|
Atari coding BBS
Re: Built in obsolescence DSP |
Posted by: mikro
|
Jan,15.2010-11:11
|
>Does it mean that in both case, (JSR or JMP) it generates a LONG interupt, but only a JSR saves the SR state, increase IPL, save the PC, ... ?
Not sure about IPL but the rest -- yes.
>In the case of this demo, if there's no RTI, the stack can't be restored, and the IPL mask is too high and will mask the next $1b interrupt.
Beware, there's nothing to restore from stack by JMP! I remember one trick used in 6502 coding and that was pushing return address to stack and executing RTS instruction which served as jump command. Maybe they do something similar (i.e. set SP to some place with meaningful information)
>They've probably chosen to generate the stack overflow to let the program restart from beginning and execute again the ANDI instruction (to allow again the interrupts).
It's possible. I'm thinking about the way how to verify it on real falcon. Does Hatari provide some disassembler dump what one could modify and recompile?
>Another littler question : is there a way on the real falcon to get traces of a program that executes inside the DSP ?
Well, directly no but you can use always some interrupt for that, I think there's vector on CPU side which DSP can "activate" but the truth is for me it was always enough to use DSPDebug but I didn't do any reverse engineering of course.
|
[All messages in this thread] [Start new thread]
What's the anti-troll code? That's your personal code to be able to add comments and messages on the dhs.nu site.
Don't have a code or forgot it? Fix it here.
|