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:   bingjingc <bingjingc@synology.com>
To:     fdmanana@kernel.org, josef@toxicpanda.com, dsterba@suse.com,
        clm@fb.com, linux-btrfs@vger.kernel.org,
        linux-kernel@vger.kernel.org
Cc:     bingjingc@synology.com, robbieko@synology.com, bxxxjxxg@gmail.com
Subject: [PATCH 0/2] btrfs: send: fix a bug that sending unlink commands for directories
Date:   Mon,  8 Aug 2022 18:27:33 +0800
Message-Id: <20220808102735.4556-1-bingjingc@synology.com>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Precedence: bulk

From: BingJing Chang <bingjingc@synology.com>

There is a bug sending unlink commands for directories. In
commit 46b2f4590aab ("Btrfs: fix send failure when root has deleted files
still open")', orphan inode issue was addressed. There're no reference
paths for these orphan inodes, so the send operation fails with an ENOENT
error. Therefore, in that patch, sctx->ignore_cur_inode was introduced to
be set if the current inode has a link count of zero for bypassing some
unnecessary steps. And a helper function btrfs_unlink_all_paths() was
introduced and called to clean up old reference paths found in the parent
snapshot. However, not only regular files but also directories can be
orphan inodes. So if it meets an orphan directory, a wrong unlink command
for this directory will be issued. Soon the unlink command fails with an
EISDIR error.


BingJing Chang (2):
  btrfs: send: refactor get_inode_info()
  btrfs: send: fix a bug that sending unlink commands for directories

 fs/btrfs/send.c | 390 +++++++++++++++++++++++-------------------------
 1 file changed, 189 insertions(+), 201 deletions(-)

-- 
2.37.1