po plugin: added todo note about optimization

[ikiwiki] / doc / security.mdwn

Let's do an ikiwiki security analysis.

If you are using ikiwiki to render pages that only you can edit, do not
generate any wrappers, and do not use the cgi, then there are no more
security issues with this program than with cat(1). If, however, you let
others edit pages in your wiki, then some possible security issues do need
to be kept in mind.

[[!toc levels=2]]

----

# Probable holes

_(The list of things to fix.)_

## commit spoofing

Anyone with direct commit access can forge "web commit from foo" and
make it appear on [[RecentChanges]] like foo committed. One way to avoid
this would be to limit web commits to those done by a certain user.

## other stuff to look at

I need to audit the git backend a bit, and have been meaning to
see if any CRLF injection type things can be done in the CGI code.

----

# Potential gotchas

_(Things not to do.)_

## image file etc attacks

If it enounters a file type it does not understand, ikiwiki just copies it
into place. So if you let users add any kind of file they like, they can
upload images, movies, windows executables, css files, etc (though not html
files). If these files exploit security holes in the browser of someone
who's viewing the wiki, that can be a security problem.

Of course nobody else seems to worry about this in other wikis, so should we?

People with direct commit access can upload such files
(and if you wanted to you could block that with a pre-commit hook).

The attachments plugin is not enabled by default. If you choose to
enable it, you should make use of its powerful abilities to filter allowed
types of attachments, and only let trusted users upload.

It is possible to embed an image in a page edited over the web, by using
`img src="data:image/png;"`. Ikiwiki's htmlscrubber only allows `data:`
urls to be used for `image/*` mime types. It's possible that some broken
browser might ignore the mime type and if the data provided is not an
image, instead run it as javascript, or something evil like that. Hopefully
not many browsers are that broken.

## multiple accessors of wiki directory

If multiple people can directly write to the source directory ikiwiki is
using, or to the destination directory it writes files to, then one can
cause trouble for the other when they run ikiwiki through symlink attacks.

So it's best if only one person can ever directly write to those directories.

## setup files

Setup files are not safe to keep in the same revision control repository
with the rest of the wiki. Just don't do it.

## page locking can be bypassed via direct commits

A locked page can only be edited on the web by an admin, but anyone who is
allowed to commit directly to the repository can bypass this. This is by
design, although a pre-commit hook could be used to prevent editing of
locked pages, if you really need to.

## web server attacks

If your web server does any parsing of special sorts of files (for example,
server parsed html files), then if you let anyone else add files to the wiki,
they can try to use this to exploit your web server.

----

# Hopefully non-holes

_(AKA, the assumptions that will be the root of most security holes...)_

## exploiting ikiwiki with bad content

Someone could add bad content to the wiki and hope to exploit ikiwiki.
Note that ikiwiki runs with perl taint checks on, so this is unlikely.

One fun thing in ikiwiki is its handling of a PageSpec, which involves
translating it into perl and running the perl. Of course, this is done
*very* carefully to guard against injecting arbitrary perl code.

## publishing cgi scripts

ikiwiki does not allow cgi scripts to be published as part of the wiki. Or
rather, the script is published, but it's not marked executable (except in
the case of "destination directory file replacement" below), so hopefully
your web server will not run it.

## suid wrappers

`ikiwiki --wrapper` is intended to generate a wrapper program that
runs ikiwiki to update a given wiki. The wrapper can in turn be made suid,
for example to be used in a [[post-commit]] hook by people who cannot write
to the html pages, etc.

If the wrapper script is made suid, then any bugs in this wrapper would be
security holes. The wrapper is written as securely as I know how, is based
on code that has a history of security use long before ikiwiki, and there's
been no problem yet.

## shell exploits

ikiwiki does not expose untrusted data to the shell. In fact it doesn't use
`system(3)` at all, and the only use of backticks is on data supplied by the
wiki admin and untainted filenames. 

Ikiwiki was developed and used for a long time with perl's taint checking
turned on as a second layer of defense against shell and other exploits. Due
to a strange [bug](http://bugs.debian.org/411786) in perl, taint checking
is currently disabled for production builds of ikiwiki.

## cgi data security

When ikiwiki runs as a cgi to edit a page, it is passed the name of the
page to edit. It has to make sure to sanitise this page, to prevent eg,
editing of ../../../foo, or editing of files that are not part of the wiki,
such as subversion dotfiles. This is done by sanitising the filename
removing unallowed characters, then making sure it doesn't start with "/"
or contain ".." or "/.svn/", etc. Annoyingly ad-hoc, this kind of code is
where security holes breed. It needs a test suite at the very least.

## CGI::Session security

I've audited this module and it is massively insecure by default. ikiwiki
uses it in one of the few secure ways; by forcing it to write to a
directory it controls (and not /tmp) and by setting a umask that makes the
file not be world readable.

## cgi password security

Login to the wiki using [[plugins/passwordauth]] involves sending a password
in cleartext over the net. Cracking the password only allows editing the wiki
as that user though. If you care, you can use https, I suppose. If you do use
https either for all of the wiki, or just the cgi access, then consider using
the sslcookie option. Using [[plugins/openid]] is a potentially better option.

## XSS holes in CGI output

ikiwiki has been audited to ensure that all cgi script input/output
is sanitised to prevent XSS attacks. For example, a user can't register
with a username containing html code (anymore).

It's difficult to know for sure if all such avenues have really been
closed though.

## HTML::Template security

If the [[plugins/template]] plugin is enabled, users can modify templates
like any other part of the wiki. This assumes that HTML::Template is secure
when used with untrusted/malicious templates. (Note that includes are not
allowed, so that's not a problem.)

----

# Plugins

The security of [[plugins]] depends on how well they're written and what
external tools they use. The plugins included in ikiwiki are all held to
the same standards as the rest of ikiwiki, but with that said, here are
some security notes for them.

* The [[plugins/img]] plugin assumes that imagemagick/perlmagick are secure
  from malformed image attacks. Imagemagick has had security holes in the
  past. To be able to exploit such a hole, a user would need to be able to
  upload images to the wiki.

----

# Fixed holes

_(Unless otherwise noted, these were discovered and immediately fixed by the
ikiwiki developers.)_

## destination directory file replacement

Any file in the destination directory that is a valid page filename can be
replaced, even if it was not originally rendered from a page. For example,
ikiwiki.cgi could be edited in the wiki, and it would write out a
replacement. File permission is preseved. Yipes!

This was fixed by making ikiwiki check if the file it's writing to exists;
if it does then it has to be a file that it's aware of creating before, or
it will refuse to create it.

Still, this sort of attack is something to keep in mind.

## symlink attacks

Could a committer trick ikiwiki into following a symlink and operating on
some other tree that it shouldn't? svn supports symlinks, so one can get
into the repo. ikiwiki uses File::Find to traverse the repo, and does not
tell it to follow symlinks, but it might be possible to race replacing a
directory with a symlink and trick it into following the link.

Also, if someone checks in a symlink to /etc/passwd, ikiwiki would read and
publish that, which could be used to expose files a committer otherwise
wouldn't see.

To avoid this, ikiwiki will skip over symlinks when scanning for pages, and
uses locking to prevent more than one instance running at a time. The lock
prevents one ikiwiki from running a svn up/git pull/etc at the wrong time
to race another ikiwiki. So only attackers who can write to the working
copy on their own can race it.

## symlink + cgi attacks

Similarly, a commit of a symlink could be made, ikiwiki ignores it
because of the above, but the symlink is still there, and then you edit the
page from the web, which follows the symlink when reading the page
(exposing the content), and again when saving the changed page (changing
the content).

This was fixed for page saving by making ikiwiki refuse to write to files
that are symlinks, or that are in subdirectories that are symlinks,
combined with the above locking.

For page editing, it's fixed by ikiwiki checking to make sure that it
already has found a page by scanning the tree, before loading it for
editing, which as described above, also is done in a way that avoids
symlink attacks.

## underlaydir override attacks

ikiwiki also scans an underlaydir for pages, this is used to provide stock
pages to all wikis w/o needing to copy them into the wiki. Since ikiwiki
internally stores only the base filename from the underlaydir or srcdir,
and searches for a file in either directory when reading a page source,
there is the potential for ikiwiki's scanner to reject a file from the
srcdir for some reason (such as it being contained in a directory that is
symlinked in), find a valid copy of the file in the underlaydir, and then
when loading the file, mistakenly load the bad file from the srcdir.

This attack is avoided by making ikiwiki refuse to add any files from the
underlaydir if a file also exists in the srcdir with the same name.

## multiple page source issues

Note that I previously worried that underlay override attacks could also be
accomplished if ikiwiki were extended to support other page markup
languages besides markdown. However, a closer look indicates that this is
not a problem: ikiwiki does preserve the file extension when storing the
source filename of a page, so a file with another extension that renders to
the same page name can't bypass the check. Ie, ikiwiki won't skip foo.rst
in the srcdir, find foo.mdwn in the underlay, decide to render page foo and
then read the bad foo.mdwn. Instead it will remember the .rst extension and
only render a file with that extension.

## XSS attacks in page content

ikiwiki supports protecting users from their own broken browsers via the
[[plugins/htmlscrubber]] plugin, which is enabled by default.

## svn commit logs

It's was possible to force a whole series of svn commits to appear to
have come just before yours, by forging svn log output. This was
guarded against by using svn log --xml.

ikiwiki escapes any html in svn commit logs to prevent other mischief.

## XML::Parser

XML::Parser is used by the aggregation plugin, and has some security holes. 
Bug #[378411](http://bugs.debian.org/378411) does not
seem to affect our use, since the data is not encoded as utf-8 at that
point. #[378412](http://bugs.debian.org/378412) could affect us, although it
doesn't seem very exploitable. It has a simple fix, and has been fixed in
Debian unstable.

## include loops

Various directives that cause one page to be included into another could
be exploited to DOS the wiki, by causing a loop. Ikiwiki has always guarded
against this one way or another; the current solution should detect all
types of loops involving preprocessor directives.

## Online editing of existing css and images

A bug in ikiwiki allowed the web-based editor to edit any file that was in
the wiki, not just files that are page sources. So an attacker (or a
genuinely helpful user, which is how the hole came to light) could edit
files like style.css. It is also theoretically possible that an attacker
could have used this hole to edit images or other files in the wiki, with
some difficulty, since all editing would happen in a textarea.

This hole was discovered on 10 Feb 2007 and fixed the same day with the
release of ikiwiki 1.42. A fix was also backported to Debian etch, as
version 1.33.1. I recommend upgrading to one of these versions if your wiki
allows web editing.

## html insertion via title

Missing html escaping of the title contents allowed a web-based editor to
insert arbitrary html inside the title tag of a page. Since that part of
the page is not processed by the htmlscrubber, evil html could be injected.

This hole was discovered on 21 March 2007 and fixed the same day (er, hour) 
with the release of ikiwiki 1.46. A fix was also backported to Debian etch,
as version 1.33.2. I recommend upgrading to one of these versions if your
wiki allows web editing or aggregates feeds.

## javascript insertion via meta tags

It was possible to use the meta plugin's meta tags to insert arbitrary
url contents, which could be used to insert stylesheet information
containing javascript. This was fixed by sanitising meta tags.

This hole was discovered on 21 March 2007 and fixed the same day
with the release of ikiwiki 1.47. A fix was also backported to Debian etch,
as version 1.33.3. I recommend upgrading to one of these versions if your
wiki can be edited by third parties.

## insufficient checking for symlinks in srcdir path

Ikiwiki did not check if path to the srcdir to contained a symlink. If an
attacker had commit access to the directories in the path, they could
change it to a symlink, causing ikiwiki to read and publish files that were
not intended to be published. (But not write to them due to other checks.)

In most configurations, this is not exploitable, because the srcdir is
checked out of revision control, but the directories leading up to it are
not. Or, the srcdir is a single subdirectory of a project in revision
control (ie, `ikiwiki/doc`), and if the subdirectory were a symlink,
ikiwiki would still typically not follow it.

There are at least two configurations where this is exploitable:

* If the srcdir is a deeper subdirectory of a project. For example if it is
  `project/foo/doc`, an an attacker can replace `foo` with a symlink to a
  directory containing a `doc` directory (not a symlink), then ikiwiki
  would follow the symlink.
* If the path to the srcdir in ikiwiki's configuration ended in "/", 
  and the srcdir is a single subdirectory of a project, (ie,
  `ikiwiki/doc/`), the srcdir could be a symlink and ikiwiki would not
  notice.

This security hole was discovered on 26 November 2007 and fixed the same
day with the release of ikiwiki 2.14. I recommend upgrading to this version
if your wiki can be committed to by third parties. Alternatively, don't use
a trailing slash in the srcdir, and avoid the (unusual) configurations that
allow the security hole to be exploited.

## javascript insertion via uris

The htmlscrubber did not block javascript in uris. This was fixed by adding
a whitelist of valid uri types, which does not include javascript. 
([[!cve CVE-2008-0809]]) Some urls specifyable by the meta plugin could also
theoretically have been used to inject javascript; this was also blocked
([[!cve CVE-2008-0808]]).

This hole was discovered on 10 February 2008 and fixed the same day
with the release of ikiwiki 2.31.1. (And a few subsequent versions..)
A fix was also backported to Debian etch, as version 1.33.4. I recommend
upgrading to one of these versions if your wiki can be edited by third
parties.

## Cross Site Request Forging

Cross Site Request Forging could be used to constuct a link that would
change a logged-in user's password or other preferences if they clicked on
the link. It could also be used to construct a link that would cause a wiki
page to be modified by a logged-in user. ([[!cve CVE-2008-0165]])

These holes were discovered on 10 April 2008 and fixed the same day with
the release of ikiwiki 2.42. A fix was also backported to Debian etch, as
version 1.33.5. I recommend upgrading to one of these versions.

## Cleartext passwords

Until version 2.48, ikiwiki stored passwords in cleartext in the `userdb`.
That risks exposing all users' passwords if the file is somehow exposed. To
pre-emtively guard against that, current versions of ikiwiki store password
hashes (using Eksblowfish).

If you use the [[plugins/passwordauth]] plugin, I recommend upgrading to
ikiwiki 2.48, installing the [[!cpan Authen::Passphrase]] perl module, and running
`ikiwiki-transition hashpassword` to replace all existing cleartext passwords
with strong blowfish hashes. 

You might also consider changing to [[plugins/openid]], which does not 
require ikiwiki deal with passwords at all, and does not involve users sending
passwords in cleartext over the net to log in, either.

## Empty password security hole

This hole allowed ikiwiki to accept logins using empty passwords, to openid
accounts that didn't use a password. It was introduced in version 1.34, and
fixed in version 2.48. The [bug](http://bugs.debian.org/483770) was
discovered on 30 May 2008 and fixed the same day. ([[!cve CVE-2008-0169]])

I recommend upgrading to 2.48 immediatly if your wiki allows both password
and openid logins.