[PATCH 1/4] blk-mq: check rq->state explicitly in blk_mq_tagset_count_completed_rqs

Ming Lei ming.lei at redhat.com
Sun Oct 18 20:50:29 EDT 2020


On Fri, Oct 16, 2020 at 10:28:08PM +0800, Ming Lei wrote:
> blk_mq_tagset_count_completed_rqs() is called from
> blk_mq_tagset_wait_completed_request() for draining requests to be
> completed remotely. What we need to do is to see request->state to
> switch to non-MQ_RQ_COMPLETE.
> 
> So check MQ_RQ_COMPLETE explicitly in blk_mq_tagset_count_completed_rqs().
> 
> Meantime mark flush request as IDLE in its .end_io() for aligning to
> end normal request because flush request may stay in inflight tags in case
> of !elevator, so we need to change its state into IDLE.
> 
> Cc: Chao Leng <lengchao at huawei.com>
> Cc: Sagi Grimberg <sagi at grimberg.me>
> Signed-off-by: Ming Lei <ming.lei at redhat.com>
> ---
>  block/blk-flush.c  | 2 ++
>  block/blk-mq-tag.c | 2 +-
>  2 files changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/block/blk-flush.c b/block/blk-flush.c
> index 53abb5c73d99..31a2ae04d3f0 100644
> --- a/block/blk-flush.c
> +++ b/block/blk-flush.c
> @@ -231,6 +231,8 @@ static void flush_end_io(struct request *flush_rq, blk_status_t error)
>  		return;
>  	}
>  
> +	WRITE_ONCE(rq->state, MQ_RQ_IDLE);

oops, the above line should have been:

	WRITE_ONCE(flush_rq->state, MQ_RQ_IDLE);

Thanks,
Ming




More information about the Linux-nvme mailing list