On Tue, Apr 16, 2013 at 04:17:15PM -0700, Mark Fasheh wrote: > On Wed, Apr 17, 2013 at 12:50:04AM +0200, Marek Otahal wrote: could you > > compare (appart from online/offline) your implementation to LiuBo's work?, > > appeared on ML a while ago: > > http://www.mail-archive.com/linux-btrfs@vger.kernel.org/msg23656.html > > Well that's the primary difference. Liu Bo's patch requires a format change > also since it's done online. My patch requires no format change. So they're > complimentary approaches in my opinion. > > There's also the possibility that some other file systems could pick up the > ioctl. Ocfs2 in particular should be able to. > > > > It would be interesting if the two approaches could share some code, and > > also confirmation that using one technique does not disregard using the > > other in future. > > Both features can exist together and probably should, I can see great uses > for both cases. > > I haven't looked at the patches but with respect to code sharing I'll take a > look. My patches don't actually add any custom code for the actual "let's > de-dupe this extent" as I re-use the code from btrfs_ioctl_clone().
In online dedup, I just make some changes in write path, as we regard dedup as a special kind of compression, doing dedup as compression way is the goal. The difference is where hash database is -- offline dedup puts it in userspace while online dedup in kernel. Although there might be no code that can be shared here, I agree that both online and offline one are useful. Good job. thanks, liubo -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
