Apple sits on critical Mac bug for 7 months (and counting)
Unix flaw fixed in OpenBSD, not OS X
Researchers have disclosed a critical vulnerability in the latest version of Mac OS X that they say Apple has sat on for almost seven months without fixing.
The buffer overflow flaw could be exploited by attackers to remotely execute malicious code, and virtually all Apple devices - including Mac computers and servers, iPhones, and even Apple TV - are susceptible, one of the researchers, Maksymilian Arciemowicz, told The Register. SecurityReason.com, the Poland-based security firm he works for, alerted Apple to the vulnerability in the middle of June and again last month, but the computer maker has yet to patch the bug.
By contrast, developers for OpenBSD, NetBSD, FreeBSD, and a variety of Mozilla applications have fixed identical vulnerabilities, in some cases within hours of notification. The bug affects all applications and operating systems that implement gdtoa floating point numbers.
"It was not that difficult to patch it," Arciemowicz wrote in an email. "It seems to us that Apple comes from the assumption that when there is no PoC or exploit given that the problem doesn't exist."
The OS X bug resides in the libc/strtod(3) and libc/gdtoa function. Arciemowicz said the vulnerability could be remotely exploited using booby-trapped PHP code on a website, among other methods.
SecurityReason has posted proof-of-concept code here that shows how the flaw can be exploited to make a machine crash. With additional work - specifically, by manipulating esi and edi registers - it is possible to remotely execute code, Arciemowicz said.
Of the 16 applications or systems known to be affected by the bug, only four remain vulnerable. In addition to OS X, they include Mozilla Sunbird, K-Meleon, and the J programming language. ®
COMMENTS
For every 1
You mean, for THE one OSX server running a public facing web site, right?
Big Deal
Well, my Macbook (Snow Leopard latest patches) doesn't crash when I test out this bug - it does crash the process that's in context though.
Your standard buffer-overflow exploit then, and nothing worth getting all fanny-flapping about.
"Not entirely Apple's fault?"
Apple may not have written the buggy code, but they do deploy it. They were told where the issue is, and in there are open source apps that have issued fixes. (You can go crib the code if you can't write your own fix.)
How is it not Apple's fault?
10.6.3...
is currently being finalised. I'd expect that this will be fixed. http://arstechnica.com/apple/news/2010/01/apple-moves-to-improve-opengl-support-in-1063-builds.ars
Based on my reading of the article...
The security firm found this exploit 7 months ago then contacted Apple (and the other affected comapnies) and has since sat on it and not released details in order to give the companies time to patch their systems. Most of those have. Apple has not.
Now in order to try and force Apple to fix the security hole, the security firm are releasing a script which (based on peoples comments here) doesnt entirely work but should be close enough that someone with insider knowledge of the system could get it to work, but it would probably take time and hopefully Apple will now use that time to plug the gap. This seems like a particularaly responsible way to go about it to me.
Still if Apple continues to do nothing, i kinda hope (and i apologise to Apple users in advance) that this security hole does get exploited so that it embarrases the hell out of Apple and makes them get around to fixing the issue. Just because no one has taken advantage of a security hole and even if that security hole is very difficult to penetrate there is no excuse for not plugging it when it appears to be a (relatively) simple fix...
