diff --git a/CVE-2025-4516-DecodeError-handler.patch b/CVE-2025-4516-DecodeError-handler.patch new file mode 100644 index 0000000..c007aef --- /dev/null +++ b/CVE-2025-4516-DecodeError-handler.patch @@ -0,0 +1,517 @@ +From a75953b347716fff694aa59a7c7c2489fa50d1f5 Mon Sep 17 00:00:00 2001 +From: Serhiy Storchaka +Date: Tue, 20 May 2025 15:46:57 +0300 +Subject: [PATCH] [3.12] gh-133767: Fix use-after-free in the unicode-escape + decoder with an error handler (GH-129648) (GH-133944) + +If the error handler is used, a new bytes object is created to set as +the object attribute of UnicodeDecodeError, and that bytes object then +replaces the original data. A pointer to the decoded data will became invalid +after destroying that temporary bytes object. So we need other way to return +the first invalid escape from _PyUnicode_DecodeUnicodeEscapeInternal(). + +_PyBytes_DecodeEscape() does not have such issue, because it does not +use the error handlers registry, but it should be changed for compatibility +with _PyUnicode_DecodeUnicodeEscapeInternal(). +(cherry picked from commit 9f69a58623bd01349a18ba0c7a9cb1dad6a51e8e) +(cherry picked from commit 6279eb8c076d89d3739a6edb393e43c7929b429d) + +Co-authored-by: Serhiy Storchaka +--- + Include/cpython/bytesobject.h | 4 + Include/cpython/unicodeobject.h | 13 ++ + Lib/test/test_codeccallbacks.py | 39 ++++++ + Lib/test/test_codecs.py | 52 ++++++-- + Misc/NEWS.d/next/Security/2025-05-09-20-22-54.gh-issue-133767.kN2i3Q.rst | 2 + Objects/bytesobject.c | 54 +++++--- + Objects/unicodeobject.c | 61 +++++++--- + Parser/string_parser.c | 26 ++-- + 8 files changed, 194 insertions(+), 57 deletions(-) + create mode 100644 Misc/NEWS.d/next/Security/2025-05-09-20-22-54.gh-issue-133767.kN2i3Q.rst + +Index: Python-3.11.12/Include/cpython/bytesobject.h +=================================================================== +--- Python-3.11.12.orig/Include/cpython/bytesobject.h 2025-04-08 14:15:29.000000000 +0000 ++++ Python-3.11.12/Include/cpython/bytesobject.h 2025-05-22 12:57:58.382969999 +0000 +@@ -25,6 +25,10 @@ + int use_bytearray); + + /* Helper for PyBytes_DecodeEscape that detects invalid escape chars. */ ++PyAPI_FUNC(PyObject*) _PyBytes_DecodeEscape2(const char *, Py_ssize_t, ++ const char *, ++ int *, const char **); ++// Export for binary compatibility. + PyAPI_FUNC(PyObject *) _PyBytes_DecodeEscape(const char *, Py_ssize_t, + const char *, const char **); + +Index: Python-3.11.12/Include/cpython/unicodeobject.h +=================================================================== +--- Python-3.11.12.orig/Include/cpython/unicodeobject.h 2025-04-08 14:15:29.000000000 +0000 ++++ Python-3.11.12/Include/cpython/unicodeobject.h 2025-05-22 12:57:58.383963607 +0000 +@@ -914,6 +914,19 @@ + ); + /* Helper for PyUnicode_DecodeUnicodeEscape that detects invalid escape + chars. */ ++PyAPI_FUNC(PyObject*) _PyUnicode_DecodeUnicodeEscapeInternal2( ++ const char *string, /* Unicode-Escape encoded string */ ++ Py_ssize_t length, /* size of string */ ++ const char *errors, /* error handling */ ++ Py_ssize_t *consumed, /* bytes consumed */ ++ int *first_invalid_escape_char, /* on return, if not -1, contain the first ++ invalid escaped char (<= 0xff) or invalid ++ octal escape (> 0xff) in string. */ ++ const char **first_invalid_escape_ptr); /* on return, if not NULL, may ++ point to the first invalid escaped ++ char in string. ++ May be NULL if errors is not NULL. */ ++// Export for binary compatibility. + PyAPI_FUNC(PyObject*) _PyUnicode_DecodeUnicodeEscapeInternal( + const char *string, /* Unicode-Escape encoded string */ + Py_ssize_t length, /* size of string */ +Index: Python-3.11.12/Lib/test/test_codeccallbacks.py +=================================================================== +--- Python-3.11.12.orig/Lib/test/test_codeccallbacks.py 2025-05-22 12:57:49.587342695 +0000 ++++ Python-3.11.12/Lib/test/test_codeccallbacks.py 2025-05-22 12:57:58.384369150 +0000 +@@ -1,6 +1,7 @@ + import codecs + import html.entities + import itertools ++import re + import sys + import unicodedata + import unittest +@@ -1124,7 +1125,7 @@ + text = 'abcghi'*n + text.translate(charmap) + +- def test_mutatingdecodehandler(self): ++ def test_mutating_decode_handler(self): + baddata = [ + ("ascii", b"\xff"), + ("utf-7", b"++"), +@@ -1159,6 +1160,42 @@ + for (encoding, data) in baddata: + self.assertEqual(data.decode(encoding, "test.mutating"), "\u4242") + ++ def test_mutating_decode_handler_unicode_escape(self): ++ decode = codecs.unicode_escape_decode ++ def mutating(exc): ++ if isinstance(exc, UnicodeDecodeError): ++ r = data.get(exc.object[:exc.end]) ++ if r is not None: ++ exc.object = r[0] + exc.object[exc.end:] ++ return ('\u0404', r[1]) ++ raise AssertionError("don't know how to handle %r" % exc) ++ ++ codecs.register_error('test.mutating2', mutating) ++ data = { ++ br'\x0': (b'\\', 0), ++ br'\x3': (b'xxx\\', 3), ++ br'\x5': (b'x\\', 1), ++ } ++ def check(input, expected, msg): ++ with self.assertWarns(DeprecationWarning) as cm: ++ self.assertEqual(decode(input, 'test.mutating2'), (expected, len(input))) ++ self.assertIn(msg, str(cm.warning)) ++ ++ check(br'\x0n\z', '\u0404\n\\z', r"invalid escape sequence '\z'") ++ check(br'\x0n\501', '\u0404\n\u0141', r"invalid octal escape sequence '\501'") ++ check(br'\x0z', '\u0404\\z', r"invalid escape sequence '\z'") ++ ++ check(br'\x3n\zr', '\u0404\n\\zr', r"invalid escape sequence '\z'") ++ check(br'\x3zr', '\u0404\\zr', r"invalid escape sequence '\z'") ++ check(br'\x3z5', '\u0404\\z5', r"invalid escape sequence '\z'") ++ check(memoryview(br'\x3z5x')[:-1], '\u0404\\z5', r"invalid escape sequence '\z'") ++ check(memoryview(br'\x3z5xy')[:-2], '\u0404\\z5', r"invalid escape sequence '\z'") ++ ++ check(br'\x5n\z', '\u0404\n\\z', r"invalid escape sequence '\z'") ++ check(br'\x5n\501', '\u0404\n\u0141', r"invalid octal escape sequence '\501'") ++ check(br'\x5z', '\u0404\\z', r"invalid escape sequence '\z'") ++ check(memoryview(br'\x5zy')[:-1], '\u0404\\z', r"invalid escape sequence '\z'") ++ + # issue32583 + def test_crashing_decode_handler(self): + # better generating one more character to fill the extra space slot +Index: Python-3.11.12/Lib/test/test_codecs.py +=================================================================== +--- Python-3.11.12.orig/Lib/test/test_codecs.py 2025-05-22 12:57:49.608177948 +0000 ++++ Python-3.11.12/Lib/test/test_codecs.py 2025-05-22 12:57:58.385050493 +0000 +@@ -1198,23 +1198,39 @@ + check(br"[\1010]", b"[A0]") + check(br"[\x41]", b"[A]") + check(br"[\x410]", b"[A0]") ++ ++ def test_warnings(self): ++ decode = codecs.escape_decode ++ check = coding_checker(self, decode) + for i in range(97, 123): + b = bytes([i]) + if b not in b'abfnrtvx': +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\%c'" % i): + check(b"\\" + b, b"\\" + b) +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\%c'" % (i-32)): + check(b"\\" + b.upper(), b"\\" + b.upper()) +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\8'"): + check(br"\8", b"\\8") + with self.assertWarns(DeprecationWarning): + check(br"\9", b"\\9") +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\\xfa'") as cm: + check(b"\\\xfa", b"\\\xfa") + for i in range(0o400, 0o1000): +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid octal escape sequence '\\%o'" % i): + check(rb'\%o' % i, bytes([i & 0o377])) + ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\z'"): ++ self.assertEqual(decode(br'\x\z', 'ignore'), (b'\\z', 4)) ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid octal escape sequence '\\501'"): ++ self.assertEqual(decode(br'\x\501', 'ignore'), (b'A', 6)) ++ + def test_errors(self): + decode = codecs.escape_decode + self.assertRaises(ValueError, decode, br"\x") +@@ -2487,24 +2503,40 @@ + check(br"[\x410]", "[A0]") + check(br"\u20ac", "\u20ac") + check(br"\U0001d120", "\U0001d120") ++ ++ def test_decode_warnings(self): ++ decode = codecs.unicode_escape_decode ++ check = coding_checker(self, decode) + for i in range(97, 123): + b = bytes([i]) + if b not in b'abfnrtuvx': +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\%c'" % i): + check(b"\\" + b, "\\" + chr(i)) + if b.upper() not in b'UN': +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\%c'" % (i-32)): + check(b"\\" + b.upper(), "\\" + chr(i-32)) +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\8'"): + check(br"\8", "\\8") + with self.assertWarns(DeprecationWarning): + check(br"\9", "\\9") +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\\xfa'") as cm: + check(b"\\\xfa", "\\\xfa") + for i in range(0o400, 0o1000): +- with self.assertWarns(DeprecationWarning): ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid octal escape sequence '\\%o'" % i): + check(rb'\%o' % i, chr(i)) + ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid escape sequence '\\z'"): ++ self.assertEqual(decode(br'\x\z', 'ignore'), ('\\z', 4)) ++ with self.assertWarnsRegex(DeprecationWarning, ++ r"invalid octal escape sequence '\\501'"): ++ self.assertEqual(decode(br'\x\501', 'ignore'), ('\u0141', 6)) ++ + def test_decode_errors(self): + decode = codecs.unicode_escape_decode + for c, d in (b'x', 2), (b'u', 4), (b'U', 4): +Index: Python-3.11.12/Misc/NEWS.d/next/Security/2025-05-09-20-22-54.gh-issue-133767.kN2i3Q.rst +=================================================================== +--- /dev/null 1970-01-01 00:00:00.000000000 +0000 ++++ Python-3.11.12/Misc/NEWS.d/next/Security/2025-05-09-20-22-54.gh-issue-133767.kN2i3Q.rst 2025-05-22 12:57:58.385668586 +0000 +@@ -0,0 +1,2 @@ ++Fix use-after-free in the "unicode-escape" decoder with a non-"strict" error ++handler. +Index: Python-3.11.12/Objects/bytesobject.c +=================================================================== +--- Python-3.11.12.orig/Objects/bytesobject.c 2025-04-08 14:15:29.000000000 +0000 ++++ Python-3.11.12/Objects/bytesobject.c 2025-05-22 12:57:58.386149592 +0000 +@@ -1057,10 +1057,11 @@ + } + + /* Unescape a backslash-escaped string. */ +-PyObject *_PyBytes_DecodeEscape(const char *s, ++PyObject *_PyBytes_DecodeEscape2(const char *s, + Py_ssize_t len, + const char *errors, +- const char **first_invalid_escape) ++ int *first_invalid_escape_char, ++ const char **first_invalid_escape_ptr) + { + int c; + char *p; +@@ -1074,7 +1075,8 @@ + return NULL; + writer.overallocate = 1; + +- *first_invalid_escape = NULL; ++ *first_invalid_escape_char = -1; ++ *first_invalid_escape_ptr = NULL; + + end = s + len; + while (s < end) { +@@ -1112,9 +1114,10 @@ + c = (c<<3) + *s++ - '0'; + } + if (c > 0377) { +- if (*first_invalid_escape == NULL) { +- *first_invalid_escape = s-3; /* Back up 3 chars, since we've +- already incremented s. */ ++ if (*first_invalid_escape_char == -1) { ++ *first_invalid_escape_char = c; ++ /* Back up 3 chars, since we've already incremented s. */ ++ *first_invalid_escape_ptr = s - 3; + } + } + *p++ = c; +@@ -1155,9 +1158,10 @@ + break; + + default: +- if (*first_invalid_escape == NULL) { +- *first_invalid_escape = s-1; /* Back up one char, since we've +- already incremented s. */ ++ if (*first_invalid_escape_char == -1) { ++ *first_invalid_escape_char = (unsigned char)s[-1]; ++ /* Back up one char, since we've already incremented s. */ ++ *first_invalid_escape_ptr = s - 1; + } + *p++ = '\\'; + s--; +@@ -1171,23 +1175,37 @@ + return NULL; + } + ++// Export for binary compatibility. ++PyObject *_PyBytes_DecodeEscape(const char *s, ++ Py_ssize_t len, ++ const char *errors, ++ const char **first_invalid_escape) ++{ ++ int first_invalid_escape_char; ++ return _PyBytes_DecodeEscape2( ++ s, len, errors, ++ &first_invalid_escape_char, ++ first_invalid_escape); ++} ++ + PyObject *PyBytes_DecodeEscape(const char *s, + Py_ssize_t len, + const char *errors, + Py_ssize_t Py_UNUSED(unicode), + const char *Py_UNUSED(recode_encoding)) + { +- const char* first_invalid_escape; +- PyObject *result = _PyBytes_DecodeEscape(s, len, errors, +- &first_invalid_escape); ++ int first_invalid_escape_char; ++ const char *first_invalid_escape_ptr; ++ PyObject *result = _PyBytes_DecodeEscape2(s, len, errors, ++ &first_invalid_escape_char, ++ &first_invalid_escape_ptr); + if (result == NULL) + return NULL; +- if (first_invalid_escape != NULL) { +- unsigned char c = *first_invalid_escape; +- if ('4' <= c && c <= '7') { ++ if (first_invalid_escape_char != -1) { ++ if (first_invalid_escape_char > 0xff) { + if (PyErr_WarnFormat(PyExc_DeprecationWarning, 1, +- "invalid octal escape sequence '\\%.3s'", +- first_invalid_escape) < 0) ++ "invalid octal escape sequence '\\%o'", ++ first_invalid_escape_char) < 0) + { + Py_DECREF(result); + return NULL; +@@ -1196,7 +1214,7 @@ + else { + if (PyErr_WarnFormat(PyExc_DeprecationWarning, 1, + "invalid escape sequence '\\%c'", +- c) < 0) ++ first_invalid_escape_char) < 0) + { + Py_DECREF(result); + return NULL; +Index: Python-3.11.12/Objects/unicodeobject.c +=================================================================== +--- Python-3.11.12.orig/Objects/unicodeobject.c 2025-04-08 14:15:29.000000000 +0000 ++++ Python-3.11.12/Objects/unicodeobject.c 2025-05-22 12:57:58.387970080 +0000 +@@ -6301,20 +6301,23 @@ + static _PyUnicode_Name_CAPI *ucnhash_capi = NULL; + + PyObject * +-_PyUnicode_DecodeUnicodeEscapeInternal(const char *s, ++_PyUnicode_DecodeUnicodeEscapeInternal2(const char *s, + Py_ssize_t size, + const char *errors, + Py_ssize_t *consumed, +- const char **first_invalid_escape) ++ int *first_invalid_escape_char, ++ const char **first_invalid_escape_ptr) + { + const char *starts = s; ++ const char *initial_starts = starts; + _PyUnicodeWriter writer; + const char *end; + PyObject *errorHandler = NULL; + PyObject *exc = NULL; + + // so we can remember if we've seen an invalid escape char or not +- *first_invalid_escape = NULL; ++ *first_invalid_escape_char = -1; ++ *first_invalid_escape_ptr = NULL; + + if (size == 0) { + if (consumed) { +@@ -6402,9 +6405,12 @@ + } + } + if (ch > 0377) { +- if (*first_invalid_escape == NULL) { +- *first_invalid_escape = s-3; /* Back up 3 chars, since we've +- already incremented s. */ ++ if (*first_invalid_escape_char == -1) { ++ *first_invalid_escape_char = ch; ++ if (starts == initial_starts) { ++ /* Back up 3 chars, since we've already incremented s. */ ++ *first_invalid_escape_ptr = s - 3; ++ } + } + } + WRITE_CHAR(ch); +@@ -6503,9 +6509,12 @@ + goto error; + + default: +- if (*first_invalid_escape == NULL) { +- *first_invalid_escape = s-1; /* Back up one char, since we've +- already incremented s. */ ++ if (*first_invalid_escape_char == -1) { ++ *first_invalid_escape_char = c; ++ if (starts == initial_starts) { ++ /* Back up one char, since we've already incremented s. */ ++ *first_invalid_escape_ptr = s - 1; ++ } + } + WRITE_ASCII_CHAR('\\'); + WRITE_CHAR(c); +@@ -6544,24 +6553,40 @@ + return NULL; + } + ++// Export for binary compatibility. ++PyObject * ++_PyUnicode_DecodeUnicodeEscapeInternal(const char *s, ++ Py_ssize_t size, ++ const char *errors, ++ Py_ssize_t *consumed, ++ const char **first_invalid_escape) ++{ ++ int first_invalid_escape_char; ++ return _PyUnicode_DecodeUnicodeEscapeInternal2( ++ s, size, errors, consumed, ++ &first_invalid_escape_char, ++ first_invalid_escape); ++} ++ + PyObject * + _PyUnicode_DecodeUnicodeEscapeStateful(const char *s, + Py_ssize_t size, + const char *errors, + Py_ssize_t *consumed) + { +- const char *first_invalid_escape; +- PyObject *result = _PyUnicode_DecodeUnicodeEscapeInternal(s, size, errors, ++ int first_invalid_escape_char; ++ const char *first_invalid_escape_ptr; ++ PyObject *result = _PyUnicode_DecodeUnicodeEscapeInternal2(s, size, errors, + consumed, +- &first_invalid_escape); ++ &first_invalid_escape_char, ++ &first_invalid_escape_ptr); + if (result == NULL) + return NULL; +- if (first_invalid_escape != NULL) { +- unsigned char c = *first_invalid_escape; +- if ('4' <= c && c <= '7') { ++ if (first_invalid_escape_char != -1) { ++ if (first_invalid_escape_char > 0xff) { + if (PyErr_WarnFormat(PyExc_DeprecationWarning, 1, +- "invalid octal escape sequence '\\%.3s'", +- first_invalid_escape) < 0) ++ "invalid octal escape sequence '\\%o'", ++ first_invalid_escape_char) < 0) + { + Py_DECREF(result); + return NULL; +@@ -6570,7 +6595,7 @@ + else { + if (PyErr_WarnFormat(PyExc_DeprecationWarning, 1, + "invalid escape sequence '\\%c'", +- c) < 0) ++ first_invalid_escape_char) < 0) + { + Py_DECREF(result); + return NULL; +Index: Python-3.11.12/Parser/string_parser.c +=================================================================== +--- Python-3.11.12.orig/Parser/string_parser.c 2025-04-08 14:15:29.000000000 +0000 ++++ Python-3.11.12/Parser/string_parser.c 2025-05-22 13:00:34.860750605 +0000 +@@ -130,13 +130,16 @@ + len = p - buf; + s = buf; + +- const char *first_invalid_escape; +- v = _PyUnicode_DecodeUnicodeEscapeInternal(s, len, NULL, NULL, &first_invalid_escape); ++ int first_invalid_escape_char; ++ const char *first_invalid_escape_ptr; ++ v = _PyUnicode_DecodeUnicodeEscapeInternal2(s, (Py_ssize_t)len, NULL, NULL, ++ &first_invalid_escape_char, ++ &first_invalid_escape_ptr); + +- if (v != NULL && first_invalid_escape != NULL) { +- if (warn_invalid_escape_sequence(parser, first_invalid_escape, t) < 0) { +- /* We have not decref u before because first_invalid_escape points +- inside u. */ ++ if (v != NULL && first_invalid_escape_ptr != NULL && t != NULL) { ++ if (warn_invalid_escape_sequence(parser, s, first_invalid_escape_ptr, t) < 0) { ++ /* We have not decref u before because first_invalid_escape_ptr ++ points inside u. */ + Py_XDECREF(u); + Py_DECREF(v); + return NULL; +@@ -149,14 +152,17 @@ + static PyObject * + decode_bytes_with_escapes(Parser *p, const char *s, Py_ssize_t len, Token *t) + { +- const char *first_invalid_escape; +- PyObject *result = _PyBytes_DecodeEscape(s, len, NULL, &first_invalid_escape); ++ int first_invalid_escape_char; ++ const char *first_invalid_escape_ptr; ++ PyObject *result = _PyBytes_DecodeEscape2(s, len, NULL, ++ &first_invalid_escape_char, ++ &first_invalid_escape_ptr); + if (result == NULL) { + return NULL; + } + +- if (first_invalid_escape != NULL) { +- if (warn_invalid_escape_sequence(p, first_invalid_escape, t) < 0) { ++ if (first_invalid_escape_ptr != NULL) { ++ if (warn_invalid_escape_sequence(p, s, first_invalid_escape_ptr, t) < 0) { + Py_DECREF(result); + return NULL; + } diff --git a/python311.changes b/python311.changes index 37954db..323f0c5 100644 --- a/python311.changes +++ b/python311.changes @@ -1,3 +1,10 @@ +------------------------------------------------------------------- +Thu May 22 13:01:17 UTC 2025 - Matej Cepl + +- Add CVE-2025-4516-DecodeError-handler.patch fixing + CVE-2025-4516 (bsc#1243273) blocking DecodeError handling + vulnerability, which could lead to DoS. + ------------------------------------------------------------------- Sat May 17 10:02:27 UTC 2025 - Matej Cepl diff --git a/python311.spec b/python311.spec index fc56507..95d07f7 100644 --- a/python311.spec +++ b/python311.spec @@ -189,6 +189,9 @@ Patch24: add-loongarch64-support.patch # PATCH-FIX-UPSTREAM gh-126572-test_ssl-no-stop-ThreadedEchoServer-OSError.patch bsc#1241067 mcepl@suse.com # don't stop ThreadedEchoServer on OSError, makes test_ssl fail with OpenSSL 3.5 Patch25: gh-126572-test_ssl-no-stop-ThreadedEchoServer-OSError.patch +# PATCH-FIX-UPSTREAM CVE-2025-4516-DecodeError-handler.patch bsc#1243273 mcepl@suse.com +# this patch makes things totally awesome +Patch26: CVE-2025-4516-DecodeError-handler.patch BuildRequires: autoconf-archive BuildRequires: automake BuildRequires: crypto-policies-scripts