[hatari-devel] Union demo abort with new EmuTOS?

[Eero Tamminen]

Hi,

On torstai 10 marraskuu 2011, Thomas Huth wrote:
> > Does the Union demo crap out with EmuTOS because one of the line-A
> > variables has a bogus value?
> 
> No, if you dump a little bit more of the history before the a000
> opcode, you'll see that the program is already executing garbage before
> this. It somehow jumps to a region in memory where no valid code has
> been stored (or the code there has been trashed somehow).
> 
> > Hatari can avoid the abort by checking that VDI_OldPC is valid (see
> > attached patch), but how I can make it to raise illegal instruction
> > exception if it isn't?
> 
> I'd rather suggest to chech m68k_getpc() being within the cartridge
> memory region when OpCode_VDI has been called.

Any particular reason why it should be done like that?   IMHO checking
VDI_OldPC is cleaner, as that must anyway be valid when this is called.


> When it is called from somewhere else, just run op_illg_1 instead of of
> VDI_Complete.

Thanks!

I guess running the fill_prefetch_0() after that is fine, nothing's
expecting things to be 100% correctly prefilled after illegal opcodes?
:-)


	- Eero