talos-op-linux/fs/autofs, branch master

Merge branch 'next.autofs' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs

2019-12-06T01:11:48+00:00

Pull autofs updates from Al Viro: "autofs misuses checks for ->d_subdirs emptiness; the cursors are in the same lists, resulting in false negatives. It's not needed anyway, since autofs maintains counter in struct autofs_info, containing 0 for removed ones, 1 for live symlinks and 1 + number of children for live directories, which is precisely what we need for those checks. This series switches to use of that counter and untangles the crap around its uses (it needs not be atomic and there's a bunch of completely pointless "defensive" checks). This fell out of dcache_readdir work; the main point is to get rid of ->d_subdirs abuses in there. I've more followup cleanups, but I hadn't run those by Ian yet, so they can go next cycle" * 'next.autofs' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: autofs: don't bother with atomics for ino->count autofs_dir_rmdir(): check ino->count for deciding whether it's empty... autofs: get rid of pointless checks around ->count handling autofs_clear_leaf_automount_flags(): use ino->count instead of ->d_subdirs

autofs: fix a leak in autofs_expire_indirect()

2019-10-25T04:03:11+00:00

if the second call of should_expire() in there ends up grabbing and returning a new reference to dentry, we need to drop it before continuing. Signed-off-by: Al Viro

autofs: don't bother with atomics for ino->count

2019-09-18T03:31:27+00:00

All writers are serialized on inode->i_rwsem. So are the readers outside of expire.c. And the readers in expire.c are in the code that really doesn't care about narrow races - it's looking for expiry candidates and its callers have to cope with the possibility of a good candidate becoming busy right under them. No point bothering with atomic operations - just use int and mark the non-serialized readers with READ_ONCE(). Signed-off-by: Al Viro

autofs_dir_rmdir(): check ino->count for deciding whether it's empty...

2019-09-18T03:28:08+00:00

Signed-off-by: Al Viro

autofs: get rid of pointless checks around ->count handling

2019-09-18T03:23:08+00:00

* IS_ROOT can't be true for unlink or rmdir victim * any positive autofs dentry has non-NULL autofs_dentry_ino() * autofs symlink can't have ->count other than 1 * autofs empty directory can't have ->count other than 1 Signed-off-by: Al Viro

autofs_clear_leaf_automount_flags(): use ino->count instead of ->d_subdirs

2019-09-18T02:16:58+00:00

We want to find out if the parent will become empty after we remove the victim of rmdir(). Checking if the victim is the only element of parent's ->d_subdirs is completely wrong - e.g. opening the parent will end up with a cursor added to its ->d_parent and fooling the check. We do maintain ino->count - 0 for anything removed, 1 + number of children for anything live. Which gives us precisely what we need for that check... Signed-off-by: Al Viro

autofs_lookup(): hold ->d_lock over playing with ->d_flags

2019-07-27T14:03:14+00:00

... as well as setting ->d_fsdata, etc. Make all of that atomic. Signed-off-by: Al Viro

get rid of autofs_info->active_count

2019-07-27T14:00:33+00:00

autofs_add_active() is always called only once (and on a dentry with freshly allocated ino, at that). autofs_del_active() is never called more than once. Signed-off-by: Al Viro

autofs: simplify get_next_positive_...(), get rid of trylocks

2019-07-25T03:02:21+00:00

* new helper: positive_after(parent, child); parent->d_lock is held by caller, grabs and returns the first thing after child in the list of children that has simple_positive() true. NULL if nothing's found; NULL child == search the entire list. * get_next_positive_subdir() loses the redundant check for d_count and switches to use of that helper. BTW, dput(NULL) is a no-op for a good reason... * get_next_positive_dentry() switched to the same helper. Logics: look for positive child in prev; if not found, look for the positive child of prev's parent following prev, etc. That way we are guaranteed that we are only moving rootwards through the ancestors of prev, which is pinned and thus not going anywhere. Since ->d_parent on autofs never changes, the same goes for the entire chain of ancestors and we don't need overlapping ->d_lock on them. Which avoids the trylock loops, in addition to simplifying the logics in there... Signed-off-by: Al Viro

treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 83

2019-05-24T15:37:52+00:00

Based on 1 normalized pattern(s): this file is part of the linux kernel and is made available under the terms of the gnu general public license version 2 or at your option any later version incorporated herein by reference extracted by the scancode license scanner the SPDX license identifier GPL-2.0-or-later has been chosen to replace the boilerplate/reference in 18 file(s). Signed-off-by: Thomas Gleixner Reviewed-by: Richard Fontana Reviewed-by: Allison Randal Reviewed-by: Armijn Hemel Reviewed-by: Kate Stewart Cc: linux-spdx@vger.kernel.org Link: https://lkml.kernel.org/r/20190520075211.321157221@linutronix.de Signed-off-by: Greg Kroah-Hartman