Re: [hatari-devel] Tests with the cycle exact fix 68030

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

ColorRunner has regressed.

It's stuck between "presents" and "color runner"
startup screens in this loop:

$000fe92c :             btst      #0,$fffc(a5)
$000fe932 :             beq.s     $fe92c

I.e. something to do with DSP.  It worked fine
in v2.0 (but wasn't playable as only joypad fire
works, not directions).


	- Eero

On 02/11/2017 04:52 PM, Nicolas Pomarède wrote:
> Le 09/02/2017 à 23:58, Eero Tamminen a écrit :
> > Hi,
> >
> > On 02/09/2017 12:54 AM, Laurent Sallafranque wrote:
> > > Hi Eero,
> > >
> > > I've tested a demo for the first time under hatari :
> > >
> > > http://www.pouet.net/prod.php?which=16805
> > >
> > > It stops after a while with double bus error.
> > >
> > >
> > > I always test with 14 Mo, no fast ram, 68030 Cycle exact + prefetch,
> > > 68882 FPU, DSP ON (SDL 1.2), RGB 50 Hz, no MMU.
> > >
> > > Maybe you'll be luckier than me ;)
> >
> > Nope, I also get double bus error.  Program tries to
> > do NULL pointer write at PC=$1cfe0.
>
> Hi
>
> at the moment the symposium 96 demo crashes, there're some cpu/dsp
> transfers, I suspect it's a case where cycles are not accurate enough yet.
>
> Nicolas