On Wed, 24 Mar 1999, Ignatios Souvatzis wrote:

> On Wed, Mar 24, 1999 at 01:25:11PM +0100, Manuel Bouyer wrote:
> > On Mar 23, Perry E. Metzger wrote
> > > 3) Eliminate endianness of FFSes -- this is more controversial. Right
> > > now, we have bi-endian support, but it would be cool not to need it.
> > 
> > I'm not sure this one is a that good idea. It may have performances
> > issues. Did someone actually check this on a slow (m68k, vax, ...) machine ?
> > And which byte order will we use ?
> 
> in case little endiannes was used: m68k swapl is 3 instructions,
> 60ns on 68060/50, 20 cycles/1 us on 20MHz 68020.
> 
> Depending on how much of that you need, affordable.

Not. I don't want my cpus wasting cycles like this. However
"short" those cycles might be they could be put to better use
elsewhere, regardless of how fast/slow the processor is.

If an endian-neutral FFS goes in the tree, please make it compile
time selectable. Otherwise I'll be keeping my own native-endian
FFS source locally.

For the _few_ occasions where endian independence is useful, this
isn't worth the effort (IMHO) to make it not compile-time
selectable. It'll just give {Free,Open}BSD and L*nux another
performance arrow to shoot us with.

Cheers, Steve