[PATCH v4 08/10] locks: Set FL_CLOSE when removing flock locks on close()

From: Benjamin Coddington <bcodding@redhat.com>
To: Trond Myklebust <trond.myklebust@primarydata.com>,
	Anna Schumaker <anna.schumaker@netapp.com>,
	Jeff Layton <jlayton@poochiereds.net>,
	"J. Bruce Fields" <bfields@fieldses.org>,
	Christoph Hellwig <hch@infradead.org>
Cc: linux-nfs@vger.kernel.org
Subject: [PATCH v4 08/10] locks: Set FL_CLOSE when removing flock locks on close()
Date: Wed, 30 Dec 2015 08:14:04 -0500	[thread overview]
Message-ID: <2c9e21196b0949087c892ffffe49d26174a54fdd.1451480826.git.bcodding@redhat.com> (raw)
In-Reply-To: <cover.1451480826.git.bcodding@redhat.com>
In-Reply-To: <cover.1451480826.git.bcodding@redhat.com>

Use FL_CLOSE in fl_flags as in locks_remove_posix() when clearing locks.
NFS will depend on this flag to properly defer an unlock until IO under the
current lock has completed.

Signed-off-by: Benjamin Coddington <bcodding@redhat.com>
---
 fs/locks.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/fs/locks.c b/fs/locks.c
index 0d2b326..9aea07a 100644
--- a/fs/locks.c
+++ b/fs/locks.c
@@ -2423,7 +2423,7 @@ locks_remove_flock(struct file *filp, struct file_lock_context *flctx)
 		.fl_owner = filp,
 		.fl_pid = current->tgid,
 		.fl_file = filp,
-		.fl_flags = FL_FLOCK,
+		.fl_flags = FL_FLOCK | FL_CLOSE,
 		.fl_type = F_UNLCK,
 		.fl_end = OFFSET_MAX,
 	};
-- 
1.7.1

