readerwriterqueue.h

Go to the documentation of this file.

// ©2013-2016 Cameron Desrochers.
// Distributed under the simplified BSD license (see the license file that
// should have come with this header).
 
#pragma once
 
#include "atomicops.h"
#include <type_traits>
#include <utility>
#include <cassert>
#include <stdexcept>
#include <new>
#include <cstdint>
#include <cstdlib>    // For malloc/free/abort & size_t
#include <memory>
#if __cplusplus > 199711L || _MSC_VER >= 1700 // C++11 or VS2012
#include <chrono>
#endif
 
 
// A lock-free queue for a single-consumer, single-producer architecture.
// The queue is also wait-free in the common path (except if more memory
// needs to be allocated, in which case malloc is called).
// Allocates memory sparingly (O(lg(n) times, amortized), and only once if
// the original maximum size estimate is never exceeded.
// Tested on x86/x64 processors, but semantics should be correct for all
// architectures (given the right implementations in atomicops.h), provided
// that aligned integer and pointer accesses are naturally atomic.
// Note that there should only be one consumer thread and producer thread;
// Switching roles of the threads, or using multiple consecutive threads for
// one role, is not safe unless properly synchronized.
// Using the queue exclusively from one thread is fine, though a bit silly.
 
#ifndef MOODYCAMEL_CACHE_LINE_SIZE
#define MOODYCAMEL_CACHE_LINE_SIZE 64
#endif
 
#ifndef MOODYCAMEL_HAS_EMPLACE
#if !defined(_MSC_VER) || _MSC_VER >= 1800 // variadic templates: either a non-MS compiler or VS >= 2013
#define MOODYCAMEL_HAS_EMPLACE    1
#endif
#endif
 
#ifdef AE_VCPP
#pragma warning(push)
#pragma warning(disable: 4324)  // structure was padded due to __declspec(align())
#pragma warning(disable: 4820)  // padding was added
#pragma warning(disable: 4127)  // conditional expression is constant
#endif
 
#define MAX_BLOCK_SIZE 512
class ReaderWriterQueue
{
  // Design: Based on a queue-of-queues. The low-level queues are just
  // circular buffers with front and tail indices indicating where the
  // next element to dequeue is and where the next element can be enqueued,
  // respectively. Each low-level queue is called a "block". Each block
  // wastes exactly one element's worth of space to keep the design simple
  // (if front == tail then the queue is empty, and can't be full).
  // The high-level queue is a circular linked list of blocks; again there
  // is a front and tail, but this time they are pointers to the blocks.
  // The front block is where the next element to be dequeued is, provided
  // the block is not empty. The back block is where elements are to be
  // enqueued, provided the block is not full.
  // The producer thread owns all the tail indices/pointers. The consumer
  // thread owns all the front indices/pointers. Both threads read each
  // other's variables, but only the owning thread updates them. E.g. After
  // the consumer reads the producer's tail, the tail may change before the
  // consumer is done dequeuing an object, but the consumer knows the tail
  // will never go backwards, only forwards.
  // If there is no room to enqueue an object, an additional block (of
  // equal size to the last block) is added. Blocks are never removed.
 
public:
  UINT8* value_type;
 
  // Constructs a queue that can hold maxSize elements without further
  // allocations. If more than MAX_BLOCK_SIZE elements are requested,
  // then several blocks of MAX_BLOCK_SIZE each are reserved (including
  // at least one extra buffer block).
  AE_NO_TSAN explicit ReaderWriterQueue(size_t maxSize = 15)
  {
    assert(maxSize > 0);
    assert(MAX_BLOCK_SIZE == ceilToPow2(MAX_BLOCK_SIZE) && "MAX_BLOCK_SIZE must be a power of 2");
    assert(MAX_BLOCK_SIZE >= 2 && "MAX_BLOCK_SIZE must be at least 2");
    
    Block* firstBlock = nullptr;
    
    largestBlockSize = ceilToPow2(maxSize + 1);   // We need a spare slot to fit maxSize elements in the block
    if (largestBlockSize > MAX_BLOCK_SIZE * 2) {
      // We need a spare block in case the producer is writing to a different block the consumer is reading from, and
      // wants to enqueue the maximum number of elements. We also need a spare element in each block to avoid the ambiguity
      // between front == tail meaning "empty" and "full".
      // So the effective number of slots that are guaranteed to be usable at any time is the block size - 1 times the
      // number of blocks - 1. Solving for maxSize and applying a ceiling to the division gives us (after simplifying):
      size_t initialBlockCount = (maxSize + MAX_BLOCK_SIZE * 2 - 3) / (MAX_BLOCK_SIZE - 1);
      largestBlockSize = MAX_BLOCK_SIZE;
      Block* lastBlock = nullptr;
      for (size_t i = 0; i != initialBlockCount; ++i) {
        Block* block = make_block(largestBlockSize);
        if (block == nullptr) {
          abort();
        }
        if (firstBlock == nullptr) {
          firstBlock = block;
        }
        else {
          lastBlock->next = block;
        }
        lastBlock = block;
        block->next = firstBlock;
      }
    }
    else {
      firstBlock = make_block(largestBlockSize);
      if (firstBlock == nullptr) {
        abort();
      }
      firstBlock->next = firstBlock;
    }
    frontBlock = firstBlock;
    tailBlock = firstBlock;
    
    // Make sure the reader/writer threads will have the initialized memory setup above:
    fence(memory_order_sync);
  }
 
 
 
  // Note: The queue should not be accessed concurrently while it's
  // being deleted. It's up to the user to synchronize this.
  AE_NO_TSAN ~ReaderWriterQueue()
  {
    // Make sure we get the latest version of all variables from other CPUs:
    fence(memory_order_sync);
 
    // Destroy any remaining objects in queue and free memory
    Block* frontBlock_ = frontBlock;
    Block* block = frontBlock_;
    do {
      Block* nextBlock = block->next;
      UINT8* rawBlock = block->rawThis;
      block->~Block();
      std::free(rawBlock);
      block = nextBlock;
    } while (block != frontBlock_);
  }
 
 
 
 
 
  // Enqueues a copy of element on the queue.
  // Allocates an additional block of memory if needed.
  // Only fails (returns false) if memory allocation fails.
  AE_FORCEINLINE bool enqueue(UINT8* const& element) AE_NO_TSAN
  {
    return inner_enqueue(CanAlloc,element);
  }
 
 
  // Attempts to dequeue an element; if the queue is empty,
  // returns false instead. If the queue has at least one element,
  // moves front to result using operator=, then returns true.
  bool try_dequeue(UINT8*& result) AE_NO_TSAN
  {
 
    // High-level pseudocode:
    // Remember where the tail block is
    // If the front block has an element in it, dequeue it
    // Else
    //     If front block was the tail block when we entered the function, return false
    //     Else advance to next block and dequeue the item there
 
    // Note that we have to use the value of the tail block from before we check if the front
    // block is full or not, in case the front block is empty and then, before we check if the
    // tail block is at the front block or not, the producer fills up the front block *and
    // moves on*, which would make us skip a filled block. Seems unlikely, but was consistently
    // reproducible in practice.
    // In order to avoid overhead in the common case, though, we do a double-checked pattern
    // where we have the fast path if the front block is not empty, then read the tail block,
    // then re-read the front block and check if it's not empty again, then check if the tail
    // block has advanced.
    
    Block* frontBlock_ = frontBlock.load();
    size_t blockTail = frontBlock_->localTail;
    size_t blockFront = frontBlock_->front.load();
    
    if (blockFront != blockTail || blockFront != (frontBlock_->localTail = frontBlock_->tail.load())) {
      fence(memory_order_acquire);
      
    non_empty_front_block:
      // Front block not empty, dequeue from here
      result = frontBlock_->data[blockFront];
      //element->~T();
 
      blockFront = (blockFront + 1) & frontBlock_->sizeMask;
 
      fence(memory_order_release);
      frontBlock_->front = blockFront;
    }
    else if (frontBlock_ != tailBlock.load()) {
      fence(memory_order_acquire);
 
      frontBlock_ = frontBlock.load();
      blockTail = frontBlock_->localTail = frontBlock_->tail.load();
      blockFront = frontBlock_->front.load();
      fence(memory_order_acquire);
      
      if (blockFront != blockTail) {
        // Oh look, the front block isn't empty after all
        goto non_empty_front_block;
      }
      
      // Front block is empty but there's another block ahead, advance to it
      Block* nextBlock = frontBlock_->next;
      // Don't need an acquire fence here since next can only ever be set on the tailBlock,
      // and we're not the tailBlock, and we did an acquire earlier after reading tailBlock which
      // ensures next is up-to-date on this CPU in case we recently were at tailBlock.
 
      size_t nextBlockFront = nextBlock->front.load();
      size_t nextBlockTail = nextBlock->localTail = nextBlock->tail.load();
      fence(memory_order_acquire);
 
      // Since the tailBlock is only ever advanced after being written to,
      // we know there's for sure an element to dequeue on it
      assert(nextBlockFront != nextBlockTail);
      AE_UNUSED(nextBlockTail);
 
      // We're done with this block, let the producer use it if it needs
      fence(memory_order_release);    // Expose possibly pending changes to frontBlock->front from last dequeue
      frontBlock = frontBlock_ = nextBlock;
 
      compiler_fence(memory_order_release); // Not strictly needed
 
      result = frontBlock_->data [nextBlockFront];
      
      //element->~T();
 
      nextBlockFront = (nextBlockFront + 1) & frontBlock_->sizeMask;
      
      fence(memory_order_release);
      frontBlock_->front = nextBlockFront;
    }
    else {
      // No elements in current block and no other block to advance to
      return false;
    }
 
    return true;
  }
 
 
  // Returns a pointer to the front element in the queue (the one that
  // would be removed next by a call to `try_dequeue` or `pop`). If the
  // queue appears empty at the time the method is called, nullptr is
  // returned instead.
  // Must be called only from the consumer thread.
  UINT8* peek() AE_NO_TSAN
  {
    // See try_dequeue() for reasoning
 
    Block* frontBlock_ = frontBlock.load();
    size_t blockTail = frontBlock_->localTail;
    size_t blockFront = frontBlock_->front.load();
    
    if (blockFront != blockTail || blockFront != (frontBlock_->localTail = frontBlock_->tail.load())) {
      fence(memory_order_acquire);
    non_empty_front_block:
      return reinterpret_cast<UINT8*>(frontBlock_->data + blockFront * sizeof(UINT8*));
    }
    else if (frontBlock_ != tailBlock.load()) {
      fence(memory_order_acquire);
      frontBlock_ = frontBlock.load();
      blockTail = frontBlock_->localTail = frontBlock_->tail.load();
      blockFront = frontBlock_->front.load();
      fence(memory_order_acquire);
      
      if (blockFront != blockTail) {
        goto non_empty_front_block;
      }
      
      Block* nextBlock = frontBlock_->next;
      
      size_t nextBlockFront = nextBlock->front.load();
      fence(memory_order_acquire);
 
      assert(nextBlockFront != nextBlock->tail.load());
      return reinterpret_cast<UINT8*>(nextBlock->data + nextBlockFront * sizeof(UINT8*));
    }
    
    return nullptr;
  }
  
  // Removes the front element from the queue, if any, without returning it.
  // Returns true on success, or false if the queue appeared empty at the time
  // `pop` was called.
  bool pop() AE_NO_TSAN
  {
    // See try_dequeue() for reasoning
    
    Block* frontBlock_ = frontBlock.load();
    size_t blockTail = frontBlock_->localTail;
    size_t blockFront = frontBlock_->front.load();
    
    if (blockFront != blockTail || blockFront != (frontBlock_->localTail = frontBlock_->tail.load())) {
      fence(memory_order_acquire);
      
    non_empty_front_block:
      auto element = reinterpret_cast<UINT8*>(frontBlock_->data + blockFront * sizeof(UINT8));
      //element->~T();
 
      blockFront = (blockFront + 1) & frontBlock_->sizeMask;
 
      fence(memory_order_release);
      frontBlock_->front = blockFront;
    }
    else if (frontBlock_ != tailBlock.load()) {
      fence(memory_order_acquire);
      frontBlock_ = frontBlock.load();
      blockTail = frontBlock_->localTail = frontBlock_->tail.load();
      blockFront = frontBlock_->front.load();
      fence(memory_order_acquire);
      
      if (blockFront != blockTail) {
        goto non_empty_front_block;
      }
      
      // Front block is empty but there's another block ahead, advance to it
      Block* nextBlock = frontBlock_->next;
      
      size_t nextBlockFront = nextBlock->front.load();
      size_t nextBlockTail = nextBlock->localTail = nextBlock->tail.load();
      fence(memory_order_acquire);
 
      assert(nextBlockFront != nextBlockTail);
      AE_UNUSED(nextBlockTail);
 
      fence(memory_order_release);
      frontBlock = frontBlock_ = nextBlock;
 
      compiler_fence(memory_order_release);
 
      auto element = reinterpret_cast<UINT8*>(frontBlock_->data + nextBlockFront * sizeof(UINT8*));
//      element->~T();
 
      nextBlockFront = (nextBlockFront + 1) & frontBlock_->sizeMask;
      
      fence(memory_order_release);
      frontBlock_->front = nextBlockFront;
    }
    else {
      // No elements in current block and no other block to advance to
      return false;
    }
 
    return true;
  }
  
  // Returns the approximate number of items currently in the queue.
  // Safe to call from both the producer and consumer threads.
  inline size_t size_approx() const AE_NO_TSAN
  {
    size_t result = 0;
    Block* frontBlock_ = frontBlock.load();
    Block* block = frontBlock_;
    do {
      fence(memory_order_acquire);
      size_t blockFront = block->front.load();
      size_t blockTail = block->tail.load();
      result += (blockTail - blockFront) & block->sizeMask;
      block = block->next.load();
    } while (block != frontBlock_);
    return result;
  }
 
 
private:
  enum AllocationMode { CanAlloc, CannotAlloc };
 
  bool inner_enqueue(AllocationMode canAlloc,UINT8* const& element) AE_NO_TSAN
  {
 
    // High-level pseudocode (assuming we're allowed to alloc a new block):
    // If room in tail block, add to tail
    // Else check next block
    //     If next block is not the head block, enqueue on next block
    //     Else create a new block and enqueue there
    //     Advance tail to the block we just enqueued to
 
    Block* tailBlock_ = tailBlock.load();
    size_t blockFront = tailBlock_->localFront;
    size_t blockTail = tailBlock_->tail.load();
 
    size_t nextBlockTail = (blockTail + 1) & tailBlock_->sizeMask;
    if (nextBlockTail != blockFront || nextBlockTail != (tailBlock_->localFront = tailBlock_->front.load())) {
      fence(memory_order_acquire);
      // This block has room for at least one more element
      tailBlock_->data[blockTail]= element;
 
 
      fence(memory_order_release);
      tailBlock_->tail = nextBlockTail;
    }
    else {
      fence(memory_order_acquire);
      if (tailBlock_->next.load() != frontBlock) {
        // Note that the reason we can't advance to the frontBlock and start adding new entries there
        // is because if we did, then dequeue would stay in that block, eventually reading the new values,
        // instead of advancing to the next full block (whose values were enqueued first and so should be
        // consumed first).
 
        fence(memory_order_acquire);    // Ensure we get latest writes if we got the latest frontBlock
 
        // tailBlock is full, but there's a free block ahead, use it
        Block* tailBlockNext = tailBlock_->next.load();
        size_t nextBlockFront = tailBlockNext->localFront = tailBlockNext->front.load();
        nextBlockTail = tailBlockNext->tail.load();
        fence(memory_order_acquire);
 
        // This block must be empty since it's not the head block and we
        // go through the blocks in a circle
        assert(nextBlockFront == nextBlockTail);
        tailBlockNext->localFront = nextBlockFront;
 
        tailBlockNext->data[nextBlockTail] = element;
 
        tailBlockNext->tail = (nextBlockTail + 1) & tailBlockNext->sizeMask;
 
        fence(memory_order_release);
        tailBlock = tailBlockNext;
      }
      else if (canAlloc == CanAlloc) {
        // tailBlock is full and there's no free block ahead; create a new block
        auto newBlockSize = largestBlockSize >= MAX_BLOCK_SIZE ? largestBlockSize : largestBlockSize * 2;
        auto newBlock = make_block(newBlockSize);
        if (newBlock == nullptr) {
          // Could not allocate a block!
          return false;
        }
        largestBlockSize = newBlockSize;
 
 
        newBlock->data[0] = element;
        assert(newBlock->front == 0);
        newBlock->tail = newBlock->localTail = 1;
 
        newBlock->next = tailBlock_->next.load();
        tailBlock_->next = newBlock;
 
        // Might be possible for the dequeue thread to see the new tailBlock->next
        // *without* seeing the new tailBlock value, but this is OK since it can't
        // advance to the next block until tailBlock is set anyway (because the only
        // case where it could try to read the next is if it's already at the tailBlock,
        // and it won't advance past tailBlock in any circumstance).
 
        fence(memory_order_release);
        tailBlock = newBlock;
      }
      else if (canAlloc == CannotAlloc) {
        // Would have had to allocate a new block to enqueue, but not allowed
        return false;
      }
      else {
        assert(false && "Should be unreachable code");
        return false;
      }
    }
 
    return true;
  }
 
 
 
  AE_FORCEINLINE static size_t ceilToPow2(size_t x)
  {
    // From http://graphics.stanford.edu/~seander/bithacks.html#RoundUpPowerOf2
    --x;
    x |= x >> 1;
    x |= x >> 2;
    x |= x >> 4;
    for (size_t i = 1; i < sizeof(size_t); i <<= 1) {
      x |= x >> (i << 3);
    }
    ++x;
    return x;
  }
  
  template<typename U>
  static AE_FORCEINLINE UINT8* align_for(UINT8* ptr) AE_NO_TSAN
  {
    const std::size_t alignment = std::alignment_of<U>::value;
    return ptr + (alignment - (reinterpret_cast<std::uintptr_t>(ptr) % alignment)) % alignment;
  }
 
private:
 
 
  struct Block
  {
    // Avoid false-sharing by putting highly contended variables on their own cache lines
    weak_atomic<size_t> front;  // (Atomic) Elements are read from here
    size_t localTail;     // An uncontended shadow copy of tail, owned by the consumer
    
    char cachelineFiller0[MOODYCAMEL_CACHE_LINE_SIZE - sizeof(weak_atomic<size_t>) - sizeof(size_t)];
    weak_atomic<size_t> tail; // (Atomic) Elements are enqueued here
    size_t localFront;
    
    char cachelineFiller1[MOODYCAMEL_CACHE_LINE_SIZE - sizeof(weak_atomic<size_t>) - sizeof(size_t)]; // next isn't very contended, but we don't want it on the same cache line as tail (which is)
    weak_atomic<Block*> next; // (Atomic)
    
    UINT8** data;   // Contents (on heap) are aligned to T's alignment
 
    const size_t sizeMask;
 
 
    // size must be a power of two (and greater than 0)
    AE_NO_TSAN Block(size_t const& _size, UINT8* _rawThis, UINT8* _data)
      : front(0), localTail(0), tail(0), localFront(0), next(nullptr), data((UINT8**)_data), sizeMask(_size - 1), rawThis(_rawThis)
    {
    }
 
  private:
    // C4512 - Assignment operator could not be generated
    Block& operator=(Block const&);
 
  public:
    UINT8* rawThis;
  };
  
  
  static Block* make_block(size_t capacity) AE_NO_TSAN
  {
    // Allocate enough memory for the block itself, as well as all the elements it will contain
    UINT size = sizeof(Block) + std::alignment_of<Block>::value - 1;
    size += sizeof(UINT8*) * capacity + std::alignment_of<UINT8*>::value - 1;
    UINT8* newBlockRaw = static_cast<UINT8*>(std::malloc(size));
    if (newBlockRaw == nullptr) {
      return nullptr;
    }
    
    UINT8* newBlockAligned = align_for<Block>(newBlockRaw);
    UINT8* newBlockData = align_for<UINT8*>(newBlockAligned + sizeof(Block));
    return new (newBlockAligned) Block(capacity, newBlockRaw, newBlockData);
  }
 
private:
  weak_atomic<Block*> frontBlock;   // (Atomic) Elements are enqueued to this block
  
  char cachelineFiller[MOODYCAMEL_CACHE_LINE_SIZE - sizeof(weak_atomic<Block*>)];
  weak_atomic<Block*> tailBlock;    // (Atomic) Elements are dequeued from this block
 
  size_t largestBlockSize;
 
};
 
// Like ReaderWriterQueue, but also providees blocking operations
class BlockingReaderWriterQueue
{
  
public:
  explicit BlockingReaderWriterQueue(size_t maxSize = 15) AE_NO_TSAN
    : inner(maxSize), sema(new spsc_sema::LightweightSemaphore())
  { }
 
  BlockingReaderWriterQueue(BlockingReaderWriterQueue&& other) AE_NO_TSAN
    : inner(std::move(other.inner)), sema(std::move(other.sema))
  { }
 
  BlockingReaderWriterQueue& operator=(BlockingReaderWriterQueue&& other) AE_NO_TSAN
  {
    std::swap(sema, other.sema);
    std::swap(inner, other.inner);
    return *this;
  }
 
  // Enqueues a copy of element on the queue.
  // Allocates an additional block of memory if needed.
  // Only fails (returns false) if memory allocation fails.
  AE_FORCEINLINE bool enqueue(UINT8* const& element) AE_NO_TSAN
  {
    if (inner.enqueue(element)) {
      sema->signal();
      return true;
    }
    return false;
  }
 
 
 
  // Attempts to dequeue an element; if the queue is empty,
  // returns false instead. If the queue has at least one element,
  // moves front to result using operator=, then returns true.
  bool try_dequeue(UINT8*& result) AE_NO_TSAN
  {
    if (sema->tryWait()) {
      bool success = inner.try_dequeue(result);
      assert(success);
      AE_UNUSED(success);
      return true;
    }
    return false;
  }
  
  
  // Attempts to dequeue an element; if the queue is empty,
  // waits until an element is available, then dequeues it.
  void wait_dequeue(UINT8*& result) AE_NO_TSAN
  {
    sema->wait();
    bool success = inner.try_dequeue(result);
    AE_UNUSED(result);
    assert(success);
    AE_UNUSED(success);
  }
 
 
  // Attempts to dequeue an element; if the queue is empty,
  // waits until an element is available up to the specified timeout,
  // then dequeues it and returns true, or returns false if the timeout
  // expires before an element can be dequeued.
  // Using a negative timeout indicates an indefinite timeout,
  // and is thus functionally equivalent to calling wait_dequeue.
  bool wait_dequeue_timed(UINT8*& result, std::int64_t timeout_usecs) AE_NO_TSAN
  {
    if (!sema->wait(timeout_usecs)) {
      return false;
    }
    bool success = inner.try_dequeue(result);
    AE_UNUSED(result);
    assert(success);
    AE_UNUSED(success);
    return true;
  }
 
 
#if __cplusplus > 199711L || _MSC_VER >= 1700
  // Attempts to dequeue an element; if the queue is empty,
  // waits until an element is available up to the specified timeout,
  // then dequeues it and returns true, or returns false if the timeout
  // expires before an element can be dequeued.
  // Using a negative timeout indicates an indefinite timeout,
  // and is thus functionally equivalent to calling wait_dequeue.
  template<typename Rep, typename Period>
  inline bool wait_dequeue_timed(UINT8*& result, std::chrono::duration<Rep, Period> const& timeout) AE_NO_TSAN
  {
        return wait_dequeue_timed(result, std::chrono::duration_cast<std::chrono::microseconds>(timeout).count());
  }
#endif
 
 
  // Returns a pointer to the front element in the queue (the one that
  // would be removed next by a call to `try_dequeue` or `pop`). If the
  // queue appears empty at the time the method is called, nullptr is
  // returned instead.
  // Must be called only from the consumer thread.
  AE_FORCEINLINE UINT8* peek() AE_NO_TSAN
  {
    return inner.peek();
  }
  
  // Removes the front element from the queue, if any, without returning it.
  // Returns true on success, or false if the queue appeared empty at the time
  // `pop` was called.
  AE_FORCEINLINE bool pop() AE_NO_TSAN
  {
    if (sema->tryWait()) {
      bool result = inner.pop();
      assert(result);
      AE_UNUSED(result);
      return true;
    }
    return false;
  }
  
  // Returns the approximate number of items currently in the queue.
  // Safe to call from both the producer and consumer threads.
  AE_FORCEINLINE size_t size_approx() const AE_NO_TSAN
  {
    return sema->availableApprox();
  }
 
 
  
private:
  ReaderWriterQueue inner;
  std::unique_ptr<spsc_sema::LightweightSemaphore> sema;
};
 
 
#ifdef AE_VCPP
#pragma warning(pop)
#endif