Yeah the biggest problem is really the fact that the exploit itself is well a glitch. I mean, the hardware works perfectly, I can get it to start to exploit the box within 20 seconds of trying , every time.
The problem is, 90/100 times the exploit crashes / locks up the ps3 / errors , resulting in the need to reboot, and restart.
Once the exploit is planted, then we start running our own kernel module to dump out the real memory. The way we we're doing it is well, unreliable and prone to massive corruption (not to mention slow)
(it looks better in a Hex Editor!)
But, with dumping memory to a file you run into other issues. You can't just use FileI/O in a kernel module any more, and you can't access lv1_peek from user mode either, so you need to make some additional code to handle it, which is what were working on now - although I'm open to any suggestions to get it done faster, its such a pain after your kernel module crashes, and having to reboot and re-exploit the PS3!