Daniel Veillard writes:
> I had a look in this direction too, but the current implementation
> of SPIM doesn't seems to emulate anything concerning the exception
> handling of the MIPS familly especially the TLB related exceptions.
No, it could be added though.. oh, it was just a thought.
> I got in touch with some people trying to add GCC support for
> 64 bit operations and got a version of GAS abble to assemble
> 64 bit specific stuff. One question concerning the Mips port concern
> the extended adressing capabilities of the Mips chips, should we
> use it or restain memory adressing to the same of the 86,
> I suppose too that maintaining compatibility with Little Endian
> mode would avoid some weird porting problems.
Some of this was discussed a while ago. I think the general consensus
was that initially should stay as close as possible to the x86 to
finish the first port as fast and as easy as possible. Later,
we can start looking into 64 bit, better MMU etc. I can't imagine any
hacker with 64 bit hardware who wouldn't want to run 64 bit binaries,
but better get something running first.
> I admit looking at commercial sources is dangerous and should be
> prohibited for working on this project, but looking at free sources
> looking for HW programming tips should be Ok.
Absolutely agreed, but be sure the source is free. Last time I looked,
some of the source in Mach was only free for use in Mach.
> Looking at Mach sources it seems that DecStations, and some AST
> machines. But I'm afraid no machine based on ARC design are supported.
> What about NetBSD ?
The (possible) differences between DecStation, AST and ARC are:
Endian, Memory configuration, CPU and devices. Did I leave anything
out? We need to write our own drivers for scsi, video and ethernet
anyway. Drew is fairly familliar with the scsi chips use (right?).
The video should be simple. I don't know about the ethernet.
Is someone out there into NetBSD?