On Wed, Oct 25, 2000 at 02:22:28PM -0400, Kevin Lawton wrote:
> Folks,
> 
> Some enhancements and bug-fixes later, and now plex86 boots Linux
> normally...kick-ass!!!

Could you run 2.4 in it and run plex86 in that and figure out why it's
not working? At least the part about getting the pages (host_reserve_page
if my memory serves correctly) is a bit wrong at the moment.

> I will commit changes to CVS soon after I clean up my current
> code.

Please do :) quite cool. Have you tried more advanced stuff like X?
Probably Windows isn't far behind... and how about, more generally,
a test suite of a bunch of combinations of assembly instructions,
and also invalid codes and other erroneous stuff, that both plex86
and bochs, and maybe a normal CPU could run and test if the worst-
behaving code works. If that succeeds, then sane code like Linux
(and I might even say Windows also, considering that it is compiled
code) should run without any hitches <g>

> Plex86 can actually run guest code in 3 modes now:
> 
>   1) Normal mode.  This is SBE (scan-before-execute) controlled.
>      Most code is run natively, some instructions are virtualized
>      and thus emulated.
> 
>   2) Emulation mode.  This emulates code in a loop much like bochs,
>      only everything happens in the monitor space as in 1).  This
>      is good for testing the emulation code in plex86.  And for
>      cosimulation with bochs.
> 
>   3) Breakpoint mode.  This is like 1), only the TF is set so that
>      only one instruction will be run.  This is useful for testing 1)
>      but with very fine-grained control of execution, and is very useful
>      for cosimulation with bochs.
> 
> I believe all 3 modes can boot Linux.  The enclosed graphic is from 1).
> The other modes seem to work, but I bailed early on execution because
> I was impatient.  :^)

There would be any reason that they _wouldn't_ boot Linux? I would suppose
it doesn't really matter in the long run, because most people will be
running in mode 1 anyway (unless they are debugging operating systems).

-- 
Kenneth Arnold <[EMAIL PROTECTED]> / kcarnold / Linux user #180115
http://arnoldnet.net/~kcarnold/

Reply via email to