rtsc_factor)}; } u64 NativeClock::GetClockCycles() { const u64 rtsc_value = GetRTSC(); - return MultiplyAndDivide64(rtsc_value, emulated_clock_frequency, rtsc_frequency); + return MultiplyHigh(rtsc_value, clock_rtsc_factor); } u64 NativeClock::GetCPUCycles() { const u64 rtsc_value = GetRTSC(); - return MultiplyAndDivide64(rtsc_value, emulated_cpu_frequency, rtsc_frequency); + return MultiplyHigh(rtsc_value, cpu_rtsc_factor); } } // namespace X64 diff --git a/src/common/x64/native_clock.h b/src/common/x64/native_clock.h index 6d1e32ac8..7cbd400d2 100644 --- a/src/common/x64/native_clock.h +++ b/src/common/x64/native_clock.h @@ -6,7 +6,6 @@ #include <optional> -#include "common/spin_lock.h" #include "common/wall_clock.h" namespace Common { @@ -32,14 +31,28 @@ public: private: u64 GetRTSC(); + union alignas(16) TimePoint { + TimePoint() : pack{} {} + u128 pack{}; + struct Inner { + u64 last_measure{}; + u64 accumulated_ticks{}; + } inner; + }; + /// value used to reduce the native clocks accuracy as some apss rely on /// undefined behavior where the level of accuracy in the clock shouldn't /// be higher. static constexpr u64 inaccuracy_mask = ~(UINT64_C(0x400) - 1); - SpinLock rtsc_serialize{}; - u64 last_measure{}; - u64 accumulated_ticks{}; + TimePoint time_point; + // factors + u64 clock_rtsc_factor{}; + u64 cpu_rtsc_factor{}; + u64 ns_rtsc_factor{}; + u64 us_rtsc_factor{}; + u64 ms_rtsc_factor{}; + u64 rtsc_frequency; }; } // namespace X64 |