Delay dropping the block_devices for partitions in del_gendisk until after the call to blk_mark_disk_dead, so that we can implementat notification of removed devices in blk_mark_disk_dead. This requires splitting a lower-level drop_partition helper out of delete_partition and using that from del_gendisk, while having a common loop for the whole device and partitions that calls remove_inode_hash, fsync_bdev and __invalidate_device before the call to blk_mark_disk_dead. Signed-off-by: Christoph Hellwig <hch@xxxxxx> --- block/blk.h | 2 +- block/genhd.c | 24 +++++++++++++++++++----- block/partitions/core.c | 19 ++++++++++++------- 3 files changed, 32 insertions(+), 13 deletions(-) diff --git a/block/blk.h b/block/blk.h index 45547bcf111938..4363052f90416a 100644 --- a/block/blk.h +++ b/block/blk.h @@ -409,7 +409,7 @@ int bdev_add_partition(struct gendisk *disk, int partno, sector_t start, int bdev_del_partition(struct gendisk *disk, int partno); int bdev_resize_partition(struct gendisk *disk, int partno, sector_t start, sector_t length); -void blk_drop_partitions(struct gendisk *disk); +void drop_partition(struct block_device *part); void bdev_set_nr_sectors(struct block_device *bdev, sector_t sectors); diff --git a/block/genhd.c b/block/genhd.c index a744daeed55318..bd4c4eca31363e 100644 --- a/block/genhd.c +++ b/block/genhd.c @@ -615,6 +615,8 @@ EXPORT_SYMBOL_GPL(blk_mark_disk_dead); void del_gendisk(struct gendisk *disk) { struct request_queue *q = disk->queue; + struct block_device *part; + unsigned long idx; might_sleep(); @@ -623,16 +625,28 @@ void del_gendisk(struct gendisk *disk) disk_del_events(disk); + /* + * Prevent new openers by unlinked the bdev inode, and write out + * dirty data before marking the disk dead and stopping all I/O. + */ mutex_lock(&disk->open_mutex); - remove_inode_hash(disk->part0->bd_inode); - blk_drop_partitions(disk); + xa_for_each(&disk->part_tbl, idx, part) { + remove_inode_hash(part->bd_inode); + fsync_bdev(part); + __invalidate_device(part, true); + } mutex_unlock(&disk->open_mutex); - fsync_bdev(disk->part0); - __invalidate_device(disk->part0, true); - blk_mark_disk_dead(disk); + /* + * Drop all partitions now that the disk is marked dead. + */ + mutex_lock(&disk->open_mutex); + xa_for_each_start(&disk->part_tbl, idx, part, 1) + drop_partition(part); + mutex_unlock(&disk->open_mutex); + if (!(disk->flags & GENHD_FL_HIDDEN)) { sysfs_remove_link(&disk_to_dev(disk)->kobj, "bdi"); diff --git a/block/partitions/core.c b/block/partitions/core.c index fa5c707fe0ad2f..31ac815d77a83c 100644 --- a/block/partitions/core.c +++ b/block/partitions/core.c @@ -263,10 +263,19 @@ struct device_type part_type = { .uevent = part_uevent, }; -static void delete_partition(struct block_device *part) +void drop_partition(struct block_device *part) { lockdep_assert_held(&part->bd_disk->open_mutex); + xa_erase(&part->bd_disk->part_tbl, part->bd_partno); + kobject_put(part->bd_holder_dir); + + device_del(&part->bd_device); + put_device(&part->bd_device); +} + +static void delete_partition(struct block_device *part) +{ /* * Remove the block device from the inode hash, so that it cannot be * looked up any more even when openers still hold references. @@ -276,11 +285,7 @@ static void delete_partition(struct block_device *part) fsync_bdev(part); __invalidate_device(part, true); - xa_erase(&part->bd_disk->part_tbl, part->bd_partno); - kobject_put(part->bd_holder_dir); - device_del(&part->bd_device); - - put_device(&part->bd_device); + drop_partition(part); } static ssize_t whole_disk_show(struct device *dev, @@ -519,7 +524,7 @@ static bool disk_unlock_native_capacity(struct gendisk *disk) return true; } -void blk_drop_partitions(struct gendisk *disk) +static void blk_drop_partitions(struct gendisk *disk) { struct block_device *part; unsigned long idx; -- 2.39.2