nfs/blocklayout: Fix premature PR key unregistration

During generic/069 runs with pNFS SCSI layouts, the NFS client emits
the following in the system journal:

kernel: pNFS: failed to open device /dev/disk/by-id/dm-uuid-mpath-0x6001405e3366f045b7949eb8e4540b51 (-2)
kernel: pNFS: using block device sdb (reservation key 0x666b60901e7b26b3)
kernel: pNFS: failed to open device /dev/disk/by-id/dm-uuid-mpath-0x6001405e3366f045b7949eb8e4540b51 (-2)
kernel: pNFS: using block device sdb (reservation key 0x666b60901e7b26b3)
kernel: sd 6:0:0:1: reservation conflict
kernel: sd 6:0:0:1: [sdb] tag#16 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_OK cmd_age=0s
kernel: sd 6:0:0:1: [sdb] tag#16 CDB: Write(10) 2a 00 00 00 00 50 00 00 08 00
kernel: reservation conflict error, dev sdb, sector 80 op 0x1:(WRITE) flags 0x0 phys_seg 1 prio class 2
kernel: sd 6:0:0:1: reservation conflict
kernel: sd 6:0:0:1: reservation conflict
kernel: sd 6:0:0:1: [sdb] tag#18 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_OK cmd_age=0s
kernel: sd 6:0:0:1: [sdb] tag#17 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_OK cmd_age=0s
kernel: sd 6:0:0:1: [sdb] tag#18 CDB: Write(10) 2a 00 00 00 00 60 00 00 08 00
kernel: sd 6:0:0:1: [sdb] tag#17 CDB: Write(10) 2a 00 00 00 00 58 00 00 08 00
kernel: reservation conflict error, dev sdb, sector 96 op 0x1:(WRITE) flags 0x0 phys_seg 1 prio class 0
kernel: reservation conflict error, dev sdb, sector 88 op 0x1:(WRITE) flags 0x0 phys_seg 1 prio class 0
systemd[1]: fstests-generic-069.scope: Deactivated successfully.
systemd[1]: fstests-generic-069.scope: Consumed 5.092s CPU time.
systemd[1]: media-test.mount: Deactivated successfully.
systemd[1]: media-scratch.mount: Deactivated successfully.
kernel: sd 6:0:0:1: reservation conflict
kernel: failed to unregister PR key.

This appears to be due to a race. bl_alloc_lseg() calls this:

561 static struct nfs4_deviceid_node *
562 bl_find_get_deviceid(struct nfs_server *server,
563                 const struct nfs4_deviceid *id, const struct cred *cred,
564                 gfp_t gfp_mask)
565 {
566         struct nfs4_deviceid_node *node;
567         unsigned long start, end;
568
569 retry:
570         node = nfs4_find_get_deviceid(server, id, cred, gfp_mask);
571         if (!node)
572                 return ERR_PTR(-ENODEV);

nfs4_find_get_deviceid() does a lookup without the spin lock first.
If it can't find a matching deviceid, it creates a new device_info
(which calls bl_alloc_deviceid_node, and that registers the device's
PR key).

Then it takes the nfs4_deviceid_lock and looks up the deviceid again.
If it finds it this time, bl_find_get_deviceid() frees the spare
(new) device_info, which unregisters the PR key for the same device.

Any subsequent I/O from this client on that device gets EBADE.

The umount later unregisters the device's PR key again.

To prevent this problem, register the PR key after the deviceid_node
lookup.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Benjamin Coddington <bcodding@redhat.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>

Author: chucklever

fs/nfs/blocklayout/blocklayout.c

@@ -564,25 +564,32 @@ bl_find_get_deviceid(struct nfs_server *server,
 		gfp_t gfp_mask)
 {
 	struct nfs4_deviceid_node *node;
-	unsigned long start, end;
+	int err = -ENODEV;
 
 retry:
 	node = nfs4_find_get_deviceid(server, id, cred, gfp_mask);
 	if (!node)
 		return ERR_PTR(-ENODEV);
 
-	if (test_bit(NFS_DEVICEID_UNAVAILABLE, &node->flags) == 0)
-		return node;
+	if (test_bit(NFS_DEVICEID_UNAVAILABLE, &node->flags)) {
+		unsigned long end = jiffies;
+		unsigned long start = end - PNFS_DEVICE_RETRY_TIMEOUT;
 
-	end = jiffies;
-	start = end - PNFS_DEVICE_RETRY_TIMEOUT;
-	if (!time_in_range(node->timestamp_unavailable, start, end)) {
-		nfs4_delete_deviceid(node->ld, node->nfs_client, id);
-		goto retry;
+		if (!time_in_range(node->timestamp_unavailable, start, end)) {
+			nfs4_delete_deviceid(node->ld, node->nfs_client, id);
+			goto retry;
+		}
+		goto out_put;
 	}
 
+	if (!bl_register_dev(container_of(node, struct pnfs_block_dev, node)))
+		goto out_put;
+
+	return node;
+
+out_put:
 	nfs4_put_deviceid_node(node);
-	return ERR_PTR(-ENODEV);
+	return ERR_PTR(err);
 }
 
 static int

fs/nfs/blocklayout/blocklayout.h

@@ -104,20 +104,26 @@ struct pnfs_block_dev {
 	u64				start;
 	u64				len;
 
+	enum pnfs_block_volume_type	type;
 	u32				nr_children;
 	struct pnfs_block_dev		*children;
 	u64				chunk_size;
 
 	struct file			*bdev_file;
 	u64				disk_offset;
+	unsigned long			flags;
 
 	u64				pr_key;
-	bool				pr_registered;
 
 	bool (*map)(struct pnfs_block_dev *dev, u64 offset,
 			struct pnfs_block_dev_map *map);
 };
 
+/* pnfs_block_dev flag bits */
+enum {
+	PNFS_BDEV_REGISTERED = 0,
+};
+
 /* sector_t fields are all in 512-byte sectors */
 struct pnfs_block_extent {
 	union {
@@ -172,6 +178,7 @@ struct bl_msg_hdr {
 #define BL_DEVICE_REQUEST_ERR          0x2 /* User level process fails */
 
 /* dev.c */
+bool bl_register_dev(struct pnfs_block_dev *d);
 struct nfs4_deviceid_node *bl_alloc_deviceid_node(struct nfs_server *server,
 		struct pnfs_device *pdev, gfp_t gfp_mask);
 void bl_free_deviceid_node(struct nfs4_deviceid_node *d);

fs/nfs/blocklayout/dev.c

@@ -13,27 +13,82 @@
 
 #define NFSDBG_FACILITY		NFSDBG_PNFS_LD
 
+static void bl_unregister_scsi(struct pnfs_block_dev *dev)
+{
+	struct block_device *bdev = file_bdev(dev->bdev_file);
+	const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
+
+	if (!test_and_clear_bit(PNFS_BDEV_REGISTERED, &dev->flags))
+		return;
+
+	if (ops->pr_register(bdev, dev->pr_key, 0, false))
+		pr_err("failed to unregister PR key.\n");
+}
+
+static bool bl_register_scsi(struct pnfs_block_dev *dev)
+{
+	struct block_device *bdev = file_bdev(dev->bdev_file);
+	const struct pr_ops *ops = bdev->bd_disk->fops->pr_ops;
+	int status;
+
+	if (test_and_set_bit(PNFS_BDEV_REGISTERED, &dev->flags))
+		return true;
+
+	status = ops->pr_register(bdev, 0, dev->pr_key, true);
+	if (status) {
+		pr_err("pNFS: failed to register key for block device %s.",
+		       bdev->bd_disk->disk_name);
+		return false;
+	}
+	return true;
+}
+
+static void bl_unregister_dev(struct pnfs_block_dev *dev)
+{
+	u32 i;
+
+	if (dev->nr_children) {
+		for (i = 0; i < dev->nr_children; i++)
+			bl_unregister_dev(&dev->children[i]);
+		return;
+	}
+
+	if (dev->type == PNFS_BLOCK_VOLUME_SCSI)
+		bl_unregister_scsi(dev);
+}
+
+bool bl_register_dev(struct pnfs_block_dev *dev)
+{
+	u32 i;
+
+	if (dev->nr_children) {
+		for (i = 0; i < dev->nr_children; i++) {
+			if (!bl_register_dev(&dev->children[i])) {
+				while (i > 0)
+					bl_unregister_dev(&dev->children[--i]);
+				return false;
+			}
+		}
+		return true;
+	}
+
+	if (dev->type == PNFS_BLOCK_VOLUME_SCSI)
+		return bl_register_scsi(dev);
+	return true;
+}
+
 static void
 bl_free_device(struct pnfs_block_dev *dev)
 {
+	bl_unregister_dev(dev);
+
 	if (dev->nr_children) {
 		int i;
 
 		for (i = 0; i < dev->nr_children; i++)
 			bl_free_device(&dev->children[i]);
 		kfree(dev->children);
 	} else {
-		if (dev->pr_registered) {
-			const struct pr_ops *ops =
-				file_bdev(dev->bdev_file)->bd_disk->fops->pr_ops;
-			int error;
-
-			error = ops->pr_register(file_bdev(dev->bdev_file),
-				dev->pr_key, 0, false);
-			if (error)
-				pr_err("failed to unregister PR key.\n");
-		}
-
 		if (dev->bdev_file)
 			fput(dev->bdev_file);
 	}
@@ -365,14 +420,6 @@ bl_parse_scsi(struct nfs_server *server, struct pnfs_block_dev *d,
 		goto out_blkdev_put;
 	}
 
-	error = ops->pr_register(file_bdev(d->bdev_file), 0, d->pr_key, true);
-	if (error) {
-		pr_err("pNFS: failed to register key for block device %s.",
-				file_bdev(d->bdev_file)->bd_disk->disk_name);
-		goto out_blkdev_put;
-	}
-
-	d->pr_registered = true;
 	return 0;
 
 out_blkdev_put:
@@ -458,7 +505,9 @@ static int
 bl_parse_deviceid(struct nfs_server *server, struct pnfs_block_dev *d,
 		struct pnfs_block_volume *volumes, int idx, gfp_t gfp_mask)
 {
-	switch (volumes[idx].type) {
+	d->type = volumes[idx].type;
+
+	switch (d->type) {
 	case PNFS_BLOCK_VOLUME_SIMPLE:
 		return bl_parse_simple(server, d, volumes, idx, gfp_mask);
 	case PNFS_BLOCK_VOLUME_SLICE:
@@ -470,7 +519,7 @@ bl_parse_deviceid(struct nfs_server *server, struct pnfs_block_dev *d,
 	case PNFS_BLOCK_VOLUME_SCSI:
 		return bl_parse_scsi(server, d, volumes, idx, gfp_mask);
 	default:
-		dprintk("unsupported volume type: %d\n", volumes[idx].type);
+		dprintk("unsupported volume type: %d\n", d->type);
 		return -EIO;
 	}
 }