Re: [hatari-devel] It seems I encounter a bug into hatari

[Nicolas Pomarède <npomarede@xxxxxxxxxxxx>]

Le 27/03/2014 18:48, Laurent Sallafranque a écrit :
> You're right again ;)
> I should have taken more time before replying.
>
> In Racer, I use byte values most of the time, that's why it works.
> And of course, it didn't work when I used long values.
>
>
> Here are the results of the 2 tests :
>
> 1st test :
>
> Under Falcon:
> move.l #$ffeeddcc,$ffff8240.w
> D1 ($ffff8240.w) = $0fee
> D2 ($ffff8242.w) = $0dcc
>
> Under hatari :
> move.l #$ffeeddcc,$ffff8240.w
> D1 ($ffff8240.w) = $ffee
> D2 ($ffff8242.w) = $ddcc
>
>
> 2nd test :
>
> Under Falcon:
> move.l #$ffffffff,$ffff8240.w
> D1 ($ffff8240.w) = $0fff
> D2 ($ffff8242.w) = $0fff
>
> Under hatari :
> move.l #$ffffffff,$ffff8240.w
> D1 ($ffff8240.w) = $ffff
> D2 ($ffff8242.w) = $ffff
>
>
> Don't hesitate to ask if you want me to make more tests or to test some
> other registers that would help improving hatari.

Yes, the function VIDEL_StColorRegsWrite() doesn't mask the value with 
0x0fff as this is done in video.c for STF/STE.

It could be added to match the behaviour of a real Falcon, but I don't 
think it's a priority, apart from your code I'm not sure many programs 
rely on this :)  You can try to adapt video.c part into videl.c if you like.

Nicolas