blk_mq_tag_to_rq() always returns a request if the tag id is in a valid range [0...max_tags). If the target replies with a tag for which we don't have a request but it's not started, the host will likely corrupt data or simply crash. Add an additional check if the a request has been started if not reset the connection. This addition check will not protected against an invalid tag which maps to a request which has been started. There is nothing we can do about this. Though it will at a least protect from crashing the host, which generally thought to be the right thing to do.
Daniel, again, there is nothing specific about this to nvme-tcp, this is a safeguard against a funky controller (or a different bug that is hidden by this). The same can happen in any other transport so I would suggest that if this is a safeguard we want to put in place, we should make it a generic one. i.e. nvme_tag_to_rq() that _all_ transports call consistently.
