On Mon, Jan 5, 2015 at 9:14 PM, Michael Brutman <mbbrut...@brutman.com>
wrote:
Options 1, 2, and 3 do not exist and are not likely to exist for a few
> years even after somebody actively starts working on them.
>
Correct. I never said this was something which could be thrown together
overnight. I know the existing FreeDOS kernel took a ton of work, and I
expect this undertaking would be no different. But it is doable.
Options 1 and 2 can not promise "100% compatibility with both DOS
> applications and the full range of PC hardware" when they are not even well
> defined. Is it a single 32 bit kernel or is it multiple kernels running in
> VDMs? I've seen so many things thrown around here so loosely ...
>
It would be a single 32-bit protected mode
<http://en.wikipedia.org/wiki/Protected_mode> kernel which sets up individual
work spaces <http://en.wikipedia.org/wiki/Virtual_8086_mode> for each
application, which can all run independently (and unaware) of each other.
These work spaces would be set up by the processor itself to mimic the way
real mode (e.g. 16-bit mode) works so that the applications can run as they
always have. This is similar to running them in a virtual machine, except
there is no emulator - the chip already contains the necessary silicon (the
MMU) to perform the memory address relocation which makes this all
possible. Interrupts are serviced by 32-bit handlers to eliminate the
cycle-expensive switch from protected mode to real mode and back again. This
<http://prodebug.sourceforge.net/pmtut.html> should explain the details of
protected mode operation more concisely than I can do here. This kernel
would (optimally) be a drop-in replacement for the existing FreeDOS kernel
to enable 32-bit operation.
It is a little silly to keep talking about a 32 bit kernel on the roadmap
> when such an option does not exist.
>
I see your point, however if that attitude was taken during the initial
formative discussions of the existing FreeDOS kernel... it never would have
been made. ;-) At that point the 16-bit FreeDOS kernel didn't exist yet
either, but discussion started, people got together and with a lot of work,
it got made.
> To be considered for the roadmap it should exist in some form. Right now
> it is not even well defined what a 32 bit kernel would be. Not even a
> specification that we can debate.
>
This is a very good point, I'll have to draw one up sometime soon.
> Let's see some concrete results on a 32 bit kernel before talking about
> putting it on any roadmap.
>
Agreed. I've never said that we should actually put it on the map or make
this huge push to create such an animal at this point. However, it
short-circuits innovation to not discuss all options and methods available
in any situation. I fully expect FD 1.2 to be 16-bit, and most likely 2.0
will be as well, but this shouldn't make us just blanket-disregard the
future possibility. And maybe, in the end, some of these goals aren't
attainable; I don't claim to be the world's foremost expert on the x86
architecture. But what I do know - about programming, OS design and the
Intel processor line in general - indicates to me that these things should
all be doable.
------------------------------------------------------------------------------
Dive into the World of Parallel Programming! The Go Parallel Website,
sponsored by Intel and developed in partnership with Slashdot Media, is your
hub for all things parallel software development, from weekly thought
leadership blogs to news, videos, case studies, tutorials and more. Take a
look and join the conversation now. http://goparallel.sourceforge.net
_______________________________________________
Freedos-devel mailing list
Freedos-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freedos-devel
