File
Blob: src/workerd/api/deferred-proxy.h
| 1 | // Copyright (c) 2017-2022 Cloudflare, Inc. |
| 2 | // Licensed under the Apache 2.0 license found in the LICENSE file or at: |
| 3 | // https://opensource.org/licenses/Apache-2.0 |
| 4 | |
| 5 | #pragma once |
| 6 | |
| 7 | #include <kj/async.h> |
| 8 | #include <kj/debug.h> |
| 9 | |
| 10 | namespace workerd::api { |
| 11 | |
| 12 | // ======================================================================================= |
| 13 | |
| 14 | // Some API methods return Promise<DeferredProxy<T>> when the task can be separated into two |
| 15 | // parts: some work that must be done with the IoContext still live, and some part that |
| 16 | // can occur after the IoContext completes, but which should still be performed before |
| 17 | // the overall task is "done". |
| 18 | // |
| 19 | // In particular, when an HTTP event ends up proxying the response body stream (or WebSocket |
| 20 | // stream) directly to/from origin, then that streaming can take place without pinning the |
| 21 | // isolate in memory, and without holding the IoContext open. So, |
| 22 | // `ServiceWorkerGlobalScope::request()` returns `Promise<DeferredProxy<void>>`. The outer |
| 23 | // Promise waits for the JavaScript work to be done, and the inner DeferredProxy<void> represents |
| 24 | // the proxying step. |
| 25 | // |
| 26 | // Note that if you're performing a task that resolves to DeferredProxy but JavaScript is |
| 27 | // actually waiting for the result of the task, then it's your responsibility to call |
| 28 | // IoContext::current().registerPendingEvent() and attach it to `proxyTask`, otherwise |
| 29 | // the request might be canceled as the proxy task won't be recognized as something that the |
| 30 | // request is waiting on. |
| 31 | template <typename T> |
| 32 | struct DeferredProxy { |
| 33 | // TODO(cleanup): Now that we have jsg::Promise, it might make sense for deferred proxying to |
| 34 | // be represented as `jsg::Promise<api::DeferredProxy<T>>`, since the outer promise is |
| 35 | // intended to represent activity that happens in JavaScript while the inner one represents |
| 36 | // pure I/O. This will require some refactoring, though. |
| 37 | |
| 38 | kj::Promise<T> proxyTask; |
| 39 | }; |
| 40 | |
| 41 | inline DeferredProxy<void> newNoopDeferredProxy() { |
| 42 | return DeferredProxy<void>{kj::READY_NOW}; |
| 43 | } |
| 44 | |
| 45 | template <typename T> |
| 46 | inline DeferredProxy<T> newNoopDeferredProxy(T&& value) { |
| 47 | return DeferredProxy<T>{kj::mv(value)}; |
| 48 | } |
| 49 | |
| 50 | // Helper method to use when you need to return `Promise<DeferredProxy<T>>` but no part of the |
| 51 | // operation you are returning is eligible to be deferred past the IoContext lifetime. |
| 52 | template <typename T> |
| 53 | inline kj::Promise<DeferredProxy<T>> addNoopDeferredProxy(kj::Promise<T> promise) { |
| 54 | co_return newNoopDeferredProxy(co_await promise); |
| 55 | } |
| 56 | inline kj::Promise<DeferredProxy<void>> addNoopDeferredProxy(kj::Promise<void> promise) { |
| 57 | co_await promise; |
| 58 | co_return newNoopDeferredProxy(); |
| 59 | } |
| 60 | |
| 61 | // --------------------------------------------------------- |
| 62 | // Deferred proxy coroutine integration |
| 63 | |
| 64 | // If a coroutine returns a kj::Promise<DeferredProxy<T>>, the coroutine implementation gains the |
| 65 | // following features: |
| 66 | // |
| 67 | // - `KJ_CO_MAGIC BEGIN_DEFERRED_PROXYING` fulfills the outer kj::Promise<DeferredProxy<T>>. The |
| 68 | // resulting DeferredProxy<T> object contains a `proxyTask` Promise which owns the coroutine. |
| 69 | // |
| 70 | // - `co_return` implicitly fulfills the outer Promise for the DeferredProxy<T> (if it has not |
| 71 | // already been fulfilled by the magic `KJ_CO_MAGIC` described above), then fulfills the inner |
| 72 | // `proxyTask`. |
| 73 | // |
| 74 | // - Unhandled exceptions reject the outer kj::Promise<DeferredProxy<T>> (if it has not already |
| 75 | // been fulfilled by the magic `KJ_CO_MAGIC` described above), then reject the inner `proxyTask`. |
| 76 | // |
| 77 | // It is not possible to write a "regular" coroutine which returns kj::Promise<DeferredProxy<T>>; |
| 78 | // that is, `co_return DeferredProxy<T> { ... }` is a compile error. You must initiate deferred |
| 79 | // proxying using `KJ_CO_MAGIC BEGIN_DEFERRED_PROXYING`. |
| 80 | |
| 81 | // The coroutine adapter class, required for the compiler to know how to create coroutines |
| 82 | // returning kj::Promise<DeferredProxy<T>>. We declare it here so we can name it in our |
| 83 | // `coroutine_traits` specialization below. |
| 84 | template <typename T, typename... Args> |
| 85 | class DeferredProxyCoroutine; |
| 86 | } // namespace workerd::api |
| 87 | |
| 88 | namespace KJ_COROUTINE_STD_NAMESPACE { |
| 89 | // Enter the `std` or `std::experimental` namespace, depending on whether we're using C++20 |
| 90 | // coroutines or the Coroutines TS. |
| 91 | |
| 92 | template <class T, class... Args> |
| 93 | struct coroutine_traits<kj::Promise<workerd::api::DeferredProxy<T>>, Args...> { |
| 94 | using promise_type = workerd::api::DeferredProxyCoroutine<T, Args...>; |
| 95 | }; |
| 96 | |
| 97 | } // namespace KJ_COROUTINE_STD_NAMESPACE |
| 98 | |
| 99 | namespace workerd::api { |
| 100 | |
| 101 | class BeginDeferredProxyingConstant final {}; |
| 102 | // A magic constant which a DeferredProxyPromise<T> coroutine can `KJ_CO_MAGIC` to indicate that the |
| 103 | // deferred proxying phase of its operation has begun. |
| 104 | constexpr BeginDeferredProxyingConstant BEGIN_DEFERRED_PROXYING{}; |
| 105 | |
| 106 | // A concept which is true if C is a coroutine adapter which supports the `co_yield` operator for |
| 107 | // type T. We could also check that the expression results in an awaitable, but that is already a |
| 108 | // compile error in other ways. |
| 109 | template <typename T, typename C> |
| 110 | concept CoroutineYieldValue = requires(T&& v, C coroutineAdapter) { |
| 111 | { coroutineAdapter.yield_value(kj::fwd<T>(v)) }; |
| 112 | }; |
| 113 | |
| 114 | // The coroutine adapter type for DeferredProxyPromise<T>. Most of the work is forwarded to the |
| 115 | // regular kj::Promise<T> coroutine adapter. |
| 116 | template <typename T, typename... Args> |
| 117 | class DeferredProxyCoroutine: public kj::_::PromiseNode, |
| 118 | public kj::_::CoroutineMixin<DeferredProxyCoroutine<T, Args...>, T> { |
| 119 | using InnerCoroutineAdapter = |
| 120 | kj::_::stdcoro::coroutine_traits<kj::Promise<T>, Args...>::promise_type; |
| 121 | |
| 122 | public: |
| 123 | using Handle = kj::_::stdcoro::coroutine_handle<DeferredProxyCoroutine>; |
| 124 | |
| 125 | DeferredProxyCoroutine(kj::SourceLocation location = {}) |
| 126 | : inner(Handle::from_promise(*this), location) {} |
| 127 | |
| 128 | kj::Promise<DeferredProxy<T>> get_return_object() { |
| 129 | // We need to return a RAII object which will destroy this (as in, `this`) coroutine adapter. |
| 130 | // The logic which calls `coroutine_handle<>::destroy()` is tucked away in our inner coroutine |
| 131 | // adapter, however, leading to the weird situation where the `inner.get_return_object()` |
| 132 | // Promise owns `this`. And `this` owns `inner.get_return_object()` transitively via `result`! |
| 133 | // |
| 134 | // Fortunately, DeferredProxyCoroutine implements the PromiseNode interface, meaning when our |
| 135 | // returned Promise is eventually dropped, our `PromiseNode::destroy()` implementation will be |
| 136 | // called. This gives us the opportunity (that is, in `destroy()`) to destroy our |
| 137 | // `inner.get_return_object()` Promise, breaking the ownership cycle and destroying `this`. |
| 138 | |
| 139 | result.value = DeferredProxy<T>{inner.get_return_object()}; |
| 140 | return kj::_::PromiseNode::to<kj::Promise<DeferredProxy<T>>>(kj::_::OwnPromiseNode(this)); |
| 141 | } |
| 142 | |
| 143 | auto initial_suspend() { |
| 144 | return inner.initial_suspend(); |
| 145 | } |
| 146 | auto final_suspend() noexcept { |
| 147 | return inner.final_suspend(); |
| 148 | } |
| 149 | // Just trivially forward these. |
| 150 | |
| 151 | void unhandled_exception() { |
| 152 | // Reject our outer promise if it hasn't yet been fulfilled, or forward to the inner |
| 153 | // implementation. |
| 154 | |
| 155 | if (!deferredProxyingHasBegun) { |
| 156 | result.addException(kj::getCaughtExceptionAsKj()); |
| 157 | onReadyEvent.arm(); |
| 158 | deferredProxyingHasBegun = true; |
| 159 | } else { |
| 160 | inner.unhandled_exception(); |
| 161 | } |
| 162 | } |
| 163 | |
| 164 | kj::_::stdcoro::suspend_never yield_value(decltype(BEGIN_DEFERRED_PROXYING)) { |
| 165 | // This allows us to write `KJ_CO_MAGIC` within a DeferredProxyPromise<T> coroutine to fulfill |
| 166 | // the coroutine's outer promise with a DeferredProxy<T>. |
| 167 | // |
| 168 | // This could alternatively be implemented as an await_transform() with a magic parameter type. |
| 169 | |
| 170 | fulfillOuterPromise(); |
| 171 | return {}; |
| 172 | } |
| 173 | |
| 174 | template <CoroutineYieldValue<InnerCoroutineAdapter> U> |
| 175 | auto yield_value(U&& value) { |
| 176 | // Forward all other `co_yield`s to the inner coroutine, if it has a `yield_value()` |
| 177 | // implementation -- it might implement some magic, too. |
| 178 | return inner.yield_value(kj::fwd<U>(value)); |
| 179 | } |
| 180 | |
| 181 | void fulfill(kj::_::FixVoid<T>&& value) { |
| 182 | // Required by CoroutineMixin implementation to implement `co_return`. |
| 183 | |
| 184 | fulfillOuterPromise(); |
| 185 | inner.fulfill(kj::mv(value)); |
| 186 | } |
| 187 | |
| 188 | template <typename U> |
| 189 | decltype(auto) await_transform(U&& awaitable) { |
| 190 | // Trivially forward everything, so we can await anything a kj::Promise<T> can. |
| 191 | return inner.await_transform(kj::fwd<U>(awaitable)); |
| 192 | } |
| 193 | |
| 194 | operator kj::_::CoroutineBase&() { |
| 195 | return inner; |
| 196 | } |
| 197 | // Required by Awaiter<T>::await_suspend() to support awaiting Promises. |
| 198 | |
| 199 | private: |
| 200 | void fulfillOuterPromise() { |
| 201 | // Fulfill the outer promise if it hasn't already settled. |
| 202 | |
| 203 | if (!deferredProxyingHasBegun) { |
| 204 | // Our `result` is put in place already by `get_return_object()`, so all we have to do is arm |
| 205 | // the event. |
| 206 | onReadyEvent.arm(); |
| 207 | deferredProxyingHasBegun = true; |
| 208 | } |
| 209 | } |
| 210 | |
| 211 | // PromiseNode implementation |
| 212 | |
| 213 | void setSelfPointer(kj::_::OwnPromiseNode* selfPtr) noexcept override { |
| 214 | this->selfPtr = selfPtr; |
| 215 | } |
| 216 | |
| 217 | void destroy() override { |
| 218 | // The promise returned by `inner.get_return_object()` is what actually owns this coroutine |
| 219 | // frame. We temporarily store that in `result` until our outer promise is fulfilled. So, to |
| 220 | // destroy ourselves, we must manually drop `result`. |
| 221 | // |
| 222 | // On the other hand, if our outer promise has already been fulfilled, then `result` has already |
| 223 | // been delivered to wherever it is going, and someone else directly owns the coroutine now, not |
| 224 | // us. In this case, this `destroy()` override will have already been called (and it will have |
| 225 | // been a no-op), because our own OwnPromiseNode will have already been dropped in `get()`. |
| 226 | |
| 227 | auto drop = kj::mv(result); |
| 228 | } |
| 229 | |
| 230 | void onReady(kj::_::Event* event) noexcept override { |
| 231 | onReadyEvent.init(event); |
| 232 | } |
| 233 | |
| 234 | void get(kj::_::ExceptionOrValue& output) noexcept override { |
| 235 | // Make sure that the outer PromiseNode (`this` one) is destroyed before the inner PromiseNode. |
| 236 | // kj-async should already provide us this guarantee, but since incorrect destruction order |
| 237 | // would cause invalid memory access, we provide a stronger guarantee. Also see the comment for |
| 238 | // the `result` data member. |
| 239 | KJ_ASSERT(selfPtr != nullptr); |
| 240 | KJ_DEFER(*selfPtr = nullptr); |
| 241 | |
| 242 | static_cast<decltype(result)&>(output) = kj::mv(result); |
| 243 | } |
| 244 | |
| 245 | void tracePromise(kj::_::TraceBuilder& builder, bool stopAtNextEvent) override { |
| 246 | // The PromiseNode we're waiting on is whatever the coroutine is waiting on. |
| 247 | static_cast<kj::_::PromiseNode&>(inner).tracePromise(builder, stopAtNextEvent); |
| 248 | |
| 249 | // Maybe returning the address of get() will give us a function name with meaningful type |
| 250 | // information. |
| 251 | builder.add(getMethodStartAddress(implicitCast<PromiseNode&>(*this), &PromiseNode::get)); |
| 252 | } |
| 253 | |
| 254 | // We defer the majority of the implementation to the regular kj::Promise<T> coroutine adapter. |
| 255 | InnerCoroutineAdapter inner; |
| 256 | |
| 257 | // Helper to arm the event which fires when the outer promise (that is, `this` PromiseNode) for |
| 258 | // the DeferredProxy<T> is ready. |
| 259 | OnReadyEvent onReadyEvent; |
| 260 | |
| 261 | // Stores the result for the outer promise. |
| 262 | // |
| 263 | // WARNING: This object owns `this` PromiseNode! If `result` is ever moved away, as is done in |
| 264 | // `get()`, we must arrange to make sure that no one ever tries to use `this` PromiseNode again. |
| 265 | // Stated another way, we must guarantee that the outer PromiseNode (for `DeferredProxy<T>`) is |
| 266 | // always destroyed before the inner PromiseNode (for `T`). kj-async always does this anyway, but |
| 267 | // we implement an additional safeguard by immediately destroying our own `OwnPromiseNode` (which |
| 268 | // we have access to via `setSelfPointer()`) when we move `result` away in `get()`. |
| 269 | kj::_::ExceptionOr<DeferredProxy<T>> result; |
| 270 | |
| 271 | // Used to drop ourselves in `get()` -- see comment for `result`. |
| 272 | kj::_::OwnPromiseNode* selfPtr = nullptr; |
| 273 | |
| 274 | // Set to true when deferred proxying has begun -- that is, when the outer DeferredProxy<T> |
| 275 | // promise is fulfilled by calling `onReadyEvent.arm()`. |
| 276 | bool deferredProxyingHasBegun = false; |
| 277 | }; |
| 278 | |
| 279 | } // namespace workerd::api |