On Sat, Oct 06, 2018 at 03:41:34AM -0700, Christoph Hellwig wrote: > On Thu, Oct 04, 2018 at 05:45:35PM -0700, Darrick J. Wong wrote: > > From: Darrick J. Wong <darrick.wong@xxxxxxxxxx> > > > > Change the clone_file_range and dedupe_file_range functions to return > > the number of bytes they operated on. This is the precursor to allowing > > fs implementations to return short clone/dedupe results to the user, > > which will enable us to obey resource limits in a graceful manner. > > For clone this doesn't make any sense, as we don't allow 'short clones'. That's correct. The overall goal is that copy_file_range will be able to call ->clone_file_range with a "I can handle a short clone" flag so that the fs (and the prep functions) know that they're allowed to shorten the request length, and we have a way to communicate the actual results back to userspace. Neither FS_IOC_CLONE nor FS_IOC_CLONERANGE will be able to take advantage of this, of course. They won't pass in the flag, and any implementation that wants to shorten the length will pass back -EINVAL instead. --D
