On Tue, Oct 25, 2022 at 11:43:19AM -0600, Keith Busch wrote: > We still need the flush_work(scan_work) prior to killing the queues. Yes. > It > looks like it could safely be moved to nvme_stop_ctrl(), which might > make it easier on everyone if it were there. Hmm. As Sagi pointed out this could become a bit more complicated due to multipath, but in general I'd love to bring some more regularity into the teardown path.
