/******************************************************************************
 * Copyright (c) 2016, NVIDIA CORPORATION.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions are met:
 *     * Redistributions of source code must retain the above copyright
 *       notice, this list of conditionu and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above copyright
 *       notice, this list of conditions and the following disclaimer in the
 *       documentation and/or other materials provided with the distribution.
 *     * Neither the name of the NVIDIA CORPORATION nor the
 *       names of its contributors may be used to endorse or promote products
 *       derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
 * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY
 * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 ******************************************************************************/
#pragma once

#include <thrust/detail/config.h>

#if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
#  pragma GCC system_header
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_CLANG)
#  pragma clang system_header
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_MSVC)
#  pragma system_header
#endif // no system header

// XXX
// this file must not be included on its own, ever,
// but must be part of include in thrust/system/cuda/detail/copy.h

#include <thrust/system/cuda/config.h>

#include <thrust/advance.h>
#include <thrust/detail/raw_pointer_cast.h>
#include <thrust/detail/temporary_array.h>
#include <thrust/distance.h>
#include <thrust/system/cuda/detail/uninitialized_copy.h>
#include <thrust/system/cuda/detail/util.h>
#include <thrust/type_traits/is_trivially_relocatable.h>

THRUST_NAMESPACE_BEGIN
namespace cuda_cub
{

namespace __copy
{

template <class H, class D, class T, class Size>
THRUST_HOST_FUNCTION void trivial_device_copy(
  thrust::cpp::execution_policy<H>&, thrust::cuda_cub::execution_policy<D>& device_s, T* dst, T const* src, Size count)
{
  cudaError status;
  status = cuda_cub::trivial_copy_to_device(dst, src, count, cuda_cub::stream(device_s));
  cuda_cub::throw_on_error(status, "__copy::trivial_device_copy H->D: failed");
}

template <class D, class H, class T, class Size>
THRUST_HOST_FUNCTION void trivial_device_copy(
  thrust::cuda_cub::execution_policy<D>& device_s, thrust::cpp::execution_policy<H>&, T* dst, T const* src, Size count)
{
  cudaError status;
  status = cuda_cub::trivial_copy_from_device(dst, src, count, cuda_cub::stream(device_s));
  cuda_cub::throw_on_error(status, "trivial_device_copy D->H failed");
}

template <class System1, class System2, class InputIt, class Size, class OutputIt>
OutputIt _CCCL_HOST cross_system_copy_n(
  thrust::execution_policy<System1>& sys1,
  thrust::execution_policy<System2>& sys2,
  InputIt begin,
  Size n,
  OutputIt result,
  thrust::detail::true_type) // trivial copy

{
  typedef typename iterator_traits<InputIt>::value_type InputTy;
  if (n > 0)
  {
    trivial_device_copy(
      derived_cast(sys1),
      derived_cast(sys2),
      reinterpret_cast<InputTy*>(thrust::raw_pointer_cast(&*result)),
      reinterpret_cast<InputTy const*>(thrust::raw_pointer_cast(&*begin)),
      n);
  }

  return result + n;
}

// non-trivial H->D copy
template <class H, class D, class InputIt, class Size, class OutputIt>
OutputIt _CCCL_HOST cross_system_copy_n(
  thrust::cpp::execution_policy<H>& host_s,
  thrust::cuda_cub::execution_policy<D>& device_s,
  InputIt first,
  Size num_items,
  OutputIt result,
  thrust::detail::false_type) // non-trivial copy
{
  // get type of the input data
  typedef typename thrust::iterator_value<InputIt>::type InputTy;

  // copy input data into host temp storage
  InputIt last = first;
  thrust::advance(last, num_items);
  thrust::detail::temporary_array<InputTy, H> temp(host_s, num_items);

  for (Size idx = 0; idx != num_items; idx++)
  {
    ::new (static_cast<void*>(temp.data().get() + idx)) InputTy(*first);
    ++first;
  }

  // allocate device temporary storage
  thrust::detail::temporary_array<InputTy, D> d_in_ptr(device_s, num_items);

  // trivial copy data from host to device
  cudaError status =
    cuda_cub::trivial_copy_to_device(d_in_ptr.data().get(), temp.data().get(), num_items, cuda_cub::stream(device_s));
  cuda_cub::throw_on_error(status, "__copy:: H->D: failed");

  // device->device copy
  OutputIt ret = cuda_cub::copy_n(device_s, d_in_ptr.data(), num_items, result);

  return ret;
}

#if THRUST_DEVICE_COMPILER == THRUST_DEVICE_COMPILER_NVCC
// non-trivial copy D->H, only supported with NVCC compiler
// because copy ctor must have  __device__ annotations, which is nvcc-only
// feature
template <class D, class H, class InputIt, class Size, class OutputIt>
OutputIt _CCCL_HOST cross_system_copy_n(
  thrust::cuda_cub::execution_policy<D>& device_s,
  thrust::cpp::execution_policy<H>& host_s,
  InputIt first,
  Size num_items,
  OutputIt result,
  thrust::detail::false_type) // non-trivial copy

{
  // get type of the input data
  typedef typename thrust::iterator_value<InputIt>::type InputTy;

  // allocate device temp storage
  thrust::detail::temporary_array<InputTy, D> d_in_ptr(device_s, num_items);

  // uninitialize copy into temp device storage
  cuda_cub::uninitialized_copy_n(device_s, first, num_items, d_in_ptr.data());

  // allocate host temp storage
  thrust::detail::temporary_array<InputTy, H> temp(host_s, num_items);

  // trivial copy from device to host
  cudaError status;
  status =
    cuda_cub::trivial_copy_from_device(temp.data().get(), d_in_ptr.data().get(), num_items, cuda_cub::stream(device_s));
  cuda_cub::throw_on_error(status, "__copy:: D->H: failed");

  // host->host copy
  OutputIt ret = thrust::copy_n(host_s, temp.data(), num_items, result);

  return ret;
}
#endif

template <class System1, class System2, class InputIt, class Size, class OutputIt>
OutputIt _CCCL_HOST cross_system_copy_n(cross_system<System1, System2> systems, InputIt begin, Size n, OutputIt result)
{
  return cross_system_copy_n(
    derived_cast(systems.sys1),
    derived_cast(systems.sys2),
    begin,
    n,
    result,
    typename is_indirectly_trivially_relocatable_to<InputIt, OutputIt>::type());
}

template <class System1, class System2, class InputIterator, class OutputIterator>
OutputIterator _CCCL_HOST
cross_system_copy(cross_system<System1, System2> systems, InputIterator begin, InputIterator end, OutputIterator result)
{
  return cross_system_copy_n(systems, begin, thrust::distance(begin, end), result);
}

} // namespace __copy

} // namespace cuda_cub
THRUST_NAMESPACE_END