I want to escape from a JSR - RTS loop. Is PLA PLA right?

[Marius]

I'm writing a tool (almost ready) ... and the user can abort a certain process by pressing START (hehe, that makes sense... press START to quit hehehe, but ok)

 

But this 'press start' routine is called IN an existing JSR - RTS loop.

 

So the QUIT routine can not end with an RTS...

 

I created a trick for this

 

PLA
PLA 
RTS

 

Is that a bad solution? Or is this rather common? Any other ideas?

 

Thanks

Marius

[Bryan]

If you PLA,PLA you'll remove the first return address and then RTS to the previous address on the stack. If there isn't one, you'll crash.

 

I assume you can't just RTS normally but with some flag set to indicate that START was pressed?

 

Here's a little example (the three dots are where your surrounding opcodes would be):

 

; The calling routine...
 ... 
 ...
 jsr ROUTINE1
 bcc continue
 jmp STARTPRESSED
continue
 ...
 ...



;called routine
ROUTINE1
 ...
 ...
 lda COLSOL
 and #01
 bne no_start
 sec          ;flag START pressed
 rts          ;immediate exit
no_start
 ...
 ...
 ...          ;keep processing
 ...
 ...
 clc          ;flag START not pressed
 rts          ;normal exit 

Edited January 6, 2011 by Bryan

[Marius]

If you PLA,PLA you'll remove the first return address and then RTS to the previous address on the stack. If there isn't one, you'll crash.

 

I assume you can't just RTS normally but with some flag set to indicate that START was pressed?

 

That is an option indeed.

 

But... there will definately be a return adress on the stack, since this "exit" routine (with the pla pla) is always called from a jsr-rts loop.

 

When I get you right, you say: set a flag when the START button is presst. And JMP to the EXIT routine BEFORE the JSR-RTS loop is called again.

 

That is the safest indeed.

I'll think it over!

 

Thanks

Marius