On 2019/8/30 8:58, Xiubo Li wrote:
On 2019/8/30 7:49, Mike Christie wrote:On 08/22/2019 02:59 AM, xiubli@redhat.com wrote:
[...]
+ test_bit(NBD_DISCONNECT_REQUESTED, &nbd->flags)) { + mutex_unlock(&nbd_index_mutex); + + /* Wait untill the recv_work exit */If that is all you need we could do a flush_workqueue(nbd->recv_workq) (you would need Jens's for next branch which has some work queue changes in nbd).Yeah, this makes sense.
This has already been done in nbd_disconnect_and_put() in the Jen's for next branch code. So here it will make no sense.
I will rebase this patch set to that branch. Thanks. BRs Xiubo