Written by Mahammad Mohsen
Monday, 15 June 2009 19:47
Apple Safari & Quicktime Denial of Service
Shameless plug :
You are invited to join the 2009 edition of HACK.LU, a small but
concentrated luxemburgish security conference.
More information : http://www.hack.lu - CFP is open, sponsorship is still
possible and warmly welcomed.
Release mode: Coordinated
Ref : [TZO-36-2009] - Apple Safari & Quicktime DoS
Vendor : http://www.apple.com
WWW : http://blog.zoller.lu/2009/05/advisory-apple-safari-quicktime-dos.html
Status : Not patched
Credit : none given (Apple can't find a place to credit)
Discovered : 18.11.2008 Zoller, 19.06.2009 Alexios Fakos (probably plenty
Security notification reaction rating : good
Notification to patch window : n+1
Disclosure Policy : http://blog.zoller.lu/2008/09/notification-and-disclosure-policy.html
- Apple Safari (all)
- Quicktime (all)
Wikipedia quote: "Apple Inc. (NASDAQ: AAPL) is an American multinational
corporation which designs and manufactures consumer electronics and
software products. The company's best-known hardware products include
Macintosh computers, the iPod and the iPhone."
A null pointer is being dereference when CFRelease() is called on NULL.
The browser will crash, your data might be lost.
IV. Proof of concept (hold your breath)
V. Disclosure timeline
18/11/2008 : Send proof of concept file and a description that failed to
give the correct impact.
25/11/2009 : Apple acknowledges receipt and reproducability :
"After investigating this issue further, we've determined
that the crash your test case triggers is caused by
dereferencing a null pointer and not from a format string issue"
20/01/2009 : Ask for an update
23/01/2009 : Apple sends an encrypted and signed PGP mail, fine, however the mail
is encrypted with their own key
23/01/2009 : Ask for the mail to be resend as I don't have Apple's private key
24/01/2009 : Apple states that "Regarding the QuickTime null dereference you
reported, this bug is still being worked on by our engineers
and is not addressed in QuickTime 7.6"
26/01/2009 : Ask apple for a fix timeline as this is an ridiculouly easy to fix
27/01/2009 : Apple statest "Regarding the QuickTime null deref issue, it is
currently set to be part of the next QuickTime update. [..]
Additionally, we do not intend to describe this crasher in our
Note: No Security advisory = no credit, should have published here.
28/01/2009: Apple states "Given that we are handling this as a crasher and
not as a security exposure, it stands to reason that you may
want to disclose it without waiting for the update that
addresses it and without further coordination with Apple.
We do appreciate the fact that you reported it to us and are
intending to address it in the next available update"
[Several discussion about CIA, why a DoS against the iPhone is worth a security
advisory, when it isn't against safari.. etc. I spare you the details]
29/01/2009 : Ask why I should hold disclosure for a DoS in a particular
portable apple product but disclose DoS in other apple products.
Asked apple to make a choice, either DoS is a security issue and
I won't disclose or it isn't and I disclose all of them,
including the one in the very portable apple product
30/01/2009 : Apple answers that
"Your QuickTime and Safari issues constitute denial of service.
We consider any denial of service issue to be security related,
and they are important to fix. We plan to fix the ones you
reported in the next available updates."
"I believe we can put credit in an appropriate place for the
WebKit/Safari change. I was not able to locate a suitable place
for crediting the QuickTime crasher"
Fast forward 5 months, and apple releases a stream of code execution bug fixes
01/06/2009 : Ask for an update and if the DoS condition has been fixed
02/06/2009 : Apple states that
"According to our bug tracking system the null-dereference crasher
issue is not yet addressed in QuickTime. We are investigating
now to see if for some reason the latest version has picked up
changes that address this issue and will send you feedback
today about it."
In summary, no credit, no advisory, and 7 months of time to (not) fix a
single line of code.
10/06/2009 : Release of this advisory