Unionfs: unionfs_lookup locking consistency
authorErez_Zadok <ezk@cs.sunysb.edu>
Tue, 11 Sep 2007 00:15:37 +0000 (20:15 -0400)
committerErez Zadok <ezk@cs.sunysb.edu>
Fri, 6 Aug 2010 03:00:56 +0000 (23:00 -0400)
commit6cda0dc851bcddec44e2e01e3fe7a9b2e50bebc5
tree663a5756a03cf18f77518c521969a2e11ee0b17b
parent6c8205f19f66e2d92f50003c91ed5929867957c6
Unionfs: unionfs_lookup locking consistency

Ensure that our lookup locking is consistent and symmetric: if a lock
existed before calling lookup_backend, it should remain so; only if
performing a lookup of a known new dentry, should lookup_backend return a
newly-locked dentry-inode info (and only if there was no error).  Document
this behavior.  This cleanup allowed us to remove two unnecessary int
declarations.

Signed-off-by: Erez Zadok <ezk@cs.sunysb.edu>
fs/unionfs/inode.c
fs/unionfs/lookup.c