This is an automated email from the ASF dual-hosted git repository. xiaoxiang781216 pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/nuttx.git
commit 1980e80b4339b14b7979ea1395e73adf36caf8ba Author: wangmingrong1 <[email protected]> AuthorDate: Mon Nov 2 16:08:34 2026 +0800 feat(gcov): add binary pack format support Add support for parsing gcov binary pack files produced by gcov_binfile_dump (COVERAGE_MINI mode). The binary format stores gcda files as [path_len(4B LE)][path+NUL][data_size(4B LE)][data] sequences. - Add is_gcov_binpack_file() heuristic detection - Add parse_gcov_binpack() to extract individual gcda files - Integrate binary pack parsing into the main file dispatch logic Signed-off-by: wangmingrong1 <[email protected]> --- tools/gcov.py | 112 ++++++++++++++++++++++++++++++++++++++++++++++++++++++---- 1 file changed, 105 insertions(+), 7 deletions(-) diff --git a/tools/gcov.py b/tools/gcov.py index 4c871ecde6c..1818bf04629 100755 --- a/tools/gcov.py +++ b/tools/gcov.py @@ -22,6 +22,7 @@ import argparse import os import re import shutil +import struct import subprocess import sys @@ -32,6 +33,13 @@ import sys PROFRAW_MAGIC_64 = b"\xff\x6c\x70\x72\x6f\x66\x72\x81" PROFRAW_MAGIC_32 = b"\xff\x6c\x70\x72\x6f\x66\x52\x81" +# gcov binary pack magic heuristic: +# First 4 bytes are path_len (LE uint32), typically 30~200 for gcda paths. +# Followed by a path string starting with a printable char (not \x00 or \xff). + +GCOV_BINPACK_PATH_LEN_MIN = 10 +GCOV_BINPACK_PATH_LEN_MAX = 4096 + def is_profraw_file(path): """Check if a file is an LLVM profraw file by magic number.""" @@ -125,6 +133,93 @@ def parse_gcda_data(path): return gcda_path +def is_gcov_binpack_file(path): + """Check if a file is a gcov binary pack file. + + Binary format: [path_len(4B LE)][path+NUL][data_size(4B LE)][data]... + Heuristic: first 4 bytes are a small uint32, followed by a printable path. + """ + + try: + with open(path, "rb") as f: + header = f.read(8) + if len(header) < 8: + return False + path_len = struct.unpack_from("<I", header, 0)[0] + if path_len < GCOV_BINPACK_PATH_LEN_MIN: + return False + if path_len > GCOV_BINPACK_PATH_LEN_MAX: + return False + first_char = header[4] + return first_char >= 0x20 and first_char < 0x7F + except (IOError, OSError): + return False + + +def parse_gcov_binpack(path): + """Parse gcov binary pack file into individual gcda files. + + Binary format: [path_len(4B LE)][path+NUL][data_size(4B LE)][data]... + """ + + with open(path, "rb") as f: + data = f.read() + + gcda_path = path + "_covert" + os.makedirs(gcda_path, exist_ok=True) + + offset = 0 + count = 0 + errors = 0 + + while offset < len(data): + if offset + 4 > len(data): + break + + path_len = struct.unpack_from("<I", data, offset)[0] + offset += 4 + + if ( + path_len == 0 + or path_len > GCOV_BINPACK_PATH_LEN_MAX + or offset + path_len > len(data) + ): + print(f"Invalid path_len {path_len} at offset {offset - 4}") + errors += 1 + break + + filepath = data[offset : offset + path_len].rstrip(b"\x00").decode("utf-8") + offset += path_len + + if offset + 4 > len(data): + print(f"Truncated at data_size for {filepath}") + errors += 1 + break + + data_size = struct.unpack_from("<I", data, offset)[0] + offset += 4 + + if offset + data_size > len(data): + print(f"Truncated data for {filepath}") + gcda_data = data[offset:] + offset = len(data) + errors += 1 + else: + gcda_data = data[offset : offset + data_size] + offset += data_size + + out_path = os.path.join(gcda_path, filepath.lstrip("/")) + os.makedirs(os.path.dirname(out_path), exist_ok=True) + + with open(out_path, "wb") as out: + out.write(gcda_data) + + count += 1 + + print(f"Parsed binary pack: {count} files ({errors} errors)") + return gcda_path + + def correct_content_path(file, shield: list, newpath): with open(file, "r", encoding="utf-8") as f: content = f.read() @@ -634,19 +729,22 @@ def main(): debug_file = os.path.join(result_dir, "debug.log") sys.stdout = open(debug_file, "w+") - # Parse hex dump files (works for both gcc and clang) + # Parse dump files (hex dump, binary pack, or profraw) gcda_dirs = [] for i in args.gcda_dir: if os.path.isfile(i): - # Check if it's a binary profraw or a text hex dump log - if is_profraw_file( - os.path.join(root_dir, i) if not os.path.isabs(i) else i - ): - # Binary profraw file, use directly + abs_path = os.path.join(root_dir, i) if not os.path.isabs(i) else i + if is_profraw_file(abs_path): + # Binary profraw file (clang), use directly gcda_dirs.append(os.path.abspath(i)) + elif is_gcov_binpack_file(abs_path): + # Binary pack file from COVERAGE_MINI gcov_binfile_dump + gcda_dirs.append(parse_gcov_binpack(abs_path)) + if args.delete: + os.remove(i) else: # Text file: try to parse as hex dump log - gcda_dirs.append(parse_gcda_data(os.path.join(root_dir, i))) + gcda_dirs.append(parse_gcda_data(abs_path)) if args.delete: os.remove(i) else:
