On Wed, Jun 14, 2017 at 9:26 AM, Amir Goldstein <amir73il@xxxxxxxxx> wrote: > Miklos, > > I've made all the changes we discussed on v2 review and now all > upper/lower hardlinks are consistent. Note that I chose to invalidate > lower indexed dentries instead of updating all aliases on first copy up > (patch 21). That seemed simpler to me. Please let me know if you see any > culprit with that approach. If the alias is open, it won't receive the st_ino updates through fstat(2) since it will still have the ref to the old dentry/old inode. So it's more correct to update the aliases, but I guess it's also more complex. > I also re-introduced ovl_update_type() > (patch 8) for easier and more compact storing of the INDEX type flag. > > The remaining work on persistent nlink accounting and orphan index > cleanup is left for a differnt posting. Without that complementary work, > the index entries will remain forever. I am not aware of any other > culprit or correctness issues that may arrise from lack of index cleanup. > > This work introduces the inodes index opt-in feature, which provides: > - Hardlinks are not broken on copy up > - Infrastructure for overlayfs NFS export > > The work is available on my ovl-hardlinks branch [1]. > For the curious, documentation about how overlay NFS export would work > with the inodes index is available in the linked commit [2]. > > The most significant change w.r.t. vfs is that with inodes index, > overlay inodes are hashed and unique throughout the lifecycle of an > overlay object (i.e. across copy up). This is required for not breaking > hardlinks on copy up. > > Hardlink copy up tests including coverage for concurrent copy up of > lower hardlinks and lower/upper hardlinks consistency are available on > my xfstests dev branch [3]. > > This work also fixes constant st_ino for samefs case for lower hardlinks, > so the few constant st_ino tests that fail on v4.12-rc1 due to copy up of > lower hardlinks now pass [3][4]. Thanks, will have a look though. Miklos
