Updated: clarified some points of contention.
Early this morning Google’s Tavis Ormandy published a vulnerability in the hcp protocol handler. It allows the attacker to run arbitrary commands as the user. In practice it created a lot of alerts and warnings for me - but the XP install I was using is somewhat locked down. So I’m not sure how practical this attack would be over any other attack that causes an alert, as the article mentions. Later his reports says it works around the alerts (I couldn’t reproduce that, but that was his intention). Either way, though, this is some pretty amazing research. However, there are some odd things about this that really struck me the wrong way.
Google has been the loudest proponent for responsible disclosure in the past. But if you look at the dates in his post, he says he reported it to Microsoft on the 5th of June (a Saturday), who responded the same day. He sent the advisory early in the morning today the 10th of June - meaning Google gave Microsoft less than 5 days
to fix it to respond to his demand to have it fixed in 60 days. Even Mozilla backed down from 10 day turn around, and they’re only running a single software suite. How is that possibly reasonable to expect a company like MS to turn around a patch in 4-5 days and then get so upset that then you must go full disclosure? (Incorrectly stated) And it’s not like Tavis was acting on his own - he credits other security researchers inside of Google for their help lcamtuf who works at Google. So apparently it’s okay for
See, here’s the big problem. Either you are all about full disclosure (which is happening less and less these days), you use it only when you know the company won’t react otherwise or has all kinds of other hinky things they do behind your back (the same reason I advocate full disclosure against Google), or you use responsible disclosure. Google says it adheres to responsible disclosure, but at the same time they give Microsoft 5 days to
fix their 0day agree to a 60 day patch cycle for exploit code that Google’s researchers themselves created! From Google’s own website:
This process of notifying a vendor before publicly releasing information is an industry standard best practice known as responsible disclosure. Responsible disclosure is important to the ecology of the Internet. It allows companies like Google to better protect our users by fixing vulnerabilities and resolving security concerns before they are brought to the attention of the bad guys. We strongly encourage anyone who is interested in researching and reporting security issues to observe the simple courtesies and protocols of responsible disclosure. Our Security team follows the same procedure when we discover and report security vulnerabilities to other companies.
… except when you don’t.
Then Tavis puts a patch up on a domain that, no offense to Tavis, is more sketchy sounding than a lot of malware sites out there (http://lock.cmpxchg8b.com). Do you really expect a billion XP users to download and run that? (Non sequitur) There is evidence that it doesn’t even work in some cases, but it does appear to work against the one PoC Tavis put up in the test I ran. I don’t know, the whole thing just rubbed me the wrong way. But at least now no one has to pretend to do responsible disclosure with Google just because it’s the right thing to do - they don’t use it themselves. Even when MS finds a vuln in Google they do so responsibly. I don’t mean to say anything bad about Tavis, because he’s probably a good guy, with a lot of skill. But let’s stop pretending Google’s team is chivalrous, shall we? Let’s see what Google does when one of their own breaks their stated policies, whether the researcher is working in their own time or not.