fix: drain archive padding before waiting on decoders

This commit is contained in:
maxjivi05 committed 2026-11-04 14:13:56 -05:00
1 parent 527fadd3b3
commit 6ea7b10d94
2 files changed
+47

No files matched your search

+9
View File
@@ -34,6 +34,12 @@ def data_file(member, magic):
return False
def finish_archive(archive):
for block in iter(lambda: archive.fileobj.read(65536), b''):
if block.strip(b'\0'):
raise ValueError('Nonzero data follows the tar end marker')
def inventory(source, zstd):
process = subprocess.Popen([zstd, '-dc', str(source)], stdout=subprocess.PIPE)
records = []
@@ -65,6 +71,7 @@ def inventory(source, zstd):
{key: value for key, value in member.pax_headers.items() if key not in
{'path', 'linkpath', 'mtime', 'size', 'uid', 'gid', 'uname', 'gname'}},
digest.hexdigest() if member.isfile() else None])
finish_archive(archive)
if process.wait():
raise ValueError('Archive decompression failed')
finally:
@@ -106,6 +113,7 @@ def prepare(source, code, data, zstd_version='1.5.5'):
encoder.stdin.write(block)
remaining -= len(block)
encoder.stdin.write(b'\0' * (-member.size % 512))
finish_archive(archive)
if kind == 'data':
encoder.stdin.write(b'\0' * 10240)
encoder.stdin.close()
@@ -140,6 +148,7 @@ def validate_data(source):
if not member.isfile() or not data_file(member, archive.extractfile(member).read(16)):
raise ValueError('Executable or unclassified content in a Play data segment')
found = True
finish_archive(archive)
if process.wait() or not found:
raise ValueError('Invalid Play data segment')
finally:
@@ -2,6 +2,7 @@ import io
import re
from pathlib import Path
import subprocess
import sys
import tarfile
import tempfile
import unittest
@@ -61,3 +62,40 @@ class PlaySplitArchiveTest(unittest.TestCase):
source = self.archive([('../font.ttf', b'\x00\x01\x00\x00font', 0o644)])
with self.assertRaisesRegex(ValueError, 'unsafe'):
inventory(source, 'zstd')
def test_padded_archives_finish_with_small_decoder_pipes(self):
source = self.archive([('program', b'\x7fELFcode', 0o755),
('font.ttf', b'\x00\x01\x00\x00font', 0o644)])
raw = subprocess.check_output(['zstd', '-dc', str(source)])
source.write_bytes(subprocess.check_output(['zstd', '-c'], input=raw + b'\0' * (1024 * 1024)))
script = '''
import subprocess
import sys
from pathlib import Path
from unittest.mock import patch
from tools.playstore.split_archive import inventory, prepare, combine, validate_data
original = subprocess.Popen
def spawn(*args, **kwargs):
if kwargs.get('stdout') == subprocess.PIPE or kwargs.get('stdin') == subprocess.PIPE:
kwargs['pipesize'] = 4096
return original(*args, **kwargs)
root = Path(sys.argv[1])
with patch.object(subprocess, 'Popen', spawn):
expected = inventory(root / 'source', 'zstd')[0]
prepare(root / 'source', root / 'code', root / 'data', sys.argv[2])
combine([root / 'code', root / 'data'], root / 'combined')
assert inventory(root / 'combined', 'zstd')[0] == expected
validate_data(root / 'data')
print('padded archive completed')
'''
result = subprocess.run([sys.executable, '-c', script, str(self.root), self.zstd_version],
capture_output=True, text=True, timeout=15)
self.assertEqual(0, result.returncode, result.stdout + result.stderr)
self.assertIn('padded archive completed', result.stdout)
def test_data_after_the_end_marker_is_rejected(self):
source = self.archive([('font.ttf', b'\x00\x01\x00\x00font', 0o644)])
raw = subprocess.check_output(['zstd', '-dc', str(source)])
source.write_bytes(subprocess.check_output(['zstd', '-c'], input=raw + b'\x7fELFhidden'))
with self.assertRaisesRegex(ValueError, 'Nonzero data'):
validate_data(source)