Re: [Squeak-VMdev] Versiojn 4 changes

[Tim R. <ti...@su...>]

In message <F8D...@in...>
          Ian Piumarta <ian...@in...> wrote:

> On 07 Apr 2004, at 01:06, Andreas Raab wrote:
> 
> > The only objection I'd have is whether this would have any effects on
> > machines that "like to give you the upper half of the address space" 
> > (e.g.,
> > all addresses must have that "tag bit" set).
> 
> I've never met one of these (outside of 32-bit address spaces).  This 
> would only be serious problem on machines that weren't consistent about 
> whether the top bit was set (consider: top N bits of virtual address 
> tell you what kind of segment the memory is allocated in).
> 
> Given that we've only (reaslistically) got 3 s/w architectures to worry 
> about, the experiment would be trivial to perform...  (I'll do so as 
> soon as I get the chance.)
RISC OS uses the upper half of memory quite happily. Mess up the VM for
my OS and I'll be popping around for a vigorous discussion...

Windows, Mac and *nix are not the only systems around and even if they
were the chances of none of them ever changing to use top-bit-set
memory addresses seems zero.


tim
--
Tim Rowledge, ti...@su..., http://sumeru.stanford.edu/tim
Strange OpCodes: SEXI: Sign EXtend Integer