Errors in decompiling

Question

Errors in decompiling

Chaprnks opened this issue 5 years ago · comments

Chaprnks commented 5 years ago

This is the only result in decompiled .pwn:

Chaprnks commented 5 years ago

Any luck?

iAmir · Answer 1 · Tue Sep 17 2019 15:56:56 GMT+0800 (China Standard Time)

@Chaprnks its probably because that compiled script is using one of anti deamx methods.

Chaprnks · Answer 2 · Thu Sep 19 2019 01:26:55 GMT+0800 (China Standard Time)

Is there any way to work around or bypass these anti deamx methods? like maybe check if a chunk looks like one, then skip those lines?

iAmir · Answer 3 · Tue Oct 01 2019 15:23:38 GMT+0800 (China Standard Time)

@Chaprnks sorry for late answer, I didnt get any email from github for your reply 🙏
It's not like there's an actual line in compiled scripts to skip it
I've been thinking about it a week ago and I might come up with an Anti-Anti-DeAMX (weird name tho 😄 ).
I'd update this if I manage to make it work.

iAmir · Answer 4 · Fri Nov 08 2019 07:21:36 GMT+0800 (China Standard Time)

@Chaprnks its kinda funny that you commented here today, and I didn't know about it.
I updated deamx just now, came here to tell you and saw your comment :P well then, go ahead, try it now

James Freund · Answer 5 · Thu Jan 09 2020 10:38:35 GMT+0800 (China Standard Time)

@AmyrAhmady getting an error thrown on this amx file.

lua53: deamx_util.lua:148: bad argument #2 to 'seek' (not an integer in proper range)
stack traceback:
        [C]: in method 'seek'
        deamx_util.lua:148: in function 'readString'
        deamx_util.lua:8: in function 'readPrefixTable'
        deamx_util.lua:43: in function 'loadAMX'
        deamx.lua:431: in main chunk
        [C]: in ?

Any thoughts on why this might happen?

iAmir · Answer 6 · Fri Jan 10 2020 17:03:56 GMT+0800 (China Standard Time)

@Quixomatic can I have your amx file for debugging and testing till I find what's wrong?

James Freund · Answer 7 · Fri Jan 10 2020 18:10:46 GMT+0800 (China Standard Time)

amx file

@AmyrAhmady , yeah, I went ahead and uploaded to google drive: battle_man.amx

James Freund · Answer 8 · Sun Jan 12 2020 01:39:04 GMT+0800 (China Standard Time)

@AmyrAhmady any luck? I noticed the offset value being fed to the seek function seemed really high when trying to debug.

James Freund · Answer 9 · Fri Apr 24 2020 22:06:10 GMT+0800 (China Standard Time)

@Chaprnks why did you close this issue?

Chaprnks · Answer 10 · Sat Apr 25 2020 01:37:27 GMT+0800 (China Standard Time)

Sorry, did it on accident.

iAmir · Answer 11 · Sat Apr 25 2020 05:30:57 GMT+0800 (China Standard Time)

Thanks for the accident actually, I totally forgot about this issue.
I just downloaded that .amx file and I kept looking for the issue for the past hour, I really don't know why this happens, sounds like a really weird issue.
It would be awesome if you could reproduce this bug by writing a pawn script and compile it into a .amx file, so we know what happened in the pawn side as well and look for the answer in a better way with source code in our hand

James Freund · Answer 12 · Sun Apr 26 2020 01:47:21 GMT+0800 (China Standard Time)

I will try to recreate the scenario and provide the source code.

Jack-Rickwood · Answer 13 · Wed Apr 29 2020 19:53:41 GMT+0800 (China Standard Time)

@AmyrAhmady I am also having this exact same issue. Im not sure if you knew, but battle_man.amx is actually from pokemon sword and shield, and so is the file i am trying to decompile, so i guess its an issue with all pokemon sword and shield scripts.
Edit: How would i even go about reverse engineering an amx file?

Jack-Rickwood · Answer 14 · Thu Apr 30 2020 11:17:54 GMT+0800 (China Standard Time)

Okay i have tried about 20 other sword and shield amx files, and all of the ones above roughly 4kb give this error, and the ones below 4kb give a different error.

Jack-Rickwood · Answer 15 · Thu Apr 30 2020 11:21:14 GMT+0800 (China Standard Time)

In case its helpful, i have uploaded 678 scripts from the game here: https://drive.google.com/file/d/1EsDS6vD6sCF9dm0z_d_lZR5a8D4wZrHJ/view?usp=sharing.

iAmir · Answer 16 · Thu Apr 30 2020 15:45:31 GMT+0800 (China Standard Time)

Ah.. I see now, wish you could tell me this sooner, this DeAMX aka AMX decompiler only works with compiled pawn 3.2 samp scripts (San Andreas Multiplayer) I don't think we can make it work for another game or newer pawn/AMX VM versions without changing the whole code

Jack-Rickwood · Answer 17 · Thu Apr 30 2020 15:59:22 GMT+0800 (China Standard Time)

I see, fair enough. If I was interested in manually reverse engineering one of these files on my own, how hard would that be and how would I go about it? Are there any disassemblers that might work on this file? @AmyrAhmady

iAmir · Answer 18 · Thu Apr 30 2020 16:19:24 GMT+0800 (China Standard Time)

The thing is, AMX changed a lot since v3.2 (obviously) so you should start reading new documentations available in pawn repository at here https://github.com/compuphase/pawn/tree/master/doc they can be useful
Also you can check AMX implementation here https://github.com/compuphase/pawn/tree/master/amx which helps A LOT for reversing your compiled scripts that basically AMX bytecodes

There are a few reliable and useful sources you can check and look into doing the RE by yourself or make use of those tools
https://forums.alliedmods.net/showthread.php?t=294847

I know that Lysis is open source, both old one and the new one which is written in Java
you can find them here
https://github.com/peace-maker/lysis-java

Jack-Rickwood · Answer 19 · Thu Apr 30 2020 17:42:24 GMT+0800 (China Standard Time)

Alright, thanks a bunch!

Edit: This is very helpful in case anyone is trying to do the same as me: https://github.com/YashasSamaga/AMX-Assembly-Docs/blob/master/DOCUMENT.md

Edit 2: It looks like pokemon swsh is NOT using pawn version 4, as when I look at the commit history, the current_file_version variable was bumped from 10 to 11 in the 4.0 commit, yet the in the swsh amx file header, the file version value is '0A' (10).

iAmir · Answer 20 · Sat May 02 2020 01:58:46 GMT+0800 (China Standard Time)

Pawn 10 doesn't even exist tho, latest is 4, pretty sure you're doing something wrong, I'll look into it too when I get home

Jack-Rickwood · Answer 21 · Sat May 02 2020 06:45:04 GMT+0800 (China Standard Time)

@AmyrAhmady Oh no I'm completely aware of that. It's actually using pawn 3.x, but that's just represented by a 10 in the file header. When I downloaded a san andreas mp script, it had an 8 in its header, meaning it uses an earlier version of pawn 3.x.