You are not logged in.
One thing I noticed when I still had my PS3 is that if you look at the censor traces around the switch they all go to a middle PCB layer thus totally obfuscating the BUS logic. This is how I know the engineers attacked their own designs from every angel, I'm sure with their minds and software simulations.
I was just bored and thought I'd make a new thread to get people riled. If you figured out where they went though, something tells me it'd be a prelude to bypassing all the security. You'd have the initialization order, and you'd know where the signature chain starts which possibly exposes unchecked instructions. Even if not it'd be cool to know how it works.
Known Aspects:
- Common Key on CELL Die(like the IBM implementation on the x360)
- Allocation randomization and page locking on RAM and potentially real time parallel hashing as with the x360
- AES, SHA1, and some light check sum encodings are used on file systems
- SELF and SPRX files are encrypted binaries/executable(SELFtoEXE and SPRXtoDLL in functionality)
- Hardware init order intentionally obfuscated with middle PCB layer to prevent reversing
- sony south bridge, flash, ram, and RSX don't do initializations. A component of the CELL or another chip contains the first instructions, and it's not the EE chip obviously. The smartest thing would be to put it in non volatile memory on the PPU which it has. A locked SPE could also invoke initialization of the main firmware and be off the rest of the session explaining thermal imaging.
Know accomplishments:
- HDD Decryption with differentials and swapping(nobody has tried putting data from other parts of the PS3 through that decryption)
- Downgrading with man in the middle attacks
- Partial NAND dumping with infectus ASIC board directly interfaced to flash chips working off bus power
- Decoding of a light encoding on file system components
- PSP control and authentication protocol to XMB reversed(there is data exposed that nobody is researching in the form of keys that aren't used by the protocol itself but are still sent.)
I'm guessing the PPU is at the forefront of initialization, and probably has a unsigned and unencrypted initialization stack in a small piece of ROM or something, or the 'locked' SPE has it in die or loads it from the PPU ROM and uses the SDRAM(the fact it only handles SIMD supposedly gives me doubt about this though; the PPU does the job of the application processor with the RISC on the PS3.) That would be the hardest place to reverse a initialization stack. Their are also like ~18 other small black chips on the 40GB model that could be doing it; they have the profiles of DAC, DSP, and serial controllers though, but I never looked any of them up.
Continued: I think IBM did with the cell what they did with the xeon and put a rom in the cpu die that inits the entire system and probably can't be dumped. That means the only remote vector is something like the upper bit field syscall attack. This means a close to identical attack vector to the x360 will have to exist, because no amount of sophistication in a shellcode payload will get past read only pages whether stack or heap, or hashed processes. The PS3 in minimalistic and proprietary in comparison too, so this could mean no go without major physical modifications and no hope of backups or even xmb. The write protection and hardware common key are the strong points. They could of used weaker algorithms and left code unencrypted and probably still be secure if it wherent for licensing issues.
I was bored...
Last edited by 0m1kr0n (2009-06-20 09:44:10)
Offline
There may not be an actual ROM in the CPU Die, but there is code, now, if you want to call it ROM, then by all means anyone can. AS I stated before, these systems do not POST like a PC does, not at all. Each important device has a static memory address and location that is expected on boot. Anything that changes, the system doesn't post. There is no PnP, no self configuration, just boot with the assumption the device is where it is supposed to be. This is with exception to the BR-D and HDD, and USB devices. AS far as CPU, RAM, and GPU - they MUST be where they are supposed to be as far as addressing. Some idiot even said they got to the XB360 through post, because it is like a PC, and has CMOS, and what not; no, it doesn't...
This new generation of consoles, are the best in digital security, with the blunder of M$s use of a standard SATA DVD drive for it's game booting and reading. Sony did not go that route, but if the weak point is anywhere, it starts with the optical portion. They don't put this much security into some of our government mainframes, you can take that to the bank. The Chinese and some other countrys cyber attack us on a daily basis, but most of those are out of OUR country based. The USAF systems in Germany are attacked pretty often, by the Chinese.
Anyhow, as I stated, these consoles won't be hacked like the original XB1 was. The PS3 will, more than likely, not be hacked to play backups at all, or, IF it does? There won't be any interest in it, as the next one will be out.. With XB360, you got the backups through DVD firmware mod, that's it. It starts at the front door, or shall I say Slot. If they can get to the optical firmware, and get the keys from it, they can get started. Too many people are trying to attack the console directly, and as I stated several times , that is THE WRONG approach.
Offline
Powerslave wrote:
There may not be an actual ROM in the CPU Die, but there is code, now, if you want to call it ROM, then by all means anyone can. AS I stated before, these systems do not POST like a PC does, not at all. Each important device has a static memory address and location that is expected on boot. Anything that changes, the system doesn't post. There is no PnP, no self configuration, just boot with the assumption the device is where it is supposed to be. This is with exception to the BR-D and HDD, and USB devices. AS far as CPU, RAM, and GPU - they MUST be where they are supposed to be as far as addressing. Some idiot even said they got to the XB360 through post, because it is like a PC, and has CMOS, and what not; no, it doesn't...
This new generation of consoles, are the best in digital security, with the blunder of M$s use of a standard SATA DVD drive for it's game booting and reading. Sony did not go that route, but if the weak point is anywhere, it starts with the optical portion. They don't put this much security into some of our government mainframes, you can take that to the bank. The Chinese and some other countrys cyber attack us on a daily basis, but most of those are out of OUR country based. The USAF systems in Germany are attacked pretty often, by the Chinese.
Anyhow, as I stated, these consoles won't be hacked like the original XB1 was. The PS3 will, more than likely, not be hacked to play backups at all, or, IF it does? There won't be any interest in it, as the next one will be out.. With XB360, you got the backups through DVD firmware mod, that's it. It starts at the front door, or shall I say Slot. If they can get to the optical firmware, and get the keys from it, they can get started. Too many people are trying to attack the console directly, and as I stated several times , that is THE WRONG approach.
Yeah I figured everything except modular components where static addressed. I was referring to the first code that is loaded to the PPU when the power circuit is closed, whatever it may be. The PPU is the only application processor there, I'm sure the south bridge has non-volatile code storage that it loads off of too, but the bus suggest the CELL is the first initialized chip; although I can't say for sure of course because of the middle pcb layer, but the south bridge goes to the other public market chips and directly under the cell.
I've also found it idiotic since the beginning that people where attacking such a system from it's strongest point. The XB1 had part of it's initialization outside of the CPU and there was a place to put custom code. The X360 has it all on CPU die which is why the DMA/syscall or a return to lib attack are literally the only vectors. I think the lib attacks fail because of real time hashing in RAM though; it hangs the system like other simpler attacks did.
Through the process of elimination it's clear DMA attacks are the only option. Which only worked on the X360 because shaders weren't encrypted on some game images. Interesting thing is I've never seen a platform where shaders wher not low level instructions that had to be loaded faster; they might haven't done it in the PS3 SDK, or even patched it in firmware after seeing the X360 exploit.
I actually still look over all the x360 research and exploits. The most intriguing part is how they found out about the upper bit flags.
Trash Talk: Also regarding all the crack team garbage and idiot approaches. If you go look at their releases(paradox etc..) you'll notice none of them had decent protection, and those crack teams are pretty much sitting on side lines now because they're too busy on 4chan looking at porn to learn how to reverse a VM and code obfuscation.
Last edited by 0m1kr0n (2009-06-20 17:47:08)
Offline