python3-django: Fix for CVE-2023-43665 and CVE-2023-46695

CVE-2023-43665:
In Django 3.2 before 3.2.22, 4.1 before 4.1.12, and 4.2 before 4.2.6, the
django.utils.text.Truncator chars() and words() methods (when used with
html=True) are subject to a potential DoS (denial of service) attack via
certain inputs with very long, potentially malformed HTML text. The chars()
and words() methods are used to implement the truncatechars_html and
truncatewords_html template filters, which are thus also vulnerable.
NOTE: this issue exists because of an incomplete fix for CVE-2019-14232.

CVE-2023-46695:
An issue was discovered in Django 3.2 before 3.2.23, 4.1 before 4.1.13, and
4.2 before 4.2.7. The NFKC normalization is slow on Windows. As a consequence,
django.contrib.auth.forms.UsernameField is subject to a potential DoS (denial of
service) attack via certain inputs with a very large number of Unicode characters.

References:
https://www.djangoproject.com/weblog/2023/oct/04/security-releases/
https://www.djangoproject.com/weblog/2023/nov/01/security-releases/

Signed-off-by: Narpat Mali <narpat.mali@windriver.com>
Signed-off-by: Armin Kuster <akuster808@gmail.com>
This commit is contained in:
Narpat Mali
2023-11-30 12:23:37 +00:00
committed by Armin Kuster
parent 8a042b540d
commit fee5560548
3 changed files with 291 additions and 0 deletions

View File

@@ -0,0 +1,199 @@
From b269a0063e9b10a6c88c92b24d1b92c7421950de Mon Sep 17 00:00:00 2001
From: Natalia <124304+nessita@users.noreply.github.com>
Date: Wed, 29 Nov 2023 12:20:01 +0000
Subject: [PATCH 1/2] Fixed CVE-2023-43665 -- Mitigated potential DoS in
django.utils.text.Truncator when truncating HTML text.
Thanks Wenchao Li of Alibaba Group for the report.
CVE: CVE-2023-43665
Upstream-Status: Backport [https://github.com/django/django/commit/ccdade1a0262537868d7ca64374de3d957ca50c5]
Signed-off-by: Narpat Mali <narpat.mali@windriver.com>
---
django/utils/text.py | 18 ++++++++++++++++-
docs/ref/templates/builtins.txt | 20 +++++++++++++++++++
docs/releases/2.2.28.txt | 20 +++++++++++++++++++
tests/utils_tests/test_text.py | 35 ++++++++++++++++++++++++---------
4 files changed, 83 insertions(+), 10 deletions(-)
diff --git a/django/utils/text.py b/django/utils/text.py
index 1fae7b2..06a377b 100644
--- a/django/utils/text.py
+++ b/django/utils/text.py
@@ -57,7 +57,14 @@ def wrap(text, width):
class Truncator(SimpleLazyObject):
"""
An object used to truncate text, either by characters or words.
+
+ When truncating HTML text (either chars or words), input will be limited to
+ at most `MAX_LENGTH_HTML` characters.
"""
+
+ # 5 million characters are approximately 4000 text pages or 3 web pages.
+ MAX_LENGTH_HTML = 5_000_000
+
def __init__(self, text):
super().__init__(lambda: str(text))
@@ -154,6 +161,11 @@ class Truncator(SimpleLazyObject):
if words and length <= 0:
return ''
+ size_limited = False
+ if len(text) > self.MAX_LENGTH_HTML:
+ text = text[: self.MAX_LENGTH_HTML]
+ size_limited = True
+
html4_singlets = (
'br', 'col', 'link', 'base', 'img',
'param', 'area', 'hr', 'input'
@@ -203,10 +215,14 @@ class Truncator(SimpleLazyObject):
# Add it to the start of the open tags list
open_tags.insert(0, tagname)
+ truncate_text = self.add_truncation_text("", truncate)
+
if current_len <= length:
+ if size_limited and truncate_text:
+ text += truncate_text
return text
+
out = text[:end_text_pos]
- truncate_text = self.add_truncation_text('', truncate)
if truncate_text:
out += truncate_text
# Close any tags still open
diff --git a/docs/ref/templates/builtins.txt b/docs/ref/templates/builtins.txt
index c4b0fa3..4faab38 100644
--- a/docs/ref/templates/builtins.txt
+++ b/docs/ref/templates/builtins.txt
@@ -2318,6 +2318,16 @@ If ``value`` is ``"<p>Joel is a slug</p>"``, the output will be
Newlines in the HTML content will be preserved.
+.. admonition:: Size of input string
+
+ Processing large, potentially malformed HTML strings can be
+ resource-intensive and impact service performance. ``truncatechars_html``
+ limits input to the first five million characters.
+
+.. versionchanged:: 2.2.28
+
+ In older versions, strings over five million characters were processed.
+
.. templatefilter:: truncatewords
``truncatewords``
@@ -2356,6 +2366,16 @@ If ``value`` is ``"<p>Joel is a slug</p>"``, the output will be
Newlines in the HTML content will be preserved.
+.. admonition:: Size of input string
+
+ Processing large, potentially malformed HTML strings can be
+ resource-intensive and impact service performance. ``truncatewords_html``
+ limits input to the first five million characters.
+
+.. versionchanged:: 2.2.28
+
+ In older versions, strings over five million characters were processed.
+
.. templatefilter:: unordered_list
``unordered_list``
diff --git a/docs/releases/2.2.28.txt b/docs/releases/2.2.28.txt
index 40eb230..6a38e9c 100644
--- a/docs/releases/2.2.28.txt
+++ b/docs/releases/2.2.28.txt
@@ -56,3 +56,23 @@ CVE-2023-41164: Potential denial of service vulnerability in ``django.utils.enco
``django.utils.encoding.uri_to_iri()`` was subject to potential denial of
service attack via certain inputs with a very large number of Unicode
characters.
+
+Backporting the CVE-2023-43665 fix on Django 2.2.28.
+
+CVE-2023-43665: Denial-of-service possibility in ``django.utils.text.Truncator``
+================================================================================
+
+Following the fix for :cve:`2019-14232`, the regular expressions used in the
+implementation of ``django.utils.text.Truncator``'s ``chars()`` and ``words()``
+methods (with ``html=True``) were revised and improved. However, these regular
+expressions still exhibited linear backtracking complexity, so when given a
+very long, potentially malformed HTML input, the evaluation would still be
+slow, leading to a potential denial of service vulnerability.
+
+The ``chars()`` and ``words()`` methods are used to implement the
+:tfilter:`truncatechars_html` and :tfilter:`truncatewords_html` template
+filters, which were thus also vulnerable.
+
+The input processed by ``Truncator``, when operating in HTML mode, has been
+limited to the first five million characters in order to avoid potential
+performance and memory issues.
diff --git a/tests/utils_tests/test_text.py b/tests/utils_tests/test_text.py
index 27e440b..cb3063d 100644
--- a/tests/utils_tests/test_text.py
+++ b/tests/utils_tests/test_text.py
@@ -1,5 +1,6 @@
import json
import sys
+from unittest.mock import patch
from django.core.exceptions import SuspiciousFileOperation
from django.test import SimpleTestCase
@@ -87,11 +88,17 @@ class TestUtilsText(SimpleTestCase):
# lazy strings are handled correctly
self.assertEqual(text.Truncator(lazystr('The quick brown fox')).chars(10), 'The quick…')
- def test_truncate_chars_html(self):
+ @patch("django.utils.text.Truncator.MAX_LENGTH_HTML", 10_000)
+ def test_truncate_chars_html_size_limit(self):
+ max_len = text.Truncator.MAX_LENGTH_HTML
+ bigger_len = text.Truncator.MAX_LENGTH_HTML + 1
+ valid_html = "<p>Joel is a slug</p>" # 14 chars
perf_test_values = [
- (('</a' + '\t' * 50000) + '//>', None),
- ('&' * 50000, '&' * 9 + '…'),
- ('_X<<<<<<<<<<<>', None),
+ ("</a" + "\t" * (max_len - 6) + "//>", None),
+ ("</p" + "\t" * bigger_len + "//>", "</p" + "\t" * 6 + "…"),
+ ("&" * bigger_len, "&" * 9 + "…"),
+ ("_X<<<<<<<<<<<>", None),
+ (valid_html * bigger_len, "<p>Joel is a…</p>"), # 10 chars
]
for value, expected in perf_test_values:
with self.subTest(value=value):
@@ -149,15 +156,25 @@ class TestUtilsText(SimpleTestCase):
truncator = text.Truncator('<p>I &lt;3 python, what about you?</p>')
self.assertEqual('<p>I &lt;3 python,…</p>', truncator.words(3, html=True))
+ @patch("django.utils.text.Truncator.MAX_LENGTH_HTML", 10_000)
+ def test_truncate_words_html_size_limit(self):
+ max_len = text.Truncator.MAX_LENGTH_HTML
+ bigger_len = text.Truncator.MAX_LENGTH_HTML + 1
+ valid_html = "<p>Joel is a slug</p>" # 4 words
perf_test_values = [
- ('</a' + '\t' * 50000) + '//>',
- '&' * 50000,
- '_X<<<<<<<<<<<>',
+ ("</a" + "\t" * (max_len - 6) + "//>", None),
+ ("</p" + "\t" * bigger_len + "//>", "</p" + "\t" * (max_len - 3) + "…"),
+ ("&" * max_len, None), # no change
+ ("&" * bigger_len, "&" * max_len + "…"),
+ ("_X<<<<<<<<<<<>", None),
+ (valid_html * bigger_len, valid_html * 12 + "<p>Joel is…</p>"), # 50 words
]
- for value in perf_test_values:
+ for value, expected in perf_test_values:
with self.subTest(value=value):
truncator = text.Truncator(value)
- self.assertEqual(value, truncator.words(50, html=True))
+ self.assertEqual(
+ expected if expected else value, truncator.words(50, html=True)
+ )
def test_wrap(self):
digits = '1234 67 9'
--
2.40.0

View File

@@ -0,0 +1,90 @@
From 32bc7fa517be1d50239827520cc13f3112d3d748 Mon Sep 17 00:00:00 2001
From: Mariusz Felisiak <felisiak.mariusz@gmail.com>
Date: Wed, 29 Nov 2023 12:49:41 +0000
Subject: [PATCH 2/2] Fixed CVE-2023-46695 -- Fixed potential DoS in
UsernameField on Windows.
Thanks MProgrammer (https://hackerone.com/mprogrammer) for the report.
CVE: CVE-2023-46695
Upstream-Status: Backport [https://github.com/django/django/commit/f9a7fb8466a7ba4857eaf930099b5258f3eafb2b]
Signed-off-by: Narpat Mali <narpat.mali@windriver.com>
---
django/contrib/auth/forms.py | 10 +++++++++-
docs/releases/2.2.28.txt | 14 ++++++++++++++
tests/auth_tests/test_forms.py | 8 +++++++-
3 files changed, 30 insertions(+), 2 deletions(-)
diff --git a/django/contrib/auth/forms.py b/django/contrib/auth/forms.py
index e6f73fe..26d3ca7 100644
--- a/django/contrib/auth/forms.py
+++ b/django/contrib/auth/forms.py
@@ -68,7 +68,15 @@ class ReadOnlyPasswordHashField(forms.Field):
class UsernameField(forms.CharField):
def to_python(self, value):
- return unicodedata.normalize('NFKC', super().to_python(value))
+ value = super().to_python(value)
+ if self.max_length is not None and len(value) > self.max_length:
+ # Normalization can increase the string length (e.g.
+ # "ff" -> "ff", "½" -> "12") but cannot reduce it, so there is no
+ # point in normalizing invalid data. Moreover, Unicode
+ # normalization is very slow on Windows and can be a DoS attack
+ # vector.
+ return value
+ return unicodedata.normalize("NFKC", value)
class UserCreationForm(forms.ModelForm):
diff --git a/docs/releases/2.2.28.txt b/docs/releases/2.2.28.txt
index 6a38e9c..c653cb6 100644
--- a/docs/releases/2.2.28.txt
+++ b/docs/releases/2.2.28.txt
@@ -76,3 +76,17 @@ filters, which were thus also vulnerable.
The input processed by ``Truncator``, when operating in HTML mode, has been
limited to the first five million characters in order to avoid potential
performance and memory issues.
+
+Backporting the CVE-2023-46695 fix on Django 2.2.28.
+
+CVE-2023-46695: Potential denial of service vulnerability in ``UsernameField`` on Windows
+=========================================================================================
+
+The :func:`NFKC normalization <python:unicodedata.normalize>` is slow on
+Windows. As a consequence, ``django.contrib.auth.forms.UsernameField`` was
+subject to a potential denial of service attack via certain inputs with a very
+large number of Unicode characters.
+
+In order to avoid the vulnerability, invalid values longer than
+``UsernameField.max_length`` are no longer normalized, since they cannot pass
+validation anyway.
diff --git a/tests/auth_tests/test_forms.py b/tests/auth_tests/test_forms.py
index bed23af..e73d4b8 100644
--- a/tests/auth_tests/test_forms.py
+++ b/tests/auth_tests/test_forms.py
@@ -6,7 +6,7 @@ from django import forms
from django.contrib.auth.forms import (
AdminPasswordChangeForm, AuthenticationForm, PasswordChangeForm,
PasswordResetForm, ReadOnlyPasswordHashField, ReadOnlyPasswordHashWidget,
- SetPasswordForm, UserChangeForm, UserCreationForm,
+ SetPasswordForm, UserChangeForm, UserCreationForm, UsernameField,
)
from django.contrib.auth.models import User
from django.contrib.auth.signals import user_login_failed
@@ -132,6 +132,12 @@ class UserCreationFormTest(TestDataMixin, TestCase):
self.assertNotEqual(user.username, ohm_username)
self.assertEqual(user.username, 'testΩ') # U+03A9 GREEK CAPITAL LETTER OMEGA
+ def test_invalid_username_no_normalize(self):
+ field = UsernameField(max_length=254)
+ # Usernames are not normalized if they are too long.
+ self.assertEqual(field.to_python("½" * 255), "½" * 255)
+ self.assertEqual(field.to_python("ff" * 254), "ff" * 254)
+
def test_duplicate_normalized_unicode(self):
"""
To prevent almost identical usernames, visually identical but differing
--
2.40.0

View File

@@ -8,6 +8,8 @@ inherit setuptools3
SRC_URI += "file://CVE-2023-31047.patch \
file://CVE-2023-36053.patch \
file://CVE-2023-41164.patch \
file://CVE-2023-43665.patch \
file://CVE-2023-46695.patch \
"
SRC_URI[sha256sum] = "0200b657afbf1bc08003845ddda053c7641b9b24951e52acd51f6abda33a7413"