I agree, but there are several minor problems - the change is not
quite trivial - more something for 1.5.9.
For a single device one would like to have a single buffer cache;
on the other hand, as it is now filesystems think that they can
determine the desired blocksize themselves, and msdos and hpfs do
sb->s_blocksize = 512. So, it seems that the filesystem code has
to be adapted to work for arbitrary blocksize, or, that there has
to be a layer below the filesystem code to take care of differences
in blocksize.
As far as user convenience is concerned, yesterday night I changed
my own mount, so that "mount /tmp/fdimage /mnt -t msdos" now works
(by finding an unused loop device and mounting over that).
Andries