block: fix locking for struct block_device size updates

Two different callers use two different mutexes for updating the
block device size, which obviously doesn't help to actually protect
against concurrent updates from the different callers.  In addition
one of the locks, bd_mutex is rather prone to deadlocks with other
parts of the block stack that use it for high level synchronization.

Switch to using a new spinlock protecting just the size updates, as
that is all we need, and make sure everyone does the update through
the proper helper.

This fixes a bug reported with the nvme revalidating disks during a
hot removal operation, which can currently deadlock on bd_mutex.

Reported-by: Xianting Tian <xianting_tian@126.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>

1 files changed, 14 insertions, 11 deletions

diff --git a/fs/block_dev.c b/fs/block_dev.c
index f52597172c8b..08158bb2e76c 100644
--- a/fs/block_dev.c
+++ b/fs/block_dev.c
@@ -876,6 +876,7 @@ struct block_device *bdget(dev_t dev)
 	bdev = &BDEV_I(inode)->bdev;
 
 	if (inode->i_state & I_NEW) {
+		spin_lock_init(&bdev->bd_size_lock);
 		bdev->bd_contains = NULL;
 		bdev->bd_super = NULL;
 		bdev->bd_inode = inode;
@@ -1290,6 +1291,7 @@ static void check_disk_size_change(struct gendisk *disk,
 {
 	loff_t disk_size, bdev_size;
 
+	spin_lock(&bdev->bd_size_lock);
 	disk_size = (loff_t)get_capacity(disk) << 9;
 	bdev_size = i_size_read(bdev->bd_inode);
 	if (disk_size != bdev_size) {
@@ -1299,11 +1301,15 @@ static void check_disk_size_change(struct gendisk *disk,
 			       disk->disk_name, bdev_size, disk_size);
 		}
 		i_size_write(bdev->bd_inode, disk_size);
-		if (bdev_size > disk_size && __invalidate_device(bdev, false))
+	}
+	bdev->bd_invalidated = 0;
+	spin_unlock(&bdev->bd_size_lock);
+
+	if (bdev_size > disk_size) {
+		if (__invalidate_device(bdev, false))
 			pr_warn("VFS: busy inodes on resized disk %s\n",
 				disk->disk_name);
 	}
-	bdev->bd_invalidated = 0;
 }
 
 /**
@@ -1328,13 +1334,10 @@ int revalidate_disk(struct gendisk *disk)
 	if (!(disk->flags & GENHD_FL_HIDDEN)) {
 		struct block_device *bdev = bdget_disk(disk, 0);
 
-		if (!bdev)
-			return ret;
-
-		mutex_lock(&bdev->bd_mutex);
-		check_disk_size_change(disk, bdev, ret == 0);
-		mutex_unlock(&bdev->bd_mutex);
-		bdput(bdev);
+		if (bdev) {
+			check_disk_size_change(disk, bdev, ret == 0);
+			bdput(bdev);
+		}
 	}
 	return ret;
 }
@@ -1373,9 +1376,9 @@ EXPORT_SYMBOL(check_disk_change);
 
 void bd_set_nr_sectors(struct block_device *bdev, sector_t sectors)
 {
-	inode_lock(bdev->bd_inode);
+	spin_lock(&bdev->bd_size_lock);
 	i_size_write(bdev->bd_inode, (loff_t)sectors << SECTOR_SHIFT);
-	inode_unlock(bdev->bd_inode);
+	spin_unlock(&bdev->bd_size_lock);
 }
 EXPORT_SYMBOL(bd_set_nr_sectors);