> Well - for the first version, there's nothing from preventing a program
> from issuing the int10 without making the ioctl().
At some point on an 8086 we have to trust programmers not to break the
rules. Without some sort of device driver lock for video, even if it's
an unenforceable lock, there are no rules. We could always revector
int 10 elsewhere, lets say to abort() and pick some other interrupt for
the video BIOS in elks, or use a syscall. It would at least encourage
following the rules.
Nothing prevents programmers from rewriting the entire interrupt vector
table. We just expect them not to.
Eric
