Re: [hatari-devel] breakpoint on CALL_SUBRETURN condition doesn't work ?

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On tiistai 08 huhtikuu 2014, Nicolas Pomarède wrote:
> Le 07/04/2014 22:48, Eero Tamminen a écrit :
> > Can you provide an exact case that doesn't work?
> > 
> > It's easiest to do with CPU disassembly enabled.
> 
> Problem is that when the breakpoint doesn't trigger, you get hundreds of
> cpu disasm, because the program never breaks at all, so it won't help.

You can output trace to a file so that you can check afterwards
whether there actually was a problem.  E.g. screen filling can
easily do a lot of instructions without any RTS in between.


> It seems to work randomly. Test with tos1.02 fr :
> 
> 1) start hatari and immediately set this breakpoint for example
> b pc=fc156a
> 
> once it breaks :
> n subreturn
> 
> CPU=$fc06cc, VBL=56, FrameCycles=168, HBL=0, LineCycles=168, DSP=N/A
> $fc06cc : 6b00 00cc                            bmi       $fc079a
> 
> -> subreturn stops at a "branch" Bcc ?

If you have changed debugger number base from decimal to
hexadecimal, I know what causes this:
Breakpoint value didn't have number base prefix, so if subreturn bit
is interpreted as hexadecimal instead of decimal, it would match also
branches.

I've commited fix for that.


	 - Eero