concurrentqueue/benchmarks/tbb/concurrent_priority_queue.h

/*
    Copyright 2005-2014 Intel Corporation.  All Rights Reserved.

    This file is part of Threading Building Blocks. Threading Building Blocks is free software;
    you can redistribute it and/or modify it under the terms of the GNU General Public License
    version 2  as  published  by  the  Free Software Foundation.  Threading Building Blocks is
    distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the
    implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
    See  the GNU General Public License for more details.   You should have received a copy of
    the  GNU General Public License along with Threading Building Blocks; if not, write to the
    Free Software Foundation, Inc.,  51 Franklin St,  Fifth Floor,  Boston,  MA 02110-1301 USA

    As a special exception,  you may use this file  as part of a free software library without
    restriction.  Specifically,  if other files instantiate templates  or use macros or inline
    functions from this file, or you compile this file and link it with other files to produce
    an executable,  this file does not by itself cause the resulting executable to be covered
    by the GNU General Public License. This exception does not however invalidate any other
    reasons why the executable file might be covered by the GNU General Public License.
*/

#ifndef __TBB_concurrent_priority_queue_H
#define __TBB_concurrent_priority_queue_H

#include "atomic.h"
#include "cache_aligned_allocator.h"
#include "tbb_exception.h"
#include "tbb_stddef.h"
#include "tbb_profiling.h"
#include "internal/_aggregator_impl.h"
#include <vector>
#include <iterator>
#include <functional>

#if __TBB_INITIALIZER_LISTS_PRESENT
    #include <initializer_list>
#endif

namespace tbb {
namespace interface5 {

using namespace tbb::internal;

//! Concurrent priority queue
template <typename T, typename Compare=std::less<T>, typename A=cache_aligned_allocator<T> >
class concurrent_priority_queue {
 public:
    //! Element type in the queue.
    typedef T value_type;

    //! Reference type
    typedef T& reference;

    //! Const reference type
    typedef const T& const_reference;

    //! Integral type for representing size of the queue.
    typedef size_t size_type;

    //! Difference type for iterator
    typedef ptrdiff_t difference_type;

    //! Allocator type
    typedef A allocator_type;

    //! Constructs a new concurrent_priority_queue with default capacity
    explicit concurrent_priority_queue(const allocator_type& a = allocator_type()) : mark(0), my_size(0), data(a)
    {
        my_aggregator.initialize_handler(my_functor_t(this));
    }

    //! Constructs a new concurrent_priority_queue with init_sz capacity
    explicit concurrent_priority_queue(size_type init_capacity, const allocator_type& a = allocator_type()) :
        mark(0), my_size(0), data(a)
    {
        data.reserve(init_capacity);
        my_aggregator.initialize_handler(my_functor_t(this));
    }

    //! [begin,end) constructor
    template<typename InputIterator>
    concurrent_priority_queue(InputIterator begin, InputIterator end, const allocator_type& a = allocator_type()) :
        mark(0), data(begin, end, a)
    {
        my_aggregator.initialize_handler(my_functor_t(this));
        heapify();
        my_size = data.size();
    }

#if __TBB_INITIALIZER_LISTS_PRESENT
    //! Constructor from std::initializer_list
    concurrent_priority_queue(std::initializer_list<T> init_list, const allocator_type &a = allocator_type()) :
        mark(0),data(init_list.begin(), init_list.end(), a)
    {
        my_aggregator.initialize_handler(my_functor_t(this));
        heapify();
        my_size = data.size();
    }
#endif //# __TBB_INITIALIZER_LISTS_PRESENT

    //! Copy constructor
    /** This operation is unsafe if there are pending concurrent operations on the src queue. */
    explicit concurrent_priority_queue(const concurrent_priority_queue& src) : mark(src.mark),
        my_size(src.my_size), data(src.data.begin(), src.data.end(), src.data.get_allocator())
    {
        my_aggregator.initialize_handler(my_functor_t(this));
        heapify();
    }

    //! Copy constructor with specific allocator
    /** This operation is unsafe if there are pending concurrent operations on the src queue. */
    concurrent_priority_queue(const concurrent_priority_queue& src, const allocator_type& a) : mark(src.mark),
        my_size(src.my_size), data(src.data.begin(), src.data.end(), a)
    {
        my_aggregator.initialize_handler(my_functor_t(this));
        heapify();
    }

    //! Assignment operator
    /** This operation is unsafe if there are pending concurrent operations on the src queue. */
    concurrent_priority_queue& operator=(const concurrent_priority_queue& src) {
        if (this != &src) {
            vector_t(src.data.begin(), src.data.end(), src.data.get_allocator()).swap(data);
            mark = src.mark;
            my_size = src.my_size;
        }
        return *this;
    }

#if __TBB_CPP11_RVALUE_REF_PRESENT
    //! Move constructor
    /** This operation is unsafe if there are pending concurrent operations on the src queue. */
    concurrent_priority_queue(concurrent_priority_queue&& src) : mark(src.mark),
        my_size(src.my_size), data(std::move(src.data))
    {
        my_aggregator.initialize_handler(my_functor_t(this));
    }

    //! Move constructor with specific allocator
    /** This operation is unsafe if there are pending concurrent operations on the src queue. */
    concurrent_priority_queue(concurrent_priority_queue&& src, const allocator_type& a) : mark(src.mark),
        my_size(src.my_size),
#if __TBB_ALLOCATOR_TRAITS_PRESENT
        data(std::move(src.data), a)
#else
    // Some early version of C++11 STL vector does not have a constructor of vector(vector&& , allocator).
    // It seems that the reason is absence of support of allocator_traits (stateful allocators).
        data(a)
#endif //__TBB_ALLOCATOR_TRAITS_PRESENT
    {
        my_aggregator.initialize_handler(my_functor_t(this));
#if !__TBB_ALLOCATOR_TRAITS_PRESENT
        if (a != src.data.get_allocator()){
            data.reserve(src.data.size());
            data.assign(std::make_move_iterator(src.data.begin()), std::make_move_iterator(src.data.end()));
        }else{
            data = std::move(src.data);
        }
#endif //!__TBB_ALLOCATOR_TRAITS_PRESENT
    }

    //! Move assignment operator
    /** This operation is unsafe if there are pending concurrent operations on the src queue. */
    concurrent_priority_queue& operator=( concurrent_priority_queue&& src) {
        if (this != &src) {
            mark = src.mark;
            my_size = src.my_size;
#if !__TBB_ALLOCATOR_TRAITS_PRESENT
            if (data.get_allocator() != src.data.get_allocator()){
                vector_t(std::make_move_iterator(src.data.begin()), std::make_move_iterator(src.data.end()), data.get_allocator()).swap(data);
            }else
#endif //!__TBB_ALLOCATOR_TRAITS_PRESENT
            {
                data = std::move(src.data);
            }
        }
        return *this;
    }
#endif //__TBB_CPP11_RVALUE_REF_PRESENT

    //! Assign the queue from [begin,end) range, not thread-safe
    template<typename InputIterator>
    void assign(InputIterator begin, InputIterator end) {
        vector_t(begin, end, data.get_allocator()).swap(data);
        mark = 0;
        my_size = data.size();
        heapify();
    }

#if __TBB_INITIALIZER_LISTS_PRESENT
    //! Assign the queue from std::initializer_list, not thread-safe
    void assign(std::initializer_list<T> il) { this->assign(il.begin(), il.end()); }

    //! Assign from std::initializer_list, not thread-safe
    concurrent_priority_queue& operator=(std::initializer_list<T> il) {
        this->assign(il.begin(), il.end());
        return *this;
    }
#endif //# __TBB_INITIALIZER_LISTS_PRESENT

    //! Returns true if empty, false otherwise
    /** Returned value may not reflect results of pending operations.
        This operation reads shared data and will trigger a race condition. */
    bool empty() const { return size()==0; }

    //! Returns the current number of elements contained in the queue
    /** Returned value may not reflect results of pending operations.
        This operation reads shared data and will trigger a race condition. */
    size_type size() const { return __TBB_load_with_acquire(my_size); }

    //! Pushes elem onto the queue, increasing capacity of queue if necessary
    /** This operation can be safely used concurrently with other push, try_pop or emplace operations. */
    void push(const_reference elem) {
        cpq_operation op_data(elem, PUSH_OP);
        my_aggregator.execute(&op_data);
        if (op_data.status == FAILED) // exception thrown
            throw_exception(eid_bad_alloc);
    }

#if __TBB_CPP11_RVALUE_REF_PRESENT
    //! Pushes elem onto the queue, increasing capacity of queue if necessary
    /** This operation can be safely used concurrently with other push, try_pop or emplace operations. */
    void push(value_type &&elem) {
        cpq_operation op_data(elem, PUSH_RVALUE_OP);
        my_aggregator.execute(&op_data);
        if (op_data.status == FAILED) // exception thrown
            throw_exception(eid_bad_alloc);
    }

#if __TBB_CPP11_VARIADIC_TEMPLATES_PRESENT
    //! Constructs a new element using args as the arguments for its construction and pushes it onto the queue */
    /** This operation can be safely used concurrently with other push, try_pop or emplace operations. */
    template<typename... Args>
    void emplace(Args&&... args) {
        push(value_type(std::forward<Args>(args)...));
    }
#endif /* __TBB_CPP11_VARIADIC_TEMPLATES_PRESENT */
#endif /* __TBB_CPP11_RVALUE_REF_PRESENT */

    //! Gets a reference to and removes highest priority element
    /** If a highest priority element was found, sets elem and returns true,
        otherwise returns false.
        This operation can be safely used concurrently with other push, try_pop or emplace operations. */
    bool try_pop(reference elem) {
        cpq_operation op_data(POP_OP);
        op_data.elem = &elem;
        my_aggregator.execute(&op_data);
        return op_data.status==SUCCEEDED;
    }

    //! Clear the queue; not thread-safe
    /** This operation is unsafe if there are pending concurrent operations on the queue.
        Resets size, effectively emptying queue; does not free space.
        May not clear elements added in pending operations. */
    void clear() {
        data.clear();
        mark = 0;
        my_size = 0;
    }

    //! Swap this queue with another; not thread-safe
    /** This operation is unsafe if there are pending concurrent operations on the queue. */
    void swap(concurrent_priority_queue& q) {
        using std::swap;
        data.swap(q.data);
        swap(mark, q.mark);
        swap(my_size, q.my_size);
    }

    //! Return allocator object
    allocator_type get_allocator() const { return data.get_allocator(); }

 private:
    enum operation_type {INVALID_OP, PUSH_OP, POP_OP, PUSH_RVALUE_OP};
    enum operation_status { WAIT=0, SUCCEEDED, FAILED };

    class cpq_operation : public aggregated_operation<cpq_operation> {
     public:
        operation_type type;
        union {
            value_type *elem;
            size_type sz;
        };
        cpq_operation(const_reference e, operation_type t) :
            type(t), elem(const_cast<value_type*>(&e)) {}
        cpq_operation(operation_type t) : type(t) {}
    };

    class my_functor_t {
        concurrent_priority_queue<T, Compare, A> *cpq;
     public:
        my_functor_t() {}
        my_functor_t(concurrent_priority_queue<T, Compare, A> *cpq_) : cpq(cpq_) {}
        void operator()(cpq_operation* op_list) {
            cpq->handle_operations(op_list);
        }
    };

    typedef tbb::internal::aggregator< my_functor_t, cpq_operation > aggregator_t;
    aggregator_t my_aggregator;
    //! Padding added to avoid false sharing
    char padding1[NFS_MaxLineSize - sizeof(aggregator_t)];
    //! The point at which unsorted elements begin
    size_type mark;
    __TBB_atomic size_type my_size;
    Compare compare;
    //! Padding added to avoid false sharing
    char padding2[NFS_MaxLineSize - (2*sizeof(size_type)) - sizeof(Compare)];
    //! Storage for the heap of elements in queue, plus unheapified elements
    /** data has the following structure:

         binary unheapified
          heap   elements
        ____|_______|____
        |       |       |
        v       v       v
        [_|...|_|_|...|_| |...| ]
         0       ^       ^       ^
                 |       |       |__capacity
                 |       |__my_size
                 |__mark

        Thus, data stores the binary heap starting at position 0 through
        mark-1 (it may be empty).  Then there are 0 or more elements
        that have not yet been inserted into the heap, in positions
        mark through my_size-1. */
    typedef std::vector<value_type, allocator_type> vector_t;
    vector_t data;

    void handle_operations(cpq_operation *op_list) {
        cpq_operation *tmp, *pop_list=NULL;

        __TBB_ASSERT(mark == data.size(), NULL);

        // First pass processes all constant (amortized; reallocation may happen) time pushes and pops.
        while (op_list) {
            // ITT note: &(op_list->status) tag is used to cover accesses to op_list
            // node. This thread is going to handle the operation, and so will acquire it
            // and perform the associated operation w/o triggering a race condition; the
            // thread that created the operation is waiting on the status field, so when
            // this thread is done with the operation, it will perform a
            // store_with_release to give control back to the waiting thread in
            // aggregator::insert_operation.
            call_itt_notify(acquired, &(op_list->status));
            __TBB_ASSERT(op_list->type != INVALID_OP, NULL);
            tmp = op_list;
            op_list = itt_hide_load_word(op_list->next);
            if (tmp->type == POP_OP) {
                if (mark < data.size() &&
                    compare(data[0], data[data.size()-1])) {
                    // there are newly pushed elems and the last one
                    // is higher than top
                    *(tmp->elem) = move(data[data.size()-1]);
                    __TBB_store_with_release(my_size, my_size-1);
                    itt_store_word_with_release(tmp->status, uintptr_t(SUCCEEDED));
                    data.pop_back();
                    __TBB_ASSERT(mark<=data.size(), NULL);
                }
                else { // no convenient item to pop; postpone
                    itt_hide_store_word(tmp->next, pop_list);
                    pop_list = tmp;
                }
            } else { // PUSH_OP or PUSH_RVALUE_OP
                __TBB_ASSERT(tmp->type == PUSH_OP || tmp->type == PUSH_RVALUE_OP, "Unknown operation" );
                __TBB_TRY{
                    if (tmp->type == PUSH_OP) {
                        data.push_back(*(tmp->elem));
                    } else {
                        data.push_back(move(*(tmp->elem)));
                    }
                    __TBB_store_with_release(my_size, my_size + 1);
                    itt_store_word_with_release(tmp->status, uintptr_t(SUCCEEDED));
                } __TBB_CATCH(...) {
                    itt_store_word_with_release(tmp->status, uintptr_t(FAILED));
                }
            }
        }

        // second pass processes pop operations
        while (pop_list) {
            tmp = pop_list;
            pop_list = itt_hide_load_word(pop_list->next);
            __TBB_ASSERT(tmp->type == POP_OP, NULL);
            if (data.empty()) {
                itt_store_word_with_release(tmp->status, uintptr_t(FAILED));
            }
            else {
                __TBB_ASSERT(mark<=data.size(), NULL);
                if (mark < data.size() &&
                    compare(data[0], data[data.size()-1])) {
                    // there are newly pushed elems and the last one is
                    // higher than top
                    *(tmp->elem) = move(data[data.size()-1]);
                    __TBB_store_with_release(my_size, my_size-1);
                    itt_store_word_with_release(tmp->status, uintptr_t(SUCCEEDED));
                    data.pop_back();
                }
                else { // extract top and push last element down heap
                    *(tmp->elem) = move(data[0]);
                    __TBB_store_with_release(my_size, my_size-1);
                    itt_store_word_with_release(tmp->status, uintptr_t(SUCCEEDED));
                    reheap();
                }
            }
        }

        // heapify any leftover pushed elements before doing the next
        // batch of operations
        if (mark<data.size()) heapify();
        __TBB_ASSERT(mark == data.size(), NULL);
    }

    //! Merge unsorted elements into heap
    void heapify() {
        if (!mark && data.size()>0) mark = 1;
        for (; mark<data.size(); ++mark) {
            // for each unheapified element under size
            size_type cur_pos = mark;
            value_type to_place = move(data[mark]);
            do { // push to_place up the heap
                size_type parent = (cur_pos-1)>>1;
                if (!compare(data[parent], to_place)) break;
                data[cur_pos] = move(data[parent]);
                cur_pos = parent;
            } while( cur_pos );
            data[cur_pos] = move(to_place);
        }
    }

    //! Re-heapify after an extraction
    /** Re-heapify by pushing last element down the heap from the root. */
    void reheap() {
        size_type cur_pos=0, child=1;

        while (child < mark) {
            size_type target = child;
            if (child+1 < mark && compare(data[child], data[child+1]))
                ++target;
            // target now has the higher priority child
            if (compare(data[target], data[data.size()-1])) break;
            data[cur_pos] = move(data[target]);
            cur_pos = target;
            child = (cur_pos<<1)+1;
        }
        if (cur_pos != data.size()-1)
            data[cur_pos] = move(data[data.size()-1]);
        data.pop_back();
        if (mark > data.size()) mark = data.size();
    }
};

} // namespace interface5

using interface5::concurrent_priority_queue;

} // namespace tbb

#endif /* __TBB_concurrent_priority_queue_H */