Received-SPF: pass (google.com: domain of linux-kernel+bounces-2509-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Message-ID: <tencent_AB577926C13364C3696965A3A2244B776605@qq.com>
From: Edward Adam Davis <eadavis@qq.com>
To: syzbot+4da851837827326a7cd4@syzkaller.appspotmail.com
Cc: axboe@kernel.dk,
	linux-block@vger.kernel.org,
	linux-kernel@vger.kernel.org,
	syzkaller-bugs@googlegroups.com
Subject: [PATCH next] nbd: fix task hung in bdev_release
Date: Sun, 17 Dec 2023 16:34:02 +0800
In-Reply-To: <000000000000098af2060b5ff161@google.com>
References: <000000000000098af2060b5ff161@google.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit

If the nbd timeout value is not set before calling nbd_start_device_ioctl(),
so the socket will be disabled disconnection, which will cause the timer to 
only be reset and cause this 143 second timeout issue.

The solution I have provided here is to set the default timeout value for nbd in
nbd_start_device() to avoid problems from occurring.

Reported-and-tested-by: syzbot+4da851837827326a7cd4@syzkaller.appspotmail.com
Signed-off-by: Edward Adam Davis <eadavis@qq.com>
---
 drivers/block/nbd.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
index b6414e1e645b..31656364d8a3 100644
--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c
@@ -1394,6 +1394,9 @@ static int nbd_start_device(struct nbd_device *nbd)
 	}
 	set_bit(NBD_RT_HAS_PID_FILE, &config->runtime_flags);
 
+	if (num_connections == 1 && !nbd->tag_set.timeout)
+		nbd->tag_set.timeout = HZ * 1024;
+
 	nbd_dev_dbg_init(nbd);
 	for (i = 0; i < num_connections; i++) {
 		struct recv_thread_args *args;
-- 
2.43.0