nvme-pci: avoid dereference of symbol from unloaded module

The 'remove_work' may be scheduled to run after nvme_remove()
returns since we can't simply cancel it in nvme_remove() for
avoiding deadlock. Once nvme_remove() returns, this module(nvme)
can be unloaded.

On the other hand, nvme_put_ctrl() calls ctr->ops->free_ctrl
which may point to nvme_pci_free_ctrl() in unloaded module.

This patch avoids this issue by queuing 'remove_work' via 'nvme_wq',
and flush this worqueue in nvme_exit() as suggested by Sagi.

Suggested-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Keith Busch <keith.busch@intel.com>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

drivers/nvme/host/pci.c

@@ -2280,7 +2280,7 @@ static void nvme_remove_dead_ctrl(struct nvme_dev *dev, int status)
 
 	nvme_get_ctrl(&dev->ctrl);
 	nvme_dev_disable(dev, false);
-	if (!schedule_work(&dev->remove_work))
+	if (!queue_work(nvme_wq, &dev->remove_work))
 		nvme_put_ctrl(&dev->ctrl);
 }
 
@@ -2703,6 +2703,7 @@ static int __init nvme_init(void)
 static void __exit nvme_exit(void)
 {
 	pci_unregister_driver(&nvme_driver);
+	flush_workqueue(nvme_wq);
 	_nvme_check_size();
 }