mfbt/SPSCQueue.h

   1 /* -*- Mode: C++; tab-width: 8; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
   2 /* vim: set ts=8 sts=2 et sw=2 tw=80: */
   3 /* This Source Code Form is subject to the terms of the Mozilla Public
   4  * License, v. 2.0. If a copy of the MPL was not distributed with this
   5  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
   6
   7 /* Single producer single consumer lock-free and wait-free queue. */
   8
   9 #ifndef mozilla_LockFreeQueue_h
  10 #define mozilla_LockFreeQueue_h
  11
  12 #include "mozilla/Assertions.h"
  13 #include "mozilla/Attributes.h"
  14 #include "mozilla/PodOperations.h"
  15 #include <algorithm>
  16 #include <atomic>
  17 #include <cstddef>
  18 #include <limits>
  19 #include <memory>
  20 #include <thread>
  21 #include <type_traits>
  22
  23 namespace mozilla {
  24
  25 namespace detail {
  26 template <typename T, bool IsPod = std::is_trivial<T>::value>
  27 struct MemoryOperations {
  28   /**
  29    * This allows zeroing (using memset) or default-constructing a number of
  30    * elements calling the constructors if necessary.
  31    */
  32   static void ConstructDefault(T* aDestination, size_t aCount);
  33   /**
  34    * This allows either moving (if T supports it) or copying a number of
  35    * elements from a `aSource` pointer to a `aDestination` pointer.
  36    * If it is safe to do so and this call copies, this uses PodCopy. Otherwise,
  37    * constructors and destructors are called in a loop.
  38    */
  39   static void MoveOrCopy(T* aDestination, T* aSource, size_t aCount);
  40 };
  41
  42 template <typename T>
  43 struct MemoryOperations<T, true> {
  44   static void ConstructDefault(T* aDestination, size_t aCount) {
  45     PodZero(aDestination, aCount);
  46   }
  47   static void MoveOrCopy(T* aDestination, T* aSource, size_t aCount) {
  48     PodCopy(aDestination, aSource, aCount);
  49   }
  50 };
  51
  52 template <typename T>
  53 struct MemoryOperations<T, false> {
  54   static void ConstructDefault(T* aDestination, size_t aCount) {
  55     for (size_t i = 0; i < aCount; i++) {
  56       aDestination[i] = T();
  57     }
  58   }
  59   static void MoveOrCopy(T* aDestination, T* aSource, size_t aCount) {
  60     std::move(aSource, aSource + aCount, aDestination);
  61   }
  62 };
  63 }  // namespace detail
  64
  65 /**
  66  * This data structure allows producing data from one thread, and consuming it
  67  * on another thread, safely and without explicit synchronization.
  68  *
  69  * The role for the producer and the consumer must be constant, i.e., the
  70  * producer should always be on one thread and the consumer should always be on
  71  * another thread.
  72  *
  73  * Some words about the inner workings of this class:
  74  * - Capacity is fixed. Only one allocation is performed, in the constructor.
  75  *   When reading and writing, the return value of the method allows checking if
  76  *   the ring buffer is empty or full.
  77  * - We always keep the read index at least one element ahead of the write
  78  *   index, so we can distinguish between an empty and a full ring buffer: an
  79  *   empty ring buffer is when the write index is at the same position as the
  80  *   read index. A full buffer is when the write index is exactly one position
  81  *   before the read index.
  82  * - We synchronize updates to the read index after having read the data, and
  83  *   the write index after having written the data. This means that the each
  84  *   thread can only touch a portion of the buffer that is not touched by the
  85  *   other thread.
  86  * - Callers are expected to provide buffers. When writing to the queue,
  87  *   elements are copied into the internal storage from the buffer passed in.
  88  *   When reading from the queue, the user is expected to provide a buffer.
  89  *   Because this is a ring buffer, data might not be contiguous in memory;
  90  *   providing an external buffer to copy into is an easy way to have linear
  91  *   data for further processing.
  92  */
  93 template <typename T>
  94 class SPSCRingBufferBase {
  95  public:
  96   /**
  97    * Constructor for a ring buffer.
  98    *
  99    * This performs an allocation on the heap, but is the only allocation that
 100    * will happen for the life time of a `SPSCRingBufferBase`.
 101    *
 102    * @param Capacity The maximum number of element this ring buffer will hold.
 103    */
 104   explicit SPSCRingBufferBase(int aCapacity)
 105       : mReadIndex(0),
 106         mWriteIndex(0)
 107         /* One more element to distinguish from empty and full buffer. */
 108         ,
 109         mCapacity(aCapacity + 1) {
 110     MOZ_ASSERT(StorageCapacity() < std::numeric_limits<int>::max() / 2,
 111                "buffer too large for the type of index used.");
 112     MOZ_ASSERT(mCapacity > 0 && aCapacity != std::numeric_limits<int>::max());
 113
 114     mData = std::make_unique<T[]>(StorageCapacity());
 115
 116     std::atomic_thread_fence(std::memory_order::memory_order_seq_cst);
 117   }
 118   /**
 119    * Push `aCount` zero or default constructed elements in the array.
 120    *
 121    * Only safely called on the producer thread.
 122    *
 123    * @param count The number of elements to enqueue.
 124    * @return The number of element enqueued.
 125    */
 126   MOZ_MUST_USE
 127   int EnqueueDefault(int aCount) { return Enqueue(nullptr, aCount); }
 128   /**
 129    * @brief Put an element in the queue.
 130    *
 131    * Only safely called on the producer thread.
 132    *
 133    * @param element The element to put in the queue.
 134    *
 135    * @return 1 if the element was inserted, 0 otherwise.
 136    */
 137   MOZ_MUST_USE
 138   int Enqueue(T& aElement) { return Enqueue(&aElement, 1); }
 139   /**
 140    * Push `aCount` elements in the ring buffer.
 141    *
 142    * Only safely called on the producer thread.
 143    *
 144    * @param elements a pointer to a buffer containing at least `count` elements.
 145    * If `elements` is nullptr, zero or default constructed elements are enqueud.
 146    * @param count The number of elements to read from `elements`
 147    * @return The number of elements successfully coped from `elements` and
 148    * inserted into the ring buffer.
 149    */
 150   MOZ_MUST_USE
 151   int Enqueue(T* aElements, int aCount) {
 152 #ifdef DEBUG
 153     AssertCorrectThread(mProducerId);
 154 #endif
 155
 156     int rdIdx = mReadIndex.load(std::memory_order::memory_order_acquire);
 157     int wrIdx = mWriteIndex.load(std::memory_order::memory_order_relaxed);
 158
 159     if (IsFull(rdIdx, wrIdx)) {
 160       return 0;
 161     }
 162
 163     int toWrite = std::min(AvailableWriteInternal(rdIdx, wrIdx), aCount);
 164
 165     /* First part, from the write index to the end of the array. */
 166     int firstPart = std::min(StorageCapacity() - wrIdx, toWrite);
 167     /* Second part, from the beginning of the array */
 168     int secondPart = toWrite - firstPart;
 169
 170     if (aElements) {
 171       detail::MemoryOperations<T>::MoveOrCopy(mData.get() + wrIdx, aElements,
 172                                               firstPart);
 173       detail::MemoryOperations<T>::MoveOrCopy(
 174           mData.get(), aElements + firstPart, secondPart);
 175     } else {
 176       detail::MemoryOperations<T>::ConstructDefault(mData.get() + wrIdx,
 177                                                     firstPart);
 178       detail::MemoryOperations<T>::ConstructDefault(mData.get(), secondPart);
 179     }
 180
 181     mWriteIndex.store(IncrementIndex(wrIdx, toWrite),
 182                       std::memory_order::memory_order_release);
 183
 184     return toWrite;
 185   }
 186   /**
 187    * Retrieve at most `count` elements from the ring buffer, and copy them to
 188    * `elements`, if non-null.
 189    *
 190    * Only safely called on the consumer side.
 191    *
 192    * @param elements A pointer to a buffer with space for at least `count`
 193    * elements. If `elements` is `nullptr`, `count` element will be discarded.
 194    * @param count The maximum number of elements to Dequeue.
 195    * @return The number of elements written to `elements`.
 196    */
 197   MOZ_MUST_USE
 198   int Dequeue(T* elements, int count) {
 199 #ifdef DEBUG
 200     AssertCorrectThread(mConsumerId);
 201 #endif
 202
 203     int wrIdx = mWriteIndex.load(std::memory_order::memory_order_acquire);
 204     int rdIdx = mReadIndex.load(std::memory_order::memory_order_relaxed);
 205
 206     if (IsEmpty(rdIdx, wrIdx)) {
 207       return 0;
 208     }
 209
 210     int toRead = std::min(AvailableReadInternal(rdIdx, wrIdx), count);
 211
 212     int firstPart = std::min(StorageCapacity() - rdIdx, toRead);
 213     int secondPart = toRead - firstPart;
 214
 215     if (elements) {
 216       detail::MemoryOperations<T>::MoveOrCopy(elements, mData.get() + rdIdx,
 217                                               firstPart);
 218       detail::MemoryOperations<T>::MoveOrCopy(elements + firstPart, mData.get(),
 219                                               secondPart);
 220     }
 221
 222     mReadIndex.store(IncrementIndex(rdIdx, toRead),
 223                      std::memory_order::memory_order_release);
 224
 225     return toRead;
 226   }
 227   /**
 228    * Get the number of available elements for consuming.
 229    *
 230    * Only safely called on the consumer thread. This can be less than the actual
 231    * number of elements in the queue, since the mWriteIndex is updated at the
 232    * very end of the Enqueue method on the producer thread, but consequently
 233    * always returns a number of elements such that a call to Dequeue return this
 234    * number of elements.
 235    *
 236    * @return The number of available elements for reading.
 237    */
 238   int AvailableRead() const {
 239 #ifdef DEBUG
 240     AssertCorrectThread(mConsumerId);
 241 #endif
 242     return AvailableReadInternal(
 243         mReadIndex.load(std::memory_order::memory_order_relaxed),
 244         mWriteIndex.load(std::memory_order::memory_order_relaxed));
 245   }
 246   /**
 247    * Get the number of available elements for writing.
 248    *
 249    * Only safely called on the producer thread. This can be less than than the
 250    * actual number of slots that are available, because mReadIndex is update at
 251    * the very end of the Deque method. It always returns a number such that a
 252    * call to Enqueue with this number will succeed in enqueuing this number of
 253    * elements.
 254    *
 255    * @return The number of empty slots in the buffer, available for writing.
 256    */
 257   int AvailableWrite() const {
 258 #ifdef DEBUG
 259     AssertCorrectThread(mProducerId);
 260 #endif
 261     return AvailableWriteInternal(
 262         mReadIndex.load(std::memory_order::memory_order_relaxed),
 263         mWriteIndex.load(std::memory_order::memory_order_relaxed));
 264   }
 265   /**
 266    * Get the total Capacity, for this ring buffer.
 267    *
 268    * Can be called safely on any thread.
 269    *
 270    * @return The maximum Capacity of this ring buffer.
 271    */
 272   int Capacity() const { return StorageCapacity() - 1; }
 273   /**
 274    * Reset the consumer and producer thread identifier, in case the threads are
 275    * being changed. This has to be externally synchronized. This is no-op when
 276    * asserts are disabled.
 277    */
 278   void ResetThreadIds() {
 279 #ifdef DEBUG
 280     mConsumerId = mProducerId = std::thread::id();
 281 #endif
 282   }
 283
 284  private:
 285   /** Return true if the ring buffer is empty.
 286    *
 287    * This can be called from the consumer or the producer thread.
 288    *
 289    * @param aReadIndex the read index to consider
 290    * @param writeIndex the write index to consider
 291    * @return true if the ring buffer is empty, false otherwise.
 292    **/
 293   bool IsEmpty(int aReadIndex, int aWriteIndex) const {
 294     return aWriteIndex == aReadIndex;
 295   }
 296   /** Return true if the ring buffer is full.
 297    *
 298    * This happens if the write index is exactly one element behind the read
 299    * index.
 300    *
 301    * This can be called from the consummer or the producer thread.
 302    *
 303    * @param aReadIndex the read index to consider
 304    * @param writeIndex the write index to consider
 305    * @return true if the ring buffer is full, false otherwise.
 306    **/
 307   bool IsFull(int aReadIndex, int aWriteIndex) const {
 308     return (aWriteIndex + 1) % StorageCapacity() == aReadIndex;
 309   }
 310   /**
 311    * Return the size of the storage. It is one more than the number of elements
 312    * that can be stored in the buffer.
 313    *
 314    * This can be called from any thread.
 315    *
 316    * @return the number of elements that can be stored in the buffer.
 317    */
 318   int StorageCapacity() const { return mCapacity; }
 319   /**
 320    * Returns the number of elements available for reading.
 321    *
 322    * This can be called from the consummer or producer thread, but see the
 323    * comment in `AvailableRead`.
 324    *
 325    * @return the number of available elements for reading.
 326    */
 327   int AvailableReadInternal(int aReadIndex, int aWriteIndex) const {
 328     if (aWriteIndex >= aReadIndex) {
 329       return aWriteIndex - aReadIndex;
 330     } else {
 331       return aWriteIndex + StorageCapacity() - aReadIndex;
 332     }
 333   }
 334   /**
 335    * Returns the number of empty elements, available for writing.
 336    *
 337    * This can be called from the consummer or producer thread, but see the
 338    * comment in `AvailableWrite`.
 339    *
 340    * @return the number of elements that can be written into the array.
 341    */
 342   int AvailableWriteInternal(int aReadIndex, int aWriteIndex) const {
 343     /* We subtract one element here to always keep at least one sample
 344      * free in the buffer, to distinguish between full and empty array. */
 345     int rv = aReadIndex - aWriteIndex - 1;
 346     if (aWriteIndex >= aReadIndex) {
 347       rv += StorageCapacity();
 348     }
 349     return rv;
 350   }
 351   /**
 352    * Increments an index, wrapping it around the storage.
 353    *
 354    * Incrementing `mWriteIndex` can be done on the producer thread.
 355    * Incrementing `mReadIndex` can be done on the consummer thread.
 356    *
 357    * @param index a reference to the index to increment.
 358    * @param increment the number by which `index` is incremented.
 359    * @return the new index.
 360    */
 361   int IncrementIndex(int aIndex, int aIncrement) const {
 362     MOZ_ASSERT(aIncrement >= 0 && aIncrement < StorageCapacity() &&
 363                aIndex < StorageCapacity());
 364     return (aIndex + aIncrement) % StorageCapacity();
 365   }
 366   /**
 367    * @brief This allows checking that Enqueue (resp. Dequeue) are always
 368    * called by the right thread.
 369    *
 370    * The role of the thread are assigned the first time they call Enqueue or
 371    * Dequeue, and cannot change, except when ResetThreadIds is called..
 372    *
 373    * @param id the id of the thread that has called the calling method first.
 374    */
 375 #ifdef DEBUG
 376   static void AssertCorrectThread(std::thread::id& aId) {
 377     if (aId == std::thread::id()) {
 378       aId = std::this_thread::get_id();
 379       return;
 380     }
 381     MOZ_ASSERT(aId == std::this_thread::get_id());
 382   }
 383 #endif
 384   /** Index at which the oldest element is. */
 385   std::atomic<int> mReadIndex;
 386   /** Index at which to write new elements. `mWriteIndex` is always at
 387    * least one element ahead of `mReadIndex`. */
 388   std::atomic<int> mWriteIndex;
 389   /** Maximum number of elements that can be stored in the ring buffer. */
 390   const int mCapacity;
 391   /** Data storage, of size `mCapacity + 1` */
 392   std::unique_ptr<T[]> mData;
 393 #ifdef DEBUG
 394   /** The id of the only thread that is allowed to read from the queue. */
 395   mutable std::thread::id mConsumerId;
 396   /** The id of the only thread that is allowed to write from the queue. */
 397   mutable std::thread::id mProducerId;
 398 #endif
 399 };
 400
 401 /**
 402  * Instantiation of the `SPSCRingBufferBase` type. This is safe to use
 403  * from two threads, one producer, one consumer (that never change role),
 404  * without explicit synchronization.
 405  */
 406 template <typename T>
 407 using SPSCQueue = SPSCRingBufferBase<T>;
 408
 409 }  // namespace mozilla
 410
 411 #endif  // mozilla_LockFreeQueue_h