s390/dcssblk: fix deadlock when adding a DCSS

After the rework from commit 1ebe2e5f9d68 ("block: remove
GENHD_FL_EXT_DEVT"), when calling device_add_disk(), dcssblk will end up
in disk_scan_partitions(), and not break out early w/o GENHD_FL_NO_PART.
This will trigger implicit open/release via blkdev_get/put_whole()
later. dcssblk_release() will then deadlock on dcssblk_devices_sem
semaphore, which is already held from dcssblk_add_store() when calling
device_add_disk().

dcssblk does not support partitions (DCSSBLK_MINORS_PER_DISK == 1), and
never scanned partitions before. Therefore restore the previous
behavior, and explicitly disallow partition scanning by setting the
GENHD_FL_NO_PART flag. This will also prevent this deadlock scenario.

Fixes: 1ebe2e5f9d68 ("block: remove GENHD_FL_EXT_DEVT")
Cc: <stable@vger.kernel.org> # 5.17+
Signed-off-by: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Acked-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>

diff --git a/drivers/s390/block/dcssblk.c b/drivers/s390/block/dcssblk.c
index 93b80da..b392b9f 100644 (file)
--- a/drivers/s390/block/dcssblk.c
+++ b/drivers/s390/block/dcssblk.c
@@ -636,6 +636,7 @@ dcssblk_add_store(struct device *dev, struct device_attribute *attr, const char
        dev_info->gd->minors = DCSSBLK_MINORS_PER_DISK;
        dev_info->gd->fops = &dcssblk_devops;
        dev_info->gd->private_data = dev_info;
+       dev_info->gd->flags |= GENHD_FL_NO_PART;
        blk_queue_logical_block_size(dev_info->gd->queue, 4096);
        blk_queue_flag_set(QUEUE_FLAG_DAX, dev_info->gd->queue);