Port-i386 archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: 80386 support



On Wed, Aug 22, 2012 at 11:21:49AM -0400, Mouse wrote:
> >> I then pointed to mail-index's archived copy of the "we're going to
> >> a three-tier paradigm" announcement.
> > I'll admit to some disappointment when I first read that announcement
> > but I have come round to the idea that it's a recognition of finite
> > developer resources.  It doesn't prevent people from working on
> > unusual ports but illustrates where current development focus is
> > likely to be.
> 
> I expect it _will_ prevent people from working on non-privileged-tier
> ports, because anyone trying to will be - probably repeatedly - screwed
> over by people making changes to the main tree that don't work, or are
> unacceptably expensive, for such ports.  I know it sure demotivates

That's not even remotely why nor how i386 CPU support was removed.  It
was removed because the i386 MMU lacks features that are important to
efficient operation of the VM system on all other CPUs, and the i386
instruction set lacks the primitive we use for atomic operations, so that
different synchronization primitives would be required for not just the
kernel but userspace too.

Conditionalizing all that stuff seriously screwed over the vast
proportion of users who have more modern hardware -- actual i386
systems were never more than a tiny, tiny number of installations
of NetBSD.  Why should everyone else suffer because of your odd
preferences in hardware?

As Andy pointed out at the time, because the incompatibilities reach
both into libc and the kernel, what would really be required for
efficient support of both i386 and newer x86 processors is a new
port of NetBSD -- not just a kernel, but a different userland ABI
too.  All the code that would be needed is still there in the CVS
history, but nobody has actually cared enough to do the work.

Thor


Home | Main Index | Thread Index | Old Index