| HellGates |
You're not disqualified haha, you do what you want. Anyway actually, I'm quite happy that AI can't solve it (yet?), so it was interesting to see that it failed |
2026-03-06 11:34 |
| HellGates |
Understood, no problems then :), good luck ! |
2026-03-05 19:21 |
| HellGates |
Okay,
I must warn that I'm not against using AI,
but this time I need to express myself entirely on this because 50% of what you sent is just wrong,
so I sadly I can't answer properly something that just doesn't exist.
I'm sad to see that AI is used this way because clearly,
if you had looked at what it gave you on the binary, you would see that half of it wrong, or at least doesn't sound right.
Sorry but let me correct some things about things "you've done":
About 1) There's no LFSR decryption, There's no JIT, just recursively encrypted shellcodes inside an sig handler.
About 2) Again no JIT, "page handlers" doesn't make make much sense to me, maybe page faults ? The list of cases seems correct though.
About 3) There's no page id stuff ? Although yeah, there's faults incorporated into the binary to avoid static analysis. About the buffers there is way more than copying buffers if you can notice, I'll let you discover what it does. (maybe you haven't digged into that part too much and it's not a problem much)
About 4) There is indeed some sort of logic gate code, but 113 is way too low, look on the 4,8MB region. Also keyboard ports ? Wtf ? Your AI is probably giving you *cough* bullshit *cough* to compensate its lack of skills.
About 5) They are actually valid instructions, give a better look on the sig handler and what it does to the "invalid instructions".
About 6) THEREISNOCOW ? C'mon, if you have tried it yourself in the LCD, you would see that it wouldn't work.
About 7) I did not use VFPGAObfuscator for this, though data.bin is a good direction to look at first, try to see what does the code between this 1GB and the sig handler in a bigger view.
About 8) NOP patches magic write to prevent infinite loops ? No this is done for something else, probably your AI hallucinating again. Also 2 tick convergence? I'm sorry, wut.
Things "you" know for "certain":
1) No.
2) Yesn't.
3) No.
4) JITJITJIT. (the memfd is actually for the music part ... I'm just speechless right now, thanks AI)
5) Yesn't.
6) Mh wut ? The circuit doesn't know/care about the keyboard, not sure why it even brings it up here, but it's interesting.
7) Maybe accurate, I didn't count how many addresses are actually faulting anyway, but maybe AI can count now, thanks god.
8) ???
10) Yes
11) Yesn't
Where "you" are stuck:
I guess if you've read, I don't need to tell you what to look on.
1) There's a lot more logic gates, indeed.
2) If you've looked what I said, I'll let you discover what it does, this is mostly an anti dumping technique.
3) Same thing as above. It does nothing of that and it doesn't use VFPGAObfuscator.
4) Forget about keyboard bits. Yes there is more logic gates.
If you want to understand how the binary without dynamic analysis, good luck.
I won't help AI doing this, I hope you understand.
You said you respect me for the time writing it but using AI to recap but I honestly feel betrayed. |
2026-03-05 16:48 |
| HellGates |
Can you send details of what you think you understood ? So I can know which hint I can give |
2026-03-04 17:10 |
| HellGates |
That's okay, let me know if you want some spoilers (I tried to send you an email but gmail stops me because my ISP is garbage and doesn't want to set reverse DNS ipv4/ipv6 to my domain, I guess I'm forced to use a SMTP relay for now :/) |
2026-02-28 10:23 |
| HellGates |
Hm sorry, why do you talk about a key ?
There's indeed a bit of "key" related stuffs but I don't see what you mean.
I double checked and there's nothing related to this so I'm not sure how did you find such result.
Can you send me the details of what you found so I can understand ? |
2026-02-27 21:54 |
| HellGates |
The LCD initialization is slow on purpose, you have to get around it somehow first to make it faster. |
2026-01-02 14:41 |