On Tue, 27 Jun 2000, Roman Zippel wrote:
> /* Unfortunately this kludge is needed for FIBMAP. Don't use it */
> int (*bmap)(struct address_space *, long);
>
> I ask that in the context of the fs specific functions of
> read_page/write_page/prepare_write, that simply call a generic function
> adding a pointer to the get_block function. If the get_block
> functionality would be integrated into address_space_operations, lots of
> these indirect calls could be removed. It's also make more sense this
> way.
No, it would not. There are two kinds of get_block() (normal semantics and
no-holes one) and they just happen to share the same prototype. And there
is a bunch of cases when operation simply does not make any sense.