On 13/11/2024 08:15, Paul Dufresne via Freedos-devel wrote:
Here is what I see (attached).
[ snip]
Now that looks like it's failing to set the RAMDAC palette once it goes
into a 256-colour graphics mode.
For the record, I get the usual Mode 13h palette in DOSBox-X, DOSBox
(which randomly decides to massively slow down near the end at max
cycles!), QEMU (std driver), and 86Box (S3 Trio64 Diamond Stealth DRAM 2MB).
The palette you got is 64 colours set (uses up 1/4 of the screen) and
looks like some weird mix of the palette that gets set for EGA
compatibility and the EGA palette remapping machinery.
There's probably an assumption that every card supports VBE 2.0 (which
is to use INT 10h AX=4F09h BL=00h) and so they might not have bothered
to support the VBE 1.2 way of setting the palette (which is to go
directly via the VGA RAMDAC ports 03C6h through 03C9h).
The EGA remapping makes this even *more* bizarre - why would it keep the
EGA mapping around? And furthermore, how would it even interact properly
with bits 4-5? Admittedly, the EGA 4->6 palette mapping is something
I've yet to fully get my head around and in my own 256-colour tests I
prefer to bypass it (by identity-mapping the whole thing).
I smell an unintentional regression in VirtualBox here.
_______________________________________________
Freedos-devel mailing list
Freedos-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freedos-devel
