Cheat Engine

[Autem]

I'd like the below script to disable itself and revert everything back to normal as soon as ONE instance runs through the modified/enabled newmem section. Whether it takes 1 hour of being activated or just 3 seconds, I only want the script to basically wait until it's used 1 time and then disable itself immediately to avoid any possible follow up data being influenced.

I've tried a few variations on how I think it's possible to call lua inline with asm but it's not working. How should I write the part that force disables everything after 1 usage?

[++METHOS]

You can use lua, or you can simply create a boolean that resets to 0 after you enable it.

[cooleko]

I'd just use LUA, condition the jump from executing the newcode on whether a flag is set, set the flag within the newcode, and have the script erase everything once the flag is set, too.

newcode:
cmp [flag], 1
je skip
//Code you want to execute once
mov [flag], 1
skip:
//Originalcode


Then have the timer the LUA code set up check the flag as well and write over the jump, clean out the code cave, reset the flag, and turn off the timer that is checking on the flag.

If it has to be in AA, it would be interesting to try:
push eax
mov eax, 0xC5F82F7334 //Note: bytes may be out of order
mov [INJECT], eax //Note: making this up
pop eax
return

I have never attempted this so I would be surprised if it did work. It won't clean up the code cave and you may not have to register any symbols since you only are using them once and you may be able to overwrite all the preceding instructions of the code cave with the last write. So there would be evidence left, but not evidence of what was done or where it was done, just some stray lines of code seeming to reference itself. You will introduce a memory leak if this runs multiple times.

[ParkourPenguin]

You're looking for {$luacode}, but that's a terrible idea. The disable section deallocates newmem. What do you think happens if the {$luacode} block returns to memory that's been deallocated?

Everything between the `code:` label and `jmp newmem` shouldn't be there. I think I see what you're trying to do, and the AutoAssembler doesn't work that way.

I'd use a breakpoint.

[Autem]

The boolean trigger idea from Methos seems to be working in my early trials. Thanks for the idea.

I wonder tho if a breakpoint like Penguin suggests would actually be better as far as eliminating any slim chance of more than 1 execution sneaking thru at the same time? Or would the result be identical either way?

[++METHOS]

[ParkourPenguin]

Technically speaking, there would be a race condition between the `cmp` and the `mov`. If thread1 executes the `cmp`, gets preempted before the `mov`, and thread2 executes the `cmp`, then both thread1 and thread2 will run the code that was only suppose to be run once.

This is a common problem in concurrent systems. If you want to know more about concurrency, I'd recommend "C++ Concurrency in Action" by Anthony Williams. It's a little dated for C++, but there's still a lot to be learned from it.

The odds of this happening in practice are negligible. It's probably single-threaded, and even if it isn't, it might happen once in several billion times.

[wayden]

I use a way following your first idea, reseting bytes before disabling
and a bit of lua to auto disable, im not sure if it's a good way to do it but it seems to work
for that i got 2 scripts

-one main containing my reset function and allocating memory for itself and any other script that i would want to inject at the same InjectionPoint
and for the "variable" used to store the ResetBytes + regestering symbols

-And the other one containing the actual scipts i want to run only once



The main one looks something like this