Skype callto:// BoF technical details

[Berend-Jan Wever]

Skype reported they've found a remotely exploitable BoF in the callto:// =
URI handler. New version has been released.
http://www.skype.com/products/skype/...changelog.html
http://secunia.com/advisories/13191/

Technical details:

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.

Cheers,
SkyLined

[Berend-Jan Wever]

It is the same bug as far as I know.

Cheers,
SkyLined

----- Original Message -----=20
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:
>=20
> >Skype reported they've found a remotely exploitable BoF in the =
callto:// URI handler. New version has been released.
> >http://www.skype.com/products/skype/...changelog.html
> >http://secunia.com/advisories/13191/
> >
> >Technical details:
> >
> >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.
> >
> >Cheers,
> >SkyLined
> >
> >
> >
> > =20
> >
> 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,=20
> too. This was fixed within the new version. Maybe your flaw is fixed,=20
> too, if not, I think it soon will be
>=20
>

[Fabian Becker]

Berend-Jan Wever wrote:

>Skype reported they've found a remotely exploitable BoF in the callto:// URI handler. New version has been released.
>http://www.skype.com/products/skype/...changelog.html
>http://secunia.com/advisories/13191/
>
>Technical details:
>
>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 b
e 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.
>
>Cheers,
>SkyLined
>
>
>
>
>
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