Hi,

> we should be able to arrange it so that the overhead of multiplies and
> divides is only incurred when actually accessing such a device, which
> I'd say is ok.  I'll argue that filesystems should never support
> non-power-of-two devices, so it would only be user processes doing physio.

As I said in another post, there are filesystems out there that do support
block sizes not being a power-of-two.  Like the CDROM-XA extension to the
ISO 9660 filesystem, where you can have in one filesystem data blocks which
are 2k in size (actually this is a bit of a simplification, but we can
ignore this for now) and (in other files, but in the same filesystem)
multi-media blocks (like sound) which use the space used for error
correction on the data blocks to store more sound bits.  Supporting this
would be quite a bit hairy, as you don't have a consistent block size
throughout the media. The computation of device block numbers from file
and offset would have to be file-specific.

Ciao,
Wolfgang
-- 
ws@TooLs.DE     (Wolfgang Solfrank, TooLs GmbH) 	+49-228-985800