Re: VFS/ext2fs - large files on the Alpha (fails for 17GB+)
David Wragg (dpw@doc.ic.ac.uk)
20 Aug 1998 21:25:59 +0000
Olivier Galibert <galibert@pobox.com> writes:
> On Wed, Aug 19, 1998 at 12:47:24PM +0000, H. Peter Anvin wrote:
> > By author: Daniel Pittman <danielp@osa.de>
> > > Well, not totally. There is the posiblity that the Intel engineers
> > > offered us for addressing more than one single 32 bit space:
> > > segmentation.
> >
> > Nope, segmentation gets mapped onto a single 4 GB linear space.
>
> 64Gb (36 bits) since the P6. But then segmented addressing sucks
> blackholes through strawpipes, and kills performance badly (increased
> register pressure et al.).
Afraid not. If that was the case, the 36-bit stuff wouldn't be nearly
so bad.
It goes:
16:32-bit segmented addresses
translated to (segmentation)
32-bit linear addresses
translated to (paging)
36-bit physical addresses
I.e. everything has to go through the 32-bit linear address space
bottleneck. Segmentation is tacked on at the front, 36-bitness is
tacked on at the back, and they never meet.
--
Dave Wragg
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.altern.org/andrebalsa/doc/lkml-faq.html