nbd: add a flush_workqueue in nbd_start_device

[ Upstream commit 5c0dd228b5fc30a3b732c7ae2657e0161ec7ed80 ]

When kzalloc fail, may cause trying to destroy the
workqueue from inside the workqueue.

If num_connections is m (2 < m), and NO.1 ~ NO.n
(1 < n < m) kzalloc are successful. The NO.(n + 1)
failed. Then, nbd_start_device will return ENOMEM
to nbd_start_device_ioctl, and nbd_start_device_ioctl
will return immediately without running flush_workqueue.
However, we still have n recv threads. If nbd_release
run first, recv threads may have to drop the last
config_refs and try to destroy the workqueue from
inside the workqueue.

To fix it, add a flush_workqueue in nbd_start_device.

Fixes: e9e006f5fcf2 ("nbd: fix max number of supported devs")
Signed-off-by: Sun Ke <sunke32@huawei.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

1 files changed, 10 insertions, 0 deletions

diff --git a/drivers/block/nbd.c b/drivers/block/nbd.c
index b9d321bdaa8a..226103af30f0 100644
--- a/drivers/block/nbd.c
+++ b/drivers/block/nbd.c
@@ -1216,6 +1216,16 @@ static int nbd_start_device(struct nbd_device *nbd)
 		args = kzalloc(sizeof(*args), GFP_KERNEL);
 		if (!args) {
 			sock_shutdown(nbd);
+			/*
+			 * If num_connections is m (2 < m),
+			 * and NO.1 ~ NO.n(1 < n < m) kzallocs are successful.
+			 * But NO.(n + 1) failed. We still have n recv threads.
+			 * So, add flush_workqueue here to prevent recv threads
+			 * dropping the last config_refs and trying to destroy
+			 * the workqueue from inside the workqueue.
+			 */
+			if (i)
+				flush_workqueue(nbd->recv_workq);
 			return -ENOMEM;
 		}
 		sk_set_memalloc(config->socks[i]->sock->sk);