|
| 1 | +// Copyright 2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved. |
| 2 | +// |
| 3 | +// Redistribution and use in source and binary forms, with or without |
| 4 | +// modification, are permitted provided that the following conditions |
| 5 | +// are met: |
| 6 | +// * Redistributions of source code must retain the above copyright |
| 7 | +// notice, this list of conditions and the following disclaimer. |
| 8 | +// * Redistributions in binary form must reproduce the above copyright |
| 9 | +// notice, this list of conditions and the following disclaimer in the |
| 10 | +// documentation and/or other materials provided with the distribution. |
| 11 | +// * Neither the name of NVIDIA CORPORATION nor the names of its |
| 12 | +// contributors may be used to endorse or promote products derived |
| 13 | +// from this software without specific prior written permission. |
| 14 | +// |
| 15 | +// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ``AS IS'' AND ANY |
| 16 | +// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 17 | +// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
| 18 | +// PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR |
| 19 | +// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, |
| 20 | +// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
| 21 | +// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR |
| 22 | +// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY |
| 23 | +// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 24 | +// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| 25 | +// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 26 | +#pragma once |
| 27 | + |
| 28 | +#include <coroutine> |
| 29 | +#include <type_traits> |
| 30 | +#include <utility> |
| 31 | + |
| 32 | +namespace triton::perfanalyzer { |
| 33 | + |
| 34 | +template <typename T = void> |
| 35 | +class Coroutine { |
| 36 | + public: |
| 37 | + struct Empty {}; |
| 38 | + typedef |
| 39 | + typename std::conditional<std::is_void<T>::value, Empty, T>::type SafeT; |
| 40 | + |
| 41 | + Coroutine() = default; |
| 42 | + Coroutine(Coroutine&& other) = default; |
| 43 | + Coroutine& operator=(Coroutine&& other) = default; |
| 44 | + Coroutine(Coroutine const&) = delete; |
| 45 | + Coroutine& operator=(Coroutine const&) = delete; |
| 46 | + |
| 47 | + class Awaiter { |
| 48 | + public: |
| 49 | + Awaiter(Awaiter&& other) = default; |
| 50 | + Awaiter& operator=(Awaiter&& other) = default; |
| 51 | + Awaiter(Awaiter const&) = default; |
| 52 | + Awaiter& operator=(Awaiter const&) = default; |
| 53 | + constexpr bool await_ready() const noexcept |
| 54 | + { |
| 55 | + bool ret = coroutine_->earlyResume_; |
| 56 | + coroutine_->earlyResume_ = false; |
| 57 | + return ret; |
| 58 | + } |
| 59 | + constexpr void await_suspend(std::coroutine_handle<> h) |
| 60 | + { |
| 61 | + coroutine_->suspended_ = true; |
| 62 | + } |
| 63 | + constexpr void await_resume() const noexcept {} |
| 64 | + |
| 65 | + private: |
| 66 | + Awaiter(Coroutine* coroutine) : coroutine_(coroutine) {} |
| 67 | + Coroutine* coroutine_; |
| 68 | + friend struct Coroutine; |
| 69 | + }; |
| 70 | + |
| 71 | + Awaiter awaiter() { return Awaiter(this); } |
| 72 | + |
| 73 | + void resume() |
| 74 | + { |
| 75 | + if (!handle_) |
| 76 | + return; |
| 77 | + if (!suspended_) { |
| 78 | + earlyResume_ = true; |
| 79 | + return; |
| 80 | + } |
| 81 | + suspended_ = false; |
| 82 | + handle_.resume(); |
| 83 | + } |
| 84 | + |
| 85 | + bool done() |
| 86 | + { |
| 87 | + if (!handle_) |
| 88 | + return true; |
| 89 | + bool isDone = handle_.done(); |
| 90 | + if (isDone) { |
| 91 | + if constexpr (!std::is_void<T>::value) { |
| 92 | + value_ = std::move(handle_.promise().value_); |
| 93 | + } |
| 94 | + handle_.destroy(); |
| 95 | + handle_ = nullptr; |
| 96 | + } |
| 97 | + return isDone; |
| 98 | + } |
| 99 | + |
| 100 | + const SafeT& value() const { return value_; } |
| 101 | + |
| 102 | + private: |
| 103 | + struct PromiseVoid { |
| 104 | + Coroutine<> get_return_object() |
| 105 | + { |
| 106 | + return Coroutine<>{ |
| 107 | + std::move(std::coroutine_handle<Promise>::from_promise(*this))}; |
| 108 | + } |
| 109 | + std::suspend_always initial_suspend() { return {}; } |
| 110 | + std::suspend_always final_suspend() noexcept { return {}; } |
| 111 | + void unhandled_exception() {} |
| 112 | + void return_void() |
| 113 | + { |
| 114 | + auto awaitingCoroutine = awaitingCoroutine_; |
| 115 | + if (awaitingCoroutine) { |
| 116 | + __builtin_coro_resume(awaitingCoroutine); |
| 117 | + } |
| 118 | + } |
| 119 | + [[no_unique_address]] Empty value_; |
| 120 | + void* awaitingCoroutine_ = nullptr; |
| 121 | + }; |
| 122 | + struct PromiseValue { |
| 123 | + Coroutine<T> get_return_object() |
| 124 | + { |
| 125 | + return Coroutine{ |
| 126 | + std::move(std::coroutine_handle<Promise>::from_promise(*this))}; |
| 127 | + } |
| 128 | + std::suspend_always initial_suspend() { return {}; } |
| 129 | + std::suspend_always final_suspend() noexcept { return {}; } |
| 130 | + void unhandled_exception() {} |
| 131 | + void return_value(T&& value) |
| 132 | + { |
| 133 | + value_ = std::move(value); |
| 134 | + auto awaitingCoroutine = awaitingCoroutine_; |
| 135 | + if (awaitingCoroutine) { |
| 136 | + __builtin_coro_resume(awaitingCoroutine); |
| 137 | + } |
| 138 | + } |
| 139 | + T value_; |
| 140 | + void* awaitingCoroutine_ = nullptr; |
| 141 | + }; |
| 142 | + typedef typename std::conditional< |
| 143 | + std::is_void<T>::value, PromiseVoid, PromiseValue>::type Promise; |
| 144 | + Coroutine(std::coroutine_handle<Promise>&& handle) |
| 145 | + : handle_(std::move(handle)) |
| 146 | + { |
| 147 | + } |
| 148 | + std::coroutine_handle<Promise> handle_; |
| 149 | + [[no_unique_address]] SafeT value_; |
| 150 | + void* awaitingCoroutine_ = nullptr; |
| 151 | + bool suspended_ = true; |
| 152 | + bool earlyResume_ = false; |
| 153 | + |
| 154 | + public: |
| 155 | + using promise_type = Promise; |
| 156 | + |
| 157 | + constexpr bool await_ready() { return handle_.done(); } |
| 158 | + template <typename U> |
| 159 | + constexpr void await_suspend(std::coroutine_handle<U> h) |
| 160 | + { |
| 161 | + auto& promise = handle_.promise(); |
| 162 | + promise.awaitingCoroutine_ = h.address(); |
| 163 | + resume(); |
| 164 | + } |
| 165 | + constexpr SafeT await_resume() |
| 166 | + { |
| 167 | + SafeT value = std::move(handle_.promise().value_); |
| 168 | + handle_.destroy(); |
| 169 | + return value; |
| 170 | + } |
| 171 | +}; |
| 172 | + |
| 173 | +} // namespace triton::perfanalyzer |
0 commit comments