I swear to the heavens above, this is probably, maybe, almost certainly the last post on this subject: David Maynor is a rather polite individual via email; I've never had the pleasure of meeting him. Likewise, his colleague Jon Ellch ("Johnny Cache"), his partner in the exploit last August that led to such gnashing of teeth. I've been corresponding with Maynor since last week when, at another Black Hat conference, he said that he and Ellch did, in fact, develop a way of gaining root control of a Macintosh running OS X 10.4.6, an system version superceded in June 2006. (Mac OS X 10.4.7 apparently had a slightly different problem; everything was fixed with patches by the 10.4.8 release.)
I have hours of conversation with colleagues since last week about this issue, with many of them stating that Maynor was still being too vague in describing what, precisely, he and Ellch had up their sleeves last August.
Via email this morning, I asked Maynor point blank: "At Black Hat last August, you had a working rootkit for Mac OS X 10.4.6 stock out of the box using Broadcom chipsets? No special software, no special configuration, just on and checking for a Wi-Fi signal, and you could gain root access and perform arbitrary actions as root?"
His very clear answer: "Yes."
Update: I later followed up with him as I was too specific. He confirmed that both PowerPC plus Broadcom (pre-Intel architecture) and a MacBook (with Intel) coupled with Atheros were targeted through two separately engineered exploits.
He also clarified that Brian Krebs did see a native exploit and a third-party exploit of Mac OS X. A Black Hat staffer also saw the exploit at that time before the talk was given.
I have spoken to multiple people who saw a demonstration of this last year before Apple released their patches, saw the email that Apple and Maynor exchanged, and/or had access to the actual code involved. (Correction: After more back and forth, I find I have not spoken to anyone who has seen the native exploit, only the third-party one.) I posted an editor's note on a previous post that Maynor should release these folks from any pledges of secrecy. He did:
"At this point anyone I have told anything to about this has been released from any obligation to me from discussing it. As far as I am concerned it is open season, no more secrets."
Anyone who has information is now welcome to confirm this with Maynor--an out-of-band confirmation is always best in security fields, you don't have to trust me that he said this--and then post in the comments below what you saw and what you know.
A colleague I spoke with nailed down the two issues at stake here, and why we keep writing about it. This was echoed in comments on the post about last week's Black Hat presentation, too.
First, did Maynor and Ellch have a root exploit, not just a way to crash 10.4.6? Maynor has now said unequivocally yes. I know, many of you are still saying, "Show me the code." Last week, Maynor wrote in comments on my site that he had released his code and it "should be showing up on websites at any time." He said he has withheld the "weaponized shellcode" that allows root access because of a presentation he's giving in a few months.
Second, did Apple lie about what Maynor and Ellch provided? Clearly, Maynor's email shows more communication than Apple stated. I have not heard back from Apple since speaking with them about this issue last week. I still believe that there were many cooks in the broth at Apple, and that statements made by, say, Lynn Fox, seem to reflect a lack of accurate information provided rather than an attempt to discredit Maynor and Ellch. There's no profit in that and a lot of potential exposure as a publicly held company. We'll see if we hear more about this.
Ladies and gentlemen, I give you the comments. Post what you know.
Glenn:
This won't ever be the last post on this subject.
I'm sure Dave Maynor is a nice guy. And so in Johnny Cache. Even Lynn Fox....
But the problem is that MacBooks (intel) don't have Broadcom wireless chipsets -- they have Atheros chipsets:
http://www.wifinetnews.com/archives/006212.html
Or at the very least, SOME MacBooks have an Atheros chipset.
In writing fiction, there is an old adage: show, don't tell.
That adage applies here. I'm way past the point that I care what anyone, especially the principals, has to say. The only thing that is useful at this point is showing the exploit.
And even then, it's possible to argue that the exploit didn't exist last year. Not showing the exploit then was a critical error in judgement that can't be fully repaired now.
By now it should be clear to everyone that Maynor can't stay on topic and tell a clear and consistent story. If he's been telling the truth, his oral and written communication skills are abysmal. And in that case, imagine what his bug reports must look like. If Apple didn't find them useful, he has only himself to blame.
The tragedy - to both David Maynor and Jon Ellch - is that they originally set out to show that ALL operating systems have/had this weakness. But then for some reason (a mistake which David has acknowledged) they chose to use a MacBook to demonstrate their claim. Maybe at the time they thought it was a good idea because it would bring the maximum coverage, who knows?
The second mistake, which David also acknowledged, was showing that set-up to a journalist who, for all we know, saw an opportunity to make a name for himself and advance his stature towards the keys of the executive toilet. Said journalist decided to go with a sensationalised angle designed to grab maximum attention - "Yes, Macs can be hijacked, and in under a minute, too!" - thereby ensuring the security duo's freefall into lasting infamy.
Under the original pretext, the journalist could have written about how the platform was not the point of the exploit - Maynor and Ellch said it would work on ANY platform. But imagine a headline that said: "Hijacking a Windows XP notebook in 60 seconds" - people would have just yawned and flipped straight to the funny pages.
"Hijacking a Linux notebook in 60 seconds" may have gotten people perked up, but again, Apple is a much juicier target because of the PR value. In this way, Maynor and Ellch could be guilty of tying their own noose. Brian Krebs only helped push.