Geohot’s discovery of PS3 console metldr keys caught a lot of attention across the net and because of this it has open a lot of doors to some unfinished keys. It does not go on pretty well with Marcan though. In the last 27th Chaos Communication Conference (27c3), he and his Fail0verflow team has presented a wonderful explanation on the PS3 total epic fail security. The 45-minute presentation revealed the methodology that made the on-die console security an irrelevance and proved beyond doubt that the Hypervisor tech – the CPU guardian that is supposed to stop unauthorized code running – was almost completely pointless, at least in the eyes of the hackers.
So, the point of this post is that without Marcan and his Fail0verflow team, Geohot doesn’t have the chance to even find the metldr keys which unlock all security levels in the PS3 console. Fail0verflow team is like the teacher who teaches Mathematic to Geohot and like what euss said, “Reversing, discovery, documenting, presentation, proof of concept AND release tools @ private keys is an unique world achievement”
Announce/publish a ridiculous break in embedded security, a few sites pick it up. Geohot uses it, it’s all over the web. Sigh.
I mean, maybe it’s just me, but I think the “calculating private keys” part deserves a bit more credit than a metldr exploit.
For the record, this an Alexa stats which compared the popularity of Geohot and Fail0verflow team respective sites. Suffice to say, it’s Fail0verflow team that is far more popular, so no, Geohot is not overrated
UPDATE: Marcan’s rants in Slashdot, the drama continues icon smile Marcan: Fail0verflow Deserved More Credit Than Geohot
We (fail0verflow) discovered and released two things:
An exploit in the revocation list parsing, enabling us to dump a bunch of loaders, and thus their decryption keys
A humongous screwup by Sony, enabling us to calculate their private signing keys for all of those loaders, and thus sign anything to be loaded by those loaders
We used these techniques to obtain encryption, public, and private keys for lv2ldr, isoldr, the spp verifier, the pkg verifier, and the revocation lists themselves. We could’ve obtained appldr, (the loader used to load games and apps), but chose not to, since we are not interested in app-level stuff and that just helps piracy. We didn’t have lv1ldr, but due to the way lv1 works, we could gain control of it early in the boot process through isoldr, so effectively we also had lv1 control.
With these keys we could decrypt firmware and sign our own firmware. And since the revocation is useless and the lame “anti-downgrade” protection is also easily bypassed, this already enables hardware-based hacks and downgrades forever. Basically, homebrew/Linux on every currently manufactured PS3, through software means now, and through hardware means (flasher/modchip) forever, regardless of what Sony tries to do with future firmwares.
The root of all of the aforementioned loaders is metldr, which remained elusive. Then Geohot announced that he had broken into metldr (with an exploit, analogous to the way we exploited lv2ldr to get its keys) and was thus able to apply our techniques one level higher in the loader chain. He has released the metldr keyset (with the private key calculated using our attack), but not the exploit method that he used.
The metldr key does break the console’s security even more (especially with respect to newer, future firmwares – and thus also piracy of newer games), and also makes some things require less workarounds. Geohot clearly did a good job finding an exploit in it, but considering a) he used our key recovery attack verbatim, and b) he found his exploit right after our talk, so he was clearly inspired by something we said when we explained ours, I think we deserve a little more credit than we’re getting for this latest bit of news.
There’s still bootldr and lv0, which are used at the earliest point during the PS3 boot process. These remain secure, but likely mean little for the PS3 security at this stage.
We published our exploits at the talk by explaining exactly how they works, and how anyone could use them. We said we’d release tools through the following month, and we already released two Git repositories containing most of the tools (that’s 4 days after the talk). We didn’t release keys due to fear of legal repercussions, but we told people exactly how to calculate them, and they did.
Geohot first released a useless signed loader to prove that he had the keys. Then he released the keys. He hasn’t released information on how he got the metldr plaintext and apparently doesn’t have plans to do so.
Personally, I think explaining things first, then a few days later releasing tools, is better than just dumping keys on the world and keeping how you got them a secret.