Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
From:   Jiachen Zhang <zhangjiachen.jaycee@bytedance.com>
To:     Miklos Szeredi <miklos@szeredi.hu>, linux-fsdevel@vger.kernel.org,
        linux-kernel@vger.kernel.org
Cc:     Andrew Morton <akpm@osdl.org>, me@jcix.top,
        Jiachen Zhang <zhangjiachen.jaycee@bytedance.com>
Subject: [PATCH 1/2] fuse: support unlock remote OFD locks on file release
Date:   Thu,  8 Jun 2023 16:46:08 +0800
Message-Id: <20230608084609.14245-2-zhangjiachen.jaycee@bytedance.com>
In-Reply-To: <20230608084609.14245-1-zhangjiachen.jaycee@bytedance.com>
References: <20230608084609.14245-1-zhangjiachen.jaycee@bytedance.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

Like flock(2), the fcntl(2) OFD locks also use struct file addresses as
the lock owner ID, and also should be unlocked on file release.

The commit 37fb3a30b462 ("fuse: fix flock") fixed the flock unlocking
issue on file release. This commit aims to fix the OFD lock by reusing
the release_flag 'FUSE_RELEASE_FLOCK_UNLOCK'. The FUSE daemons should
unlock both OFD locks and flocks in the FUSE_RELEASE handler.

To make it more clear, rename 'ff->flock' to 'ff->unlock_on_release', as
it would be used for both flock and OFD lock. It will be set true if the
value of fl->fl_owner equals to the struct file address.

Fixes: 37fb3a30b462 ("fuse: fix flock")
Signed-off-by: Jiachen Zhang <zhangjiachen.jaycee@bytedance.com>
---
 fs/fuse/file.c   | 17 ++++++++++++++---
 fs/fuse/fuse_i.h |  2 +-
 2 files changed, 15 insertions(+), 4 deletions(-)

diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index de37a3a06a71..7fe9d405969e 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -312,7 +312,7 @@ void fuse_file_release(struct inode *inode, struct fuse_file *ff,
 
 	fuse_prepare_release(fi, ff, open_flags, opcode);
 
-	if (ff->flock) {
+	if (ff->unlock_on_release) {
 		ra->inarg.release_flags |= FUSE_RELEASE_FLOCK_UNLOCK;
 		ra->inarg.lock_owner = fuse_lock_owner_id(ff->fm->fc, id);
 	}
@@ -2650,8 +2650,19 @@ static int fuse_file_lock(struct file *file, int cmd, struct file_lock *fl)
 	} else {
 		if (fc->no_lock)
 			err = posix_lock_file(file, fl, NULL);
-		else
+		else {
+			/*
+			 * Like flock, the OFD lock also uses the struct
+			 * file address as the fl_owner, and should be
+			 * unlocked on file release.
+			 */
+			if (file == fl->fl_owner) {
+				struct fuse_file *ff = file->private_data;
+
+				ff->unlock_on_release = true;
+			}
 			err = fuse_setlk(file, fl, 0);
+		}
 	}
 	return err;
 }
@@ -2668,7 +2679,7 @@ static int fuse_file_flock(struct file *file, int cmd, struct file_lock *fl)
 		struct fuse_file *ff = file->private_data;
 
 		/* emulate flock with POSIX locks */
-		ff->flock = true;
+		ff->unlock_on_release = true;
 		err = fuse_setlk(file, fl, 1);
 	}
 
diff --git a/fs/fuse/fuse_i.h b/fs/fuse/fuse_i.h
index 9b7fc7d3c7f1..574f67bd5684 100644
--- a/fs/fuse/fuse_i.h
+++ b/fs/fuse/fuse_i.h
@@ -225,7 +225,7 @@ struct fuse_file {
 	wait_queue_head_t poll_wait;
 
 	/** Has flock been performed on this file? */
-	bool flock:1;
+	bool unlock_on_release:1;
 };
 
 /** One input argument of a request */
-- 
2.20.1