Re: [RFC PATCH 00/22] loop: Issue O_DIRECT aio with pages

[Christoph Hellwig]

I had a very brief look over the series, and in general I like it.

But there is one thing that needs a major revision, and that is the
filesyste, interface.

For one you make ->aio_read/write trivial wrappers around ->read_iter
and ->write_iter.  Instead of keeping this duplication around please
make sure to entirely kill ->aio_read/write and always use your new
methods.  Without that we'll get into a complete mess like the old
->aio_read/write vs ->readv/writev again.

A similar thing applies to the ->direct_IO/direct_IO_bvec interface -
instead of duplicating it I'd rather change the ->direct_IO interface to:

	ssize_t (*direct_IO)(int rw, struct kiocb *iocb,
			     struct iov_iter *iter, loff_t pos)

and let only fs/direct-io.c care about the difference when using user vs
kernel pages.

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/