[Ocfs2-devel] [PATCH 16/25] vfs: make remapping to source file eof more explicit

[Darrick J. Wong]

On Sun, Oct 14, 2018 at 10:24:33AM -0700, Christoph Hellwig wrote:
> On Fri, Oct 12, 2018 at 05:07:37PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong at oracle.com>
> > 
> > Create a RFR_TO_SRC_EOF flag to explicitly declare that the caller wants
> > the remap implementation to remap to the end of the source file, once
> > the files are locked.
> 
> The name looks like a cat threw up on your keyboard :)

Yeah... :(

> From reading the code this seems to ask for a whole file remap, right?

Nope.  In the original btrfs clonerange ioctl, length == 0 meant "to EOF".
If you made a call like this:

struct btrfs_ioctl_clone_range_args x = {
	.src_offset = 16384,
	.src_length = 0,
	.dest_offset = 0,
	.src_fd = whatever,
};
ftruncate(dest_fd, 0);
ioctl(dest_fd, BTRFS_IOC_CLONE, &x);

Then dest_fd ends up with the contents of [16k...EOF] from src_fd.
It's annoying to have the magic length number (and no flags?) but we're
stuck with this quirk of the interface.

> Why not put that in the name to make it more descriptive?

I'm all ears for better suggestions. :)

--D