1 files changed, 226 insertions, 212 deletions

diff --git a/drivers/block/loop.c b/drivers/block/loop.c
index f8145499da38..c05fe27a96b6 100644
--- a/drivers/block/loop.c
+++ b/drivers/block/loop.c
@@ -68,7 +68,6 @@ struct loop_device {
 	struct list_head        idle_worker_list;
 	struct rb_root          worker_tree;
 	struct timer_list       timer;
-	bool			use_dio;
 	bool			sysfs_inited;
 
 	struct request_queue	*lo_queue;
@@ -173,7 +172,7 @@ static loff_t get_loop_size(struct loop_device *lo, struct file *file)
 static bool lo_bdev_can_use_dio(struct loop_device *lo,
 		struct block_device *backing_bdev)
 {
-	unsigned short sb_bsize = bdev_logical_block_size(backing_bdev);
+	unsigned int sb_bsize = bdev_logical_block_size(backing_bdev);
 
 	if (queue_logical_block_size(lo->lo_queue) < sb_bsize)
 		return false;
@@ -182,44 +181,44 @@ static bool lo_bdev_can_use_dio(struct loop_device *lo,
 	return true;
 }
 
-static void __loop_update_dio(struct loop_device *lo, bool dio)
+static bool lo_can_use_dio(struct loop_device *lo)
 {
-	struct file *file = lo->lo_backing_file;
-	struct inode *inode = file->f_mapping->host;
-	struct block_device *backing_bdev = NULL;
-	bool use_dio;
+	struct inode *inode = lo->lo_backing_file->f_mapping->host;
 
-	if (S_ISBLK(inode->i_mode))
-		backing_bdev = I_BDEV(inode);
-	else if (inode->i_sb->s_bdev)
-		backing_bdev = inode->i_sb->s_bdev;
+	if (!(lo->lo_backing_file->f_mode & FMODE_CAN_ODIRECT))
+		return false;
 
-	use_dio = dio && (file->f_mode & FMODE_CAN_ODIRECT) &&
-		(!backing_bdev || lo_bdev_can_use_dio(lo, backing_bdev));
+	if (S_ISBLK(inode->i_mode))
+		return lo_bdev_can_use_dio(lo, I_BDEV(inode));
+	if (inode->i_sb->s_bdev)
+		return lo_bdev_can_use_dio(lo, inode->i_sb->s_bdev);
+	return true;
+}
 
-	if (lo->use_dio == use_dio)
-		return;
+/*
+ * Direct I/O can be enabled either by using an O_DIRECT file descriptor, or by
+ * passing in the LO_FLAGS_DIRECT_IO flag from userspace.  It will be silently
+ * disabled when the device block size is too small or the offset is unaligned.
+ *
+ * loop_get_status will always report the effective LO_FLAGS_DIRECT_IO flag and
+ * not the originally passed in one.
+ */
+static inline void loop_update_dio(struct loop_device *lo)
+{
+	bool dio_in_use = lo->lo_flags & LO_FLAGS_DIRECT_IO;
 
-	/* flush dirty pages before changing direct IO */
-	vfs_fsync(file, 0);
+	lockdep_assert_held(&lo->lo_mutex);
+	WARN_ON_ONCE(lo->lo_state == Lo_bound &&
+		     lo->lo_queue->mq_freeze_depth == 0);
 
-	/*
-	 * The flag of LO_FLAGS_DIRECT_IO is handled similarly with
-	 * LO_FLAGS_READ_ONLY, both are set from kernel, and losetup
-	 * will get updated by ioctl(LOOP_GET_STATUS)
-	 */
-	if (lo->lo_state == Lo_bound)
-		blk_mq_freeze_queue(lo->lo_queue);
-	lo->use_dio = use_dio;
-	if (use_dio) {
-		blk_queue_flag_clear(QUEUE_FLAG_NOMERGES, lo->lo_queue);
+	if (lo->lo_backing_file->f_flags & O_DIRECT)
 		lo->lo_flags |= LO_FLAGS_DIRECT_IO;
-	} else {
-		blk_queue_flag_set(QUEUE_FLAG_NOMERGES, lo->lo_queue);
+	if ((lo->lo_flags & LO_FLAGS_DIRECT_IO) && !lo_can_use_dio(lo))
 		lo->lo_flags &= ~LO_FLAGS_DIRECT_IO;
-	}
-	if (lo->lo_state == Lo_bound)
-		blk_mq_unfreeze_queue(lo->lo_queue);
+
+	/* flush dirty pages before starting to issue direct I/O */
+	if ((lo->lo_flags & LO_FLAGS_DIRECT_IO) && !dio_in_use)
+		vfs_fsync(lo->lo_backing_file, 0);
 }
 
 /**
@@ -302,6 +301,28 @@ static int lo_read_simple(struct loop_device *lo, struct request *rq,
 	return 0;
 }
 
+static void loop_clear_limits(struct loop_device *lo, int mode)
+{
+	struct queue_limits lim = queue_limits_start_update(lo->lo_queue);
+
+	if (mode & FALLOC_FL_ZERO_RANGE)
+		lim.max_write_zeroes_sectors = 0;
+
+	if (mode & FALLOC_FL_PUNCH_HOLE) {
+		lim.max_hw_discard_sectors = 0;
+		lim.discard_granularity = 0;
+	}
+
+	/*
+	 * XXX: this updates the queue limits without freezing the queue, which
+	 * is against the locking protocol and dangerous.  But we can't just
+	 * freeze the queue as we're inside the ->queue_rq method here.  So this
+	 * should move out into a workqueue unless we get the file operations to
+	 * advertise if they support specific fallocate operations.
+	 */
+	queue_limits_commit_update(lo->lo_queue, &lim);
+}
+
 static int lo_fallocate(struct loop_device *lo, struct request *rq, loff_t pos,
 			int mode)
 {
@@ -320,6 +341,14 @@ static int lo_fallocate(struct loop_device *lo, struct request *rq, loff_t pos,
 	ret = file->f_op->fallocate(file, mode, pos, blk_rq_bytes(rq));
 	if (unlikely(ret && ret != -EINVAL && ret != -EOPNOTSUPP))
 		return -EIO;
+
+	/*
+	 * We initially configure the limits in a hope that fallocate is
+	 * supported and clear them here if that turns out not to be true.
+	 */
+	if (unlikely(ret == -EOPNOTSUPP))
+		loop_clear_limits(lo, mode);
+
 	return ret;
 }
 
@@ -445,9 +474,9 @@ static int lo_rw_aio(struct loop_device *lo, struct loop_cmd *cmd,
 	cmd->iocb.ki_ioprio = IOPRIO_PRIO_VALUE(IOPRIO_CLASS_NONE, 0);
 
 	if (rw == ITER_SOURCE)
-		ret = call_write_iter(file, &cmd->iocb, &iter);
+		ret = file->f_op->write_iter(&cmd->iocb, &iter);
 	else
-		ret = call_read_iter(file, &cmd->iocb, &iter);
+		ret = file->f_op->read_iter(&cmd->iocb, &iter);
 
 	lo_rw_aio_do_completion(cmd);
 
@@ -500,12 +529,6 @@ static int do_req_filebacked(struct loop_device *lo, struct request *rq)
 	}
 }
 
-static inline void loop_update_dio(struct loop_device *lo)
-{
-	__loop_update_dio(lo, (lo->lo_backing_file->f_flags & O_DIRECT) |
-				lo->use_dio);
-}
-
 static void loop_reread_partitions(struct loop_device *lo)
 {
 	int rc;
@@ -563,6 +586,7 @@ static int loop_change_fd(struct loop_device *lo, struct block_device *bdev,
 {
 	struct file *file = fget(arg);
 	struct file *old_file;
+	unsigned int memflags;
 	int error;
 	bool partscan;
 	bool is_loop;
@@ -600,14 +624,14 @@ static int loop_change_fd(struct loop_device *lo, struct block_device *bdev,
 
 	/* and ... switch */
 	disk_force_media_change(lo->lo_disk);
-	blk_mq_freeze_queue(lo->lo_queue);
+	memflags = blk_mq_freeze_queue(lo->lo_queue);
 	mapping_set_gfp_mask(old_file->f_mapping, lo->old_gfp_mask);
 	lo->lo_backing_file = file;
 	lo->old_gfp_mask = mapping_gfp_mask(file->f_mapping);
 	mapping_set_gfp_mask(file->f_mapping,
 			     lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
 	loop_update_dio(lo);
-	blk_mq_unfreeze_queue(lo->lo_queue);
+	blk_mq_unfreeze_queue(lo->lo_queue, memflags);
 	partscan = lo->lo_flags & LO_FLAGS_PARTSCAN;
 	loop_global_unlock(lo, is_loop);
 
@@ -750,12 +774,12 @@ static void loop_sysfs_exit(struct loop_device *lo)
 				   &loop_attribute_group);
 }
 
-static void loop_config_discard(struct loop_device *lo)
+static void loop_get_discard_config(struct loop_device *lo,
+				    u32 *granularity, u32 *max_discard_sectors)
 {
 	struct file *file = lo->lo_backing_file;
 	struct inode *inode = file->f_mapping->host;
-	struct request_queue *q = lo->lo_queue;
-	u32 granularity, max_discard_sectors;
+	struct kstatfs sbuf;
 
 	/*
 	 * If the backing device is a block device, mirror its zeroing
@@ -765,38 +789,18 @@ static void loop_config_discard(struct loop_device *lo)
 	 * file-backed loop devices: discarded regions read back as zero.
 	 */
 	if (S_ISBLK(inode->i_mode)) {
-		struct request_queue *backingq = bdev_get_queue(I_BDEV(inode));
+		struct block_device *bdev = I_BDEV(inode);
 
-		max_discard_sectors = backingq->limits.max_write_zeroes_sectors;
-		granularity = bdev_discard_granularity(I_BDEV(inode)) ?:
-			queue_physical_block_size(backingq);
+		*max_discard_sectors = bdev_write_zeroes_sectors(bdev);
+		*granularity = bdev_discard_granularity(bdev);
 
 	/*
 	 * We use punch hole to reclaim the free space used by the
 	 * image a.k.a. discard.
 	 */
-	} else if (!file->f_op->fallocate) {
-		max_discard_sectors = 0;
-		granularity = 0;
-
-	} else {
-		struct kstatfs sbuf;
-
-		max_discard_sectors = UINT_MAX >> 9;
-		if (!vfs_statfs(&file->f_path, &sbuf))
-			granularity = sbuf.f_bsize;
-		else
-			max_discard_sectors = 0;
-	}
-
-	if (max_discard_sectors) {
-		q->limits.discard_granularity = granularity;
-		blk_queue_max_discard_sectors(q, max_discard_sectors);
-		blk_queue_max_write_zeroes_sectors(q, max_discard_sectors);
-	} else {
-		q->limits.discard_granularity = 0;
-		blk_queue_max_discard_sectors(q, 0);
-		blk_queue_max_write_zeroes_sectors(q, 0);
+	} else if (file->f_op->fallocate && !vfs_statfs(&file->f_path, &sbuf)) {
+		*max_discard_sectors = UINT_MAX >> 9;
+		*granularity = sbuf.f_bsize;
 	}
 }
 
@@ -927,24 +931,6 @@ static void loop_free_idle_workers_timer(struct timer_list *timer)
 	return loop_free_idle_workers(lo, false);
 }
 
-static void loop_update_rotational(struct loop_device *lo)
-{
-	struct file *file = lo->lo_backing_file;
-	struct inode *file_inode = file->f_mapping->host;
-	struct block_device *file_bdev = file_inode->i_sb->s_bdev;
-	struct request_queue *q = lo->lo_queue;
-	bool nonrot = true;
-
-	/* not all filesystems (e.g. tmpfs) have a sb->s_bdev */
-	if (file_bdev)
-		nonrot = bdev_nonrot(file_bdev);
-
-	if (nonrot)
-		blk_queue_flag_set(QUEUE_FLAG_NONROT, q);
-	else
-		blk_queue_flag_clear(QUEUE_FLAG_NONROT, q);
-}
-
 /**
  * loop_set_status_from_info - configure device from loop_info
  * @lo: struct loop_device to configure
@@ -982,21 +968,62 @@ loop_set_status_from_info(struct loop_device *lo,
 
 	memcpy(lo->lo_file_name, info->lo_file_name, LO_NAME_SIZE);
 	lo->lo_file_name[LO_NAME_SIZE-1] = 0;
-	lo->lo_flags = info->lo_flags;
 	return 0;
 }
 
+static unsigned int loop_default_blocksize(struct loop_device *lo,
+		struct block_device *backing_bdev)
+{
+	/* In case of direct I/O, match underlying block size */
+	if ((lo->lo_backing_file->f_flags & O_DIRECT) && backing_bdev)
+		return bdev_logical_block_size(backing_bdev);
+	return SECTOR_SIZE;
+}
+
+static void loop_update_limits(struct loop_device *lo, struct queue_limits *lim,
+		unsigned int bsize)
+{
+	struct file *file = lo->lo_backing_file;
+	struct inode *inode = file->f_mapping->host;
+	struct block_device *backing_bdev = NULL;
+	u32 granularity = 0, max_discard_sectors = 0;
+
+	if (S_ISBLK(inode->i_mode))
+		backing_bdev = I_BDEV(inode);
+	else if (inode->i_sb->s_bdev)
+		backing_bdev = inode->i_sb->s_bdev;
+
+	if (!bsize)
+		bsize = loop_default_blocksize(lo, backing_bdev);
+
+	loop_get_discard_config(lo, &granularity, &max_discard_sectors);
+
+	lim->logical_block_size = bsize;
+	lim->physical_block_size = bsize;
+	lim->io_min = bsize;
+	lim->features &= ~(BLK_FEAT_WRITE_CACHE | BLK_FEAT_ROTATIONAL);
+	if (file->f_op->fsync && !(lo->lo_flags & LO_FLAGS_READ_ONLY))
+		lim->features |= BLK_FEAT_WRITE_CACHE;
+	if (backing_bdev && !bdev_nonrot(backing_bdev))
+		lim->features |= BLK_FEAT_ROTATIONAL;
+	lim->max_hw_discard_sectors = max_discard_sectors;
+	lim->max_write_zeroes_sectors = max_discard_sectors;
+	if (max_discard_sectors)
+		lim->discard_granularity = granularity;
+	else
+		lim->discard_granularity = 0;
+}
+
 static int loop_configure(struct loop_device *lo, blk_mode_t mode,
 			  struct block_device *bdev,
 			  const struct loop_config *config)
 {
 	struct file *file = fget(config->fd);
-	struct inode *inode;
 	struct address_space *mapping;
+	struct queue_limits lim;
 	int error;
 	loff_t size;
 	bool partscan;
-	unsigned short bsize;
 	bool is_loop;
 
 	if (!file)
@@ -1029,22 +1056,16 @@ static int loop_configure(struct loop_device *lo, blk_mode_t mode,
 		goto out_unlock;
 
 	mapping = file->f_mapping;
-	inode = mapping->host;
 
 	if ((config->info.lo_flags & ~LOOP_CONFIGURE_SETTABLE_FLAGS) != 0) {
 		error = -EINVAL;
 		goto out_unlock;
 	}
 
-	if (config->block_size) {
-		error = blk_validate_block_size(config->block_size);
-		if (error)
-			goto out_unlock;
-	}
-
 	error = loop_set_status_from_info(lo, &config->info);
 	if (error)
 		goto out_unlock;
+	lo->lo_flags = config->info.lo_flags;
 
 	if (!(file->f_mode & FMODE_WRITE) || !(mode & BLK_OPEN_WRITE) ||
 	    !file->f_op->write_iter)
@@ -1066,29 +1087,18 @@ static int loop_configure(struct loop_device *lo, blk_mode_t mode,
 	disk_force_media_change(lo->lo_disk);
 	set_disk_ro(lo->lo_disk, (lo->lo_flags & LO_FLAGS_READ_ONLY) != 0);
 
-	lo->use_dio = lo->lo_flags & LO_FLAGS_DIRECT_IO;
 	lo->lo_device = bdev;
 	lo->lo_backing_file = file;
 	lo->old_gfp_mask = mapping_gfp_mask(mapping);
 	mapping_set_gfp_mask(mapping, lo->old_gfp_mask & ~(__GFP_IO|__GFP_FS));
 
-	if (!(lo->lo_flags & LO_FLAGS_READ_ONLY) && file->f_op->fsync)
-		blk_queue_write_cache(lo->lo_queue, true, false);
-
-	if (config->block_size)
-		bsize = config->block_size;
-	else if ((lo->lo_backing_file->f_flags & O_DIRECT) && inode->i_sb->s_bdev)
-		/* In case of direct I/O, match underlying block size */
-		bsize = bdev_logical_block_size(inode->i_sb->s_bdev);
-	else
-		bsize = 512;
-
-	blk_queue_logical_block_size(lo->lo_queue, bsize);
-	blk_queue_physical_block_size(lo->lo_queue, bsize);
-	blk_queue_io_min(lo->lo_queue, bsize);
+	lim = queue_limits_start_update(lo->lo_queue);
+	loop_update_limits(lo, &lim, config->block_size);
+	/* No need to freeze the queue as the device isn't bound yet. */
+	error = queue_limits_commit_update(lo->lo_queue, &lim);
+	if (error)
+		goto out_unlock;
 
-	loop_config_discard(lo);
-	loop_update_rotational(lo);
 	loop_update_dio(lo);
 	loop_sysfs_init(lo);
 
@@ -1129,22 +1139,12 @@ out_putf:
 	return error;
 }
 
-static void __loop_clr_fd(struct loop_device *lo, bool release)
+static void __loop_clr_fd(struct loop_device *lo)
 {
+	struct queue_limits lim;
 	struct file *filp;
 	gfp_t gfp = lo->old_gfp_mask;
 
-	if (test_bit(QUEUE_FLAG_WC, &lo->lo_queue->queue_flags))
-		blk_queue_write_cache(lo->lo_queue, false, false);
-
-	/*
-	 * Freeze the request queue when unbinding on a live file descriptor and
-	 * thus an open device.  When called from ->release we are guaranteed
-	 * that there is no I/O in progress already.
-	 */
-	if (!release)
-		blk_mq_freeze_queue(lo->lo_queue);
-
 	spin_lock_irq(&lo->lo_lock);
 	filp = lo->lo_backing_file;
 	lo->lo_backing_file = NULL;
@@ -1154,9 +1154,19 @@ static void __loop_clr_fd(struct loop_device *lo, bool release)
 	lo->lo_offset = 0;
 	lo->lo_sizelimit = 0;
 	memset(lo->lo_file_name, 0, LO_NAME_SIZE);
-	blk_queue_logical_block_size(lo->lo_queue, 512);
-	blk_queue_physical_block_size(lo->lo_queue, 512);
-	blk_queue_io_min(lo->lo_queue, 512);
+
+	/*
+	 * Reset the block size to the default.
+	 *
+	 * No queue freezing needed because this is called from the final
+	 * ->release call only, so there can't be any outstanding I/O.
+	 */
+	lim = queue_limits_start_update(lo->lo_queue);
+	lim.logical_block_size = SECTOR_SIZE;
+	lim.physical_block_size = SECTOR_SIZE;
+	lim.io_min = SECTOR_SIZE;
+	queue_limits_commit_update(lo->lo_queue, &lim);
+
 	invalidate_disk(lo->lo_disk);
 	loop_sysfs_exit(lo);
 	/* let user-space know about this change */
@@ -1164,8 +1174,6 @@ static void __loop_clr_fd(struct loop_device *lo, bool release)
 	mapping_set_gfp_mask(filp->f_mapping, gfp);
 	/* This is safe: open() is still holding a reference. */
 	module_put(THIS_MODULE);
-	if (!release)
-		blk_mq_unfreeze_queue(lo->lo_queue);
 
 	disk_force_media_change(lo->lo_disk);
 
@@ -1180,11 +1188,7 @@ static void __loop_clr_fd(struct loop_device *lo, bool release)
 		 * must be at least one and it can only become zero when the
 		 * current holder is released.
 		 */
-		if (!release)
-			mutex_lock(&lo->lo_disk->open_mutex);
 		err = bdev_disk_changed(lo->lo_disk, false);
-		if (!release)
-			mutex_unlock(&lo->lo_disk->open_mutex);
 		if (err)
 			pr_warn("%s: partition scan of loop%d failed (rc=%d)\n",
 				__func__, lo->lo_number, err);
@@ -1233,24 +1237,16 @@ static int loop_clr_fd(struct loop_device *lo)
 		return -ENXIO;
 	}
 	/*
-	 * If we've explicitly asked to tear down the loop device,
-	 * and it has an elevated reference count, set it for auto-teardown when
-	 * the last reference goes away. This stops $!~#$@ udev from
-	 * preventing teardown because it decided that it needs to run blkid on
-	 * the loopback device whenever they appear. xfstests is notorious for
-	 * failing tests because blkid via udev races with a losetup
-	 * <dev>/do something like mkfs/losetup -d <dev> causing the losetup -d
-	 * command to fail with EBUSY.
+	 * Mark the device for removing the backing device on last close.
+	 * If we are the only opener, also switch the state to roundown here to
+	 * prevent new openers from coming in.
 	 */
-	if (disk_openers(lo->lo_disk) > 1) {
-		lo->lo_flags |= LO_FLAGS_AUTOCLEAR;
-		loop_global_unlock(lo, true);
-		return 0;
-	}
-	lo->lo_state = Lo_rundown;
+
+	lo->lo_flags |= LO_FLAGS_AUTOCLEAR;
+	if (disk_openers(lo->lo_disk) == 1)
+		lo->lo_state = Lo_rundown;
 	loop_global_unlock(lo, true);
 
-	__loop_clr_fd(lo, false);
 	return 0;
 }
 
@@ -1258,9 +1254,9 @@ static int
 loop_set_status(struct loop_device *lo, const struct loop_info64 *info)
 {
 	int err;
-	int prev_lo_flags;
 	bool partscan = false;
 	bool size_changed = false;
+	unsigned int memflags;
 
 	err = mutex_lock_killable(&lo->lo_mutex);
 	if (err)
@@ -1277,21 +1273,18 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info)
 		invalidate_bdev(lo->lo_device);
 	}
 
-	/* I/O need to be drained during transfer transition */
-	blk_mq_freeze_queue(lo->lo_queue);
-
-	prev_lo_flags = lo->lo_flags;
+	/* I/O needs to be drained before changing lo_offset or lo_sizelimit */
+	memflags = blk_mq_freeze_queue(lo->lo_queue);
 
 	err = loop_set_status_from_info(lo, info);
 	if (err)
 		goto out_unfreeze;
 
-	/* Mask out flags that can't be set using LOOP_SET_STATUS. */
-	lo->lo_flags &= LOOP_SET_STATUS_SETTABLE_FLAGS;
-	/* For those flags, use the previous values instead */
-	lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_SETTABLE_FLAGS;
-	/* For flags that can't be cleared, use previous values too */
-	lo->lo_flags |= prev_lo_flags & ~LOOP_SET_STATUS_CLEARABLE_FLAGS;
+	partscan = !(lo->lo_flags & LO_FLAGS_PARTSCAN) &&
+		(info->lo_flags & LO_FLAGS_PARTSCAN);
+
+	lo->lo_flags &= ~LOOP_SET_STATUS_CLEARABLE_FLAGS;
+	lo->lo_flags |= (info->lo_flags & LOOP_SET_STATUS_SETTABLE_FLAGS);
 
 	if (size_changed) {
 		loff_t new_size = get_size(lo->lo_offset, lo->lo_sizelimit,
@@ -1299,17 +1292,13 @@ loop_set_status(struct loop_device *lo, const struct loop_info64 *info)
 		loop_set_size(lo, new_size);
 	}
 
-	/* update dio if lo_offset or transfer is changed */
-	__loop_update_dio(lo, lo->use_dio);
+	/* update the direct I/O flag if lo_offset changed */
+	loop_update_dio(lo);
 
 out_unfreeze:
-	blk_mq_unfreeze_queue(lo->lo_queue);
-
-	if (!err && (lo->lo_flags & LO_FLAGS_PARTSCAN) &&
-	     !(prev_lo_flags & LO_FLAGS_PARTSCAN)) {
+	blk_mq_unfreeze_queue(lo->lo_queue, memflags);
+	if (partscan)
 		clear_bit(GD_SUPPRESS_PART_SCAN, &lo->lo_disk->state);
-		partscan = true;
-	}
 out_unlock:
 	mutex_unlock(&lo->lo_mutex);
 	if (partscan)
@@ -1458,41 +1447,52 @@ static int loop_set_capacity(struct loop_device *lo)
 
 static int loop_set_dio(struct loop_device *lo, unsigned long arg)
 {
-	int error = -ENXIO;
-	if (lo->lo_state != Lo_bound)
-		goto out;
+	bool use_dio = !!arg;
+	unsigned int memflags;
 
-	__loop_update_dio(lo, !!arg);
-	if (lo->use_dio == !!arg)
+	if (lo->lo_state != Lo_bound)
+		return -ENXIO;
+	if (use_dio == !!(lo->lo_flags & LO_FLAGS_DIRECT_IO))
 		return 0;
-	error = -EINVAL;
- out:
-	return error;
+
+	if (use_dio) {
+		if (!lo_can_use_dio(lo))
+			return -EINVAL;
+		/* flush dirty pages before starting to use direct I/O */
+		vfs_fsync(lo->lo_backing_file, 0);
+	}
+
+	memflags = blk_mq_freeze_queue(lo->lo_queue);
+	if (use_dio)
+		lo->lo_flags |= LO_FLAGS_DIRECT_IO;
+	else
+		lo->lo_flags &= ~LO_FLAGS_DIRECT_IO;
+	blk_mq_unfreeze_queue(lo->lo_queue, memflags);
+	return 0;
 }
 
 static int loop_set_block_size(struct loop_device *lo, unsigned long arg)
 {
+	struct queue_limits lim;
+	unsigned int memflags;
 	int err = 0;
 
 	if (lo->lo_state != Lo_bound)
 		return -ENXIO;
 
-	err = blk_validate_block_size(arg);
-	if (err)
-		return err;
-
 	if (lo->lo_queue->limits.logical_block_size == arg)
 		return 0;
 
 	sync_blockdev(lo->lo_device);
 	invalidate_bdev(lo->lo_device);
 
-	blk_mq_freeze_queue(lo->lo_queue);
-	blk_queue_logical_block_size(lo->lo_queue, arg);
-	blk_queue_physical_block_size(lo->lo_queue, arg);
-	blk_queue_io_min(lo->lo_queue, arg);
+	lim = queue_limits_start_update(lo->lo_queue);
+	loop_update_limits(lo, &lim, arg);
+
+	memflags = blk_mq_freeze_queue(lo->lo_queue);
+	err = queue_limits_commit_update(lo->lo_queue, &lim);
 	loop_update_dio(lo);
-	blk_mq_unfreeze_queue(lo->lo_queue);
+	blk_mq_unfreeze_queue(lo->lo_queue, memflags);
 
 	return err;
 }
@@ -1719,25 +1719,43 @@ static int lo_compat_ioctl(struct block_device *bdev, blk_mode_t mode,
 }
 #endif
 
+static int lo_open(struct gendisk *disk, blk_mode_t mode)
+{
+	struct loop_device *lo = disk->private_data;
+	int err;
+
+	err = mutex_lock_killable(&lo->lo_mutex);
+	if (err)
+		return err;
+
+	if (lo->lo_state == Lo_deleting || lo->lo_state == Lo_rundown)
+		err = -ENXIO;
+	mutex_unlock(&lo->lo_mutex);
+	return err;
+}
+
 static void lo_release(struct gendisk *disk)
 {
 	struct loop_device *lo = disk->private_data;
+	bool need_clear = false;
 
 	if (disk_openers(disk) > 0)
 		return;
+	/*
+	 * Clear the backing device information if this is the last close of
+	 * a device that's been marked for auto clear, or on which LOOP_CLR_FD
+	 * has been called.
+	 */
 
 	mutex_lock(&lo->lo_mutex);
-	if (lo->lo_state == Lo_bound && (lo->lo_flags & LO_FLAGS_AUTOCLEAR)) {
+	if (lo->lo_state == Lo_bound && (lo->lo_flags & LO_FLAGS_AUTOCLEAR))
 		lo->lo_state = Lo_rundown;
-		mutex_unlock(&lo->lo_mutex);
-		/*
-		 * In autoclear mode, stop the loop thread
-		 * and remove configuration after last close.
-		 */
-		__loop_clr_fd(lo, true);
-		return;
-	}
+
+	need_clear = (lo->lo_state == Lo_rundown);
 	mutex_unlock(&lo->lo_mutex);
+
+	if (need_clear)
+		__loop_clr_fd(lo);
 }
 
 static void lo_free_disk(struct gendisk *disk)
@@ -1754,6 +1772,7 @@ static void lo_free_disk(struct gendisk *disk)
 
 static const struct block_device_operations lo_fops = {
 	.owner =	THIS_MODULE,
+	.open =         lo_open,
 	.release =	lo_release,
 	.ioctl =	lo_ioctl,
 #ifdef CONFIG_COMPAT
@@ -1832,6 +1851,7 @@ static const struct kernel_param_ops loop_hw_qdepth_param_ops = {
 device_param_cb(hw_queue_depth, &loop_hw_qdepth_param_ops, &hw_queue_depth, 0444);
 MODULE_PARM_DESC(hw_queue_depth, "Queue depth for each hardware queue. Default: " __stringify(LOOP_DEFAULT_HW_Q_DEPTH));
 
+MODULE_DESCRIPTION("Loopback device support");
 MODULE_LICENSE("GPL");
 MODULE_ALIAS_BLOCKDEV_MAJOR(LOOP_MAJOR);
 
@@ -1854,7 +1874,7 @@ static blk_status_t loop_queue_rq(struct blk_mq_hw_ctx *hctx,
 		cmd->use_aio = false;
 		break;
 	default:
-		cmd->use_aio = lo->use_dio;
+		cmd->use_aio = lo->lo_flags & LO_FLAGS_DIRECT_IO;
 		break;
 	}
 
@@ -1982,6 +2002,12 @@ static const struct blk_mq_ops loop_mq_ops = {
 
 static int loop_add(int i)
 {
+	struct queue_limits lim = {
+		/*
+		 * Random number picked from the historic block max_sectors cap.
+		 */
+		.max_hw_sectors		= 2560u,
+	};
 	struct loop_device *lo;
 	struct gendisk *disk;
 	int err;
@@ -2017,32 +2043,20 @@ static int loop_add(int i)
 	lo->tag_set.queue_depth = hw_queue_depth;
 	lo->tag_set.numa_node = NUMA_NO_NODE;
 	lo->tag_set.cmd_size = sizeof(struct loop_cmd);
-	lo->tag_set.flags = BLK_MQ_F_SHOULD_MERGE | BLK_MQ_F_STACKING |
-		BLK_MQ_F_NO_SCHED_BY_DEFAULT;
+	lo->tag_set.flags = BLK_MQ_F_STACKING | BLK_MQ_F_NO_SCHED_BY_DEFAULT;
 	lo->tag_set.driver_data = lo;
 
 	err = blk_mq_alloc_tag_set(&lo->tag_set);
 	if (err)
 		goto out_free_idr;
 
-	disk = lo->lo_disk = blk_mq_alloc_disk(&lo->tag_set, lo);
+	disk = lo->lo_disk = blk_mq_alloc_disk(&lo->tag_set, &lim, lo);
 	if (IS_ERR(disk)) {
 		err = PTR_ERR(disk);
 		goto out_cleanup_tags;
 	}
 	lo->lo_queue = lo->lo_disk->queue;
 
-	/* random number picked from the history block max_sectors cap */
-	blk_queue_max_hw_sectors(lo->lo_queue, 2560u);
-
-	/*
-	 * By default, we do buffer IO, so it doesn't make sense to enable
-	 * merge because the I/O submitted to backing file is handled page by
-	 * page. For directio mode, merge does help to dispatch bigger request
-	 * to underlayer disk. We will enable merge once directio is enabled.
-	 */
-	blk_queue_flag_set(QUEUE_FLAG_NOMERGES, lo->lo_queue);
-
 	/*
 	 * Disable partition scanning by default. The in-kernel partition
 	 * scanning can be requested individually per-device during its