Re: linux-next: manual merge of the vfs tree with the overlayfs tree

From: Miklos Szeredi
Date: Mon May 02 2016 - 04:30:25 EST

On Mon, May 2, 2016 at 3:08 AM, Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> On Mon, May 02, 2016 at 10:59:43AM +1000, Stephen Rothwell wrote:
>> Hi Al,
>> Today's linux-next merge of the vfs tree got a conflict in:
>> fs/overlayfs/super.c
>> between commit:
>> d478d6a8b8b7 ("ovl: ignore permissions on underlying lookup")
>> from the overlayfs tree and commit:
>> 5cf3e7fecb43 ("ovl_lookup_real(): use lookup_one_len_unlocked()")
>> from the vfs tree.
>> I fixed it up (I used the overlayfs version, since I don't know the
>> locking consequences of teh change from lookup_one_len() to lookup_hash())
>> and can carry the fix as necessary. This is now fixed as far as linux-next
>> is concerned, but any non trivial conflicts should be mentioned to your
>> upstream maintainer when your tree is submitted for merging. You may
>> also want to consider cooperating with the maintainer of the conflicting
>> tree to minimise any particularly complex conflicts.
> Should use lookup_one_len_unlocked(), actually. lookup_hash() is
> a microoptimization, losing a lot more on excessive i_mutex contention.
> Either variant works, though.

No, here it's not an optimization:

"More specifically using lookup_one_len() causes a problem when the lower
directory lacks search permission for a specific user while the upper
directory does have search permission. Since lookups are cached, this
causes inconsistency in behavior: success depends on who did the first