On Wed, Jun 13, 2018 at 05:09:02AM -0700, Christoph Hellwig wrote: > On Wed, Jun 13, 2018 at 05:41:50PM +1000, Dave Chinner wrote: > > > So my proposal was to change the docs to make it clear that Eric > > > Sandeen's reading (that either way is fine) is the correct > > > interpretation. > > > > Ok, we're saying the same things - it wasn't clear to me that your > > proposal was to document both behaviours as valid... > > Even if both are valid we should come to a conclusion which behavior > make more sense and switch everyone to it. > > Right now we have two users of iomap_fiemap (gfs2 and xfs), and three > users of generic_block_fiemap (ext2, ext4 and hpfs), out of which two > already have iomap infrastructure and could be converted to the iomap > variant trivially, and three entirely open coded instances (f2fs, nilfs, > ocfs) which look like they could benefit a lot from using common code. > > It doesn't make sense to have different implementations and different > behavior for no good reason. Thanks, that's exactly my point, I'm working on the removal of the ->bmap inteface, and I'd be happy if we could reach some agreement in this case ;) Although I tend to agree that returning the beginning of the extent, makes more sense to me, even though the required offset passed into FIEMAP is after the beginning of the extent. Cheers -- Carlos
