Resource Allocation for Option Rom

[Eric W. Biederman]

Stefan Reinauer <stepan at openbios.org> writes:

> * Eric W. Biederman <ebiederman at lnxi.com> [040529 05:29]:
> > Stefan Reinauer <stepan at openbios.org> writes:
> > > indeed true. But for x86 hardware, putting all of the x86 emulator in
> > > rom is less elegant than vm86 stuff, even though I have no clue about
> > > the later.
> > 
> > Last I heard some Option Roms switch to 32bit mode.  You can't do
> > that in vm86 mode.  In general using vm86 is a way to put your fate
> > in someone else's hands.
>  
> Uh, this was the kind of thing I was scared would be the case. what
> happens in such a case? Is there any way to trap this and at least
> safely state that the device can't be supported?

Yes.  What happens with a vm86 based implementation is you need an
emulator for all of the slow path cases.  At which point a pure
emulator is likely simpler.  Plus vm86 is currently deprecated.
 
> > We could possibly call the option roms directly but unless the space
> > cost is high an open source emulator is a good choice.
> 
> So the goal should rather be to reduce the size of the emulator than to
> introduce different code paths for different platforms it seems.

Stefan you should not need different code paths for different platforms.
And code size is secondary to correctness, but it is and will remain
an important consideration.

Eric