Re: [hatari-devel] regression in debugger / disasm ?

[ Thread Index | Date Index | More Archives ]

Le 25/12/2013 12:07, Thomas Huth a écrit :

PS: Something is currently also wrong with the colors in the latest hg
tip. Each time I boot, the desktop appears with a slightly different
white as background (using TOS 2.06).

That's fixed, that was due to adding "randomness" in STF color regs for unused bits (change #4846)

The way ioMem functions work is that you need to modify value at $ff8240-$ff825e, you can't return a modified version without altering ioMem sapce (handlers in ioMemTab return void, maybe this could be improved, but as we can return byte/word/long, it would be more complicated with the same handler declaration (require more type casts)).

So the value was read later with non "0" bits when building the palette, which created the slight color change in hue.

I could see it too with some of my STF test programs that save the palette, which triggered the random bits.


Mail converted by MHonArc 2.6.19+