Hello,
I wonder if manipulating the attribute info.branch.taken allows internally to trigger solver in order
to pursue exploration of both branches.
To clarify:
test esi, esi
mov eax, 336F9B7Fh
mov ecx, 0CA6A7046h
cmovz eax, ecx
mov ecx, 0E9213AF1h
There is an extract of the script:
def cb_path(m:MaatEngine):
if take_snapshot:
m.take_snapshot()
return ACTION.HALT
return ACTION.CONTINUE
def exec(start:int):
global m
stop_r = m.run_from(start)
take_snapshot = False
m.restore_snapshot(remove=False)
m.info.branch.taken = True
m.run()
print(f"first solution: {m.cpu.eax}")
m.restore_snapshot(remove=False)
m.info.branch.taken = False
m.run()
print(f"first solution: {m.cpu.eax}")
The output
python exec_bb.py -b bin -s 8405
[Info] Run 0x8405: TEST ESI,ESI
[Info] Run 0x8407: MOV EAX,0x336f9b7f
[Info] Run 0x840c: MOV ECX,0xca6a7046
[Info] Run 0x8411: CMOVZ EAX,ECX
[Error] Purely symbolic branch condition
[Error] Unexpected error when processing IR instruction, aborting...
[Info] Run 0x8411: CMOVZ EAX,ECX
[Error] Purely symbolic branch condition
[Error] Unexpected error when processing IR instruction, aborting...
first solution: 0x336f9b7f
[Info] Run 0x8411: CMOVZ EAX,ECX
[Error] Purely symbolic branch condition
[Error] Unexpected error when processing IR instruction, aborting...
second solution: 0x336f9b7f
What is weird is that in the first exec, it manages to inject model to get the first value of eax.
But during the second execution it seems to use equivalent model and fails to found the second possible value (0xca6a7046).
I wonder if it is a bug, or I'm miss something on how to manipulate such info attribute.
reproduce.zip
Cheers,
Anthony
Hello,
I wonder if manipulating the attribute
info.branch.takenallows internally to trigger solver in orderto pursue exploration of both branches.
To clarify:
There is an extract of the script:
The output
What is weird is that in the first exec, it manages to inject model to get the first value of eax.
But during the second execution it seems to use equivalent model and fails to found the second possible value (0xca6a7046).
I wonder if it is a bug, or I'm miss something on how to manipulate such info attribute.
reproduce.zip
Cheers,
Anthony