oss-sec mailing list archives

Re: CVE Request: KDE Pim


From: Vincent Danen <vdanen () redhat com>
Date: Tue, 17 Jul 2012 10:06:09 -0600

* [2012-07-17 10:18:06 +0200] laurent Montel wrote:

Le lundi 16 juillet 2012 11:47:59 Vincent Danen a écrit :
* [2012-07-13 10:41:33 -0600] Kurt Seifried wrote:
>On 07/13/2012 06:25 AM, Marc Deslauriers wrote:
>> Hello,
>>
>> Could a CVE please be assigned to the following issue:
>>
>> Javascript and external images were being loaded while rendering
>> HTML email in kmail. The downloaded Javascript was then being
>> interpreted.
>>
>> See:
>>
>> https://projects.kde.org/projects/kde/kdepim/repository/revisions/dbb2f72
>> f4745e00f53031965a9c10b2d6862bd54>>
>>  https://bugs.launchpad.net/ubuntu/+source/kdepim/+bug/1022690
>>
>> Thanks,
>>
>> Marc.
>
>This seems like a security hardening issue to me, but I'm not a KDE
>person so did kdepim advertise itself as not executing JavaScript/etc?

Doing some digging, it looks like this was introduced in kdepim 4.4, and
would not affect earlier versions.  Can anyone confirm this?

No it was added in 4.6 or 4.7 when we ported to akonadi2
In 4.4 it didn't use *WebKit*

and it didn't use same code :
it used DOM::

'  try {
  // Create a DOM Document from the HTML source
  DOM::HTMLDocument doc;
  doc.open();
  doc.write( htmlSource );
  doc.close();

  mIsQuotedLine = false;
  mIsFirstTextNodeInLine = true;
  processNode( doc.documentElement() );
  return doc.toString().string();
}
'

So for me it's just kmail > 4.6 (we released an old 4.6-akonadi)

not necessary to try to fix in 4.4

I CC David if he has more infos.

I hope that it helps.

Security problem is that we allows to use javascript.
In 4.4 we don't have it.

Thanks for the info, Laurent (and thanks for the testcase, David).

I'm still unsure whether or not a CVE is required, as per Kurt's
comments about whether or not kmail was ever advertised as not
executing javascript.

I think because, at one time, it did not execute javascript, and now it
does (which is clearly bad for a mail client to do), it probably does
require a CVE name but before assigning one and getting negative
feedback, I'd like to make sure that assigning one is indeed
appropriate.

--
Vincent Danen / Red Hat Security Response Team

Current thread: