Technological discussion on PS3 security and crack.*

this is really awsome. I haven't used my ps3 in about 4 months since it broke. But if i can bypass the disc drive and still play games that be awsome
 
So, if that random number was actually random then they wouldn't hack PS3, am i getting it right?

Well, supposedly they had already broken the chain of trust, though it's not clear if they'd be able to get a working exploit from that. Them getting the private key might mean there's not much Sony can do about it.
 
Presumably Sony can put out a new release of the PS3 that has a fixed authentication system for the lower levels of the system. A new hypervisor, a revised L2, a new signing key (and proper use of the ECDSA algorithm this time) for the firmware custom to the new hardware. They can close a lot of the lower level things down and produce parallel firmware releases for the new hardware from here on out, but the horses are out of the barn for all existing PS3 hardware.

Whatever Sony does for remediation won't be cheap, or it won't be effective.

If the ability to freely run custom firmware on the current hardware base somehow threatens the ongoing integrity of the game signing keys, the situation is even worse.
 
What's the next security update going to have?

My guess is that firmware 3.56 is going to have a whitelist of runnable games ~ 3000 executables in total
(1000 per region?). Sony will have to have a new rsa / md5 hashing + some secret sauce run on all 3000 already released games and then have the next firmware download all of them to the hard drive. Then when ever a new game starts ups, the game OS will have to run this new rsa or md5 + secret sauce on the executable, compare it to it's whitelist, and if it passes, execute the game.
 
Presumably Sony can put out a new release of the PS3 that has a fixed authentication system for the lower levels of the system. A new hypervisor, a revised L2, a new signing key (and proper use of the ECDSA algorithm this time) for the firmware custom to the new hardware. They can close a lot of the lower level things down and produce parallel firmware releases for the new hardware from here on out, but the horses are out of the barn for all existing PS3 hardware.

Whatever Sony does for remediation won't be cheap, or it won't be effective.

If the ability to freely run custom firmware on the current hardware base somehow threatens the ongoing integrity of the game signing keys, the situation is even worse.

I guess i'm clueless, but is the private key for the actual firmware? or just for game apps that run on the gameOS?
 
What's the next security update going to have?

My guess is that firmware 3.56 is going to have a whitelist of runnable games ~ 3000 executables in total
(1000 per region?). Sony will have to have a new rsa / md5 hashing + some secret sauce run on all 3000 already released games and then have the next firmware download all of them to the hard drive. Then when ever a new game starts ups, the game OS will have to run this new rsa or md5 + secret sauce on the executable, compare it to it's whitelist, and if it passes, execute the game.

Sony already changes their game signing keys from time to time. The problem would be if the custom firmware could be modified to simply skip the check. From the video, the PS3 can easily be induced to run signed code off of the hard drive rather than a Blu-Ray, which already obviates all the per-game crypto.

No one has said anything to indicate that the game signing keys have been made vulnerable by these hijinks. If Sony is doing ECDSA for game signing the way they did for firmware, they'd need the whitelist, but if they are doing it correctly, the risk wouldn't be people signing their own games and running them on consoles running Sony firmware, the risk would be people running burned or USB copies of games on custom firmware that Sony can't block from the hardware.

Presumably they'll be able to put firmware checking into the SDK for new games so they can be made not to work if the firmware isn't trusted, but then it's a matter of hacking the games one by one, like in the old Apple 2 days. I would suppose.
 
I guess i'm clueless, but is the private key for the actual firmware? or just for game apps that run on the gameOS?

My understanding is that the signing keys used by the hypervisor to validate the root mode and the root mode to validate the bulk of Game OS are compromised. This team have not compromised any signing keys for games, and it's not clear that they have the ability to do so. That's assuming Sony didn't make the same crypto mistake at that level.

But if you can make and install your own custom firmware, you can always bypass the signing checks on games.
 
Presumably they'll be able to put firmware checking into the SDK for new games so they can be made not to work if the firmware isn't trusted, but then it's a matter of hacking the games one by one, like in the old Apple 2 days. I would suppose.

Can they implement firmware checking via PSN, so the hacked guys can't get online and ruin the multiplayer experience?
 
So, if that random number was actually random then they wouldn't hack PS3, am i getting it right?

They wouldn't be able to calculate the private keys easily.

No wonder I heard some rumblings about rechecking random key generator at work in my organization. ^_^
 
Can they implement firmware checking via PSN, so the hacked guys can't get online and ruin the multiplayer experience?

I would imagine they could, but I suspect the games would have to have the logic to do that built into them. I kind of guess that Sony did not make provisions for allowing PSN to order arbitrary games to do arbitrary checks of the system. In fact, given the hypervisor, I'm not even sure if games are supposed to have the power to run code to look at memory managed by the hypervisor for the boot loader, code validator, and etc.

So, they might be able to do something, or they might not. It would depend on how much foresight they put into the rest of their system.
 
Soughted? Is that a new word? :LOL:

It's a fast fingers word. ;-)


can this lead to serial number for online play like PC? Fight used games market, piracy and cheaters etc?

Online check is one way to address piracy, but it may not be very effective and requires constant struggle with the pirates. They will need a multi-pronged approach. They should keep improving XMB with better features (online and offline), so that users prefer to stay with the official launcher. On PSP, IMHO, the biggest problem was Sony neglecting PSP development initially; the homebrew folks end up supporting the users.

Sony may also want to come up with strategies to sell as many PS3s as possible to scale the numbers, plus better deals for software.
 
I would imagine they could, but I suspect the games would have to have the logic to do that built into them. I kind of guess that Sony did not make provisions for allowing PSN to order arbitrary games to do arbitrary checks of the system. In fact, given the hypervisor, I'm not even sure if games are supposed to have the power to run code to look at memory managed by the hypervisor for the boot loader, code validator, and etc.

So, they might be able to do something, or they might not. It would depend on how much foresight they put into the rest of their system.

It wouldn't work, or would just be reasonably trivial to work around. If the lower level code is compromised to the extent that this seems to be, then the higher level code can't trust anything you do with it. If you ask to read kernel or hypervisor memory (which wouldn't normally be allowed!) how do you know what you are actually getting is what is being executed and not just a copy of the real unmodified code? You can't.

In retrospect Sony shouldn't have decided to run games at a different run level to the kernel. The current setup makes it significantly more difficult for exploits in games to gain access to the system, but it turns out that hasn't been an issue. By setting things up the way they are, games can't directly call into the hypervisor and can't verify the kernel hasn't been comprimised.

This situation makes Sony look like n00bs in the world of computer security. It shouldn't happen like this!
 
It wouldn't work, or would just be reasonably trivial to work around. If the lower level code is compromised to the extent that this seems to be, then the higher level code can't trust anything you do with it. If you ask to read kernel or hypervisor memory (which wouldn't normally be allowed!) how do you know what you are actually getting is what is being executed and not just a copy of the real unmodified code? You can't.

Sure, true enough. The quality of their security work revealed in this presentation makes it seem unlikely that Sony did themselves any favors in preparing for mitigation, but one can always dream.

In retrospect Sony shouldn't have decided to run games at a different run level to the kernel. The current setup makes it significantly more difficult for exploits in games to gain access to the system, but it turns out that hasn't been an issue. By setting things up the way they are, games can't directly call into the hypervisor and can't verify the kernel hasn't been comprimised.

This situation makes Sony look like n00bs in the world of computer security. It shouldn't happen like this!

Yup. I've been reading through the archives at http://rdist.root.org/, which is a research blog published by a guy who does embedded security design. He was on the team that designed BD+ which is the only thing currently protecting Blu-Ray movie content at all.

I've learned a lot over the last couple of hours from it. Too bad Sony didn't spend the money on outside consultation on the PS3's internal security to go along with the security given the BD format.
 
Back
Top