>
> Over the weekend, an idea occurred to me that I would be interested in
> developing. Unfortunately, I know squat about kernel programming --
> hence, I am looking for opinions.
>
> Basically, the idea would be to write a version of Linux that would run
> under Windows* as a regular process. Conceptually, this would be a lot
> like running under a micro-kernel, except that the kernel would be Win32,
> not Mach or whatever.
>
> The obvious first question: is this possible? I'm sure something could be
> done with ELKS or similar, but I don't want a dumbed down version -- I
> want something that I could run say WordPerfect for Linux on. What
> restrictions would something like this be subject to?
Not ELKS: it's designed for running under a segmented architecture.
uClinux uses a flat memory model, and should be much more straightforward
to use. Linux-on-Linux has been on my ever-overflowing to-do list for some
time. This would, trivially, work under any host OS.
Less trivial is writing drivers to let Linux use the OS native facilities.
> The obvious second question: why would I want this? Basically, I work in
> an organization that is very MS-centric and would like to have access to
> my preferred environment at work. Unfortunately, I have to run several
> Microsoft packages, cannot afford to reboot from linux, and can't get a
> second computer. On the other hand, if this concept could be made to work
> well, I could run Linux in a Window on my NT box and switch back and forth
> seemlessly.
That's another matter altogether. _Real_ desktop-style Linux needs more
capabilities then uClinux can provide. (Little things like an MMU.) Maybe
it would be possible to build something on top of mmap() that would
accomplish something useful, but I've not explored this.
> Another good use would be to demonstrate linux for newbies. This would be
> very nice -- most people, once they try Linux, don't go back. The problem
> is that just trying it can be a stiff hurdle for the less technically
> adept.
Demo boot disks already exist.
> Finally, what would be involved in doing this? Obviously, if I'm going to
> have to re-write, from scratch, the scheduler, the mm code, etc. then its
> hardly worth-while. On the other hand, if I just need to implement a fake
> frame buffer, use UMSDOS and give up on direct hardware, that's doable.
>
> Opinions? (No flames, please)
It's not a complete re-write, just a port.
-- Kenneth Albanowski (kjahds@kjahds.com, CIS: 70705,126)
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/