John Regehr wrote:
> Yeah, same-machine serial debugging is a big win for virtual machine
> software: I know a guy who did a sizable kernel-mode project in NetBSD
> with only one physical machine and VMWare.
Sure, I think we can make plex86 useful for OS/kernel development.
Serial debugging is one way. The guy who made the serial emulation
for bochs used it for this. But there is potential to do non-intrusive
debugging and instrumentation also. When you have complete control
of the VM, you can do this.
Debugging is a whole todo catagory for plex86. We could get fancy
and make a python script driven debug environment. I'll draw from
my experience with bochs, and make the debugger cosimulation
capable. It's best to design from the ground-up for this
capability.
I've even had it in mind to add the ability to optionally
check for cosimulation criteria in the branch handlers (every
branch could call these handlers if we want them to), so low instrumentation
debugging and cosimulation would be possible. Even while
running DT code!
Anyways, there's some things on the todo list before hammering
in a VM level debugger. But, just to get the juices flowing...
-Kevin
--
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Kevin Lawton [EMAIL PROTECTED]
MandrakeSoft, Inc. Plex86 developer
http://www.linux-mandrake.com/ http://www.plex86.org/
