[TI-OS] A bug?

driesguldolf · Post by **driesguldolf** » Mon 20 Aug, 2007 12:21 pm

Yay found an insignificant bug in tios.

When you're trying to make some art on the graph screen weird stuff can happen: sometimes when you switch from text to pt-change (not as commands but when you use them when the graph screen is on, could be the other way around (not sure)) it draws a pixel at the point where the cursor is...
It does not happen on all the areas of the screen...

Strange... Has this happened to anyone else?

Bug found in ti83+ latest os

King Harold · Post by **King Harold** » Mon 20 Aug, 2007 4:14 pm

It hasn't happened to me, I'll mess around some more and see if I can make it happen as well

Oh and upon seeing that quiz: depending on whether SP was odd or even to start with, you will end up CALLing $9D9D and crash (JP $CA9D) or you will end up executing SBC A,L and then JP $CA9D where you can not execute so crash. To top it all, before the crash you will have killed all data between the old SP and $9DCA.
At least, that is what I think would happen, and it only works this way when you can not execute from $C000 to $FFFF as is normally the case.
edit: fixed typos

driesguldolf · Post by **driesguldolf** » Tue 21 Aug, 2007 10:08 am

It was a while ago that I found that bug (the period that I made graphics during class to use them in never finshed projects, ya know...)

King Harold wrote:Oh and upon seeing that quiz: depending on whether SP was odd or even to start with, you will end up CALLing $9D9D and crash (can not execute there) or you will end up executing SBC A,L JP $CA9D where you can not execute so crash. To top it all, before the crash you will have killed all data between the old SP and $9DCA.
At least, that is what I think would happen, and it only works this way when you can not execute from $8000 to $FFFF as is normally the case.

I wanted to have some sort of instruction that could survive a stack overflow but as I see now, It is not possible... I'll think of a better one.

And by the way, it is perfectly safe to execute from $0000 to $C000, how else could you execute an asm prog?

King Harold · Post by **King Harold** » Tue 21 Aug, 2007 10:50 am

Oh yes wasn't really awake..
It will still crash: depending on whether SP was odd or even to start with, you will end up CALLing $9D9D and crash (JP $CA9D) or you will end up executing SBC A,L and then JP $CA9D where you can not execute so crash. To top it all, before the crash you will have killed all data between the old SP and $9DCA.

It seems to me that it should be possible to have an instruction that survives it's own stack overflow..

KermMartian · Post by **KermMartian** » Tue 21 Aug, 2007 1:14 pm

But if you execute at or above $c000, you get an instant crash anyway...

driesguldolf · Post by **driesguldolf** » Tue 21 Aug, 2007 4:48 pm

Ok I think we got a bit off topic,

hmmm I realized... I don't think this really belongs in programming help does it? Sorry, my mistake, I think it fits better in general ti discussion. Could someone move it please?

Damn got off topic...

Goplat · Post by **Goplat** » Tue 21 Aug, 2007 8:34 pm

9D95: di
9D96: ld bc,$E9C5 ; push bc \ jp (hl)
9D99: ld hl,$9D97
9D9C: jp (hl)
This will overwrite itself, but the 2-byte loop stays the same and keeps going.

King Harold · Post by **King Harold** » Tue 21 Aug, 2007 9:17 pm

That is insane

way to go man, how on earth did you think of it?

Post by **Dwedit** » Wed 22 Aug, 2007 12:56 am

not that hard, just make sure it overwrites the two bytes with the same two bytes.