On Tue, Nov 16, 2010 at 06:22:47PM -0600, Andreas Dilger wrote:
> On 2010-11-16, at 07:14, Jan Kara wrote:
> >> Yeah I went back and forth on this. KEEP_SIZE won't change the
> >> behavior of PUNCH_HOLE since PUNCH_HOLE implicitly means keep
> >> the size. I figured since its "mode" and not "flags" it would
> >> be ok to make either way accepted, but if you prefer PUNCH_HOLE
> >> means you have to have KEEP_SIZE set then I'm cool with that,
> >> just let me know one way or the other.
> > So we call it "mode" but speak about "flags"? Seems a bit
> > inconsistent. I'd maybe lean a bit at the "flags" side and just
> > make sure that only one of FALLOC_FL_KEEP_SIZE,
> > FALLOC_FL_PUNCH_HOLE is set (interpreting FALLOC_FL_KEEP_SIZE as
> > allocate blocks beyond i_size). But I'm not sure what others
> > think.
> IMHO, it makes more sense for consistency and "get what users
> expect" that these be treated as flags. Some users will want
> KEEP_SIZE, but in other cases it may make sense that a hole punch
> at the end of a file should shrink the file (i.e. the opposite of
> an append).
What's wrong with ftruncate() for this?
There's plenty of open questions about the interface if we allow
hole punching to change the file size. e.g. where do we set the EOF
(offset or offset+len)? What do we do with the rest of the blocks
that are now beyond EOF? We weren't asked to punch them out, so do
we leave them behind? What if we are leaving written blocks beyond
EOF - does any filesystem other than XFS support that (i.e. are we
introducing different behaviour on different filesystems)? And what
happens if the offset is beyond EOF? Do we extend the file, and if
so why wouldn't you just use ftruncate() instead?
IMO, allowing hole punching to change the file size makes it much
more complicated and hence less likely to simply do what the user
expects. It also is harder to implement and testing becomes much
more intricate. From that perspective, it does not seem desirable to