Django

==========================

*February 19, 2013*

Django 1.3.6 fixes four security issues present in previous Django releases in
the 1.3 series.

This is the sixth bugfix/security release in the Django 1.3 series.


Host header poisoning
=====================

Some parts of Django -- independent of end-user-written applications -- make
use of full URLs, including domain name, which are generated from the HTTP Host
header. Django's documentation has for some time contained notes advising users
on how to configure web servers to ensure that only valid Host headers can reach
the Django application. However, it has been reported to us that even with the
recommended web server configurations there are still techniques available for
tricking many common web servers into supplying the application with an
incorrect and possibly malicious Host header.

For this reason, Django 1.3.6 adds a new setting, ``ALLOWED_HOSTS``, which
should contain an explicit list of valid host/domain names for this site. A
request with a Host header not matching an entry in this list will raise
``SuspiciousOperation`` if ``request.get_host()`` is called. For full details
see the documentation for the :setting:`ALLOWED_HOSTS` setting.

The default value for this setting in Django 1.3.6 is ``['*']`` (matching any
host), for backwards-compatibility, but we strongly encourage all sites to set
a more restrictive value.

This host validation is disabled when ``DEBUG`` is ``True`` or when running tests.


XML deserialization
===================

The XML parser in the Python standard library is vulnerable to a number of
attacks via external entities and entity expansion. Django uses this parser for
deserializing XML-formatted database fixtures. The fixture deserializer is not
intended for use with untrusted data, but in order to err on the side of safety
in Django 1.3.6 the XML deserializer refuses to parse an XML document with a
DTD (DOCTYPE definition), which closes off these attack avenues.

These issues in the Python standard library are CVE-2013-1664 and
CVE-2013-1665. More information available `from the Python security team`_.

Django's XML serializer does not create documents with a DTD, so this should
not cause any issues with the typical round-trip from ``dumpdata`` to
``loaddata``, but if you feed your own XML documents to the ``loaddata``
management command, you will need to ensure they do not contain a DTD.

.. _from the Python security team: https://blog.python.org/2013/02/announcing-defusedxml-fixes-for-xml.html


Formset memory exhaustion
=========================

Previous versions of Django did not validate or limit the form-count data
provided by the client in a formset's management form, making it possible to
exhaust a server's available memory by forcing it to create very large numbers
of forms.

In Django 1.3.6, all formsets have a strictly-enforced maximum number of forms
(1000 by default, though it can be set higher via the ``max_num`` formset
factory argument).


Admin history view information leakage
======================================

In previous versions of Django, an admin user without change permission on a
model could still view the Unicode representation of instances via their admin
history log. Django 1.3.6 now limits the admin history log view for an object
to users with change permission for that model.


==========================

==========================

*December 10, 2012*

Django 1.3.5 addresses two security issues present in previous Django releases
in the 1.3 series.

Please be aware that this security release is slightly different from previous
ones. Both issues addressed here have been dealt with in prior security updates
to Django. In one case, we have received ongoing reports of problems, and in
the other we've chosen to take further steps to tighten up Django's code in
response to independent discovery of potential problems from multiple sources.

Host header poisoning
=====================

Several earlier Django security releases focused on the issue of poisoning the
HTTP Host header, causing Django to generate URLs pointing to arbitrary,
potentially-malicious domains.

In response to further input received and reports of continuing issues
following the previous release, we're taking additional steps to tighten Host
header validation. Rather than attempt to accommodate all features HTTP
supports here, Django's Host header validation attempts to support a smaller,
but far more common, subset:

* Hostnames must consist of characters ``[A-Za-z0-9]`` plus hyphen ('-') or dot
('.').
* IP addresses -- both IPv4 and IPv6 -- are permitted.
* Port, if specified, is numeric.

Any deviation from this will now be rejected, raising the exception
:exc:`django.core.exceptions.SuspiciousOperation`.

Redirect poisoning
==================

Also following up on a previous issue: in July of this year, we made changes to
Django's HTTP redirect classes, performing additional validation of the scheme
of the URL to redirect to (since, both within Django's own supplied
applications and many third-party applications, accepting a user-supplied
redirect target is a common pattern).

Since then, two independent audits of the code turned up further potential
problems. So, similar to the Host-header issue, we are taking steps to provide
tighter validation in response to reported problems (primarily with third-party
applications, but to a certain extent also within Django itself). This comes in
two parts:

1. A new utility function, ``django.utils.http.is_safe_url``, is added; this
function takes a URL and a hostname, and checks that the URL is either
relative, or if absolute matches the supplied hostname. This function is
intended for use whenever user-supplied redirect targets are accepted, to
ensure that such redirects cannot lead to arbitrary third-party sites.

2. All of Django's own built-in views -- primarily in the authentication system
-- which allow user-supplied redirect targets now use ``is_safe_url`` to
validate the supplied URL.


==========================

==========================

*October 17, 2012*

This is the fourth release in the Django 1.3 series.

Host header poisoning
=====================

Some parts of Django -- independent of end-user-written applications -- make
use of full URLs, including domain name, which are generated from the HTTP Host
header. Some attacks against this are beyond Django's ability to control, and
require the web server to be properly configured; Django's documentation has
for some time contained notes advising users on such configuration.

Django's own built-in parsing of the Host header is, however, still vulnerable,
as was reported to us recently. The Host header parsing in Django 1.3.3 and
Django 1.4.1 -- specifically, ``django.http.HttpRequest.get_host()`` -- was
incorrectly handling username/password information in the header. Thus, for
example, the following Host header would be accepted by Django when running on
``validsite.com``:

.. code-block:: text

Host: validsite.com:randomevilsite.com

Using this, an attacker can cause parts of Django -- particularly the
password-reset mechanism -- to generate and display arbitrary URLs to users.

To remedy this, the parsing in ``HttpRequest.get_host()`` is being modified;
Host headers which contain potentially dangerous content (such as
username/password pairs) now raise the exception
:exc:`django.core.exceptions.SuspiciousOperation`.

Details of this issue were initially posted online as a `security advisory`_.

.. _security advisory: https://www.djangoproject.com/weblog/2012/oct/17/security/


==========================

==========================

*August 1, 2012*

Following Monday's security release of :doc:`Django 1.3.2 </releases/1.3.2>`,
we began receiving reports that one of the fixes applied was breaking Python
2.4 compatibility for Django 1.3. Since Python 2.4 is a supported Python
version for that release series, this release fixes compatibility with
Python 2.4.


==========================

==========================

*July 30, 2012*

This is the second security release in the Django 1.3 series, fixing several
security issues in Django 1.3. Django 1.3.2 is a recommended upgrade for
all users of Django 1.3.

For a full list of issues addressed in this release, see the `security
advisory`_.

.. _security advisory: https://www.djangoproject.com/weblog/2012/jul/30/security-releases-issued/


==========================

==========================

*September 9, 2011*

Welcome to Django 1.3.1!

This is the first security release in the Django 1.3 series, fixing several
security issues in Django 1.3. Django 1.3.1 is a recommended upgrade for
all users of Django 1.3.

For a full list of issues addressed in this release, see the `security
advisory`_.

.. _security advisory: https://www.djangoproject.com/weblog/2011/sep/09/security-releases-issued/


========================