/****************************************************************************** * Python Remote Debugging Module - Binary Writer Implementation * * High-performance binary file writer for profiling data with optional zstd * compression. ******************************************************************************/ #ifndef Py_BUILD_CORE_MODULE # define Py_BUILD_CORE_MODULE #endif #include "binary_io.h" #include "_remote_debugging.h" #include "pycore_bitutils.h" // _Py_bit_length(), _Py_popcount32() #include "pycore_opcode_utils.h" // MAX_REAL_OPCODE #include #include #ifdef HAVE_ZSTD #include #endif #ifdef _Py_MEMORY_SANITIZER # include #endif /* ============================================================================ * CONSTANTS FOR BINARY FORMAT SIZES * ============================================================================ */ #define MAX_VARINT_SIZE 10 /* Maximum bytes for a varint64 */ #define MAX_VARINT_SIZE_U32 5 /* Maximum bytes for a varint32 */ #define MAX_BLOCK_HEADER_SIZE \ (BLOCK_MAGIC_SIZE + MAX_VARINT_SIZE_U32 + MAX_VARINT_SIZE + \ 3 * sizeof(uint8_t) + \ COLUMN_COUNT * (2 * MAX_VARINT_SIZE_U32 + sizeof(uint8_t))) #define ZSTD_MIN_SIZE 32 #define STACK_REUSE_MIN_FRAMES 3 #define MAX_SAMPLE_SIZE \ (STREAM_ENTRY_SIZE + 5 * MAX_VARINT_SIZE + sizeof(uint8_t) + \ MAX_STACK_DEPTH * sizeof(uint32_t) + sizeof(Repeat)) #define MAX_BITMAP_PADDING (CHAR_BIT - 1) #define BLOCK_FLUSH_SIZE \ (MAX_BLOCK_SIZE - MAX_SAMPLE_SIZE - MAX_BITMAP_PADDING) static int buffer_append(ByteBuffer *buffer, const void *data, size_t size) { if (buffer->size > MAX_BLOCK_SIZE || size > MAX_BLOCK_SIZE - buffer->size) { PyErr_SetString(PyExc_RuntimeError, "sample block buffer exceeds the writer safety limit"); return -1; } if (buffer_reserve(buffer, buffer->size + size) < 0) { return -1; } if (size != 0) { memcpy(buffer->data + buffer->size, data, size); } buffer->size += size; return 0; } static int buffer_append_varint(ByteBuffer *buffer, uint64_t value) { uint8_t bytes[MAX_VARINT_SIZE]; size_t size = encode_varint_u64(bytes, value); return buffer_append(buffer, bytes, size); } static int column_write(SampleBlock *block, unsigned column, const void *data, size_t size) { if (buffer_append(&block->columns[column], data, size) < 0) { return -1; } block->size += size; return 0; } static int column_write_varint(SampleBlock *block, unsigned column, uint64_t value) { uint8_t bytes[MAX_VARINT_SIZE]; size_t size = encode_varint_u64(bytes, value); return column_write(block, column, bytes, size); } /* Helper macro: convert PyLong to int32, using default_val if conversion fails */ #define PYLONG_TO_INT32_OR_DEFAULT(obj, var, default_val) \ do { \ (var) = (int32_t)PyLong_AsLong(obj); \ if (UNLIKELY(PyErr_Occurred() != NULL)) { \ PyErr_Clear(); \ (var) = (default_val); \ } \ } while (0) /* ============================================================================ * WRITER-SPECIFIC UTILITY HELPERS * ============================================================================ */ /* Grow two parallel arrays together (e.g., strings and string_lengths). * Returns 0 on success, -1 on error (sets PyErr). * On error, original arrays are preserved (truly atomic update). */ static inline int grow_parallel_arrays(void **array1, void **array2, size_t *capacity, size_t elem_size1, size_t elem_size2) { size_t old_cap = *capacity; if (old_cap > SIZE_MAX / 2) { PyErr_SetString(PyExc_OverflowError, "Array capacity overflow"); return -1; } size_t new_cap = old_cap * 2; if (new_cap > SIZE_MAX / elem_size1 || new_cap > SIZE_MAX / elem_size2) { PyErr_SetString(PyExc_OverflowError, "Array allocation size overflow"); return -1; } size_t new_size1 = new_cap * elem_size1; size_t new_size2 = new_cap * elem_size2; size_t old_size1 = old_cap * elem_size1; size_t old_size2 = old_cap * elem_size2; /* Allocate fresh memory blocks (not realloc) to ensure atomicity. * If either allocation fails, original arrays are completely unchanged. */ void *new_array1 = PyMem_Malloc(new_size1); if (!new_array1) { PyErr_NoMemory(); return -1; } void *new_array2 = PyMem_Malloc(new_size2); if (!new_array2) { /* Second allocation failed - free first and return with no state change */ PyMem_Free(new_array1); PyErr_NoMemory(); return -1; } /* Both allocations succeeded - copy data and update pointers atomically */ memcpy(new_array1, *array1, old_size1); memcpy(new_array2, *array2, old_size2); PyMem_Free(*array1); PyMem_Free(*array2); *array1 = new_array1; *array2 = new_array2; *capacity = new_cap; return 0; } /* Checked fwrite with GIL release - returns 0 on success, -1 on error (sets PyErr). * This version releases the GIL during the write operation to allow other Python * threads to run during potentially blocking I/O. */ static inline int fwrite_checked_allow_threads(const void *data, size_t size, FILE *fp) { size_t written; Py_BEGIN_ALLOW_THREADS written = fwrite(data, 1, size, fp); Py_END_ALLOW_THREADS if (written != size) { int err = errno; if (ferror(fp) && err != 0) { errno = err; PyErr_SetFromErrno(PyExc_IOError); } else { PyErr_Format(PyExc_IOError, "short write: wrote %zu of %zu bytes", written, size); } return -1; } return 0; } /* ============================================================================ * UTILITY FUNCTIONS * ============================================================================ */ int binary_io_zstd_available(void) { #ifdef HAVE_ZSTD return 1; #else return 0; #endif } int binary_io_get_best_compression(void) { #ifdef HAVE_ZSTD return COMPRESSION_ZSTD; #else return COMPRESSION_NONE; #endif } /* ============================================================================ * BINARY WRITER IMPLEMENTATION * ============================================================================ */ static int writer_init_zstd(BinaryWriter *writer) { #ifdef HAVE_ZSTD writer->zstd.cctx = ZSTD_createCCtx(); if (!writer->zstd.cctx) { PyErr_SetString(PyExc_MemoryError, "Failed to create zstd compression context"); return -1; } /* Compression level 3: good ratio without slowing down each block flush */ size_t result = ZSTD_CCtx_setParameter(writer->zstd.cctx, ZSTD_c_compressionLevel, 3); if (ZSTD_isError(result)) { PyErr_Format(PyExc_RuntimeError, "Failed to set zstd compression level: %s", ZSTD_getErrorName(result)); ZSTD_freeCCtx(writer->zstd.cctx); writer->zstd.cctx = NULL; return -1; } return 0; #else PyErr_SetString(PyExc_RuntimeError, "zstd compression requested but not available (HAVE_ZSTD not defined)"); return -1; #endif } /* ============================================================================ * HASH TABLE SUPPORT FUNCTIONS (using _Py_hashtable) * ============================================================================ */ /* Hash function for Python strings - uses Python's cached hash */ static Py_uhash_t string_hash_func(const void *key) { PyObject *str = (PyObject *)key; Py_hash_t hash = PyObject_Hash(str); if (hash == -1) { PyErr_Clear(); return 0; } return (Py_uhash_t)hash; } static int string_compare_func(const void *key1, const void *key2) { PyObject *str1 = (PyObject *)key1; PyObject *str2 = (PyObject *)key2; if (str1 == str2) { return 1; } int result = PyObject_RichCompareBool(str1, str2, Py_EQ); if (result == -1) { PyErr_Clear(); return 0; } return result; } static void string_key_destroy(void *key) { Py_XDECREF((PyObject *)key); } static Py_uhash_t frame_key_hash_func(const void *key) { const FrameKey *fk = (const FrameKey *)key; /* FNV-1a style hash combining all fields */ Py_uhash_t hash = 2166136261u; hash ^= fk->filename_idx; hash *= 16777619u; hash ^= fk->funcname_idx; hash *= 16777619u; hash ^= (uint32_t)fk->lineno; hash *= 16777619u; hash ^= (uint32_t)fk->end_lineno; hash *= 16777619u; hash ^= (uint32_t)fk->column; hash *= 16777619u; hash ^= (uint32_t)fk->end_column; hash *= 16777619u; hash ^= fk->opcode; hash *= 16777619u; return hash; } static int frame_key_compare_func(const void *key1, const void *key2) { const FrameKey *fk1 = (const FrameKey *)key1; const FrameKey *fk2 = (const FrameKey *)key2; return (fk1->filename_idx == fk2->filename_idx && fk1->funcname_idx == fk2->funcname_idx && fk1->lineno == fk2->lineno && fk1->end_lineno == fk2->end_lineno && fk1->column == fk2->column && fk1->end_column == fk2->end_column && fk1->opcode == fk2->opcode); } static void frame_key_destroy(void *key) { PyMem_Free(key); } static inline int writer_intern_string(BinaryWriter *writer, PyObject *string, uint32_t *index) { void *existing = _Py_hashtable_get(writer->string_hash, string); if (existing != NULL) { *index = (uint32_t)(uintptr_t)existing - 1; /* index+1 stored to distinguish from NULL */ return 0; } if (writer->string_count >= UINT32_MAX) { writer->state = BINARY_WRITER_LIMIT_REACHED; PyErr_SetString(PyExc_OverflowError, "too many strings for binary format"); return -1; } if (writer->string_count >= writer->string_capacity) { if (grow_parallel_arrays((void **)&writer->strings, (void **)&writer->string_lengths, &writer->string_capacity, sizeof(char *), sizeof(size_t)) < 0) { if (PyErr_ExceptionMatches(PyExc_OverflowError)) { writer->state = BINARY_WRITER_LIMIT_REACHED; } return -1; } } Py_ssize_t str_len; const char *str_data = PyUnicode_AsUTF8AndSize(string, &str_len); if (!str_data) { return -1; } if ((uintmax_t)str_len > UINT32_MAX) { writer->state = BINARY_WRITER_LIMIT_REACHED; PyErr_Format(PyExc_OverflowError, "string length %zd exceeds binary format maximum %u", str_len, UINT32_MAX); return -1; } char *str_copy = PyMem_Malloc(str_len + 1); if (!str_copy) { PyErr_NoMemory(); return -1; } memcpy(str_copy, str_data, str_len + 1); *index = (uint32_t)writer->string_count; /* Add to hash table FIRST to ensure atomic rollback on failure */ Py_INCREF(string); if (_Py_hashtable_set(writer->string_hash, string, (void *)(uintptr_t)(*index + 1)) < 0) { Py_DECREF(string); PyMem_Free(str_copy); PyErr_NoMemory(); return -1; } writer->strings[writer->string_count] = str_copy; writer->string_lengths[writer->string_count] = str_len; writer->string_count++; return 0; } static inline int writer_intern_frame(BinaryWriter *writer, const FrameEntry *entry, uint32_t *index) { FrameKey lookup_key = { entry->filename_idx, entry->funcname_idx, entry->lineno, entry->end_lineno, entry->column, entry->end_column, entry->opcode }; void *existing = _Py_hashtable_get(writer->frame_hash, &lookup_key); if (existing != NULL) { *index = (uint32_t)(uintptr_t)existing - 1; /* index+1 stored to distinguish from NULL */ return 0; } if (writer->frame_count >= UINT32_MAX) { writer->state = BINARY_WRITER_LIMIT_REACHED; PyErr_SetString(PyExc_OverflowError, "too many frames for binary format"); return -1; } if (GROW_ARRAY(writer->frame_entries, writer->frame_count, writer->frame_capacity, FrameEntry) < 0) { if (PyErr_ExceptionMatches(PyExc_OverflowError)) { writer->state = BINARY_WRITER_LIMIT_REACHED; } return -1; } FrameKey *key = PyMem_Malloc(sizeof(FrameKey)); if (!key) { PyErr_NoMemory(); return -1; } *key = lookup_key; *index = (uint32_t)writer->frame_count; writer->frame_entries[writer->frame_count] = *entry; if (_Py_hashtable_set(writer->frame_hash, key, (void *)(uintptr_t)(*index + 1)) < 0) { PyMem_Free(key); PyErr_NoMemory(); return -1; } writer->frame_count++; return 0; } /* Get or create the entry for a (thread, interpreter) pair. */ static ThreadEntry * writer_get_or_create_thread_entry(BinaryWriter *writer, uint64_t thread_id, uint32_t interpreter_id) { /* Linear search is OK for small number of threads. * Key is (thread_id, interpreter_id) since same thread_id can exist in different interpreters. */ for (size_t i = 0; i < writer->thread_count; i++) { if (writer->thread_entries[i].thread_id == thread_id && writer->thread_entries[i].interpreter_id == interpreter_id) { return &writer->thread_entries[i]; } } if (writer->thread_count >= UINT32_MAX) { writer->state = BINARY_WRITER_LIMIT_REACHED; PyErr_SetString(PyExc_OverflowError, "too many threads for binary format"); return NULL; } if (GROW_ARRAY(writer->thread_entries, writer->thread_count, writer->thread_capacity, ThreadEntry) < 0) { if (PyErr_ExceptionMatches(PyExc_OverflowError)) { writer->state = BINARY_WRITER_LIMIT_REACHED; } return NULL; } ThreadEntry *entry = &writer->thread_entries[writer->thread_count]; memset(entry, 0, sizeof(ThreadEntry)); entry->thread_id = thread_id; entry->interpreter_id = interpreter_id; entry->block_stream_index = UINT32_MAX; writer->thread_count++; return entry; } static int encode_column(BinaryWriter *writer, const ByteBuffer *decoded, ColumnDescriptor *descriptor) { SampleBlock *block = writer->block; const uint8_t *data = decoded->data; size_t size = decoded->size; uint8_t encoding = COMPRESSION_NONE; if (size > 1 && memcmp(data, data + 1, size - 1) == 0) { encoding = COLUMN_CONSTANT; size = 1; } #ifdef HAVE_ZSTD if (encoding == COMPRESSION_NONE && size >= ZSTD_MIN_SIZE && writer->compression_type == COMPRESSION_ZSTD) { size_t compress_bound = ZSTD_compressBound(size); if (buffer_reserve(&block->compressed, compress_bound) < 0) { return -1; } size_t compressed_size = ZSTD_compress2( writer->zstd.cctx, block->compressed.data, compress_bound, data, size); if (ZSTD_isError(compressed_size)) { PyErr_Format(PyExc_IOError, "zstd compression error: %s", ZSTD_getErrorName(compressed_size)); return -1; } _Py_MSAN_UNPOISON(block->compressed.data, compressed_size); uint8_t stored_size[MAX_VARINT_SIZE]; if (compressed_size + encode_varint_u64(stored_size, compressed_size) < size) { data = block->compressed.data; size = compressed_size; encoding = COMPRESSION_ZSTD; } } #endif descriptor->decoded_size = (uint32_t)decoded->size; descriptor->stored_size = (uint32_t)size; descriptor->encoding = encoding; if (buffer_reserve(&block->output, block->output.size + size) < 0) { return -1; } if (size != 0) { memcpy(block->output.data + block->output.size, data, size); } block->output.size += size; return 0; } static ByteBuffer * transform_statuses(SampleBlock *block, uint8_t bitmap_flags) { size_t plane_size = status_bitmap_size(block->sample_count); unsigned plane_count = (unsigned)_Py_popcount32(bitmap_flags); if (buffer_reserve(&block->transformed, plane_size * plane_count) < 0) { return NULL; } const uint8_t *statuses = block->columns[COLUMN_STATUS_BITMAPS].data; block->transformed.size = 0; for (unsigned bit = 0; bit < CHAR_BIT; bit++) { if (!(bitmap_flags & (1U << bit))) { continue; } uint8_t *plane = block->transformed.data + block->transformed.size; memset(plane, 0, plane_size); for (uint32_t sample = 0; sample < block->sample_count; sample++) { plane[sample / CHAR_BIT] |= ((statuses[sample] >> bit) & 1U) << (sample % CHAR_BIT); } block->transformed.size += plane_size; } return &block->transformed; } static ByteBuffer * transform_frame_ids(SampleBlock *block, unsigned width) { ByteBuffer *frames = &block->columns[COLUMN_FRAME_IDS]; size_t frame_count = frames->size / sizeof(uint32_t); if (buffer_reserve(&block->transformed, frame_count * width) < 0) { return NULL; } const uint32_t *frame_ids = (const uint32_t *)frames->data; for (unsigned byte = 0; frame_count && byte < width; byte++) { uint8_t *plane = block->transformed.data + (size_t)byte * frame_count; for (uint32_t frame = 0; frame < frame_count; frame++) { plane[frame] = (uint8_t)(frame_ids[frame] >> (byte * CHAR_BIT)); } } block->transformed.size = frame_count * width; return &block->transformed; } static ByteBuffer * transform_stack_instructions(SampleBlock *block) { ByteBuffer *instructions = &block->columns[COLUMN_STACK_INSTRUCTIONS]; size_t repeat_count = block->repeats.size / sizeof(Repeat); if (repeat_count == 0) { return instructions; } const Repeat *repeats = (const Repeat *)block->repeats.data; size_t copied_offset = 0; block->transformed.size = 0; for (size_t i = 0; i < repeat_count; i++) { if (buffer_append(&block->transformed, instructions->data + copied_offset, repeats[i].offset - copied_offset) < 0 || buffer_append_varint( &block->transformed, STACK_REPEAT) < 0 || buffer_append_varint( &block->transformed, repeats[i].count) < 0) { return NULL; } copied_offset = repeats[i].offset; } if (buffer_append(&block->transformed, instructions->data + copied_offset, instructions->size - copied_offset) < 0) { return NULL; } return &block->transformed; } static void block_reset(BinaryWriter *writer) { SampleBlock *block = writer->block; /* Only threads with a stream in this block need resetting */ for (uint32_t i = 0; i < block->stream_count; i++) { size_t thread = block->streams[i].thread_index; writer->thread_entries[thread].block_stream_index = UINT32_MAX; } for (unsigned column = 0; column < COLUMN_COUNT; column++) { block->columns[column].size = 0; } block->sample_count = 0; block->stream_count = 0; block->repeats.size = 0; block->max_frame = 0; block->size = 0; memset(block->cache, 0, sizeof(block->cache)); } static unsigned block_frame_width(const SampleBlock *block) { if (block->max_frame == 0) { return 1; } return (unsigned)(_Py_bit_length(block->max_frame) + CHAR_BIT - 1) / CHAR_BIT; } static void block_status_flags(const SampleBlock *block, uint8_t *common_flags, uint8_t *bitmap_flags) { uint8_t common_status = UINT8_MAX; uint8_t any_status = 0; const uint8_t *statuses = block->columns[COLUMN_STATUS_BITMAPS].data; for (uint32_t sample = 0; sample < block->sample_count; sample++) { common_status &= statuses[sample]; any_status |= statuses[sample]; } *common_flags = common_status; *bitmap_flags = any_status ^ common_status; } static size_t write_block_header(const BlockHeader *header, uint8_t *bytes) { /* Magic "TBLK", sample count and base timestamp (varints), then frame * ID width and the two status flag masks (1 byte each) */ size_t size = BLOCK_MAGIC_SIZE; memcpy(bytes, BLOCK_MAGIC, BLOCK_MAGIC_SIZE); size += encode_varint_u32(bytes + size, header->sample_count); size += encode_varint_u64(bytes + size, header->base_timestamp); bytes[size++] = header->frame_width; bytes[size++] = header->common_status_flags; bytes[size++] = header->status_bitmap_flags; /* Each column: decoded size (varint), encoding (1 byte), and the stored * size (varint) for zstd only */ for (unsigned column = 0; column < COLUMN_COUNT; column++) { const ColumnDescriptor *descriptor = &header->columns[column]; size += encode_varint_u32(bytes + size, descriptor->decoded_size); bytes[size++] = descriptor->encoding; if (descriptor->encoding == COMPRESSION_ZSTD) { size += encode_varint_u32(bytes + size, descriptor->stored_size); } } return size; } static int block_flush(BinaryWriter *writer) { SampleBlock *block = writer->block; if (block->sample_count == 0) { return 0; } BlockHeader header = {0}; header.sample_count = block->sample_count; header.base_timestamp = block->base_timestamp; header.frame_width = (uint8_t)block_frame_width(block); block_status_flags(block, &header.common_status_flags, &header.status_bitmap_flags); /* Convert each column to its file form, then encode it */ /* Leave room for the header, which is written once its size is known */ if (buffer_reserve(&block->output, MAX_BLOCK_HEADER_SIZE) < 0) { return -1; } block->output.size = MAX_BLOCK_HEADER_SIZE; size_t decoded_payload_size = 0; for (unsigned column = 0; column < COLUMN_COUNT; column++) { ByteBuffer *decoded = &block->columns[column]; switch (column) { case COLUMN_STATUS_BITMAPS: decoded = transform_statuses(block, header.status_bitmap_flags); break; case COLUMN_STACK_INSTRUCTIONS: decoded = transform_stack_instructions(block); break; case COLUMN_FRAME_IDS: decoded = transform_frame_ids(block, header.frame_width); break; } if (decoded == NULL) { return -1; } if (decoded->size > MAX_BLOCK_SIZE - decoded_payload_size) { PyErr_SetString(PyExc_RuntimeError, "sample block exceeds the decoded size limit"); return -1; } decoded_payload_size += decoded->size; if (encode_column(writer, decoded, &header.columns[column]) < 0) { return -1; } } /* Put the header right before the columns and write the block */ uint8_t header_bytes[MAX_BLOCK_HEADER_SIZE]; size_t header_size = write_block_header(&header, header_bytes); size_t unused = MAX_BLOCK_HEADER_SIZE - header_size; uint8_t *header_start = block->output.data + unused; memcpy(header_start, header_bytes, header_size); if (fwrite_checked_allow_threads(header_start, block->output.size - unused, writer->fp) < 0) { return -1; } writer->stats.bytes_written += decoded_payload_size; block_reset(writer); return 0; } /* FNV-1a over the frame indices, the same in every process */ static Py_uhash_t stack_hash(const uint32_t *stack, size_t depth) { uint64_t hash = 0xcbf29ce484222325; for (size_t i = 0; i < depth; i++) { hash = (hash ^ stack[i]) * 0x100000001b3; } /* Mix the high bits into the low bits that select the cache bucket */ hash ^= hash >> 33; hash *= 0xff51afd7ed558ccd; hash ^= hash >> 33; return (Py_uhash_t)hash; } static size_t common_outer_frames(const BlockStream *stream, const uint32_t *stack, size_t depth) { size_t shared = 0; while (shared < depth && shared < stream->prev_stack_depth && stack[depth - 1 - shared] == stream->prev_stack[stream->prev_stack_depth - 1 - shared]) { shared++; } return shared; } static int cached_stack_matches(const StackCacheEntry *cached, const ByteBuffer *frames, Py_uhash_t hash, const uint32_t *stack, size_t depth) { if (cached->hash != hash || cached->depth != depth) { return 0; } const uint32_t *frame_ids = (const uint32_t *)frames->data; for (size_t frame = 0; frame < depth; frame++) { if (stack[frame] != frame_ids[cached->offset + depth - 1 - frame]) { return 0; } } return 1; } static StackCacheEntry * stack_cache_lookup(StackCacheEntry *bucket, const ByteBuffer *frames, Py_uhash_t hash, const uint32_t *stack, size_t depth) { for (size_t i = 0; i < STACK_CACHE_BUCKET_SIZE; i++) { if (cached_stack_matches(&bucket[i], frames, hash, stack, depth)) { return &bucket[i]; } } return NULL; } static void stack_cache_insert(StackCacheEntry *bucket, StackCacheEntry entry) { memmove(&bucket[1], &bucket[0], (STACK_CACHE_BUCKET_SIZE - 1) * sizeof(*bucket)); bucket[0] = entry; } static int write_stack_repeat(BinaryWriter *writer, BlockStream *stream, size_t depth) { SampleBlock *block = writer->block; if (stream->current_repeat == UINT32_MAX) { Repeat repeat = { (uint32_t)block->columns[COLUMN_STACK_INSTRUCTIONS].size, 1}; if (buffer_append(&block->repeats, &repeat, sizeof(repeat)) < 0) { return -1; } block->size += sizeof(repeat); stream->current_repeat = (uint32_t)(block->repeats.size / sizeof(repeat) - 1); writer->stats.repeat_records++; } else { Repeat *repeats = (Repeat *)block->repeats.data; repeats[stream->current_repeat].count++; } writer->stats.repeat_samples++; writer->stats.frames_saved += depth; return 0; } static int write_stack_reuse(BinaryWriter *writer, const StackCacheEntry *cached, size_t depth) { SampleBlock *block = writer->block; if (column_write_varint(block, COLUMN_STACK_INSTRUCTIONS, STACK_REUSE) < 0 || column_write_varint(block, COLUMN_STACK_INSTRUCTIONS, cached->offset) < 0 || column_write_varint(block, COLUMN_STACK_INSTRUCTIONS, depth) < 0) { return -1; } writer->stats.reuse_records++; writer->stats.frames_saved += depth; return 0; } static int write_stack_replace_suffix(BinaryWriter *writer, StackCacheEntry *bucket, Py_uhash_t hash, const uint32_t *stack, size_t depth, size_t shared) { SampleBlock *block = writer->block; ByteBuffer *frames = &block->columns[COLUMN_FRAME_IDS]; size_t new_frames = depth - shared; if (column_write_varint(block, COLUMN_STACK_INSTRUCTIONS, STACK_REPLACE_SUFFIX) < 0 || column_write_varint(block, COLUMN_STACK_INSTRUCTIONS, shared) < 0 || column_write_varint(block, COLUMN_STACK_INSTRUCTIONS, new_frames) < 0) { return -1; } /* Append the new inner frames, outermost first */ size_t stack_offset = frames->size / sizeof(uint32_t); if (new_frames != 0) { if (buffer_reserve( frames, frames->size + new_frames * sizeof(uint32_t)) < 0) { return -1; } uint32_t *new_frame_ids = (uint32_t *)(frames->data + frames->size); for (size_t i = 0; i < new_frames; i++) { uint32_t frame = stack[new_frames - 1 - i]; new_frame_ids[i] = frame; if (frame > block->max_frame) { block->max_frame = frame; } } frames->size += new_frames * sizeof(uint32_t); block->size += new_frames * sizeof(uint32_t); } /* Complete stacks are cached so later samples can REUSE them */ if (shared == 0) { if (bucket != NULL) { stack_cache_insert(bucket, (StackCacheEntry){ hash, (uint32_t)stack_offset, (uint32_t)depth}); } writer->stats.full_records++; } else { writer->stats.suffix_records++; } writer->stats.total_frames_written += new_frames; writer->stats.frames_saved += shared; return 0; } static int write_stack(BinaryWriter *writer, BlockStream *stream, int new_stream, const uint32_t *stack, size_t depth) { SampleBlock *block = writer->block; size_t shared = common_outer_frames(stream, stack, depth); /* Same stack as this stream's previous sample: extend its REPEAT */ if (!new_stream && shared == depth && shared == stream->prev_stack_depth) { return write_stack_repeat(writer, stream, depth); } /* Enough new frames to be worth looking up an earlier complete stack */ Py_uhash_t hash = 0; StackCacheEntry *bucket = NULL; StackCacheEntry *cached = NULL; if (depth - shared >= STACK_REUSE_MIN_FRAMES) { hash = stack_hash(stack, depth); bucket = block->cache[hash % STACK_CACHE_BUCKET_COUNT]; cached = stack_cache_lookup(bucket, &block->columns[COLUMN_FRAME_IDS], hash, stack, depth); } int result; if (cached != NULL) { /* Stack already stored in this block: point to its frames */ result = write_stack_reuse(writer, cached, depth); } else { /* Keep the shared outer frames and write only the new inner ones */ result = write_stack_replace_suffix(writer, bucket, hash, stack, depth, shared); } if (result < 0) { return -1; } stream->current_repeat = UINT32_MAX; memcpy(stream->prev_stack, stack, depth * sizeof(uint32_t)); stream->prev_stack_depth = depth; return 0; } size_t binary_writer_buffered_size(BinaryWriter *writer) { SampleBlock *block = writer->block; if (block == NULL || block->sample_count == 0) { return 0; } size_t size = block->columns[COLUMN_STREAM_TABLE].size + block->columns[COLUMN_SAMPLE_STREAM_IDS].size + block->columns[COLUMN_TIMESTAMP_RESIDUALS].size + block->columns[COLUMN_STACK_INSTRUCTIONS].size; const Repeat *repeats = (const Repeat *)block->repeats.data; size_t repeat_count = block->repeats.size / sizeof(Repeat); uint8_t bytes[MAX_VARINT_SIZE]; for (size_t i = 0; i < repeat_count; i++) { size += encode_varint_u64(bytes, STACK_REPEAT) + encode_varint_u64(bytes, repeats[i].count); } uint8_t common_flags, bitmap_flags; block_status_flags(block, &common_flags, &bitmap_flags); size += (size_t)_Py_popcount32(bitmap_flags) * status_bitmap_size(block->sample_count); size += block->columns[COLUMN_FRAME_IDS].size / sizeof(uint32_t) * block_frame_width(block); return size; } static int block_add_stream(BinaryWriter *writer, ThreadEntry *entry) { SampleBlock *block = writer->block; if (GROW_ARRAY(block->streams, block->stream_count, block->stream_capacity, BlockStream) < 0) { return -1; } /* Stream table entry: thread_id (8 bytes) + interpreter_id (4 bytes) */ uint8_t stream_entry[STREAM_ENTRY_SIZE]; memcpy(stream_entry + STM_OFF_THREAD_ID, &entry->thread_id, sizeof(entry->thread_id)); memcpy(stream_entry + STM_OFF_INTERPRETER_ID, &entry->interpreter_id, sizeof(entry->interpreter_id)); if (column_write(block, COLUMN_STREAM_TABLE, stream_entry, sizeof(stream_entry)) < 0) { return -1; } BlockStream *stream = &block->streams[block->stream_count]; stream->thread_index = (size_t)(entry - writer->thread_entries); stream->prev_timestamp = block->base_timestamp - writer->sample_interval_us; stream->current_repeat = UINT32_MAX; stream->prev_stack_depth = 0; entry->block_stream_index = block->stream_count++; return 0; } static int block_add_sample(BinaryWriter *writer, ThreadEntry *entry, uint64_t timestamp, uint8_t status, const uint32_t *stack, size_t depth) { static_assert(BLOCK_FLUSH_SIZE + MAX_SAMPLE_SIZE + MAX_BITMAP_PADDING == MAX_BLOCK_SIZE, "block flush size needs headroom for one complete sample"); SampleBlock *block = writer->block; /* Start a new block if this sample might not fit */ if (block->size >= BLOCK_FLUSH_SIZE || (entry->block_stream_index == UINT32_MAX && block->stream_count == MAX_BLOCK_STREAMS)) { if (block_flush(writer) < 0) { return -1; } } if (block->sample_count == 0) { block->base_timestamp = timestamp; } int new_stream = entry->block_stream_index == UINT32_MAX; if (new_stream && block_add_stream(writer, entry) < 0) { return -1; } BlockStream *stream = &block->streams[entry->block_stream_index]; /* Store the difference from the expected timestamp */ uint64_t residual = timestamp - stream->prev_timestamp - writer->sample_interval_us; uint8_t residual_bytes[MAX_VARINT_SIZE]; size_t residual_size = encode_varint_i64(residual_bytes, (int64_t)residual); if (column_write_varint(block, COLUMN_SAMPLE_STREAM_IDS, entry->block_stream_index) < 0 || column_write(block, COLUMN_TIMESTAMP_RESIDUALS, residual_bytes, residual_size) < 0 || column_write(block, COLUMN_STATUS_BITMAPS, &status, sizeof(status)) < 0) { return -1; } if (write_stack(writer, stream, new_stream, stack, depth) < 0) { return -1; } stream->prev_timestamp = timestamp; block->sample_count++; writer->total_samples++; return 0; } static void block_destroy(SampleBlock *block) { if (block == NULL) { return; } for (unsigned i = 0; i < COLUMN_COUNT; i++) { buffer_clear(&block->columns[i]); } buffer_clear(&block->repeats); buffer_clear(&block->transformed); buffer_clear(&block->compressed); buffer_clear(&block->output); PyMem_Free(block->streams); PyMem_Free(block); } BinaryWriter * binary_writer_create(PyObject *path, uint64_t sample_interval_us, int compression_type, uint64_t start_time_us, int profiling_mode, int capture_features) { if (compression_type != COMPRESSION_NONE && compression_type != COMPRESSION_ZSTD) { PyErr_SetString(PyExc_ValueError, "invalid compression type"); return NULL; } BinaryWriter *writer = PyMem_Calloc(1, sizeof(BinaryWriter)); if (!writer) { PyErr_NoMemory(); return NULL; } writer->start_time_us = start_time_us; writer->sample_interval_us = sample_interval_us; writer->compression_type = compression_type; writer->profiling_mode = profiling_mode; writer->capture_features = capture_features; writer->block = PyMem_Calloc(1, sizeof(SampleBlock)); if (writer->block == NULL) { PyErr_NoMemory(); goto error; } writer->block->streams = PyMem_Malloc( INITIAL_BLOCK_STREAM_CAPACITY * sizeof(BlockStream)); if (writer->block->streams == NULL) { PyErr_NoMemory(); goto error; } writer->block->stream_capacity = INITIAL_BLOCK_STREAM_CAPACITY; writer->string_hash = _Py_hashtable_new_full( string_hash_func, string_compare_func, string_key_destroy, /* Key destroy: decref the Python string */ NULL, /* Value destroy: values are just indices, not pointers */ NULL /* Use default allocator */ ); if (!writer->string_hash) { PyErr_NoMemory(); goto error; } writer->strings = PyMem_Malloc(INITIAL_STRING_CAPACITY * sizeof(char *)); if (!writer->strings) { PyErr_NoMemory(); goto error; } writer->string_lengths = PyMem_Malloc(INITIAL_STRING_CAPACITY * sizeof(size_t)); if (!writer->string_lengths) { PyErr_NoMemory(); goto error; } writer->string_capacity = INITIAL_STRING_CAPACITY; writer->frame_hash = _Py_hashtable_new_full( frame_key_hash_func, frame_key_compare_func, frame_key_destroy, /* Key destroy: free the FrameKey */ NULL, /* Value destroy: values are just indices, not pointers */ NULL /* Use default allocator */ ); if (!writer->frame_hash) { PyErr_NoMemory(); goto error; } writer->frame_entries = PyMem_Malloc(INITIAL_FRAME_CAPACITY * sizeof(FrameEntry)); if (!writer->frame_entries) { PyErr_NoMemory(); goto error; } writer->frame_capacity = INITIAL_FRAME_CAPACITY; writer->thread_entries = PyMem_Malloc(INITIAL_THREAD_CAPACITY * sizeof(ThreadEntry)); if (!writer->thread_entries) { PyErr_NoMemory(); goto error; } writer->thread_capacity = INITIAL_THREAD_CAPACITY; if (compression_type == COMPRESSION_ZSTD) { if (writer_init_zstd(writer) < 0) { goto error; } } writer->fp = Py_fopen(path, "wb"); if (!writer->fp) { goto error; } /* Hint sequential write pattern to kernel for better I/O scheduling */ #if defined(__linux__) && defined(POSIX_FADV_SEQUENTIAL) { int fd = fileno(writer->fp); if (fd >= 0) { (void)posix_fadvise(fd, 0, 0, POSIX_FADV_SEQUENTIAL); } } #endif uint8_t header[FILE_HEADER_PLACEHOLDER_SIZE] = {0}; if (fwrite_checked_allow_threads(header, FILE_HEADER_PLACEHOLDER_SIZE, writer->fp) < 0) { goto error; } return writer; error: binary_writer_destroy(writer); return NULL; } /* Build a frame stack from Python frame list by interning all strings and frames. * Returns 0 on success, -1 on error. */ static int build_frame_stack(BinaryWriter *writer, PyObject *frame_list, uint32_t *curr_stack, size_t *curr_depth) { Py_ssize_t stack_depth = PyList_Size(frame_list); *curr_depth = (stack_depth < MAX_STACK_DEPTH) ? stack_depth : MAX_STACK_DEPTH; for (Py_ssize_t k = 0; k < (Py_ssize_t)*curr_depth; k++) { /* Use unchecked accessors since we control the data structures */ PyObject *frame_info = PyList_GET_ITEM(frame_list, k); /* Get filename, location, funcname, opcode from FrameInfo using unchecked access */ PyObject *filename = PyStructSequence_GET_ITEM(frame_info, 0); PyObject *location = PyStructSequence_GET_ITEM(frame_info, 1); PyObject *funcname = PyStructSequence_GET_ITEM(frame_info, 2); PyObject *opcode_obj = PyStructSequence_GET_ITEM(frame_info, 3); /* Extract location fields (can be None for synthetic frames) */ int32_t lineno = LOCATION_NOT_AVAILABLE; int32_t end_lineno = LOCATION_NOT_AVAILABLE; int32_t column = LOCATION_NOT_AVAILABLE; int32_t end_column = LOCATION_NOT_AVAILABLE; if (location != Py_None) { /* LocationInfo is a struct sequence or tuple with: * (lineno, end_lineno, column, end_column) */ PyObject *lineno_obj = PyTuple_Check(location) ? PyTuple_GET_ITEM(location, 0) : PyStructSequence_GET_ITEM(location, 0); PyObject *end_lineno_obj = PyTuple_Check(location) ? PyTuple_GET_ITEM(location, 1) : PyStructSequence_GET_ITEM(location, 1); PyObject *column_obj = PyTuple_Check(location) ? PyTuple_GET_ITEM(location, 2) : PyStructSequence_GET_ITEM(location, 2); PyObject *end_column_obj = PyTuple_Check(location) ? PyTuple_GET_ITEM(location, 3) : PyStructSequence_GET_ITEM(location, 3); PYLONG_TO_INT32_OR_DEFAULT(lineno_obj, lineno, LOCATION_NOT_AVAILABLE); PYLONG_TO_INT32_OR_DEFAULT(end_lineno_obj, end_lineno, LOCATION_NOT_AVAILABLE); PYLONG_TO_INT32_OR_DEFAULT(column_obj, column, LOCATION_NOT_AVAILABLE); PYLONG_TO_INT32_OR_DEFAULT(end_column_obj, end_column, LOCATION_NOT_AVAILABLE); } /* Extract opcode (can be None) */ uint8_t opcode = OPCODE_NONE; if (opcode_obj != Py_None) { long opcode_long = PyLong_AsLong(opcode_obj); if (UNLIKELY(PyErr_Occurred() != NULL)) { PyErr_Clear(); opcode = OPCODE_NONE; } else if (opcode_long >= 0 && opcode_long <= MAX_REAL_OPCODE) { opcode = (uint8_t)opcode_long; } } /* Intern filename */ uint32_t filename_idx; if (writer_intern_string(writer, filename, &filename_idx) < 0) { return -1; } /* Intern funcname */ uint32_t funcname_idx; if (writer_intern_string(writer, funcname, &funcname_idx) < 0) { return -1; } /* Intern frame with full location info */ FrameEntry frame_entry = { .filename_idx = filename_idx, .funcname_idx = funcname_idx, .lineno = lineno, .end_lineno = end_lineno, .column = column, .end_column = end_column, .opcode = opcode }; uint32_t frame_idx; if (writer_intern_frame(writer, &frame_entry, &frame_idx) < 0) { return -1; } curr_stack[k] = frame_idx; } return 0; } /* Process a single thread's sample. * Returns 0 on success, -1 on error. */ static int process_thread_sample(BinaryWriter *writer, PyObject *thread_info, uint32_t interpreter_id, uint64_t timestamp_us) { if (writer->total_samples == UINT64_MAX) { writer->state = BINARY_WRITER_LIMIT_REACHED; PyErr_SetString(PyExc_OverflowError, "too many samples for binary format"); return -1; } PyObject *thread_id_obj = PyStructSequence_GET_ITEM(thread_info, 0); PyObject *status_obj = PyStructSequence_GET_ITEM(thread_info, 1); PyObject *frame_list = PyStructSequence_GET_ITEM(thread_info, 2); uint64_t thread_id = PyLong_AsUnsignedLongLong(thread_id_obj); if (thread_id == (uint64_t)-1 && PyErr_Occurred()) { return -1; } long status_long = PyLong_AsLong(status_obj); if (status_long == -1 && PyErr_Occurred()) { return -1; } if (status_long < 0 || status_long > UINT8_MAX) { writer->state = BINARY_WRITER_LIMIT_REACHED; PyErr_Format(PyExc_OverflowError, "thread status %ld exceeds the binary format range", status_long); return -1; } uint8_t status = (uint8_t)status_long; ThreadEntry *entry = writer_get_or_create_thread_entry( writer, thread_id, interpreter_id); if (!entry) { return -1; } /* Process frames and build current stack */ uint32_t curr_stack[MAX_STACK_DEPTH]; size_t curr_depth; if (build_frame_stack(writer, frame_list, curr_stack, &curr_depth) < 0) { return -1; } return block_add_sample( writer, entry, timestamp_us, status, curr_stack, curr_depth); } int binary_writer_write_sample(BinaryWriter *writer, PyObject *stack_frames, uint64_t timestamp_us) { if (!PyList_Check(stack_frames)) { PyErr_SetString(PyExc_TypeError, "stack_frames must be a list"); return -1; } Py_ssize_t num_interpreters = PyList_GET_SIZE(stack_frames); for (Py_ssize_t i = 0; i < num_interpreters; i++) { PyObject *interp_info = PyList_GET_ITEM(stack_frames, i); PyObject *interp_id_obj = PyStructSequence_GET_ITEM(interp_info, 0); PyObject *threads = PyStructSequence_GET_ITEM(interp_info, 1); unsigned long long interp_id_long = PyLong_AsUnsignedLongLong(interp_id_obj); if (interp_id_long == (unsigned long long)-1 && PyErr_Occurred()) { return -1; } /* Bounds check: interpreter_id is stored as uint32_t in binary format */ if (interp_id_long > UINT32_MAX) { writer->state = BINARY_WRITER_LIMIT_REACHED; PyErr_Format(PyExc_OverflowError, "interpreter_id %llu exceeds maximum value %u", interp_id_long, UINT32_MAX); return -1; } uint32_t interpreter_id = (uint32_t)interp_id_long; Py_ssize_t num_threads = PyList_GET_SIZE(threads); for (Py_ssize_t j = 0; j < num_threads; j++) { PyObject *thread_info = PyList_GET_ITEM(threads, j); if (process_thread_sample(writer, thread_info, interpreter_id, timestamp_us) < 0) { return -1; } } } return 0; } int binary_writer_finalize(BinaryWriter *writer) { if (block_flush(writer) < 0) { return -1; } /* Use 64-bit file position for >2GB files */ file_offset_t string_table_offset = FTELL64(writer->fp); if (string_table_offset < 0) { PyErr_SetFromErrno(PyExc_IOError); return -1; } /* Release GIL during potentially large writes */ for (size_t i = 0; i < writer->string_count; i++) { uint8_t len_buf[10]; size_t len_size = encode_varint_u32(len_buf, (uint32_t)writer->string_lengths[i]); if (fwrite_checked_allow_threads(len_buf, len_size, writer->fp) < 0 || fwrite_checked_allow_threads(writer->strings[i], writer->string_lengths[i], writer->fp) < 0) { return -1; } } file_offset_t frame_table_offset = FTELL64(writer->fp); if (frame_table_offset < 0) { PyErr_SetFromErrno(PyExc_IOError); return -1; } for (size_t i = 0; i < writer->frame_count; i++) { FrameEntry *entry = &writer->frame_entries[i]; uint8_t buf[64]; /* Increased buffer for additional fields */ size_t pos = encode_varint_u32(buf, entry->filename_idx); pos += encode_varint_u32(buf + pos, entry->funcname_idx); pos += encode_varint_i32(buf + pos, entry->lineno); /* Delta encode end_lineno: store (end_lineno - lineno) as zigzag. * When lineno is -1, store delta as 0 (result will be -1). */ int32_t end_lineno_delta = 0; if (entry->lineno != LOCATION_NOT_AVAILABLE && entry->end_lineno != LOCATION_NOT_AVAILABLE) { end_lineno_delta = entry->end_lineno - entry->lineno; } pos += encode_varint_i32(buf + pos, end_lineno_delta); pos += encode_varint_i32(buf + pos, entry->column); /* Delta encode end_column: store (end_column - column) as zigzag. * When column is -1, store delta as 0 (result will be -1). */ int32_t end_column_delta = 0; if (entry->column != LOCATION_NOT_AVAILABLE && entry->end_column != LOCATION_NOT_AVAILABLE) { end_column_delta = entry->end_column - entry->column; } pos += encode_varint_i32(buf + pos, end_column_delta); buf[pos++] = entry->opcode; if (fwrite_checked_allow_threads(buf, pos, writer->fp) < 0) { return -1; } } if (writer->has_profile_stats) { uint8_t profile_stats[PROFILE_STATS_SIZE] = {0}; uint32_t version = PROFILE_STATS_VERSION; uint32_t size = PROFILE_STATS_SIZE; memcpy(profile_stats + PST_OFF_DURATION, &writer->duration_sec, PST_SIZE_DURATION); memcpy(profile_stats + PST_OFF_SAMPLE_RATE, &writer->sample_rate, PST_SIZE_SAMPLE_RATE); memcpy(profile_stats + PST_OFF_ERROR_RATE, &writer->error_rate, PST_SIZE_ERROR_RATE); memcpy(profile_stats + PST_OFF_MISSED_SAMPLES, &writer->missed_samples, PST_SIZE_MISSED_SAMPLES); memcpy(profile_stats + PST_OFF_PRESENT, &writer->profile_stats_present, PST_SIZE_PRESENT); memcpy(profile_stats + PST_OFF_MAGIC, PROFILE_STATS_MAGIC, PROFILE_STATS_MAGIC_SIZE); memcpy(profile_stats + PST_OFF_VERSION, &version, PST_SIZE_VERSION); memcpy(profile_stats + PST_OFF_SIZE, &size, PST_SIZE_SIZE); if (fwrite_checked_allow_threads( profile_stats, PROFILE_STATS_SIZE, writer->fp) < 0) { return -1; } } /* Footer: string_count(4) + frame_count(4) + file_size(8) + checksum(16) */ file_offset_t footer_offset = FTELL64(writer->fp); if (footer_offset < 0) { PyErr_SetFromErrno(PyExc_IOError); return -1; } uint64_t file_size = (uint64_t)footer_offset + FILE_FOOTER_SIZE; uint8_t footer[FILE_FOOTER_SIZE] = {0}; /* Cast size_t to uint32_t before memcpy to ensure correct bytes are copied * on both little-endian and big-endian systems (size_t is 8 bytes on 64-bit) */ uint32_t string_count_u32 = (uint32_t)writer->string_count; uint32_t frame_count_u32 = (uint32_t)writer->frame_count; memcpy(footer + FTR_OFF_STRINGS, &string_count_u32, FTR_SIZE_STRINGS); memcpy(footer + FTR_OFF_FRAMES, &frame_count_u32, FTR_SIZE_FRAMES); memcpy(footer + FTR_OFF_FILE_SIZE, &file_size, FTR_SIZE_FILE_SIZE); /* checksum (FTR_OFF_CHECKSUM..FILE_FOOTER_SIZE-1): placeholder zeros */ if (fwrite_checked_allow_threads(footer, FILE_FOOTER_SIZE, writer->fp) < 0) { return -1; } if (FSEEK64(writer->fp, 0, SEEK_SET) < 0) { PyErr_SetFromErrno(PyExc_IOError); return -1; } /* Convert file offsets and counts to fixed-width types for portable header format. * This ensures correct behavior on both little-endian and big-endian systems. */ uint64_t string_table_offset_u64 = (uint64_t)string_table_offset; uint64_t frame_table_offset_u64 = (uint64_t)frame_table_offset; uint32_t thread_count_u32 = (uint32_t)writer->thread_count; uint32_t compression_type_u32 = (uint32_t)writer->compression_type; uint32_t profiling_config_u32 = (uint32_t)(writer->profiling_mode + 1); if (writer->capture_features >= 0) { profiling_config_u32 |= PROFILING_CONFIG_FEATURES_KNOWN; profiling_config_u32 |= (uint32_t)writer->capture_features << PROFILING_CONFIG_FEATURES_SHIFT; } uint8_t header[FILE_HEADER_SIZE] = {0}; uint32_t magic = BINARY_FORMAT_MAGIC; uint32_t version = BINARY_FORMAT_VERSION; memcpy(header + HDR_OFF_MAGIC, &magic, HDR_SIZE_MAGIC); memcpy(header + HDR_OFF_VERSION, &version, HDR_SIZE_VERSION); header[HDR_OFF_PY_MAJOR] = PY_MAJOR_VERSION; header[HDR_OFF_PY_MINOR] = PY_MINOR_VERSION; header[HDR_OFF_PY_MICRO] = PY_MICRO_VERSION; memcpy(header + HDR_OFF_START_TIME, &writer->start_time_us, HDR_SIZE_START_TIME); memcpy(header + HDR_OFF_INTERVAL, &writer->sample_interval_us, HDR_SIZE_INTERVAL); memcpy(header + HDR_OFF_SAMPLES, &writer->total_samples, HDR_SIZE_SAMPLES); memcpy(header + HDR_OFF_THREADS, &thread_count_u32, HDR_SIZE_THREADS); memcpy(header + HDR_OFF_STR_TABLE, &string_table_offset_u64, HDR_SIZE_STR_TABLE); memcpy(header + HDR_OFF_FRAME_TABLE, &frame_table_offset_u64, HDR_SIZE_FRAME_TABLE); memcpy(header + HDR_OFF_COMPRESSION, &compression_type_u32, HDR_SIZE_COMPRESSION); memcpy(header + HDR_OFF_CONFIG, &profiling_config_u32, HDR_SIZE_CONFIG); if (fwrite_checked_allow_threads(header, FILE_HEADER_SIZE, writer->fp) < 0) { return -1; } if (Py_fclose(writer->fp) != 0) { writer->fp = NULL; PyErr_SetFromErrno(PyExc_IOError); return -1; } writer->fp = NULL; return 0; } int binary_writer_set_stats(BinaryWriter *writer, double duration_sec, double sample_rate, double error_rate, double missed_samples, uint32_t present) { if (!isfinite(duration_sec) || duration_sec < 0.0) { PyErr_SetString(PyExc_ValueError, "duration must be a finite non-negative value"); return -1; } if (!isfinite(sample_rate) || sample_rate < 0.0) { PyErr_SetString(PyExc_ValueError, "sample rate must be a finite non-negative value"); return -1; } if ((present & PROFILE_STATS_ERROR_RATE) && (!isfinite(error_rate) || error_rate < 0.0)) { PyErr_SetString(PyExc_ValueError, "error rate must be a finite non-negative value"); return -1; } if ((present & PROFILE_STATS_MISSED) && !isfinite(missed_samples)) { PyErr_SetString(PyExc_ValueError, "missed samples must be a finite value"); return -1; } writer->duration_sec = duration_sec; writer->sample_rate = sample_rate; writer->error_rate = error_rate; writer->missed_samples = missed_samples; writer->profile_stats_present = present; writer->has_profile_stats = 1; return 0; } void binary_writer_destroy(BinaryWriter *writer) { if (!writer) { return; } if (writer->fp) { Py_fclose(writer->fp); } block_destroy(writer->block); #ifdef HAVE_ZSTD if (writer->zstd.cctx) { ZSTD_freeCCtx(writer->zstd.cctx); } #endif if (writer->string_hash) { _Py_hashtable_destroy(writer->string_hash); } if (writer->strings) { for (size_t i = 0; i < writer->string_count; i++) { PyMem_Free(writer->strings[i]); } PyMem_Free(writer->strings); } PyMem_Free(writer->string_lengths); if (writer->frame_hash) { _Py_hashtable_destroy(writer->frame_hash); } PyMem_Free(writer->frame_entries); PyMem_Free(writer->thread_entries); PyMem_Free(writer); } #undef PYLONG_TO_INT32_OR_DEFAULT