[Prev][Next][Index][Thread]
Re: Is OSKit suitable for embedded real time?
> From: Jay Lepreau <lepreau@cs.utah.edu>
> Subject: Re: Is OSKit suitable for embedded real time?
>
> Re doc on HAL layer and porting:
> Someone else may elaborate, but I'll take a stab. There is no
> explicit porting documentation, but just look for 'x86' and 'arm32'
> directories and you'll see what needs to be done. The arm32
> (StrongARM) port was done with modest effort (a couple of weeks or a
> month?).
The Digital Shark port took about a month or so. The core architecture
stuff was easy. The I/O architecture is so much like a PC, that huge
portions of x86 machine dependent code and headers could be used almost
exactly as is. The hardest part was mucking with the OFW, and writing the
OFW/multiboot boot adaptor.
> Re MMU: it's not fundamentally required as far as I know, but others
> need to comment on which parts of the code would need to be tweaked to
> run without one.
An MMU is not required. You can run lots of x86 kernels without ever
turning it on. The Shark port does require that the MMU be turned on, but
thats because it made the port easier to do. So, an MMU is not required to
accomplish a port, although both ports provide hooks to make use of the
MMU. See the "pdir_" routines in thr oskit/kern/x86 directory (or arm32).
Lbs
---------------------------------------------------------------------------
Leigh B. Stoller Computer Science - Flux Research Group
stoller@cs.utah.edu University of Utah
http://www.cs.utah.edu/~stoller Salt Lake City, Utah 84112
Voice: (541) 929-2666 FAX: (801) 585-3743
---------------------------------------------------------------------------