dotnet6.0/check-debug-symbols.py

137 lines
4.5 KiB
Python
Executable File

#!/usr/bin/python3
"""
Check debug symbols are present in shared object and can identify
code.
It starts scanning from a directory and recursively scans all ELF
files found in it for various symbols to ensure all debuginfo is
present and nothing has been stripped.
Usage:
./check-debug-symbols /path/of/dir/to/scan/
Example:
./check-debug-symbols /usr/lib64
"""
# This technique was explained to me by Mark Wielaard (mjw).
import collections
import os
import re
import subprocess
import sys
ScanResult = collections.namedtuple('ScanResult',
'file_name debug_info debug_abbrev file_symbols gnu_debuglink')
def scan_file(file):
"Scan the provided file and return a ScanResult containing results of the scan."
# Test for .debug_* sections in the shared object. This is the main test.
# Stripped objects will not contain these.
readelf_S_result = subprocess.run(['eu-readelf', '-S', file],
stdout=subprocess.PIPE, encoding='utf-8', check=True)
has_debug_info = any(line for line in readelf_S_result.stdout.split('\n') if '] .debug_info' in line)
has_debug_abbrev = any(line for line in readelf_S_result.stdout.split('\n') if '] .debug_abbrev' in line)
# Test FILE symbols. These will most likely be removed by anyting that
# manipulates symbol tables because it's generally useless. So a nice test
# that nothing has messed with symbols.
def contains_file_symbols(line):
parts = line.split()
if len(parts) < 8:
return False
return \
parts[2] == '0' and parts[3] == 'FILE' and parts[4] == 'LOCAL' and parts[5] == 'DEFAULT' and \
parts[6] == 'ABS' and re.match(r'((.*/)?[-_a-zA-Z0-9]+\.(c|cc|cpp|cxx))?', parts[7])
readelf_s_result = subprocess.run(["eu-readelf", '-s', file],
stdout=subprocess.PIPE, encoding='utf-8', check=True)
has_file_symbols = any(line for line in readelf_s_result.stdout.split('\n') if contains_file_symbols(line))
# Test that there are no .gnu_debuglink sections pointing to another
# debuginfo file. There shouldn't be any debuginfo files, so the link makes
# no sense either.
has_gnu_debuglink = any(line for line in readelf_s_result.stdout.split('\n') if '] .gnu_debuglink' in line)
return ScanResult(file, has_debug_info, has_debug_abbrev, has_file_symbols, has_gnu_debuglink)
def is_elf(file):
result = subprocess.run(['file', file], stdout=subprocess.PIPE, encoding='utf-8', check=True)
return re.search(r'ELF 64-bit [LM]SB (?:pie )?(?:executable|shared object)', result.stdout)
def scan_file_if_sensible(file):
if is_elf(file):
# print(file)
return scan_file(file)
return None
def scan_dir(dir):
results = []
for root, _, files in os.walk(dir):
for name in files:
result = scan_file_if_sensible(os.path.join(root, name))
if result:
results.append(result)
return results
def scan(file):
file = os.path.abspath(file)
if os.path.isdir(file):
return scan_dir(file)
elif os.path.isfile(file):
return [scan_file_if_sensible(file)]
def is_bad_result(result):
return not result.debug_info or not result.debug_abbrev or not result.file_symbols or result.gnu_debuglink
def print_scan_results(results, verbose):
# print(results)
for result in results:
file_name = result.file_name
found_issue = False
if not result.debug_info:
found_issue = True
print('error: missing .debug_info section in', file_name)
if not result.debug_abbrev:
found_issue = True
print('error: missing .debug_abbrev section in', file_name)
if not result.file_symbols:
found_issue = True
print('error: missing FILE symbols in', file_name)
if result.gnu_debuglink:
found_issue = True
print('error: unexpected .gnu_debuglink section in', file_name)
if verbose and not found_issue:
print('OK: ', file_name)
def main(args):
verbose = False
files = []
for arg in args:
if arg == '--verbose' or arg == '-v':
verbose = True
else:
files.append(arg)
results = []
for file in files:
results.extend(scan(file))
print_scan_results(results, verbose)
if any(is_bad_result(result) for result in results):
return 1
return 0
if __name__ == '__main__':
sys.exit(main(sys.argv[1:]))