|
| 1 | +From f6e8423fca302558e250b6538d092678b2a27003 Mon Sep 17 00:00:00 2001 |
| 2 | +From: Alexander Urieles < [email protected]> |
| 3 | +Date: Mon, 28 Jul 2025 17:37:26 +0200 |
| 4 | +Subject: [PATCH] gh-130577: tarfile now validates archives to ensure member |
| 5 | + offsets are non-negative (GH-137027) (cherry picked from commit |
| 6 | + 7040aa54f14676938970e10c5f74ea93cd56aa38) |
| 7 | + |
| 8 | +Co-authored-by: Alexander Urieles < [email protected]> |
| 9 | +Co-authored-by: Gregory P. Smith < [email protected]> |
| 10 | +Signed-off-by: Azure Linux Security Servicing Account < [email protected]> |
| 11 | +Upstream-reference: https://github.com/python/cpython/pull/137171.patch |
| 12 | +--- |
| 13 | + Lib/tarfile.py | 3 + |
| 14 | + Lib/test/test_tarfile.py | 156 ++++++++++++++++++ |
| 15 | + ...-07-23-00-35-29.gh-issue-130577.c7EITy.rst | 3 + |
| 16 | + 3 files changed, 162 insertions(+) |
| 17 | + create mode 100644 Misc/NEWS.d/next/Library/2025-07-23-00-35-29.gh-issue-130577.c7EITy.rst |
| 18 | + |
| 19 | +diff --git a/Lib/tarfile.py b/Lib/tarfile.py |
| 20 | +index 9999a99..59d3f6e 100755 |
| 21 | +--- a/Lib/tarfile.py |
| 22 | ++++ b/Lib/tarfile.py |
| 23 | +@@ -1615,6 +1615,9 @@ class TarInfo(object): |
| 24 | + """Round up a byte count by BLOCKSIZE and return it, |
| 25 | + e.g. _block(834) => 1024. |
| 26 | + """ |
| 27 | ++ # Only non-negative offsets are allowed |
| 28 | ++ if count < 0: |
| 29 | ++ raise InvalidHeaderError("invalid offset") |
| 30 | + blocks, remainder = divmod(count, BLOCKSIZE) |
| 31 | + if remainder: |
| 32 | + blocks += 1 |
| 33 | +diff --git a/Lib/test/test_tarfile.py b/Lib/test/test_tarfile.py |
| 34 | +index 1c598e1..a6925bf 100644 |
| 35 | +--- a/Lib/test/test_tarfile.py |
| 36 | ++++ b/Lib/test/test_tarfile.py |
| 37 | +@@ -50,6 +50,7 @@ bz2name = os.path.join(TEMPDIR, "testtar.tar.bz2") |
| 38 | + xzname = os.path.join(TEMPDIR, "testtar.tar.xz") |
| 39 | + tmpname = os.path.join(TEMPDIR, "tmp.tar") |
| 40 | + dotlessname = os.path.join(TEMPDIR, "testtar") |
| 41 | ++SPACE = b" " |
| 42 | + |
| 43 | + sha256_regtype = ( |
| 44 | + "e09e4bc8b3c9d9177e77256353b36c159f5f040531bbd4b024a8f9b9196c71ce" |
| 45 | +@@ -4485,6 +4486,161 @@ class OverwriteTests(archiver_tests.OverwriteTests, unittest.TestCase): |
| 46 | + ar.extractall(self.testdir, filter='fully_trusted') |
| 47 | + |
| 48 | + |
| 49 | ++class OffsetValidationTests(unittest.TestCase): |
| 50 | ++ tarname = tmpname |
| 51 | ++ invalid_posix_header = ( |
| 52 | ++ # name: 100 bytes |
| 53 | ++ tarfile.NUL * tarfile.LENGTH_NAME |
| 54 | ++ # mode, space, null terminator: 8 bytes |
| 55 | ++ + b"000755" + SPACE + tarfile.NUL |
| 56 | ++ # uid, space, null terminator: 8 bytes |
| 57 | ++ + b"000001" + SPACE + tarfile.NUL |
| 58 | ++ # gid, space, null terminator: 8 bytes |
| 59 | ++ + b"000001" + SPACE + tarfile.NUL |
| 60 | ++ # size, space: 12 bytes |
| 61 | ++ + b"\xff" * 11 + SPACE |
| 62 | ++ # mtime, space: 12 bytes |
| 63 | ++ + tarfile.NUL * 11 + SPACE |
| 64 | ++ # chksum: 8 bytes |
| 65 | ++ + b"0011407" + tarfile.NUL |
| 66 | ++ # type: 1 byte |
| 67 | ++ + tarfile.REGTYPE |
| 68 | ++ # linkname: 100 bytes |
| 69 | ++ + tarfile.NUL * tarfile.LENGTH_LINK |
| 70 | ++ # magic: 6 bytes, version: 2 bytes |
| 71 | ++ + tarfile.POSIX_MAGIC |
| 72 | ++ # uname: 32 bytes |
| 73 | ++ + tarfile.NUL * 32 |
| 74 | ++ # gname: 32 bytes |
| 75 | ++ + tarfile.NUL * 32 |
| 76 | ++ # devmajor, space, null terminator: 8 bytes |
| 77 | ++ + tarfile.NUL * 6 + SPACE + tarfile.NUL |
| 78 | ++ # devminor, space, null terminator: 8 bytes |
| 79 | ++ + tarfile.NUL * 6 + SPACE + tarfile.NUL |
| 80 | ++ # prefix: 155 bytes |
| 81 | ++ + tarfile.NUL * tarfile.LENGTH_PREFIX |
| 82 | ++ # padding: 12 bytes |
| 83 | ++ + tarfile.NUL * 12 |
| 84 | ++ ) |
| 85 | ++ invalid_gnu_header = ( |
| 86 | ++ # name: 100 bytes |
| 87 | ++ tarfile.NUL * tarfile.LENGTH_NAME |
| 88 | ++ # mode, null terminator: 8 bytes |
| 89 | ++ + b"0000755" + tarfile.NUL |
| 90 | ++ # uid, null terminator: 8 bytes |
| 91 | ++ + b"0000001" + tarfile.NUL |
| 92 | ++ # gid, space, null terminator: 8 bytes |
| 93 | ++ + b"0000001" + tarfile.NUL |
| 94 | ++ # size, space: 12 bytes |
| 95 | ++ + b"\xff" * 11 + SPACE |
| 96 | ++ # mtime, space: 12 bytes |
| 97 | ++ + tarfile.NUL * 11 + SPACE |
| 98 | ++ # chksum: 8 bytes |
| 99 | ++ + b"0011327" + tarfile.NUL |
| 100 | ++ # type: 1 byte |
| 101 | ++ + tarfile.REGTYPE |
| 102 | ++ # linkname: 100 bytes |
| 103 | ++ + tarfile.NUL * tarfile.LENGTH_LINK |
| 104 | ++ # magic: 8 bytes |
| 105 | ++ + tarfile.GNU_MAGIC |
| 106 | ++ # uname: 32 bytes |
| 107 | ++ + tarfile.NUL * 32 |
| 108 | ++ # gname: 32 bytes |
| 109 | ++ + tarfile.NUL * 32 |
| 110 | ++ # devmajor, null terminator: 8 bytes |
| 111 | ++ + tarfile.NUL * 8 |
| 112 | ++ # devminor, null terminator: 8 bytes |
| 113 | ++ + tarfile.NUL * 8 |
| 114 | ++ # padding: 167 bytes |
| 115 | ++ + tarfile.NUL * 167 |
| 116 | ++ ) |
| 117 | ++ invalid_v7_header = ( |
| 118 | ++ # name: 100 bytes |
| 119 | ++ tarfile.NUL * tarfile.LENGTH_NAME |
| 120 | ++ # mode, space, null terminator: 8 bytes |
| 121 | ++ + b"000755" + SPACE + tarfile.NUL |
| 122 | ++ # uid, space, null terminator: 8 bytes |
| 123 | ++ + b"000001" + SPACE + tarfile.NUL |
| 124 | ++ # gid, space, null terminator: 8 bytes |
| 125 | ++ + b"000001" + SPACE + tarfile.NUL |
| 126 | ++ # size, space: 12 bytes |
| 127 | ++ + b"\xff" * 11 + SPACE |
| 128 | ++ # mtime, space: 12 bytes |
| 129 | ++ + tarfile.NUL * 11 + SPACE |
| 130 | ++ # chksum: 8 bytes |
| 131 | ++ + b"0010070" + tarfile.NUL |
| 132 | ++ # type: 1 byte |
| 133 | ++ + tarfile.REGTYPE |
| 134 | ++ # linkname: 100 bytes |
| 135 | ++ + tarfile.NUL * tarfile.LENGTH_LINK |
| 136 | ++ # padding: 255 bytes |
| 137 | ++ + tarfile.NUL * 255 |
| 138 | ++ ) |
| 139 | ++ valid_gnu_header = tarfile.TarInfo("filename").tobuf(tarfile.GNU_FORMAT) |
| 140 | ++ data_block = b"\xff" * tarfile.BLOCKSIZE |
| 141 | ++ |
| 142 | ++ def _write_buffer(self, buffer): |
| 143 | ++ with open(self.tarname, "wb") as f: |
| 144 | ++ f.write(buffer) |
| 145 | ++ |
| 146 | ++ def _get_members(self, ignore_zeros=None): |
| 147 | ++ with open(self.tarname, "rb") as f: |
| 148 | ++ with tarfile.open( |
| 149 | ++ mode="r", fileobj=f, ignore_zeros=ignore_zeros |
| 150 | ++ ) as tar: |
| 151 | ++ return tar.getmembers() |
| 152 | ++ |
| 153 | ++ def _assert_raises_read_error_exception(self): |
| 154 | ++ with self.assertRaisesRegex( |
| 155 | ++ tarfile.ReadError, "file could not be opened successfully" |
| 156 | ++ ): |
| 157 | ++ self._get_members() |
| 158 | ++ |
| 159 | ++ def test_invalid_offset_header_validations(self): |
| 160 | ++ for tar_format, invalid_header in ( |
| 161 | ++ ("posix", self.invalid_posix_header), |
| 162 | ++ ("gnu", self.invalid_gnu_header), |
| 163 | ++ ("v7", self.invalid_v7_header), |
| 164 | ++ ): |
| 165 | ++ with self.subTest(format=tar_format): |
| 166 | ++ self._write_buffer(invalid_header) |
| 167 | ++ self._assert_raises_read_error_exception() |
| 168 | ++ |
| 169 | ++ def test_early_stop_at_invalid_offset_header(self): |
| 170 | ++ buffer = self.valid_gnu_header + self.invalid_gnu_header + self.valid_gnu_header |
| 171 | ++ self._write_buffer(buffer) |
| 172 | ++ members = self._get_members() |
| 173 | ++ self.assertEqual(len(members), 1) |
| 174 | ++ self.assertEqual(members[0].name, "filename") |
| 175 | ++ self.assertEqual(members[0].offset, 0) |
| 176 | ++ |
| 177 | ++ def test_ignore_invalid_archive(self): |
| 178 | ++ # 3 invalid headers with their respective data |
| 179 | ++ buffer = (self.invalid_gnu_header + self.data_block) * 3 |
| 180 | ++ self._write_buffer(buffer) |
| 181 | ++ members = self._get_members(ignore_zeros=True) |
| 182 | ++ self.assertEqual(len(members), 0) |
| 183 | ++ |
| 184 | ++ def test_ignore_invalid_offset_headers(self): |
| 185 | ++ for first_block, second_block, expected_offset in ( |
| 186 | ++ ( |
| 187 | ++ (self.valid_gnu_header), |
| 188 | ++ (self.invalid_gnu_header + self.data_block), |
| 189 | ++ 0, |
| 190 | ++ ), |
| 191 | ++ ( |
| 192 | ++ (self.invalid_gnu_header + self.data_block), |
| 193 | ++ (self.valid_gnu_header), |
| 194 | ++ 1024, |
| 195 | ++ ), |
| 196 | ++ ): |
| 197 | ++ self._write_buffer(first_block + second_block) |
| 198 | ++ members = self._get_members(ignore_zeros=True) |
| 199 | ++ self.assertEqual(len(members), 1) |
| 200 | ++ self.assertEqual(members[0].name, "filename") |
| 201 | ++ self.assertEqual(members[0].offset, expected_offset) |
| 202 | ++ |
| 203 | ++ |
| 204 | + def setUpModule(): |
| 205 | + os_helper.unlink(TEMPDIR) |
| 206 | + os.makedirs(TEMPDIR) |
| 207 | +diff --git a/Misc/NEWS.d/next/Library/2025-07-23-00-35-29.gh-issue-130577.c7EITy.rst b/Misc/NEWS.d/next/Library/2025-07-23-00-35-29.gh-issue-130577.c7EITy.rst |
| 208 | +new file mode 100644 |
| 209 | +index 0000000..342cabb |
| 210 | +--- /dev/null |
| 211 | ++++ b/Misc/NEWS.d/next/Library/2025-07-23-00-35-29.gh-issue-130577.c7EITy.rst |
| 212 | +@@ -0,0 +1,3 @@ |
| 213 | ++:mod:`tarfile` now validates archives to ensure member offsets are |
| 214 | ++non-negative. (Contributed by Alexander Enrique Urieles Nieto in |
| 215 | ++:gh:`130577`.) |
| 216 | +-- |
| 217 | +2.45.4 |
| 218 | + |
0 commit comments