David Murn writes:
>
> On Fri, 30 Apr 1999, Alistair Riddoch wrote:
>
> > We could really do with a tool that looks at the vector interrupt table
> > so we can try and sort out how to boot ELKS on these strange palmtop
> > machines. It may be possible to load ELKS from DOS without replacing the
> > VII on some machines.
>
> I can't see why the VII even comes into play here. We simply load the
> kernel in exactly the same way the ELKS loader loads it from disk in
> bootsect.S, then jump to it. Why should it matter if there's info in the
> VII at 0x21 or whatever that DOS has setup? Nothing will use that, and if
> we're worried, we can just clear the vector table in the ELKS code.
>
The problem is that the only disk driver we have working yet is the one
that uses BIOS routines to access the disk. After DOS is loaded the int 13h
entry in the VII which initially points to this BIOS routine is replaced.
BootELKS gets round this by having an image of the VII before DOS was
loaded, and loading this into the VII before the kernel is started, giving
the kernel access to the BIOS routines.
Al
