/*---------------------------------------------------------------------------*
  Project:  Horizon
  File:     os_InterCoreBlockingQueue.cpp

  Copyright (C)2009 Nintendo Co., Ltd.  All rights reserved.

  These coded instructions, statements, and computer programs contain
  proprietary information of Nintendo of America Inc. and/or Nintendo
  Company Ltd., and are protected by Federal copyright law. They may
  not be disclosed to third parties or copied or duplicated in any form,
  in whole or in part, without the prior written consent of Nintendo.

  $Rev: 33107 $
 *---------------------------------------------------------------------------*/

#include <nn/assert.h>
#include <nn/config.h>
#include <nn/util/util_Result.h>
#include <nn/os/os_InterCoreBlockingQueue.h>
#include <nn/os/os_InterCoreCriticalSection.h>
#include <nn/fnd/fnd_Interlocked.h>
#include <nn/util/detail/util_InitializationTransaction.h>

#include <climits>

//---------------------------------------------------------------------------

using namespace nn;
using namespace nn::fnd;
using namespace nn::svc;
using namespace nn::os;
using namespace nn::util;

namespace nn{ namespace os{

namespace detail {

template <class Locker>
InterCoreBlockingQueueBase<Locker>::~InterCoreBlockingQueueBase()
{
    NN_TASSERT_(m_WaitingEnqueueCount == 0 && m_WaitingDequeueCount == 0);
    Finalize();
}

template <class Locker>
void InterCoreBlockingQueueBase<Locker>::Initialize(uptr buffer[], size_t size)
{
    m_ppBuffer      = buffer;
    m_size          = size;
    m_firstIndex    = 0;
    m_usedCount     = 0;
    m_WaitingEnqueueCount = 0;
    m_WaitingDequeueCount = 0;
    m_EnqueueSemaphore.Initialize(0);
    m_DequeueSemaphore.Initialize(0);
    m_cs.Initialize();
}

template <class Locker>
nn::Result InterCoreBlockingQueueBase<Locker>::TryInitialize(uptr buffer[], size_t size)
{
    m_ppBuffer      = buffer;
    m_size          = size;
    m_firstIndex    = 0;
    m_usedCount     = 0;
    m_WaitingEnqueueCount = 0;
    m_WaitingDequeueCount = 0;

    m_EnqueueSemaphore.Initialize(0);
    m_DequeueSemaphore.Initialize(0);

    NN_UTIL_RETURN_IF_FAILED(m_cs.TryInitialize());
    return ResultSuccess();
}

template <class Locker>
void InterCoreBlockingQueueBase<Locker>::Finalize()
{
    m_cs.Finalize();
    m_DequeueSemaphore.Finalize();
    m_EnqueueSemaphore.Finalize();
}


template <class Locker>
inline void InterCoreBlockingQueueBase<Locker>::NotifyEnqueue() const
{
    if (m_WaitingEnqueueCount > 0)
    {
        m_EnqueueSemaphore.Release();
    }
}

template <class Locker>
inline void InterCoreBlockingQueueBase<Locker>::NotifyDequeue() const
{
    if (m_WaitingDequeueCount > 0)
    {
        m_DequeueSemaphore.Release();
    }
}

template <class Locker>
bool InterCoreBlockingQueueBase<Locker>::TryEnqueue(uptr data)
{
    // Locks while inserting data in the queue so that other threads cannot perform operations.
    ScopedLock locker(m_cs);

    if (m_size > m_usedCount)
    {
        s32 lastIndex = (m_firstIndex + m_usedCount) % m_size;
        m_ppBuffer[lastIndex] = data;
        m_usedCount++;

        // Wakes up thread waiting for data insertion.
        NotifyEnqueue();
        return true;
    }
    else
    {
        return false;
    }
}

template <class Locker>
bool InterCoreBlockingQueueBase<Locker>::ForceEnqueue(uptr data, uptr* pOut)
{
    // Locks while inserting data in the queue so that other threads cannot perform operations.
    ScopedLock locker(m_cs);
    bool bReturn;
    s32 lastIndex = (m_firstIndex + m_usedCount) % m_size;
    if (m_size > m_usedCount)
    {
        m_usedCount++;
        bReturn = true;
    }
    else
    {
        if (pOut)
        {
            *pOut = m_ppBuffer[lastIndex];
        }
        m_firstIndex = (m_firstIndex + 1) % m_size;
        bReturn = false;
    }

    m_ppBuffer[lastIndex] = data;
    // Wakes up thread waiting for data insertion.
    NotifyEnqueue();
    return bReturn;
}

template <class Locker>
void InterCoreBlockingQueueBase<Locker>::Enqueue(uptr data)
{
    // Updates the number of threads during the process to insert in queue.
    ++m_WaitingDequeueCount;
    DataSynchronizationBarrier();
    for(;;)
    {
        if (TryEnqueue(data))
        {
            break;
        }

        // Waits until there is space in the queue.
        m_DequeueSemaphore.Acquire();
    }
    --m_WaitingDequeueCount;
    DataSynchronizationBarrier();
}

template <class Locker>
bool InterCoreBlockingQueueBase<Locker>::TryJam(uptr data)
{
    // Locks while inserting data in the queue so that other threads cannot perform operations.
    ScopedLock locker(m_cs);

    if (m_size > m_usedCount)
    {
        m_firstIndex = (m_firstIndex + m_size - 1) % m_size;
        m_ppBuffer[m_firstIndex] = data;
        m_usedCount++;

        // Wakes up thread waiting for data insertion.
        NotifyEnqueue();
        return true;
    }
    else
    {
        return false;
    }
}

template <class Locker>
void InterCoreBlockingQueueBase<Locker>::Jam(uptr data)
{
    // Updates the number of threads during the process to insert in queue.
    ++m_WaitingDequeueCount;
    DataSynchronizationBarrier();
    for(;;)
    {
        if (TryJam(data))
        {
            break;
        }

        // Waits until there is space in the queue.
        m_DequeueSemaphore.Acquire();
    }
    --m_WaitingDequeueCount;
    DataSynchronizationBarrier();
}

template <class Locker>
bool InterCoreBlockingQueueBase<Locker>::TryDequeue(uptr* pOut)
{
    // Locks while retrieving data from the queue so that other threads cannot perform operations.
    ScopedLock locker(m_cs);

    if (0 < m_usedCount)
    {
        *pOut = m_ppBuffer[m_firstIndex];
        m_firstIndex = (m_firstIndex + 1) % m_size;
        m_usedCount--;

        // Wakes up the thread waiting for a space in the queue.
        NotifyDequeue();
        return true;
    }
    else
    {
        return false;
    }
}

template <class Locker>
uptr InterCoreBlockingQueueBase<Locker>::Dequeue()
{
    // Updates the number of threads during the process to retrieve data from the queue.
    ++m_WaitingEnqueueCount;
    DataSynchronizationBarrier();
    uptr data;
    for(;;)
    {
        if (TryDequeue(&data))
        {
            break;
        }

        // Waits when the queue content is empty.
        m_EnqueueSemaphore.Acquire();
    }
    --m_WaitingEnqueueCount;
    DataSynchronizationBarrier();
    return data;
}

template <class Locker>
bool InterCoreBlockingQueueBase<Locker>::TryGetFront(uptr* pOut) const
{
    // Locks while retrieving data from the queue so that other threads cannot perform operations.
    ScopedLock locker(m_cs);

    if (0 < m_usedCount)
    {
        *pOut = m_ppBuffer[m_firstIndex];

        return true;
    }
    else
    {
        return false;
    }
}

template <class Locker>
uptr InterCoreBlockingQueueBase<Locker>::GetFront() const
{
    // Updates the number of threads during the process to retrieve data from the queue.
    ++m_WaitingEnqueueCount;
    DataSynchronizationBarrier();
    uptr data;
    for(;;)
    {
        if (TryGetFront(&data))
        {
            break;
        }

        // Waits when the queue content is empty.
        m_EnqueueSemaphore.Acquire();
    }
    --m_WaitingEnqueueCount;
    DataSynchronizationBarrier();
    return data;
}

template class InterCoreBlockingQueueBase<nn::os::InterCoreCriticalSection>;

} // namespace detail

}} // namespace nn::os


#include <new>

using namespace nn::os;

extern "C" {

void nnosInterCoreBlockingQueueInitialize(nnosInterCoreBlockingQueue* this_, uptr buffer[], size_t size)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    new (pInterCoreBlockingQueue) InterCoreBlockingQueue();
    pInterCoreBlockingQueue->Initialize(buffer, size);
}

bool nnosInterCoreBlockingQueueTryInitialize(nnosInterCoreBlockingQueue* this_, uptr buffer[], size_t size)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    new (pInterCoreBlockingQueue) InterCoreBlockingQueue();
    Result result = pInterCoreBlockingQueue->TryInitialize(buffer, size);
    return result.IsSuccess();
}

void nnosInterCoreBlockingQueueFinalize(nnosInterCoreBlockingQueue* this_)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    pInterCoreBlockingQueue->Finalize();
    pInterCoreBlockingQueue->~InterCoreBlockingQueue();
}

bool nnosInterCoreBlockingQueueTryEnqueue(nnosInterCoreBlockingQueue* this_, uptr data)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    return pInterCoreBlockingQueue->TryEnqueue(data);
}

void nnosInterCoreBlockingQueueEnqueue(nnosInterCoreBlockingQueue* this_, uptr data)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    pInterCoreBlockingQueue->Enqueue(data);
}

bool nnosInterCoreBlockingQueueTryJam(nnosInterCoreBlockingQueue* this_, uptr data)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    return pInterCoreBlockingQueue->TryJam(data);
}

void nnosInterCoreBlockingQueueJam(nnosInterCoreBlockingQueue* this_, uptr data)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    pInterCoreBlockingQueue->Jam(data);
}

bool nnosInterCoreBlockingQueueTryDequeue(nnosInterCoreBlockingQueue* this_, uptr* pOut)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    return pInterCoreBlockingQueue->TryDequeue(pOut);
}

uptr nnosInterCoreBlockingQueueDequeue(nnosInterCoreBlockingQueue* this_)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    return pInterCoreBlockingQueue->Dequeue();
}

bool nnosInterCoreBlockingQueueTryGetFront(nnosInterCoreBlockingQueue* this_, uptr* pOut)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    return pInterCoreBlockingQueue->TryGetFront(pOut);
}

uptr nnosInterCoreBlockingQueueGetFront(nnosInterCoreBlockingQueue* this_)
{
    InterCoreBlockingQueue* pInterCoreBlockingQueue = reinterpret_cast<InterCoreBlockingQueue*>(this_);
    return pInterCoreBlockingQueue->GetFront();
}

}