[hatari-devel] Moai demo as regressed a lot since hatari 1.4

Laurent Sallafranque laurent.sallafranque at free.fr
Fri Jul 15 00:30:21 CEST 2011


Hi,

I've found, fixed and committed the path to allow the STOP instruction 
to work.

I just had forgotten to add the cycles computed by do_cycles to 
M68000_AddCycles.

STOP instruction works well now (I'll do more tests with the programs 
that were freezing before).

Regards

Laurent



Le 15/07/2011 00:18, Laurent Sallafranque a écrit :
> Hi Nicolas,
>
> You're right. there's a problem with the stop instruction.
> So let's see this after the release.
>
> For the DMA, I get exactly the value 50066 that I wrote to crossbar.c 
> (ste mode compatible).
>
> I'd like to know the value a real Falcon would give in exactly the 
> same conditions :
> Your program compiled with genst3 in 68030 instructions mode.
> Launch directly from the desktop in ST med res compatibility mode.
>
> Maybe Anders, you could do it for me ?
>
> Regards
>
> Laurent
>
>
>
> Le 15/07/2011 00:10, Nicolas Pomarède a écrit :
>> Le 15/07/2011 00:03, Laurent Sallafranque a écrit :
>>> Hi,
>>>
>>> I've tried the program. I've just compiled it under genst (68030 mode).
>>>
>>> When I test it with hatari old cpu, st med resolution, I get $c392
>>>
>>> More interresting : when I launch it with newcpu, st med resolution,
>>> hatari freezes (the same way I've already seen with some demos).
>>> I have to kill the terminal.
>>>
>>> it seems to loop forever in do_specialties.
>>>
>>> Maybe this first bug should be removed before the release, as it 
>>> freezes
>>> completely hatari.
>>> (except if you consider the new CPU as highly experimental, which is
>>> what it is for now).
>>>
>>> Regards
>>>
>>> Laurent
>>
>>
>> The only "special" instruction here is "stop 2300", it could be what 
>> causes the problem in do_specialties. Maybe it's not frequently used 
>> in falcon demos.
>>
>> Try to comment the "stop" to see if the program runs without locking.
>>
>> But I would not consider this to block hatari 1.5's release. Fixing 
>> "stop" (if it's really the problem) involves a lot of work on 
>> interrupts/exceptions and would take too much time for now.
>>
>> Nicolas
>>
>
> _______________________________________________
> hatari-devel mailing list
> hatari-devel at lists.berlios.de
> https://lists.berlios.de/mailman/listinfo/hatari-devel
>




More information about the hatari-devel mailing list