mailing list archives
Re: Skype callto:// BoF technical details
From: "Berend-Jan Wever" <skylined () edup tudelft nl>
Date: Tue, 16 Nov 2004 21:35:44 +0100
It is the same bug as far as I know.
----- Original Message -----
From: "Fabian Becker" <neonomicus () gmx de>
To: "Berend-Jan Wever" <skylined () edup tudelft nl>
Cc: <bugtraq () securityfocus com>; <full-disclosure () lists netsys com>
Sent: Tuesday, November 16, 2004 20:50
Subject: Re: Skype callto:// BoF technical details
Berend-Jan Wever wrote:
Skype reported they've found a remotely exploitable BoF in the callto:// URI handler. New version has been released.
The bufferoverflow happens when a skype user clicks on a "callto://username" link with a username longer then 4096
characters that does not exist: An error message is created and put into a buffer without correct size checks. The
errormessage and buffer are unicode but unicode characters are filtered out and replaced with '?'. Only printable
ascii characters seem to get through. A return address can be overwritten as well as the SEH. Exploitation is
complicated by the fact that return addresses have to be in range 0x00??00??.
Webbrowsers like MSIE do not support URI's long enough to trigger the BoF. To exploit it, one could send a skype
user a callto:// link in a private message and trick him/her into clicking it.
If one would want to, one could write a skype worm with this. User interaction would be required: they'd have to
click the link.
They fixed it without knowing of the callto:// thing I suppose cause I
wrote them an email saying that the quick-call field is exploitable,
too. This was fixed within the new version. Maybe your flaw is fixed,
too, if not, I think it soon will be :)
Full-Disclosure - We believe in it.