On Sun, Mar 26, 2023 at 04:01:46PM +0300, Sagi Grimberg wrote: > On 3/25/23 00:28, Keith Busch wrote: > > + if (blk_rq_is_poll(req)) > > + WRITE_ONCE(ioucmd->cookie, req); > > Why aren't we always setting the cookie to point at the req? As in unconditionally set the cookie even for non-polled requests? We need to see that the cookie is NULL to prevent polling interrupt queues when an io_uring polled command is dispatched to a device without polled hctx's.
