dm: track per-add_disk holder relations in DM

dm is a bit special in that it opens the underlying devices.  Commit
89f871af1b26 ("dm: delay registering the gendisk") tried to accommodate
that by allowing to add the holder to the list before add_gendisk and
then just add them to sysfs once add_disk is called.  But that leads to
really odd lifetime problems and error handling problems as we can't
know the state of the kobjects and don't unwind properly.  To fix this
switch to just registering all existing table_devices with the holder
code right after add_disk, and remove them before calling del_gendisk.

Fixes: 89f871af1b26 ("dm: delay registering the gendisk")
Reported-by: Yu Kuai <yukuai3@huawei.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Mike Snitzer <snitzer@kernel.org>
Link: https://lore.kernel.org/r/20221115141054.1051801-7-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>

1 files changed, 39 insertions, 10 deletions

diff --git a/drivers/md/dm.c b/drivers/md/dm.c
index 3728b56b364b..e1ea3a7bd9d9 100644
--- a/drivers/md/dm.c
+++ b/drivers/md/dm.c
@@ -751,9 +751,16 @@ static struct table_device *open_table_device(struct mapped_device *md,
 		goto out_free_td;
 	}
 
-	r = bd_link_disk_holder(bdev, dm_disk(md));
-	if (r)
-		goto out_blkdev_put;
+	/*
+	 * We can be called before the dm disk is added.  In that case we can't
+	 * register the holder relation here.  It will be done once add_disk was
+	 * called.
+	 */
+	if (md->disk->slave_dir) {
+		r = bd_link_disk_holder(bdev, md->disk);
+		if (r)
+			goto out_blkdev_put;
+	}
 
 	td->dm_dev.mode = mode;
 	td->dm_dev.bdev = bdev;
@@ -774,7 +781,8 @@ out_free_td:
  */
 static void close_table_device(struct table_device *td, struct mapped_device *md)
 {
-	bd_unlink_disk_holder(td->dm_dev.bdev, dm_disk(md));
+	if (md->disk->slave_dir)
+		bd_unlink_disk_holder(td->dm_dev.bdev, md->disk);
 	blkdev_put(td->dm_dev.bdev, td->dm_dev.mode | FMODE_EXCL);
 	put_dax(td->dm_dev.dax_dev);
 	list_del(&td->list);
@@ -1951,7 +1959,13 @@ static void cleanup_mapped_device(struct mapped_device *md)
 		md->disk->private_data = NULL;
 		spin_unlock(&_minor_lock);
 		if (dm_get_md_type(md) != DM_TYPE_NONE) {
+			struct table_device *td;
+
 			dm_sysfs_exit(md);
+			list_for_each_entry(td, &md->table_devices, list) {
+				bd_unlink_disk_holder(td->dm_dev.bdev,
+						      md->disk);
+			}
 
 			/*
 			 * Hold lock to make sure del_gendisk() won't concurrent
@@ -2291,6 +2305,7 @@ int dm_setup_md_queue(struct mapped_device *md, struct dm_table *t)
 {
 	enum dm_queue_mode type = dm_table_get_type(t);
 	struct queue_limits limits;
+	struct table_device *td;
 	int r;
 
 	switch (type) {
@@ -2329,16 +2344,30 @@ int dm_setup_md_queue(struct mapped_device *md, struct dm_table *t)
 	if (r)
 		return r;
 
-	r = dm_sysfs_init(md);
-	if (r) {
-		mutex_lock(&md->table_devices_lock);
-		del_gendisk(md->disk);
-		mutex_unlock(&md->table_devices_lock);
-		return r;
+	/*
+	 * Register the holder relationship for devices added before the disk
+	 * was live.
+	 */
+	list_for_each_entry(td, &md->table_devices, list) {
+		r = bd_link_disk_holder(td->dm_dev.bdev, md->disk);
+		if (r)
+			goto out_undo_holders;
 	}
 
+	r = dm_sysfs_init(md);
+	if (r)
+		goto out_undo_holders;
+
 	md->type = type;
 	return 0;
+
+out_undo_holders:
+	list_for_each_entry_continue_reverse(td, &md->table_devices, list)
+		bd_unlink_disk_holder(td->dm_dev.bdev, md->disk);
+	mutex_lock(&md->table_devices_lock);
+	del_gendisk(md->disk);
+	mutex_unlock(&md->table_devices_lock);
+	return r;
 }
 
 struct mapped_device *dm_get_md(dev_t dev)