How can I load a constant value into the floating point register st0?

Question

I believe the correct way to do this is the following:

const1:     dq  1.2345
...
     fld    const1

However, I'm using Cheat Engine to reverse engineer a game (so I can understand it better). And it shows the following error:

Can someone please tell me what's wrong here? Ideally I would like the second command to be:

fstp dword [esi+ 3C]

But before I do this I need to load 93.5 value into the st(0) register. How can I achieve this?

Answer 1

There are a series of syntax related errors causing your problem.

"this instruction cannot be compiled" You're putting data in a code section. Define your variable outside of the newmem block and inside of a separate memory block. You're trying to define a memory block with "val:" but you haven't allocated that memory. You also can't reference it without registering a symbol.

To define a regular float you use a 4 byte variable not a 8 byte variable so you use "dd" not "dq", secondly you need to "cast" it to a float.

Your instruction you want to push a float onto the FPU stack is written like so: fld dword ptr [val]

Create the injection template as you have done previously and then insert this code at the top and continue what you were doing, it has everything I have noted in the answer and I tested it working:

alloc(newmem,2048)
label(returnhere)
label(originalcode)
label(exit)

alloc(val,8)

val:
dd (float)93.5

registersymbol(val)

newmem:
fld dword ptr [val]

Keep in mind you will need to use the rest of the generated template, this is just the code to fix the errors seen in your question.