change the type of the owner of offsets and buff_addrs

Signed-off-by: Jigao Luo <[email protected]>

Author: JigaoLuo

cpp/src/io/parquet/page_data.cu

@@ -579,20 +579,9 @@ void write_final_offsets(host_span<size_type const> offsets,
                          host_span<size_type* const> buff_addrs,
                          rmm::cuda_stream_view stream)
 {
-  auto host_pinned_offsets = cudf::detail::make_host_vector<size_type>(offsets.size(), stream);
-  auto host_pinned_buff_addrs = cudf::detail::make_host_vector<size_type*>(buff_addrs.size(), stream);
-  CUDF_CUDA_TRY(cudaMemcpyAsync(host_pinned_offsets.data(),
-                                 offsets.data(),
-                                 offsets.size_bytes(),
-                                 cudaMemcpyHostToHost,
-                                 stream.value()));
-  CUDF_CUDA_TRY(cudaMemcpyAsync(host_pinned_buff_addrs.data(),
-                                 buff_addrs.data(),
-                                 buff_addrs.size_bytes(),
-                                 cudaMemcpyHostToHost,
-                                 stream.value()));
+  // Copy offsets to device and create an iterator
   auto d_src_data = cudf::detail::make_device_uvector_async(
-    host_pinned_offsets, stream, cudf::get_current_device_resource_ref());
+    offsets, stream, cudf::get_current_device_resource_ref());
   // Iterator for the source (scalar) data
   auto src_iter = thrust::make_transform_iterator(
     thrust::make_counting_iterator<std::size_t>(0),
@@ -601,7 +590,7 @@ void write_final_offsets(host_span<size_type const> offsets,
 
   // Copy buffer addresses to device and create an iterator
   auto d_dst_addrs = cudf::detail::make_device_uvector_async(
-    host_pinned_buff_addrs, stream, cudf::get_current_device_resource_ref());
+    buff_addrs, stream, cudf::get_current_device_resource_ref());
   // size_iter is simply a constant iterator of sizeof(size_type) bytes.
   auto size_iter = thrust::make_constant_iterator(sizeof(size_type));
 

cpp/src/io/parquet/reader_impl.cpp

@@ -467,8 +467,8 @@ void reader_impl::decode_page_data(read_mode mode, size_t skip_rows, size_t num_
   // that it is difficult/impossible for a given page to know that it is writing the very
   // last value that should then be followed by a terminator (because rows can span
   // page boundaries).
-  std::vector<size_type*> out_buffers;
-  std::vector<size_type> final_offsets;
+  auto out_buffers   = cudf::detail::make_host_vector<size_type*>(_input_columns.size(), _stream);
+  auto final_offsets = cudf::detail::make_host_vector<size_type>(_input_columns.size(), _stream);
   out_buffers.reserve(_input_columns.size());
   final_offsets.reserve(_input_columns.size());
   for (size_t idx = 0; idx < _input_columns.size(); idx++) {
@@ -486,14 +486,14 @@ void reader_impl::decode_page_data(read_mode mode, size_t skip_rows, size_t num_
 
         // the final offset for a list at level N is the size of it's child
         size_type const offset = child.type.id() == type_id::LIST ? child.size - 1 : child.size;
-        out_buffers.emplace_back(static_cast<size_type*>(out_buf.data()) + (out_buf.size - 1));
-        final_offsets.emplace_back(offset);
+        out_buffers.push_back(static_cast<size_type*>(out_buf.data()) + (out_buf.size - 1));
+        final_offsets.push_back(offset);
         out_buf.user_data |= PARQUET_COLUMN_BUFFER_FLAG_LIST_TERMINATED;
       } else if (out_buf.type.id() == type_id::STRING) {
         // only if it is not a large strings column
         if (std::cmp_less_equal(col_string_sizes[idx], strings::detail::get_offset64_threshold())) {
-          out_buffers.emplace_back(static_cast<size_type*>(out_buf.data()) + out_buf.size);
-          final_offsets.emplace_back(static_cast<size_type>(col_string_sizes[idx]));
+          out_buffers.push_back(static_cast<size_type*>(out_buf.data()) + out_buf.size);
+          final_offsets.push_back(static_cast<size_type>(col_string_sizes[idx]));
         }
         // Nested large strings column
         else if (input_col.nesting_depth() > 0) {