Fix Doxygen @param entries in /src

cpp/src/ast/operators.cpp

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2021-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2021-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 #include <cudf/ast/detail/operators.cuh>
@@ -151,6 +151,7 @@ struct type_dispatch_binary_op {
  *
  * @tparam F Type of forwarded functor.
  * @tparam Ts Parameter pack of forwarded arguments.
+ * @param op The binary operator to dispatch
  * @param lhs_type Type of left input data.
  * @param rhs_type Type of right input data.
  * @param f Forwarded functor to be called.
@@ -218,6 +219,7 @@ struct type_dispatch_unary_op {
  *
  * @tparam F Type of forwarded functor.
  * @tparam Ts Parameter pack of forwarded arguments.
+ * @param op The unary operator to dispatch
  * @param input_type Type of input data.
  * @param f Forwarded functor to be called.
  * @param args Forwarded arguments to `operator()` of `f`.

cpp/src/copying/concatenate.cu

@@ -454,7 +454,8 @@ void traverse_children::operator()<cudf::list_view>(host_span<column_view const>
  * @brief Verifies that the sum of the sizes of all the columns to be concatenated
  * will not exceed the max value of size_type, and verifies all column types match
  *
- * @param columns_to_concat Span of columns to check
+ * @param cols Span of columns to check
+ * @param stream CUDA stream used for device memory operations and kernel launches
  *
  * @throws cudf::logic_error if the total length of the concatenated columns would
  * exceed the max value of size_type

cpp/src/copying/contiguous_split.cu

@@ -149,14 +149,8 @@ struct dst_buf_info {
  * @param dst Destination buffer
  * @param src Source buffer
  * @param t Thread index
- * @param num_elements Number of elements to copy
- * @param element_size Size of each element in bytes
- * @param src_element_index Element index to start copying at
+ * @param dst_info Destination buffer info containing element count, sizes, shifts, and validity
  * @param stride Size of the kernel block
- * @param value_shift Shift incoming 4-byte offset values down by this amount
- * @param bit_shift Shift incoming data right by this many bits
- * @param num_rows Number of rows being copied
- * @param valid_count Optional pointer to a value to store count of set bits
  */
 template <int block_size, bool is_offsets, typename offset_type = int32_t>
 __device__ void copy_buffer(uint8_t* __restrict__ dst,
@@ -464,6 +458,7 @@ size_type count_src_bufs(InputIter begin, InputIter end)
  * @param end End of input columns
  * @param head Beginning of source buffer info array
  * @param current Current source buffer info to be written to
+ * @param stream CUDA stream used for device memory operations and kernel launches
  * @param offset_stack_pos Integer representing our current offset nesting depth
  * (how many list or string levels deep we are)
  * @param parent_offset_index Index into src_buf_info output array indicating our nearest
@@ -799,6 +794,7 @@ std::tuple<size_t, int64_t, int64_t, size_type> build_output_column_metadata(
  * copied buffer
  * @param out_begin Output iterator of column views
  * @param base_ptr Pointer to the base address of copied data for the working partition
+ * @param mb Memory block for the output columns
  *
  * @returns new dst_buf_info iterator after processing this range of input columns
  */

cpp/src/groupby/sort/group_reductions.hpp

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2019-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2019-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -304,6 +304,7 @@ std::unique_ptr<column> group_var(column_view const& values,
  * @param values Grouped and sorted (within group) values to get quantiles from
  * @param group_sizes Number of valid elements per group
  * @param group_offsets Offsets of groups' starting points within @p values
+ * @param num_groups Number of groups
  * @param quantiles List of quantiles q where q lies in [0,1]
  * @param interp Method to use when desired value lies between data points
  * @param stream CUDA stream used for device memory operations and kernel launches.

cpp/src/groupby/sort/group_scan.hpp

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2021-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2021-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -169,6 +169,7 @@ std::unique_ptr<column> average_rank_scan(column_view const& grouped_values,
  * @brief Internal API to calculate groupwise dense rank value
  *
  * @param grouped_values column or struct column that rows within a group are sorted by
+ * @param value_order Order of values within groups
  * @param group_labels ID of group that the corresponding value belongs to
  * @param group_offsets group index offsets with group ID indices
  * @param stream CUDA stream used for device memory operations and kernel launches

cpp/src/io/avro/avro_gpu.cu

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2019-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2019-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 #include "avro_gpu.hpp"
@@ -52,7 +52,7 @@ static inline int64_t __device__ avro_decode_zigzag_varint(uint8_t const*& cur,
  *                       destination data.
  * @param[in] cur Current input data pointer
  * @param[in] end End of input data
- * @param[in] global_Dictionary Global dictionary entries
+ * @param[in] global_dictionary Global dictionary entries
  * @param[out] skipped_row Whether the row was skipped; set to false
  *                         if the row was saved (caller should ensure
  *                         this is initialized to true)
@@ -305,8 +305,8 @@ avro_decode_row(schemadesc_s const* schema,
  * @brief Decode column data
  *
  * @param[in] blocks Data block descriptions
- * @param[in] schema Schema description
- * @param[in] global_Dictionary Global dictionary entries
+ * @param[in] schema_g Schema description
+ * @param[in] global_dictionary Global dictionary entries
  * @param[in] avro_data Raw block data
  * @param[in] schema_len Number of entries in schema
  * @param[in] min_row_size Minimum size in bytes of a row

cpp/src/io/comp/gpuinflate.hpp

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2018-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2018-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -86,8 +86,6 @@ size_t get_gpu_debrotli_scratch_size(int max_num_inputs = 0);
  * @param[in] inputs List of input buffers
  * @param[out] outputs List of output buffers
  * @param[out] results List of output status structures
- * @param[in] scratch Temporary memory for intermediate work
- * @param[in] scratch_size Size in bytes of the temporary memory
  * @param[in] stream CUDA stream to use
  */
 CUDF_EXPORT

cpp/src/io/csv/csv_gpu.hpp

@@ -184,6 +184,7 @@ device_span<uint64_t> remove_blank_rows(cudf::io::parse_options_view const& opti
  * @param[in] data The row-column data
  * @param[in] column_flags Flags that control individual column parsing
  * @param[in] row_offsets List of row data start positions (offsets)
+ * @param[in] num_active_columns Number of active columns
  * @param[in] stream CUDA stream to use
  *
  * @return stats Histogram of each dtypes' occurrence for each column

cpp/src/io/fst/lookup_tables.cuh

@@ -80,7 +80,7 @@ class SingleSymbolSmemLUT {
    *
    * @param symbol_strings Array of strings, where the i-th string holds all symbols
    * (characters!) that correspond to the i-th symbol group index
-   * @param stream The stream used to copy the lookup table to device
+   * @param pre_map_op Function object that transforms a symbol to a symbol group id
    * @return
    */
   template <typename SymbolGroupItT>
@@ -863,7 +863,7 @@ class Dfa {
    * output symbols is written
    * @tparam OffsetT A type large enough to index into either of both: (a) the input symbols and
    * (b) the output symbols
-   * @param d_chars Pointer to the input string of symbols
+   * @param d_chars_it Pointer to the input string of symbols
    * @param num_chars The total number of input symbols to process
    * @param d_out_it Random-access output iterator to which the transduced output is
    * written

cpp/src/io/json/column_tree_construction.cu

@@ -86,6 +86,8 @@ struct parent_nodeids_to_colids {
  *
  * @param node_tree Node tree representation of JSON string
  * @param original_col_ids Column ids of nodes
+ * @param sorted_col_ids Sorted column IDs
+ * @param ordered_node_ids Ordered node IDs
  * @param row_offsets Row offsets of nodes
  * @param is_array_of_arrays Whether the tree is an array of arrays
  * @param row_array_parent_col_id Column id of row array, if is_array_of_arrays is true

cpp/src/io/orc/orc_gpu.hpp

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2019-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2019-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -308,7 +308,6 @@ void decode_nulls_and_string_dictionaries(column_desc* chunks,
  * @param[in] num_stripes Number of stripes
  * @param[in] first_row Crop all rows below first_row
  * @param[in] tz_table Timezone translation table
- * @param[in] tz_len Length of timezone translation table
  * @param[in] row_groups Optional row index data [rowgroup][column]
  * @param[in] num_rowgroups Number of row groups in row index data
  * @param[in] rowidx_stride Row index stride

cpp/src/io/orc/reader_impl_decode.cu

@@ -52,6 +52,7 @@ namespace {
  * @param loaded_stripe_range Range of stripes that are already loaded in memory
  * @param stream_range Range of streams to be decoded
  * @param num_decode_stripes Number of stripes that the decoding streams belong to
+ * @param compinfo Compression information for each stripe
  * @param compinfo_map A map to lookup compression info of streams
  * @param decompressor Block decompressor
  * @param stripe_data List of source stripe column data
@@ -346,7 +347,7 @@ void update_null_mask(cudf::detail::hostdevice_2dvector<column_desc>& chunks,
  * @param skip_rows Number of rows to offset from start
  * @param row_index_stride Distance between each row index
  * @param level Current nesting level being processed
- * @param tz_table Local time to UTC conversion table
+ * @param d_tz_table Local time to UTC conversion table
  * @param chunks Vector of list of column chunk descriptors
  * @param row_groups Vector of list of row index descriptors
  * @param out_buffers Output columns' device buffers

cpp/src/io/orc/stats_enc.cu

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2020-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2020-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -477,6 +477,7 @@ void orc_init_statistics_buffersize(statistics_merge_group* groups,
  * @param[in,out] groups Statistics merge groups
  * @param[in,out] chunks Statistics data
  * @param[in] statistics_count Number of statistics buffers
+ * @param[in] stream CUDA stream used for device memory operations and kernel launches
  */
 void orc_encode_statistics(uint8_t* blob_bfr,
                            statistics_merge_group* groups,

cpp/src/io/orc/stripe_data.cu

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2019-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2019-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -378,7 +378,6 @@ inline __device__ uint32_t bytestream_readu32(orc_bytestream_s* bs, int pos)
  *
  * @param[in] bs Byte stream input
  * @param[in] pos Position in byte stream
- * @param[in] numbits number of bits
  * @return bits
  */
 inline __device__ uint64_t bytestream_readu64(orc_bytestream_s* bs, int pos)
@@ -1165,10 +1164,12 @@ static const __device__ __constant__ int64_t kPow5i[28] = {1,
  * @brief ORC Decimal decoding (unbounded base-128 varints)
  *
  * @param[in] bs Input byte stream
+ * @param[in] scratch Scratch buffer for intermediate results
  * @param[in,out] vals on input: scale from secondary stream, on output: value
  * @param[in] val_scale Scale of each value
- * @param[in] col_scale Scale from schema to which value will be adjusted
  * @param[in] numvals Number of values to decode
+ * @param[in] dtype_id Data type identifier for the decimal column
+ * @param[in] col_scale Scale from schema to which value will be adjusted
  * @param[in] t thread id
  *
  * @return number of values decoded
@@ -2071,14 +2072,15 @@ void __host__ decode_nulls_and_string_dictionaries(column_desc* chunks,
  *
  * @param[in] chunks column_desc device array [stripe][column]
  * @param[in] global_dictionary Global dictionary device array
+ * @param[in] row_groups Optional row index data [row_group][column]
  * @param[in] num_columns Number of columns
  * @param[in] num_stripes Number of stripes
  * @param[in] first_row Crop all rows below first_row
  * @param[in] tz_table Timezone translation table
- * @param[in] row_groups Optional row index data [row_group][column]
  * @param[in] num_rowgroups Number of row groups in row index data
  * @param[in] rowidx_stride Row index stride
  * @param[in] level nesting level being processed
+ * @param[in] error_count Pointer to error count output
  * @param[in] stream CUDA stream used for device memory operations and kernel launches
  */
 void __host__ decode_column_data(column_desc* chunks,

cpp/src/io/orc/stripe_enc.cu

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2019-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2019-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -340,7 +340,6 @@ static inline __device__ void StoreBitsBigEndian(
  * @param[in] inbuf base input buffer
  * @param[in] inpos position in input buffer
  * @param[in] numvals max number of values to encode
- * @param[in] flush encode all remaining values if nonzero
  * @param[in] t thread id
  * @param[in] temp_storage shared memory storage to perform block reduce
  *

cpp/src/io/orc/writer_impl.cu

@@ -442,6 +442,7 @@ namespace {
  * @param columns List of columns
  * @param rowgroup_bounds Ranges of rows in each rowgroup [rowgroup][column]
  * @param max_stripe_size Maximum size of each stripe, both in bytes and in rows
+ * @param stream CUDA stream used for device memory operations and kernel launches
  * @return List of stripe descriptors
  */
 file_segmentation calculate_segmentation(host_span<orc_column_view const> columns,
@@ -536,6 +537,9 @@ size_t rle_stream_size(TypeKind kind, size_t count)
  * @param[in,out] columns List of columns
  * @param[in] segmentation stripe and rowgroup ranges
  * @param[in] decimal_column_sizes Sizes of encoded decimal columns
+ * @param[in] enable_dictionary Whether dictionary encoding is enabled
+ * @param[in] compression Compression type to use
+ * @param[in] write_mode The write mode (single or chunked)
  * @return List of stream descriptors
  */
 orc_streams create_streams(host_span<orc_column_view> columns,
@@ -1202,7 +1206,7 @@ cudf::detail::hostdevice_vector<uint8_t> allocate_and_encode_blobs(
 /**
  * @brief Returns column statistics in an intermediate format.
  *
- * @param statistics_freq Frequency of statistics to be included in the output file
+ * @param stats_freq Frequency of statistics to be included in the output file
  * @param orc_table Table information to be written
  * @param segmentation stripe and rowgroup ranges
  * @param stream CUDA stream used for device memory operations and kernel launches
@@ -1326,8 +1330,8 @@ intermediate_statistics gather_statistic_blobs(statistics_freq const stats_freq,
 /**
  * @brief Returns column statistics encoded in ORC protobuf format stored in the footer.
  *
- * @param num_stripes number of stripes in the data
- * @param incoming_stats intermediate statistics returned from `gather_statistic_blobs`
+ * @param footer ORC footer containing stripe and type information
+ * @param per_chunk_stats Persisted per-chunk statistics from `gather_statistic_blobs`
  * @param stream CUDA stream used for device memory operations and kernel launches
  * @return The encoded statistic blobs
  */

cpp/src/io/parquet/decode_fixed.cu

@@ -272,11 +272,11 @@ __device__ inline void decode_fixed_width_split_values(
  *
  * @tparam decode_block_size Size of the thread block
  * @tparam level_t Definition level type
- * @tparam is_nested Whether the type is nested
  *
  * @param target_value_count The target value count to process
  * @param s Pointer to page state
  * @param def Pointer to the definition levels
+ * @param is_nested Whether the column is nested
  * @param t Thread index
  *
  * @return Maximum depth valid count after skipping
@@ -548,9 +548,10 @@ __device__ int update_validity_and_row_indices_flat(
  * @tparam state_buf State buffer type
  *
  * @param target_value_count The target value count to process
- * @param s Pointer to  page state
- * @param sb Pointer to  state buffer
+ * @param s Pointer to page state
+ * @param sb Pointer to state buffer
  * @param def Pointer to the definition levels
+ * @param rep Repetition level decoder
  * @param t Thread index
  *
  * @return Maximum depth valid count after processing

cpp/src/io/parquet/decode_preprocess.cu

@@ -42,6 +42,7 @@ using unused_state_buf = page_state_buffers_s<0, 0, 0>;
  * @param rep Repetition level buffer
  * @param def Definition level buffer
  * @param bounds_set Boolean indicating whether min/max row bounds have been set
+ * @param block The cooperative thread block
  */
 template <typename level_t>
 __device__ void update_page_sizes(page_state_s* s,

cpp/src/io/parquet/error.hpp

@@ -1,5 +1,5 @@
 /*
- * SPDX-FileCopyrightText: Copyright (c) 2023-2025, NVIDIA CORPORATION.
+ * SPDX-FileCopyrightText: Copyright (c) 2023-2026, NVIDIA CORPORATION.
  * SPDX-License-Identifier: Apache-2.0
  */
 
@@ -38,7 +38,7 @@ class kernel_error {
    * @throws `rmm::bad_alloc` if allocating the device memory for `initial_value` fails.
    * @throws `rmm::cuda_error` if copying `initial_value` to device memory fails.
    *
-   * @param CUDA stream to use
+   * @param stream CUDA stream to use
    */
   kernel_error(rmm::cuda_stream_view stream) : _error_code(1, stream)
   {

cpp/src/io/parquet/experimental/deletion_vectors.cu

@@ -337,8 +337,8 @@ std::unique_ptr<cudf::column> compute_row_mask_column(
  * vectors
  *
  * @param row_index_column View of the row index column
- * @param deletion_vector_refs Queue of roaring bitmap wrappers
- * @param deletion_vector_row_counts Queue of number of rows in eachdeletion vector
+ * @param deletion_vectors Queue of roaring bitmap wrappers
+ * @param deletion_vector_row_counts Queue of number of rows in each deletion vector
  * @param start_row Starting row index of the current table chunk
  * @param stream CUDA stream for kernel launches and data transfers
  * @param mr Device memory resource to allocate device memory for the row mask column

cpp/src/io/parquet/experimental/dictionary_page_filter.cu

@@ -1007,6 +1007,8 @@ struct dictionary_caster {
    * @brief Build BOOL8 columns from dictionary membership results device buffers
    *
    * @param results_buffers Vector of dictionary membership results device buffers
+   * @param stream CUDA stream used for device memory operations and kernel launches
+   * @param mr Device memory resource used to allocate the returned device memory
    *
    * @return A vector of BOOL8 columns
    */