|
The iPhone Wiki is no longer updated. Visit this article on The Apple Wiki for current information. |
Difference between revisions of "Research: Pwnage Patches"
ChronicDev (talk | contribs) |
m (Disambiguation.) |
||
| (20 intermediate revisions by 8 users not shown) | |||
| Line 3: | Line 3: | ||
Thanks to CPICH for helping out! |
Thanks to CPICH for helping out! |
||
| + | ==The Patches== |
||
| − | == 2.0 (5A347) iBoot == |
||
| + | There are three core patches in Pwnage. |
||
| − | === |
+ | ===[[iBoot (Bootloader)|iBoot]]=== |
| − | There is only 1 patch made to the iBoot, iBEC, iBSS, and WTF |
+ | There is only 1 patch made to the [[iBoot (Bootloader)|iBoot]], [[LLB]], [[iBEC]], [[iBSS]], and [[WTF]]. It simply patches the RSA check to return success when an error occurs. |
| + | The patch is simple. This is the default non-patched area: |
||
| − | Here is a snippet of it from IDA: |
||
| + | <pre> |
||
| + | ROM:00002636 _fail ; CODE XREF: rsaCheck+3C�j |
||
| + | ROM:00002636 ; rsaCheck+84�j |
||
| + | ROM:00002636 ; rsaCheck+A0�j |
||
| + | ROM:00002636 ; rsaCheck+B4�j |
||
| + | ROM:00002636 ; rsaCheck+F0�j |
||
| + | ROM:00002636 ; rsaCheck+100�j |
||
| + | ROM:00002636 ; rsaCheck+10C�j |
||
| + | ROM:00002636 ; rsaCheck+110�j |
||
| + | ROM:00002636 ; rsaCheck+118�j |
||
| + | ROM:00002636 ; rsaCheck+11C�j ... |
||
| + | ROM:00002636 454 01 20 MOVS R0, #1 |
||
| + | ROM:00002638 454 40 42 NEGS R0, R0 |
||
| + | </pre> |
||
| + | Now, all that they do is change the "40 42" you see at 0x2638 to "00 20" (negs r0,r0->movs r0,#0). |
||
| − | ROM:1800587C 01 20 MOVS R0, #1 ; R1 = 1 |
||
| − | ROM:1800587E 40 42 NEGS R0, R0 ; PWNAGE PATCH |
||
| − | ROM:1800587E ; Change 40 42 > 00 20 |
||
| − | ROM:1800587E ; That will make it: |
||
| − | ROM:1800587E ; MOVS R0 = #0 |
||
| − | ROM:1800587E ; |
||
| − | ROM:1800587E ; R0 (unpatched) = -1 |
||
| − | ROM:1800587E ; R0 (patched) = 0 |
||
| + | ===DeviceTree=== |
||
| − | === Why does this help us? === |
||
| + | This simply patches "secure-root-prefix" and "function-disable_keys". It seems that you could bypass this patch by simply patching the [[iBoot (Bootloader)|iBoot]] flags to 0xffffffff, but I have personally never verified that the decrypted [[KBAG]] was correct, so anyone that is feeling adventurous and wants to verify, please do. |
||
| − | Well, this is a bit later on... |
||
| + | ===[[Kernel]]=== |
||
| − | ROM:180058C4 00 28 CMP R0, #0 ; Does R0 = 0? |
||
| + | Haven't really looked into this too far, but I know that the kernel is patched for codesign as well as a write and execution patch to allow addons such as [[Mobile Substrate]] to execute on / at the kernel |
||
| − | ROM:180058C6 3C D1 BNE loc_18005942 ; if R0 does not = 0 |
||
| − | ROM:180058C6 ; jump to 0x18005942 |
||
| − | ROM:180058C6 ; |
||
| − | ROM:180058C6 ; Pwned iBoot not jump |
||
| − | ROM:180058C6 ; Unpwned iBoot will |
||
| − | |||
| − | As you can see from my comments, it makes it so that it does not jump. I am no 1337 reverser, so I have no clue what is done when there is no jump, but I do see the fact that a pwned [[iBoot]] will not jump to 0x18005942, and an unpwned [[iBoot]] will not. This kind of interests me here, because with a Branch statement, usually you would be able to get around something that you want to by changing a certain Branch from BEQ to B, or something of the like. This one is a bit trickier, and you can't really do something like that...probably a common thing, but I am just throwing it out there since it is the first time I have seen something like this done. You may say "But Chronic, why not just NOP the BNE? Isn't that easier?". Well, for some reason, more than a few times, a check is made to see if R0 = 0 throughout the routine that the patch makes us not skip. Also, I might be missing something, and there could be an important reason for them to have R0 = 0. Whatever the case is, even if they don't have an important reason, 1 patch for the '''entire file''' is better than doing, well, anything more than 1 patch in this case :) |
||
| − | |||
| − | == 2.0 (5A347) [[Lockdownd]] == |
||
| − | This may actually confuse some people. You see, there is 'technically' two patches, but in reality, there is only one. The second one is the rehashed signature done with ldid, because you must remember that this is a userland binary, not a lower level one like the [[iBoot]], which resides in the [[NOR]]. These files on the main filesystem must cohere to the demands of the [[kernel]], and according to a [[devteam]] member, the patches to not require this were to complex and it would just be much easier to use ldid to take care of it. So that is what they did here. They took the original file, then one with the one patch that they needed, rehashed the patched one, BsDiff'd them, and then as you can now tell, the .patch tile contains the actual patch + the new sig :) |
||
| − | |||
| − | (Be back in a little bit with actual snippets from IDA showing the actual patch done, I want to go through the actual low level stuff first) |
||
Latest revision as of 04:42, 8 November 2010
If you have IDA Pro and you are at least semi-handy with ARM please contribute :)
Thanks to CPICH for helping out!
Contents
The Patches
There are three core patches in Pwnage.
iBoot
There is only 1 patch made to the iBoot, LLB, iBEC, iBSS, and WTF. It simply patches the RSA check to return success when an error occurs.
The patch is simple. This is the default non-patched area:
ROM:00002636 _fail ; CODE XREF: rsaCheck+3C�j ROM:00002636 ; rsaCheck+84�j ROM:00002636 ; rsaCheck+A0�j ROM:00002636 ; rsaCheck+B4�j ROM:00002636 ; rsaCheck+F0�j ROM:00002636 ; rsaCheck+100�j ROM:00002636 ; rsaCheck+10C�j ROM:00002636 ; rsaCheck+110�j ROM:00002636 ; rsaCheck+118�j ROM:00002636 ; rsaCheck+11C�j ... ROM:00002636 454 01 20 MOVS R0, #1 ROM:00002638 454 40 42 NEGS R0, R0
Now, all that they do is change the "40 42" you see at 0x2638 to "00 20" (negs r0,r0->movs r0,#0).
DeviceTree
This simply patches "secure-root-prefix" and "function-disable_keys". It seems that you could bypass this patch by simply patching the iBoot flags to 0xffffffff, but I have personally never verified that the decrypted KBAG was correct, so anyone that is feeling adventurous and wants to verify, please do.
Kernel
Haven't really looked into this too far, but I know that the kernel is patched for codesign as well as a write and execution patch to allow addons such as Mobile Substrate to execute on / at the kernel
