NVMe: Fix FS mount issue (hot-remove followed by hot-add)

After Hot-remove of a device with a mounted partition, when the device is hot-added again, the new node reappears as nvme0n1. Mounting this new node fails with the error:

mount: mount /dev/nvme0n1p1 on /mnt failed: File exists.

The old nodes's FS entries still exist and the kernel can't re-create procfs and sysfs entries for the new node with the same name. The patch fixes this issue.

Signed-off-by Indraneel M <indraneel.m at samsung.com>

drivers/block/nvme-core.c +1 -1

index 00fa5d2..5c151d7 100644
@@ -2735,6 +2735,7 @@ static void nvme_free_dev(struct kref *kref)
 
 	pci_dev_put(dev->pci_dev);
 	nvme_free_namespaces(dev);
+	nvme_release_instance(dev);
 	free_percpu(dev->io_queue);
 	kfree(dev->queues);
 	kfree(dev->entry);

@@ -2995,7 +2996,6 @@ static void nvme_remove(struct pci_dev *pdev)
 	nvme_dev_shutdown(dev);
 	nvme_free_queues(dev, 0);
 	nvme_dev_remove(dev);
-	nvme_release_instance(dev);
 	nvme_release_prp_pools(dev);
 	kref_put(&dev->kref, nvme_free_dev);
 }

Quoted 10 lines · Indraneel M

After Hot-remove of a device with a mounted partition, when the device is hot-added again, the new node reappears as nvme0n1. Mounting this new node fails with the error:

mount: mount /dev/nvme0n1p1 on /mnt failed: File exists.

The old nodes's FS entries still exist and the kernel can't re-create procfs and sysfs entries for the new node with the same name. The patch fixes this issue.

Quoted 1 line

Signed-off-by Indraneel M <indraneel.m at samsung.com>

Acked-by Keith Busch <keith.busch at intel.com>

Now where is this going? It won't apply to linux-block/for-next (though it is a trivial manual merge), which I think is where the 3.19 pull request to mainline will come from. I hope these trees get synced up soon so we don't have these fragmented drivers.