Merge tag 'v5.7-rc6' into rdma.git for-next

Linux 5.7-rc6

Conflict in drivers/net/ethernet/mellanox/mlx5/core/steering/dr_send.c
resolved by deleting dr_cq_event, matching how netdev resolved it.

Required for dependencies in the following patches.

Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

5 files changed, 181 insertions, 75 deletions

diff --git a/drivers/block/null_blk.h b/drivers/block/null_blk.h
index 62b660821dbc..81b311c9d781 100644
--- a/drivers/block/null_blk.h
+++ b/drivers/block/null_blk.h
@@ -85,26 +85,35 @@ struct nullb {
 	char disk_name[DISK_NAME_LEN];
 };
 
+blk_status_t null_process_cmd(struct nullb_cmd *cmd,
+			      enum req_opf op, sector_t sector,
+			      unsigned int nr_sectors);
+
 #ifdef CONFIG_BLK_DEV_ZONED
-int null_zone_init(struct nullb_device *dev);
-void null_zone_exit(struct nullb_device *dev);
+int null_init_zoned_dev(struct nullb_device *dev, struct request_queue *q);
+int null_register_zoned_dev(struct nullb *nullb);
+void null_free_zoned_dev(struct nullb_device *dev);
 int null_report_zones(struct gendisk *disk, sector_t sector,
 		      unsigned int nr_zones, report_zones_cb cb, void *data);
-blk_status_t null_handle_zoned(struct nullb_cmd *cmd,
-				enum req_opf op, sector_t sector,
-				sector_t nr_sectors);
+blk_status_t null_process_zoned_cmd(struct nullb_cmd *cmd,
+				    enum req_opf op, sector_t sector,
+				    sector_t nr_sectors);
 size_t null_zone_valid_read_len(struct nullb *nullb,
 				sector_t sector, unsigned int len);
 #else
-static inline int null_zone_init(struct nullb_device *dev)
+static inline int null_init_zoned_dev(struct nullb_device *dev,
+				      struct request_queue *q)
 {
 	pr_err("CONFIG_BLK_DEV_ZONED not enabled\n");
 	return -EINVAL;
 }
-static inline void null_zone_exit(struct nullb_device *dev) {}
-static inline blk_status_t null_handle_zoned(struct nullb_cmd *cmd,
-					     enum req_opf op, sector_t sector,
-					     sector_t nr_sectors)
+static inline int null_register_zoned_dev(struct nullb *nullb)
+{
+	return -ENODEV;
+}
+static inline void null_free_zoned_dev(struct nullb_device *dev) {}
+static inline blk_status_t null_process_zoned_cmd(struct nullb_cmd *cmd,
+			enum req_opf op, sector_t sector, sector_t nr_sectors)
 {
 	return BLK_STS_NOTSUPP;
 }
diff --git a/drivers/block/null_blk_main.c b/drivers/block/null_blk_main.c
index 4e1c0712278e..8efd8778e209 100644
--- a/drivers/block/null_blk_main.c
+++ b/drivers/block/null_blk_main.c
@@ -580,7 +580,7 @@ static void null_free_dev(struct nullb_device *dev)
 	if (!dev)
 		return;
 
-	null_zone_exit(dev);
+	null_free_zoned_dev(dev);
 	badblocks_exit(&dev->badblocks);
 	kfree(dev);
 }
@@ -1276,6 +1276,25 @@ static inline void nullb_complete_cmd(struct nullb_cmd *cmd)
 	}
 }
 
+blk_status_t null_process_cmd(struct nullb_cmd *cmd,
+			      enum req_opf op, sector_t sector,
+			      unsigned int nr_sectors)
+{
+	struct nullb_device *dev = cmd->nq->dev;
+	blk_status_t ret;
+
+	if (dev->badblocks.shift != -1) {
+		ret = null_handle_badblocks(cmd, sector, nr_sectors);
+		if (ret != BLK_STS_OK)
+			return ret;
+	}
+
+	if (dev->memory_backed)
+		return null_handle_memory_backed(cmd, op);
+
+	return BLK_STS_OK;
+}
+
 static blk_status_t null_handle_cmd(struct nullb_cmd *cmd, sector_t sector,
 				    sector_t nr_sectors, enum req_opf op)
 {
@@ -1294,17 +1313,11 @@ static blk_status_t null_handle_cmd(struct nullb_cmd *cmd, sector_t sector,
 		goto out;
 	}
 
-	if (nullb->dev->badblocks.shift != -1) {
-		cmd->error = null_handle_badblocks(cmd, sector, nr_sectors);
-		if (cmd->error != BLK_STS_OK)
-			goto out;
-	}
-
-	if (dev->memory_backed)
-		cmd->error = null_handle_memory_backed(cmd, op);
-
-	if (!cmd->error && dev->zoned)
-		cmd->error = null_handle_zoned(cmd, op, sector, nr_sectors);
+	if (dev->zoned)
+		cmd->error = null_process_zoned_cmd(cmd, op,
+						    sector, nr_sectors);
+	else
+		cmd->error = null_process_cmd(cmd, op, sector, nr_sectors);
 
 out:
 	nullb_complete_cmd(cmd);
@@ -1605,19 +1618,12 @@ static int null_gendisk_register(struct nullb *nullb)
 	disk->queue		= nullb->q;
 	strncpy(disk->disk_name, nullb->disk_name, DISK_NAME_LEN);
 
-#ifdef CONFIG_BLK_DEV_ZONED
 	if (nullb->dev->zoned) {
-		if (queue_is_mq(nullb->q)) {
-			int ret = blk_revalidate_disk_zones(disk);
-			if (ret)
-				return ret;
-		} else {
-			blk_queue_chunk_sectors(nullb->q,
-					nullb->dev->zone_size_sects);
-			nullb->q->nr_zones = blkdev_nr_zones(disk);
-		}
+		int ret = null_register_zoned_dev(nullb);
+
+		if (ret)
+			return ret;
 	}
-#endif
 
 	add_disk(disk);
 	return 0;
@@ -1773,14 +1779,9 @@ static int null_add_dev(struct nullb_device *dev)
 	}
 
 	if (dev->zoned) {
-		rv = null_zone_init(dev);
+		rv = null_init_zoned_dev(dev, nullb->q);
 		if (rv)
 			goto out_cleanup_blk_queue;
-
-		nullb->q->limits.zoned = BLK_ZONED_HM;
-		blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, nullb->q);
-		blk_queue_required_elevator_features(nullb->q,
-						ELEVATOR_F_ZBD_SEQ_WRITE);
 	}
 
 	nullb->q->queuedata = nullb;
@@ -1809,8 +1810,7 @@ static int null_add_dev(struct nullb_device *dev)
 
 	return 0;
 out_cleanup_zone:
-	if (dev->zoned)
-		null_zone_exit(dev);
+	null_free_zoned_dev(dev);
 out_cleanup_blk_queue:
 	blk_cleanup_queue(nullb->q);
 out_cleanup_tags:
diff --git a/drivers/block/null_blk_zoned.c b/drivers/block/null_blk_zoned.c
index 673618d8222a..9e4bcdad1a80 100644
--- a/drivers/block/null_blk_zoned.c
+++ b/drivers/block/null_blk_zoned.c
@@ -13,7 +13,7 @@ static inline unsigned int null_zone_no(struct nullb_device *dev, sector_t sect)
 	return sect >> ilog2(dev->zone_size_sects);
 }
 
-int null_zone_init(struct nullb_device *dev)
+int null_init_zoned_dev(struct nullb_device *dev, struct request_queue *q)
 {
 	sector_t dev_size = (sector_t)dev->size * 1024 * 1024;
 	sector_t sector = 0;
@@ -61,10 +61,27 @@ int null_zone_init(struct nullb_device *dev)
 		sector += dev->zone_size_sects;
 	}
 
+	q->limits.zoned = BLK_ZONED_HM;
+	blk_queue_flag_set(QUEUE_FLAG_ZONE_RESETALL, q);
+	blk_queue_required_elevator_features(q, ELEVATOR_F_ZBD_SEQ_WRITE);
+
+	return 0;
+}
+
+int null_register_zoned_dev(struct nullb *nullb)
+{
+	struct request_queue *q = nullb->q;
+
+	if (queue_is_mq(q))
+		return blk_revalidate_disk_zones(nullb->disk);
+
+	blk_queue_chunk_sectors(q, nullb->dev->zone_size_sects);
+	q->nr_zones = blkdev_nr_zones(nullb->disk);
+
 	return 0;
 }
 
-void null_zone_exit(struct nullb_device *dev)
+void null_free_zoned_dev(struct nullb_device *dev)
 {
 	kvfree(dev->zones);
 }
@@ -126,11 +143,16 @@ static blk_status_t null_zone_write(struct nullb_cmd *cmd, sector_t sector,
 	struct nullb_device *dev = cmd->nq->dev;
 	unsigned int zno = null_zone_no(dev, sector);
 	struct blk_zone *zone = &dev->zones[zno];
+	blk_status_t ret;
+
+	trace_nullb_zone_op(cmd, zno, zone->cond);
+
+	if (zone->type == BLK_ZONE_TYPE_CONVENTIONAL)
+		return null_process_cmd(cmd, REQ_OP_WRITE, sector, nr_sectors);
 
 	switch (zone->cond) {
 	case BLK_ZONE_COND_FULL:
 		/* Cannot write to a full zone */
-		cmd->error = BLK_STS_IOERR;
 		return BLK_STS_IOERR;
 	case BLK_ZONE_COND_EMPTY:
 	case BLK_ZONE_COND_IMP_OPEN:
@@ -143,19 +165,18 @@ static blk_status_t null_zone_write(struct nullb_cmd *cmd, sector_t sector,
 		if (zone->cond != BLK_ZONE_COND_EXP_OPEN)
 			zone->cond = BLK_ZONE_COND_IMP_OPEN;
 
+		ret = null_process_cmd(cmd, REQ_OP_WRITE, sector, nr_sectors);
+		if (ret != BLK_STS_OK)
+			return ret;
+
 		zone->wp += nr_sectors;
 		if (zone->wp == zone->start + zone->len)
 			zone->cond = BLK_ZONE_COND_FULL;
-		break;
-	case BLK_ZONE_COND_NOT_WP:
-		break;
+		return BLK_STS_OK;
 	default:
 		/* Invalid zone condition */
 		return BLK_STS_IOERR;
 	}
-
-	trace_nullb_zone_op(cmd, zno, zone->cond);
-	return BLK_STS_OK;
 }
 
 static blk_status_t null_zone_mgmt(struct nullb_cmd *cmd, enum req_opf op,
@@ -216,8 +237,8 @@ static blk_status_t null_zone_mgmt(struct nullb_cmd *cmd, enum req_opf op,
 	return BLK_STS_OK;
 }
 
-blk_status_t null_handle_zoned(struct nullb_cmd *cmd, enum req_opf op,
-			       sector_t sector, sector_t nr_sectors)
+blk_status_t null_process_zoned_cmd(struct nullb_cmd *cmd, enum req_opf op,
+				    sector_t sector, sector_t nr_sectors)
 {
 	switch (op) {
 	case REQ_OP_WRITE:
@@ -229,6 +250,6 @@ blk_status_t null_handle_zoned(struct nullb_cmd *cmd, enum req_opf op,
 	case REQ_OP_ZONE_FINISH:
 		return null_zone_mgmt(cmd, op, sector);
 	default:
-		return BLK_STS_OK;
+		return null_process_cmd(cmd, op, sector, nr_sectors);
 	}
 }
diff --git a/drivers/block/rbd.c b/drivers/block/rbd.c
index 1e0a6b19ae0d..67d65ac785e9 100644
--- a/drivers/block/rbd.c
+++ b/drivers/block/rbd.c
@@ -3754,11 +3754,7 @@ static int __rbd_notify_op_lock(struct rbd_device *rbd_dev,
 static void rbd_notify_op_lock(struct rbd_device *rbd_dev,
 			       enum rbd_notify_op notify_op)
 {
-	struct page **reply_pages;
-	size_t reply_len;
-
-	__rbd_notify_op_lock(rbd_dev, notify_op, &reply_pages, &reply_len);
-	ceph_release_page_vector(reply_pages, calc_pages_for(0, reply_len));
+	__rbd_notify_op_lock(rbd_dev, notify_op, NULL, NULL);
 }
 
 static void rbd_notify_acquired_lock(struct work_struct *work)
@@ -4527,6 +4523,10 @@ static void cancel_tasks_sync(struct rbd_device *rbd_dev)
 	cancel_work_sync(&rbd_dev->unlock_work);
 }
 
+/*
+ * header_rwsem must not be held to avoid a deadlock with
+ * rbd_dev_refresh() when flushing notifies.
+ */
 static void rbd_unregister_watch(struct rbd_device *rbd_dev)
 {
 	cancel_tasks_sync(rbd_dev);
@@ -6894,9 +6894,10 @@ static void rbd_print_dne(struct rbd_device *rbd_dev, bool is_snap)
 
 static void rbd_dev_image_release(struct rbd_device *rbd_dev)
 {
-	rbd_dev_unprobe(rbd_dev);
-	if (rbd_dev->opts)
+	if (!rbd_is_ro(rbd_dev))
 		rbd_unregister_watch(rbd_dev);
+
+	rbd_dev_unprobe(rbd_dev);
 	rbd_dev->image_format = 0;
 	kfree(rbd_dev->spec->image_id);
 	rbd_dev->spec->image_id = NULL;
@@ -6907,6 +6908,9 @@ static void rbd_dev_image_release(struct rbd_device *rbd_dev)
  * device.  If this image is the one being mapped (i.e., not a
  * parent), initiate a watch on its header object before using that
  * object to get detailed information about the rbd image.
+ *
+ * On success, returns with header_rwsem held for write if called
+ * with @depth == 0.
  */
 static int rbd_dev_image_probe(struct rbd_device *rbd_dev, int depth)
 {
@@ -6936,11 +6940,14 @@ static int rbd_dev_image_probe(struct rbd_device *rbd_dev, int depth)
 		}
 	}
 
+	if (!depth)
+		down_write(&rbd_dev->header_rwsem);
+
 	ret = rbd_dev_header_info(rbd_dev);
 	if (ret) {
 		if (ret == -ENOENT && !need_watch)
 			rbd_print_dne(rbd_dev, false);
-		goto err_out_watch;
+		goto err_out_probe;
 	}
 
 	/*
@@ -6985,10 +6992,11 @@ static int rbd_dev_image_probe(struct rbd_device *rbd_dev, int depth)
 	return 0;
 
 err_out_probe:
-	rbd_dev_unprobe(rbd_dev);
-err_out_watch:
+	if (!depth)
+		up_write(&rbd_dev->header_rwsem);
 	if (need_watch)
 		rbd_unregister_watch(rbd_dev);
+	rbd_dev_unprobe(rbd_dev);
 err_out_format:
 	rbd_dev->image_format = 0;
 	kfree(rbd_dev->spec->image_id);
@@ -7050,12 +7058,9 @@ static ssize_t do_rbd_add(struct bus_type *bus,
 		goto err_out_rbd_dev;
 	}
 
-	down_write(&rbd_dev->header_rwsem);
 	rc = rbd_dev_image_probe(rbd_dev, 0);
-	if (rc < 0) {
-		up_write(&rbd_dev->header_rwsem);
+	if (rc < 0)
 		goto err_out_rbd_dev;
-	}
 
 	if (rbd_dev->opts->alloc_size > rbd_dev->layout.object_size) {
 		rbd_warn(rbd_dev, "alloc_size adjusted to %u",
diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index f9b1e70f1b31..9d21bf0f155e 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -15,6 +15,7 @@
 #include <linux/blk-mq.h>
 #include <linux/blk-mq-virtio.h>
 #include <linux/numa.h>
+#include <uapi/linux/virtio_ring.h>
 
 #define PART_BITS 4
 #define VQ_NAME_LEN 16
@@ -32,6 +33,15 @@ struct virtio_blk_vq {
 } ____cacheline_aligned_in_smp;
 
 struct virtio_blk {
+	/*
+	 * This mutex must be held by anything that may run after
+	 * virtblk_remove() sets vblk->vdev to NULL.
+	 *
+	 * blk-mq, virtqueue processing, and sysfs attribute code paths are
+	 * shut down before vblk->vdev is set to NULL and therefore do not need
+	 * to hold this mutex.
+	 */
+	struct mutex vdev_mutex;
 	struct virtio_device *vdev;
 
 	/* The disk structure for the kernel. */
@@ -43,6 +53,13 @@ struct virtio_blk {
 	/* Process context for config space updates */
 	struct work_struct config_work;
 
+	/*
+	 * Tracks references from block_device_operations open/release and
+	 * virtio_driver probe/remove so this object can be freed once no
+	 * longer in use.
+	 */
+	refcount_t refs;
+
 	/* What host tells us, plus 2 for header & tailer. */
 	unsigned int sg_elems;
 
@@ -294,10 +311,55 @@ out:
 	return err;
 }
 
+static void virtblk_get(struct virtio_blk *vblk)
+{
+	refcount_inc(&vblk->refs);
+}
+
+static void virtblk_put(struct virtio_blk *vblk)
+{
+	if (refcount_dec_and_test(&vblk->refs)) {
+		ida_simple_remove(&vd_index_ida, vblk->index);
+		mutex_destroy(&vblk->vdev_mutex);
+		kfree(vblk);
+	}
+}
+
+static int virtblk_open(struct block_device *bd, fmode_t mode)
+{
+	struct virtio_blk *vblk = bd->bd_disk->private_data;
+	int ret = 0;
+
+	mutex_lock(&vblk->vdev_mutex);
+
+	if (vblk->vdev)
+		virtblk_get(vblk);
+	else
+		ret = -ENXIO;
+
+	mutex_unlock(&vblk->vdev_mutex);
+	return ret;
+}
+
+static void virtblk_release(struct gendisk *disk, fmode_t mode)
+{
+	struct virtio_blk *vblk = disk->private_data;
+
+	virtblk_put(vblk);
+}
+
 /* We provide getgeo only to please some old bootloader/partitioning tools */
 static int virtblk_getgeo(struct block_device *bd, struct hd_geometry *geo)
 {
 	struct virtio_blk *vblk = bd->bd_disk->private_data;
+	int ret = 0;
+
+	mutex_lock(&vblk->vdev_mutex);
+
+	if (!vblk->vdev) {
+		ret = -ENXIO;
+		goto out;
+	}
 
 	/* see if the host passed in geometry config */
 	if (virtio_has_feature(vblk->vdev, VIRTIO_BLK_F_GEOMETRY)) {
@@ -313,11 +375,15 @@ static int virtblk_getgeo(struct block_device *bd, struct hd_geometry *geo)
 		geo->sectors = 1 << 5;
 		geo->cylinders = get_capacity(bd->bd_disk) >> 11;
 	}
-	return 0;
+out:
+	mutex_unlock(&vblk->vdev_mutex);
+	return ret;
 }
 
 static const struct block_device_operations virtblk_fops = {
 	.owner  = THIS_MODULE,
+	.open = virtblk_open,
+	.release = virtblk_release,
 	.getgeo = virtblk_getgeo,
 };
 
@@ -654,6 +720,10 @@ static int virtblk_probe(struct virtio_device *vdev)
 		goto out_free_index;
 	}
 
+	/* This reference is dropped in virtblk_remove(). */
+	refcount_set(&vblk->refs, 1);
+	mutex_init(&vblk->vdev_mutex);
+
 	vblk->vdev = vdev;
 	vblk->sg_elems = sg_elems;
 
@@ -819,8 +889,6 @@ out:
 static void virtblk_remove(struct virtio_device *vdev)
 {
 	struct virtio_blk *vblk = vdev->priv;
-	int index = vblk->index;
-	int refc;
 
 	/* Make sure no work handler is accessing the device. */
 	flush_work(&vblk->config_work);
@@ -830,18 +898,21 @@ static void virtblk_remove(struct virtio_device *vdev)
 
 	blk_mq_free_tag_set(&vblk->tag_set);
 
+	mutex_lock(&vblk->vdev_mutex);
+
 	/* Stop all the virtqueues. */
 	vdev->config->reset(vdev);
 
-	refc = kref_read(&disk_to_dev(vblk->disk)->kobj.kref);
+	/* Virtqueues are stopped, nothing can use vblk->vdev anymore. */
+	vblk->vdev = NULL;
+
 	put_disk(vblk->disk);
 	vdev->config->del_vqs(vdev);
 	kfree(vblk->vqs);
-	kfree(vblk);
 
-	/* Only free device id if we don't have any users */
-	if (refc == 1)
-		ida_simple_remove(&vd_index_ida, index);
+	mutex_unlock(&vblk->vdev_mutex);
+
+	virtblk_put(vblk);
 }
 
 #ifdef CONFIG_PM_SLEEP