Re: [stella] RIOT RAM test

Subject: Re: [stella] RIOT RAM test
From: Manuel Polik <cybergoth@xxxxxxxx>
Date: Sat, 30 Nov 2002 15:28:26 +0100
Hi Nicolás!

> If you start to execute code on a random location, I 
> guess it's more likely that the game doesn't work at 
> all, than to produce a consitent slightly different 
> game. 

Well, what if only the LSB of the PC gets corrupted? 

Assuming we safely land on the starting page, which 
should most likely contain all executable code.

In that case chances of hitting a valid instruction 
within that page should be slightly better than 1 out of 
3 tries.

Now assuming that it's probably only single PC bits that 
get corrupted, so you might get "similar scenarios" 
quite more regularly than 1:255 ... :-)

Another thought: How about dumping a fried game?

> Maybe eventually a BRK will be executed, but that
> will execute the RAM initialization code that every 
> game has, so it should run the "normal" game.

Hey! Wow! Is that true? I ran into that on accident more 
then once. I even thought about just using 


instead of some 

	JMP Reset

code, saving two or more bytes. Only: I got the 
impression that the RAM was _not_ cleared and thought 
it'd *just* bankswitch back to the intro bank.

I wasn't 100% sure wether I could safely exploit that as 
described for a "Game Over" _without_ resetting the 
current RAM content (Polynomial/Hi-Score/etc...)

Do you think doing BRK would be fine?

> I stick to the RAM theory ;-D

Time will tell. :)


Archives (includes files) at
Unsub & more at

Current Thread