PS3Hax user miniv1 has posted some very interesting news today, after doing some research he seems to have figured out how recreated the PS3 JIG – now before you get to excited, understand that this is still in development and is currently a theory, but a pretty damn good one (confirmations from various PS3 devs). It still requires testing and dumping, but its a good start. His method requires 3 things: dongle master key, Lv2diag.self, and challenge/response which ultimately allows us to get a dump of lpar1 with a modified (existing or new) dump code. Add the 3 together in a new payload and you’ve got yourself the PS3 service mode .
Been doing a lot of searching the net found lots of usefull info and think I’ve discovered the way to recreate the jig.
Originally Posted by Mathieulh @psx-scene
That’s not about it, it’s about the fact that even if you manage to reproduce the dongle by dumping the decrypted “dongle master key” and reversing the challenge/response algorithm, you’d still need to use a signed/copyrighted self renamed as lv2diag.self from /dev_usb000/ The product mode flag being of no use on its own. The fact that some people know how this process works and the whole theory behind it doesn’t mean they care or have any interest whatsoever in this procedure especially as there is certainly no great mystery about it (at least not as far as we are concerned). Also the other problem that occurs when it comes to using signed selfs (besides the obvious copyright issue) is the self revocation.
Which says we need 3 things
The dongle master key
Originally Posted by http://ps3wiki.lan.st/index.php/Hypervisor_Reverse_Engineering#0x24000_-_USB_Dongle_Authenticator
0×24000 – USB Dongle Authenticator
Packet ID Description
0×24001 Generate Challenge
0×24002 Verify Response
0×24001 – Generate Challenge
* I have got access to this service through DM and tested it
* The service expects no input parameters except those in SS packet header
* It uses 0×5003 service (Generate Random Number) to generate random numbers that are used in challenge body
* The length of a challnge body is always 23 bytes, first 3 bytes are always the same: 0x2E 0×02 0×01
Here are hexdumps of some challenge bodies i let 0×24001 service generate:
2E 02 01 72 3A 0A 76 BB 81 CB 29 BC E7 B5 D6 62 7C 0E EE 23 18 A9 1D
2E 02 01 F0 DA 78 D4 1D CB D7 C9 C7 F0 32 F4 2E 92 39 BD 3F 32 93 AA
2E 02 01 3B B2 9D FD A8 83 AF 9A C0 E9 13 BB AE D5 6C 8C 45 2E DE 13
0×24002 – Verify Response
* I have got access to this service and tested it with PSGroove
* The response body is 25 bytes large
* The first 3 bytes have to be 0x2E 0×02 0×02 or else the check fails
* The 16 bit at offset 3 is a dongle ID
* The dongle ID is checked if it’s revoked or not
* When the verification succeedes then product mode is set to 1
* The service calculates USB Dongle Key from USB Dongle ID and USB Dongle Master Key by using HMAC SHA-1
* The service uses HMAC SHA-1 to calculate the correct response body from the challenge body and USB Dongle Key
* After that the service compares the calculated response body with the given one that was sent to the service
* It seems that laid and paid from SS packet header are used in decryption process
USB Dongle Master Key
* USB Dongle Master Key is stored encrypted in Process 6
* The encrypted key is 64 bytes large
* The decrypted key is 20 bytes large
* The USB Dongle Master Key is decrypted first time the service 0×24002 is used
* The USB Dongle Master Key is decrypted by using the service 0x200E (Decrpyt Master) of Vitual TRM Manager
* The decrypted USB Dongle Master Key is stored in Process 6 in clear text (after first usage of this service)
* When decrpyption of USB Dongle Master Key fails then a dummy key is used
* Unfortunately, in the HV dump 3.15 the USB Dongle Master Key was not decrypted at the moment of dumping
Here is the encrypted USB Dongle Master Key from HV 3.15:
0×22 0xD5 0xD1 0x8C 0xFF 0xE2 0x4F 0xAC 0xEC 0×72 0xA2 0×42 0xA7 0×18 0×98 0×10
0×25 0×33 0xE0 0×96 0xF2 0xC1 0×91 0x0D 0×15 0×23 0xD3 0×07 0×74 0xE7 0x2B 0×72
0xDF 0xA6 0xDD 0xE9 0×68 0x8B 0×76 0x2A 0x6A 0×87 0×51 0x7F 0×85 0×39 0x0B 0xD4
0×20 0x3F 0×46 0×89 0×04 0×82 0xB7 0×30 0×84 0×89 0x4B 0xCC 0x9D 0xB1 0×24 0x7C
Here is the USB Dongle Master Dummy Key from HV 3.15:
0xD1 0xFC 0×57 0×55 0xBF 0×20 0xFA 0xB2 0xD4 0xA5 0x4A 0x0A 0x0C 0x5D 0×52 0x8E
0xDF 0×66 0xCD 0×74
USB Dongle ID Revoke List
* Process 6 contains a revoke list for USB Dongle IDs
* The revoke list is 0×2000 bytes large. It’s a bitmap.
* Each bit represents a USB Dongle ID. If bit is 0 then USB Dongle ID is revoked.
The following USB Dongle IDs are revoked in HV 3.15:
0, 2, 13, 32, 34, 176, 241
Originally Posted by http://www.ps3-hacks.com/2010/11/16/ps-downgrade-real-confirmed-working-100/#comments
And the challenge/response
Originally Posted by http://twitter.com/#!/Mathieulh 15th Nov
@ldgchad it’s a reverse of the dongle authentication challenge/response from the ps3 side. If you can dump lpar1 it can be done.
Then I stumbled upon
Hi guys, I used an Atmega8 running at 16Mhz (I had a couple lying about from the BT Vision project I was working on) and knocked up a small prog to do the same as the other chips and dump out the PS3 Hypervisor and Bootloader.
I was quite surprised, It actually worked fairly straight away! I only had one pulse going everytime I pressed the button at first but not a lot was happening.
So I did what xorloser did, and modded it so it pulsed every 100ms while the switch is pressed.
After about 30-40 seconds… I got a hit with the exploit code posted here. Then I used the dumper (posted here) to dump the 10mb bin.
Just having a look through the dump, lots of strings in there.. I haven’t dropped it into IDA yet tho…
This is the source and hex (for those who dont want to compile it) for the Atmega8 which I glitched my PS3 with. The Chip I used was the Atmega8-16pu. You will also need a 16mhz Crystal, and 2 x 22pf Capacitors.
Grounding pin 14 on the chip will produce a pulse on Pins 2 of the chip (infact it does all of PORTD) This should then go to the memory bus point on the ps3. See Circuit diagram (below).
I used ponyprog to program my chip, with CKOPT ticked in the fuse settings, everything else was unticked.
Which should be enough to get a dump of lpar1 with a modified (existing or new) dump code.
Add the 3 together in a nice new payload and you’ve got service mode.