On Wed, Jul 28, 2021 at 02:56:48PM +0800, Hou Tao wrote:
Hi,
Thanks for your review.
Now I am not sure whether or not it is a better idea that doing synchronous or asynchronous del_gendisk()
according to NBD_CFLAG_DESTROY_ON_DISCONNECT, and will wait for some feedback from Josef.
The other option would be to add a variant of del_gendisk that does
expect the open_mutex to be held over the whole operation. But I'd be
happy if we could avoid that.