The nvme request generation counter is intended to catch duplicate
completions. Incrementing the counter on submission means duplicates can
only be caught if the request tag is reallocated and dispatched prior to
the driver observing the corrupted CQE. Incrementing on completion
removes this window, making it possible to detect duplicate completions
in consecutive entries.

OK, It just occurred to me that blk_mq_tag_to_rq will still return the
request after it was completed as tags->rqs[x] is not cleared when
ending the request.

So this looks good.
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>