On Tue, Apr 10, 2012 at 4:25 PM, David Miller <davem@xxxxxxxxxxxxx> wrote:
>
> Although that would be a neat trick, making the hashing function work
> in a way such that if we simply rewound the initial pointer to be
> aligned, put 0xff into the alignment bytes, and the hash would still
> compute properly.
I cannot imagine any worthwhile hash that works like that.
Any hash that magically works with aligned data implies basically
throwing away all the actual real bits of information in a file name,
notably all the "where are the letters" information.
So the hashing really *does* need unaligned loads (or the
shift-things-by-hand to emulate it).
Sparc has "faligndata", no? Power generally does unaligned loads
"reasonably ok", although I seem to recall some horrible 7-cycle
micro-trap on at least some micro-architecture (and a page-crossing
fault fixup in software, but that's not going to happen in practice:
the pathname is allocated from a page in getname())
Linus
--
To unsubscribe from this list: send the line "unsubscribe linux-arch" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
