[Full-disclosure] URL Spoofing vulnerability in different browsers
mustlive at websecurity.com.ua
Fri Jul 22 16:36:51 BST 2011
I want to warn you about URL Spoofing vulnerability in Mozilla Firefox,
Internet Explorer, Google Chrome, Opera and other browsers. I found it long
time ago, at 6th of February 2008, just after finding of built-in CSRF
vulnerability in Mozilla and Firefox (it's funky CSRF attack via prefetching
functionality), which I described at my site in March.
Vulnerable are all browsers which support Basic/Digest Authentication. It's
all modern browsers and many from old browsers. In particular affected are
Mozilla Firefox 3.0.19, 3.5.11, 3.6.8, Firefox 4.0b2 (and Mozilla and all
other Gecko-based browsers), Internet Explorer 6, 7, 8, Google Chrome
184.108.40.206 and Opera 10.62 and previous and next versions of these browsers.
And other browsers which support Basic/Digest Authentication.
In March, after my informing, Mozilla opened Bug 647010 in Bugzilla
Among four browsers developers informed by me only Mozilla said, that they
are planning to fix this vulnerability (without specifying the time). Google
even didn't answer me, but in June they informed in their blog
that they fixed this vulnerability in browsers Chrome 13 (it's now beta
version) and higher.
This is better to call attack, then vulnerability, because it's using
built-in browsers functionality (and its intended behavior) to attack users
of web sites. This attack allows to conduct phishing attacks on users of web
sites - in this case phishing is doing not at other (phishing) sites, not
with using of holes of target sites (like reflected XSS or persistent XSS),
but with using of browsers functionality (and allowed functionality of
target sites to place external content).
I called this attack as Onsite phishing (or Inline phishing). It can be used
(including by phishers) for stealing of logins and passwords of users of web
As I've tested, a lot of different methods (with using of tags and CSS),
which allow to make cross-site requests, can be used to conduct this attack.
Except prefetching (in all Gecko-based browsers which support prefetching
functionality), which doesn't show Authentication window at receiving of 401
response from web server. The next methods can be used:
Tags img, script, iframe, frame, embed, link (css) - Mozilla, Firefox, IE,
Google Chrome and Opera.
Tag object - Internet Explorer, Google Chrome and Opera.
CSS (inline, in html files, in external css files): such
as -moz-binding:url - Mozilla and Firefox < 3.0, such as
background-image:url - in all browsers.
Here are screenshots of the attack in different browsers (in Firefox 3.0.19,
3.5.x, 3.6.x. 4.0b2 the dialog window looks almost equally):
The attack can be made as reflected at target site, as persistent (with
using of allowed functionality at target site, which allows to put some
tags, like img tag). The persistent attack is more dangerous (and such type
of attack is showed on screenshots). And there are millions of web sites
which allow such user generated content (like img tags) which can lead to
such persistent attacks.
2011.03.26 - announced at my site.
2011.03.31 - informed Mozilla, Microsoft, Google and Opera.
2011.04.01 - Mozilla answered and opened entry in Bugzilla
2011.04.01 - Microsoft answered and asked for more details.
2011.04.03 - gave additional details for Microsoft. But they ignored to fix,
like Google and Opera did.
2011.06.14 - Google hiddenly and lamerly fixed this hole in Chrome 12 beta
(and future versions), without answering and thanking me for informing.
Which is lame behavior and I don't respect companies with such behavior. But
this Google's step should force other browsers developers to fix this
vulnerability in their products.
2011.07.21 - disclosed at my site.
I mentioned about this vulnerability at my site
Best wishes & regards,
Administrator of Websecurity web site
Full-Disclosure is hosted and sponsored by Secunia.