Re: [hatari-devel] VDI resolution limits

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On 05/11/2018 05:27 PM, Christian Zietz wrote:
> Eero Tamminen schrieb:
>
> > Any idea why these mouse related reads happen only with VDI mode,
>
> Do the reads really happen only with VDI mode?

That's the only mode where phystop & memtop have non-standard
values.

And as I've mentioned earlier, the crashes happen only with
older TOS versions (<= v2.x), and only with >= 4MB of RAM.


> After all, in standard
> resolutions on the ST, TOS keeps 768 bytes of distance from the end of
> RAM. It sets the screen base to RAM-0x8000, but the screen only needs
> 32000 (=0x7D00) bytes. So even if TOS reads past the screen buffer, it
> does not cause a bus error.

Good point, I forgot about this.

Hatari uses 0x8000 as minimum size for phystop & memtop distance.


At some point I've made memtop to round down to closest 512 dividable
value, otherwise there were some other crashes.  Unfortunately those
crashes weren't documented, so I wonder whether they also were caused
by this ancient mouse cursor issue... :-/

(There are crashes that happen when moving mouse to bottom of screen,
so crashes that additionally require moving mouse across 1-pixel wide
and 16-pixel hight area in screen, may have been noticed, or not.)


Btw. I vaguely remember some other programs using / relying on this
area, but I'm not sure whether they were something that one would
expect to work in GEM mode (e.g. harddisk driver?).


	- Eero