On Wed, 16 Jul 2025, Amir Goldstein wrote: > On Wed, Jul 16, 2025 at 2:47 AM NeilBrown <neil@xxxxxxxxxx> wrote: > > > > More excellent review feedback - more patches :-) > > > > I've chosen to use ovl_parent_lock() here as a temporary and leave the > > debate over naming for the VFS version of the function until all the new > > names are introduced later. > > Perfect. > > Please push v3 patches to branch pdirops, or to a clean branch > based on vfs-6.17.file, so I can test them. There is a branch "ovl" which is based on vfs.all as I depend on a couple of other vfs changes. Thanks, NeilBrown > > Thanks, > Amir. > > > > > > > > Original description: > > > > This series of patches for overlayfs is primarily focussed on preparing > > for some proposed changes to directory locking. In the new scheme we > > will lock individual dentries in a directory rather than the whole > > directory. > > > > ovl currently will sometimes lock a directory on the upper filesystem > > and do a few different things while holding the lock. This is > > incompatible with the new scheme. > > > > This series narrows the region of code protected by the directory lock, > > taking it multiple times when necessary. This theoretically open up the > > possibilty of other changes happening on the upper filesytem between the > > unlock and the lock. To some extent the patches guard against that by > > checking the dentries still have the expect parent after retaking the > > lock. In general, I think ovl would have trouble if upperfs were being > > changed independantly, and I don't think the changes here increase the > > problem in any important way. > > > > After this series (with any needed changes) lands I will resubmit my > > change to vfs_rmdir() behaviour to have it drop the lock on error. ovl > > will be much better positioned to handle that change. It will come with > > the new "lookup_and_lock" API that I am proposing. > > > > Thanks, > > NeilBrown > > > > [PATCH v3 01/21] ovl: simplify an error path in ovl_copy_up_workdir() > > [PATCH v3 02/21] ovl: change ovl_create_index() to take dir locks > > [PATCH v3 03/21] ovl: Call ovl_create_temp() without lock held. > > [PATCH v3 04/21] ovl: narrow the locked region in > > [PATCH v3 05/21] ovl: narrow locking in ovl_create_upper() > > [PATCH v3 06/21] ovl: narrow locking in ovl_clear_empty() > > [PATCH v3 07/21] ovl: narrow locking in ovl_create_over_whiteout() > > [PATCH v3 08/21] ovl: simplify gotos in ovl_rename() > > [PATCH v3 09/21] ovl: narrow locking in ovl_rename() > > [PATCH v3 10/21] ovl: narrow locking in ovl_cleanup_whiteouts() > > [PATCH v3 11/21] ovl: narrow locking in ovl_cleanup_index() > > [PATCH v3 12/21] ovl: narrow locking in ovl_workdir_create() > > [PATCH v3 13/21] ovl: narrow locking in ovl_indexdir_cleanup() > > [PATCH v3 14/21] ovl: narrow locking in ovl_workdir_cleanup_recurse() > > [PATCH v3 15/21] ovl: change ovl_workdir_cleanup() to take dir lock > > [PATCH v3 16/21] ovl: narrow locking on ovl_remove_and_whiteout() > > [PATCH v3 17/21] ovl: change ovl_cleanup_and_whiteout() to take > > [PATCH v3 18/21] ovl: narrow locking in ovl_whiteout() > > [PATCH v3 19/21] ovl: narrow locking in ovl_check_rename_whiteout() > > [PATCH v3 20/21] ovl: change ovl_create_real() to receive dentry > > [PATCH v3 21/21] ovl: rename ovl_cleanup_unlocked() to ovl_cleanup() >
