Quote:
Originally Posted by kaffer
Yes I am surprised by this, if the problem is CIABPRB trashing, I would expect UAE to see the bug too.
|
Maybe is only a slightly different timing sequence?
During intro unpacking system is fully alive and trackdisk.device free to do his job with CIABPRB.
When scheduler is freezed (at unpredictable times) a hacky 'motor off' sequence is used (I also use this, but only before a non-system loader
):
all PRB bits forced at 1, not preserving old values. Could be this?
When I was writing my loader I noticed that KS1.x is picky about head position and if you change something a read error appears.
You need to be consistent about position and buffered data.