Re: [hatari-devel] defaulting to SMALL_MEM

[Thomas Huth <th.huth@xxxxxxxxx>]

Am Sun, 13 Feb 2022 12:46:06 +0200
schrieb Eero Tamminen <oak@xxxxxxxxxxxxxx>:

> Hi,
> 
> On 12.2.2022 22.15, Christian Zietz wrote:
> > Thomas Huth schrieb:  
> >> I've now committed a fix, please check whether it's working for
> >> you, too.  
> > 
> > I just tested commit 0b353ce1 and can confirm the crash with TOS
> > 4.04 is fixed.
>  >
> > You also found and successfully fixed the comparable issue (i.e.,
> > Shifter reading past the end of ST RAM) in ST mode, which I noticed
> > yesterday but didn't report yet.  
> 
> I'm not so sure the fix for video.c is correct.
> 
> It returns from the Video_CopyScreenLine*() functions _without_
> internal variables being updated. E.g. "pVideoRaster" is used to
> calculate some emulation values and just skipping its update when
> wrong ST-RAM area would be read seems wrong.
> 
> Especially considering that non-existing ST-RAM areas can be *read* 
> without exceptions, at least in some cases...

Question is: What do you expect to see on the screen in those cases? I
haven't tried on a real ST, but it's certainly only garbage or nothing
at all. So if you know of a program that depends on that behavior, we
can certainly try to emulate it, otherwise, let's just be happy that
Hatari does not crash anymore.

What was missing though: Video_CopyScreenLineMono() should still update
the pSTScreen pointer at the end, otherwise the host screen does not
get updates in the bottom part anymore. I fixed that now, too.

 Thomas