Subject: Re: HEADS UP: migration to fully dynamic linked "base" system
From: David Laight <firstname.lastname@example.org>
Date: 08/27/2002 13:43:25
On Tue, Aug 27, 2002 at 01:19:43PM +0100, Richard Earnshaw wrote:
> > > Not really, it will give you a 24-bit offset relative to the PC with
> > > "garbage" bits in the upper 8 bits that will be ignored. [Or useful to
> > > Microsoft to keep a byte value in pointers stored in memory and breaking
> > > the program when upgrading to a a CPU with real 32 bit addressing :-)]
> > > The 68000/68010 only has a 24 bit address bus.
> > So the actual problem is that the shared libc is rather too large
> > (since it contains acres of stuff the average program dosn't need).
> That's hardly a problem, since the 'acres of unneeded stuff' will not get
> paged in.
Except that you are likely to get small functions sat in an entire
page. Also I got the impression that there was an address space
limit, not a physical memory limit. The mmu on a 68000/010/020
system isn't embedded in the cpu.
(I can't remember whether the 030 or 040 has one.)
David Laight: email@example.com