@foone I've never gotten around to looking into how DOSBox actually does it, but from my last foray into SDL, I recall it had some weird ideas about graphics updates due to the path dependency of how it implemented multiple-buffering. On classic PC during ordinary operations, this might tie into the VGA's frame refresh interrupt; during human-stepping debugging, obviously, the target program just can't have access to all the frame refreshes, and something in DOSBox's emulation architecture might be doing something clever about it that just happens to break things.
Obviously, DOSBox's idea of how frame refresh works doesn't have to have anything to do with actual target-visible interrupts, but it might potentially give you a bit of information to let the software think it has a CGA, not a VGA. CGA, IIRC, didn't have support for proper frame interrupt, and a different synchronisation mechanism (status register polling?) might fail in a different, and perhaps illuminating, way.
=> More informations about this toot | View the thread | More toots from riley@toot.cat
=> View foone@digipres.club profile
text/geminiThis content has been proxied by September (3851b).