Prevent accidental copies of expensive-to-copy object types (#20226)

vuule · web-flow · commit 572dbce1d69a · 2025-10-13T17:04:15.000Z
Deleted copy constructor of classes that are expensive to copy. Some of these classes have const data members, so they previously had the move constructors implicitly deleted, making accidental copies even more likely. Also made the non-default move constructors `noexcept` to avoid implicit copies in container like `std::vector`. Authors: - Vukasin Milovanovic (https://github.com/vuule) Approvers: - Bradley Dice (https://github.com/bdice) - Lawrence Mitchell (https://github.com/wence-) - Nghia Truong (https://github.com/ttnghia) URL: #20226
diff --git a/cpp/include/cudf/detail/utilities/stream_pool.hpp b/cpp/include/cudf/detail/utilities/stream_pool.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2023-2024, NVIDIA CORPORATION.
+ * Copyright (c) 2023-2025, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -32,7 +32,11 @@ class cuda_stream_pool {
   // matching type used in rmm::cuda_stream_pool::get_stream(stream_id)
   using stream_id_type = std::size_t;
 
-  virtual ~cuda_stream_pool() = default;
+  virtual ~cuda_stream_pool()                          = default;
+  cuda_stream_pool(cuda_stream_pool const&)            = delete;
+  cuda_stream_pool(cuda_stream_pool&&)                 = delete;
+  cuda_stream_pool& operator=(cuda_stream_pool const&) = delete;
+  cuda_stream_pool& operator=(cuda_stream_pool&&)      = delete;
 
   /**
    * @brief Get a `cuda_stream_view` of a stream in the pool.
@@ -76,6 +80,9 @@ class cuda_stream_pool {
    * @return the number of stream objects in the pool
    */
   [[nodiscard]] virtual std::size_t get_stream_pool_size() const = 0;
+
+ protected:
+  cuda_stream_pool() = default;
 };
 
 /**
diff --git a/cpp/include/cudf/join/sort_merge_join.hpp b/cpp/include/cudf/join/sort_merge_join.hpp
@@ -44,6 +44,12 @@ namespace CUDF_EXPORT cudf {
  */
 class sort_merge_join {
  public:
+  sort_merge_join()                                  = delete;
+  sort_merge_join(sort_merge_join const&)            = delete;
+  sort_merge_join(sort_merge_join&&)                 = delete;
+  sort_merge_join& operator=(sort_merge_join const&) = delete;
+  sort_merge_join& operator=(sort_merge_join&&)      = delete;
+
   /**
    * @brief Construct a sort-merge join object that pre-processes the right table
    * on creation, and can be used on subsequent join operations with multiple
diff --git a/cpp/include/cudf/strings/regex/regex_program.hpp b/cpp/include/cudf/strings/regex/regex_program.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2022-2024, NVIDIA CORPORATION.
+ * Copyright (c) 2022-2025, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -54,22 +54,24 @@ struct regex_program {
                                                regex_flags flags      = regex_flags::DEFAULT,
                                                capture_groups capture = capture_groups::EXTRACT);
 
-  regex_program() = delete;
+  regex_program()                                = delete;
+  regex_program(regex_program const&)            = delete;
+  regex_program& operator=(regex_program const&) = delete;
 
   /**
    * @brief Move constructor
    *
    * @param other Object to move from
    */
-  regex_program(regex_program&& other);
+  regex_program(regex_program&& other) noexcept;
 
   /**
    * @brief Move operator assignment
    *
    * @param other Object to move from
    * @return this object
    */
-  regex_program& operator=(regex_program&& other);
+  regex_program& operator=(regex_program&& other) noexcept;
 
   /**
    * @brief Return the pattern used to create this instance
diff --git a/cpp/src/io/avro/reader_impl.cu b/cpp/src/io/avro/reader_impl.cu
@@ -117,6 +117,11 @@ class metadata : public file_metadata {
  public:
   explicit metadata(datasource* const src) : source(src) {}
 
+  metadata(metadata const&)            = delete;
+  metadata& operator=(metadata const&) = delete;
+  metadata(metadata&&)                 = delete;
+  metadata& operator=(metadata&&)      = delete;
+
   /**
    * @brief Initializes the parser and filters down to a subset of rows
    *
diff --git a/cpp/src/io/json/nested_json.hpp b/cpp/src/io/json/nested_json.hpp
@@ -144,6 +144,11 @@ struct device_json_column {
   // Type used to count number of rows
   using row_offset_t = size_type;
 
+  device_json_column(device_json_column const&)            = delete;
+  device_json_column& operator=(device_json_column const&) = delete;
+  device_json_column(device_json_column&&)                 = default;
+  device_json_column& operator=(device_json_column&&)      = default;
+
   // The inferred type of this column (list, struct, or value/string column)
   json_col_t type = json_col_t::Unknown;
 
diff --git a/cpp/src/io/orc/aggregate_orc_metadata.hpp b/cpp/src/io/orc/aggregate_orc_metadata.hpp
@@ -64,6 +64,11 @@ class aggregate_orc_metadata {
   aggregate_orc_metadata(std::vector<std::unique_ptr<datasource>> const& sources,
                          rmm::cuda_stream_view stream);
 
+  aggregate_orc_metadata(aggregate_orc_metadata const&)            = delete;
+  aggregate_orc_metadata& operator=(aggregate_orc_metadata const&) = delete;
+  aggregate_orc_metadata(aggregate_orc_metadata&&)                 = delete;
+  aggregate_orc_metadata& operator=(aggregate_orc_metadata&&)      = delete;
+
   [[nodiscard]] auto get_col_type(int col_idx) const
   {
     return per_file_metadata[0].ff.types[col_idx];
diff --git a/cpp/src/io/orc/orc.hpp b/cpp/src/io/orc/orc.hpp
@@ -614,6 +614,11 @@ class metadata {
  public:
   explicit metadata(datasource* const src, rmm::cuda_stream_view stream);
 
+  metadata(metadata const&)            = delete;
+  metadata& operator=(metadata const&) = delete;
+  metadata(metadata&&)                 = default;
+  metadata& operator=(metadata&&)      = default;
+
   [[nodiscard]] auto get_total_rows() const { return ff.numberOfRows; }
   [[nodiscard]] size_type get_num_stripes() const { return ff.stripes.size(); }
   [[nodiscard]] size_type get_num_columns() const { return ff.types.size(); }
@@ -671,7 +676,7 @@ class metadata {
   Metadata md;
   std::vector<StripeFooter> stripefooters;
   std::unique_ptr<orc_decompressor> decompressor;
-  datasource* const source;
+  datasource* source;
 
  private:
   struct column_parent {
diff --git a/cpp/src/io/orc/reader_impl.hpp b/cpp/src/io/orc/reader_impl.hpp
@@ -80,6 +80,11 @@ class reader_impl {
                        rmm::cuda_stream_view stream,
                        rmm::device_async_resource_ref mr);
 
+  reader_impl(reader_impl const&)            = delete;
+  reader_impl& operator=(reader_impl const&) = delete;
+  reader_impl(reader_impl&&)                 = delete;
+  reader_impl& operator=(reader_impl&&)      = delete;
+
   /**
    * @copydoc cudf::io::orc::detail::reader::read
    */
diff --git a/cpp/src/io/orc/reader_impl_chunking.hpp b/cpp/src/io/orc/reader_impl_chunking.hpp
@@ -128,6 +128,13 @@ struct orc_stream_info {
  * @brief Struct storing intermediate processing data loaded from data sources.
  */
 struct file_intermediate_data {
+  file_intermediate_data() = default;
+
+  file_intermediate_data(file_intermediate_data const&)            = delete;
+  file_intermediate_data& operator=(file_intermediate_data const&) = delete;
+  file_intermediate_data(file_intermediate_data&&)                 = default;
+  file_intermediate_data& operator=(file_intermediate_data&&)      = default;
+
   int64_t rows_to_skip;
   int64_t rows_to_read;
   std::vector<metadata::orc_stripe_info> selected_stripes;
diff --git a/cpp/src/io/orc/writer_impl.hpp b/cpp/src/io/orc/writer_impl.hpp
@@ -249,6 +249,11 @@ class writer::impl {
    */
   ~impl();
 
+  impl(impl const&)            = delete;
+  impl& operator=(impl const&) = delete;
+  impl(impl&&)                 = delete;
+  impl& operator=(impl&&)      = delete;
+
   /**
    * @brief Writes a single subtable as part of a larger ORC file/table write.
    *
diff --git a/cpp/src/io/parquet/reader_impl.hpp b/cpp/src/io/parquet/reader_impl.hpp
@@ -113,6 +113,11 @@ class reader_impl {
                        rmm::cuda_stream_view stream,
                        rmm::device_async_resource_ref mr);
 
+  reader_impl(reader_impl const&)            = delete;
+  reader_impl& operator=(reader_impl const&) = delete;
+  reader_impl(reader_impl&&)                 = delete;
+  reader_impl& operator=(reader_impl&&)      = delete;
+
   /**
    * @copydoc cudf::io::chunked_parquet_reader::has_next
    */
diff --git a/cpp/src/io/parquet/reader_impl_chunking.hpp b/cpp/src/io/parquet/reader_impl_chunking.hpp
@@ -27,6 +27,13 @@ namespace cudf::io::parquet::detail {
  * all passes/chunks in the file.
  */
 struct file_intermediate_data {
+  file_intermediate_data() = default;
+
+  file_intermediate_data(file_intermediate_data const&)            = delete;
+  file_intermediate_data& operator=(file_intermediate_data const&) = delete;
+  file_intermediate_data(file_intermediate_data&&)                 = default;
+  file_intermediate_data& operator=(file_intermediate_data&&)      = default;
+
   // all row groups to read
   std::vector<row_group_info> row_groups{};
 
@@ -77,6 +84,13 @@ struct row_range {
  * @brief Passes are broken down into subpasses based on temporary memory constraints.
  */
 struct subpass_intermediate_data {
+  subpass_intermediate_data() = default;
+
+  subpass_intermediate_data(subpass_intermediate_data const&)            = delete;
+  subpass_intermediate_data& operator=(subpass_intermediate_data const&) = delete;
+  subpass_intermediate_data(subpass_intermediate_data&&)                 = default;
+  subpass_intermediate_data& operator=(subpass_intermediate_data&&)      = default;
+
   rmm::device_buffer decomp_page_data;
 
   rmm::device_buffer level_decode_data{};
@@ -118,6 +132,13 @@ struct subpass_intermediate_data {
  * rowgroups may represent less than all of the rowgroups to be read for the file.
  */
 struct pass_intermediate_data {
+  pass_intermediate_data() = default;
+
+  pass_intermediate_data(pass_intermediate_data const&)            = delete;
+  pass_intermediate_data& operator=(pass_intermediate_data const&) = delete;
+  pass_intermediate_data(pass_intermediate_data&&)                 = default;
+  pass_intermediate_data& operator=(pass_intermediate_data&&)      = default;
+
   std::vector<rmm::device_buffer> raw_page_data;
 
   // rowgroup, chunk and page information for the current pass.
diff --git a/cpp/src/io/parquet/reader_impl_helpers.hpp b/cpp/src/io/parquet/reader_impl_helpers.hpp
@@ -337,6 +337,11 @@ class aggregate_reader_metadata {
                             bool has_cols_from_mismatched_srcs,
                             bool read_page_indexes = true);
 
+  aggregate_reader_metadata(aggregate_reader_metadata const&)            = delete;
+  aggregate_reader_metadata& operator=(aggregate_reader_metadata const&) = delete;
+  aggregate_reader_metadata(aggregate_reader_metadata&&)                 = delete;
+  aggregate_reader_metadata& operator=(aggregate_reader_metadata&&)      = delete;
+
   [[nodiscard]] RowGroup const& get_row_group(size_type row_group_index, size_type src_idx) const;
 
   /**
diff --git a/cpp/src/io/parquet/writer_impl.hpp b/cpp/src/io/parquet/writer_impl.hpp
@@ -82,6 +82,11 @@ class writer::impl {
    */
   ~impl();
 
+  impl(impl const&)            = delete;
+  impl& operator=(impl const&) = delete;
+  impl(impl&&)                 = delete;
+  impl& operator=(impl&&)      = delete;
+
   /**
    * @brief Initializes the states before writing.
    */
diff --git a/cpp/src/io/text/bgzip_data_chunk_source.cu b/cpp/src/io/text/bgzip_data_chunk_source.cu
@@ -87,6 +87,11 @@ class bgzip_data_chunk_reader : public data_chunk_reader {
     static constexpr std::size_t default_offset_alloc =
       1 << 16;  // 64k offset allocation, resized on demand
 
+    decompression_blocks(decompression_blocks const&)            = delete;
+    decompression_blocks& operator=(decompression_blocks const&) = delete;
+    decompression_blocks(decompression_blocks&&)                 = default;
+    decompression_blocks& operator=(decompression_blocks&&)      = default;
+
     cudaEvent_t event;
     cudf::detail::host_vector<char> h_compressed_blocks;
     cudf::detail::host_vector<std::size_t> h_compressed_offsets;
diff --git a/cpp/src/io/text/device_data_chunks.hpp b/cpp/src/io/text/device_data_chunks.hpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2022, NVIDIA CORPORATION.
+ * Copyright (c) 2022-2025, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -36,6 +36,11 @@ class device_uvector_data_chunk : public device_data_chunk {
  public:
   device_uvector_data_chunk(rmm::device_uvector<char>&& data) : _data(std::move(data)) {}
 
+  device_uvector_data_chunk(device_uvector_data_chunk const&)            = delete;
+  device_uvector_data_chunk& operator=(device_uvector_data_chunk const&) = delete;
+  device_uvector_data_chunk(device_uvector_data_chunk&&)                 = default;
+  device_uvector_data_chunk& operator=(device_uvector_data_chunk&&)      = default;
+
   [[nodiscard]] char const* data() const override { return _data.data(); }
   [[nodiscard]] std::size_t size() const override { return _data.size(); }
   operator device_span<char const>() const override { return _data; }
diff --git a/cpp/src/strings/regex/regex_program.cpp b/cpp/src/strings/regex/regex_program.cpp
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2022-2023, NVIDIA CORPORATION.
+ * Copyright (c) 2022-2025, NVIDIA CORPORATION.
  *
  * Licensed under the Apache License, Version 2.0 (the "License");
  * you may not use this file except in compliance with the License.
@@ -32,9 +32,9 @@ std::unique_ptr<regex_program> regex_program::create(std::string_view pattern,
   return std::unique_ptr<regex_program>(p);
 }
 
-regex_program::~regex_program()                                = default;
-regex_program::regex_program(regex_program&& other)            = default;
-regex_program& regex_program::operator=(regex_program&& other) = default;
+regex_program::~regex_program()                                         = default;
+regex_program::regex_program(regex_program&& other) noexcept            = default;
+regex_program& regex_program::operator=(regex_program&& other) noexcept = default;
 
 regex_program::regex_program(std::string_view pattern, regex_flags flags, capture_groups capture)
   : _pattern(pattern),
diff --git a/java/src/main/native/src/DataSourceHelperJni.cpp b/java/src/main/native/src/DataSourceHelperJni.cpp
@@ -66,7 +66,7 @@ class host_buffer_done_callback {
   explicit host_buffer_done_callback(JavaVM* jvm, jobject ds, long id) : jvm(jvm), ds(ds), id(id) {}
 
   host_buffer_done_callback(host_buffer_done_callback const& other) = delete;
-  host_buffer_done_callback(host_buffer_done_callback&& other)
+  host_buffer_done_callback(host_buffer_done_callback&& other) noexcept
     : jvm(other.jvm), ds(other.ds), id(other.id)
   {
     other.jvm = nullptr;
