# Re: \[PATCH 15/15\] nvme-multipath: enable polled I/O

- From: Sagi Grimberg \<sagi@grimberg.me\>
- Date: 2021-05-12 23:12:15 
- List: linux-block
- Page: https://ratatoskr.run/linux-block/2021/05/205906
- Thread: https://ratatoskr.run/linux-block/2021/05/205819/t.md

```text

> Set the poll queue flag to enable polling, given that the multipath
> node just dispatches the bios to a lower queue.
>
> Signed-off-by: Christoph Hellwig <hch@lst.de>
> ---
>   drivers/nvme/host/multipath.c | 9 +++++++++
>   1 file changed, 9 insertions(+)
>
> diff --git a/drivers/nvme/host/multipath.c b/drivers/nvme/host/multipath.c
> index 516fe977606d..e95b93655d06 100644
> --- a/drivers/nvme/host/multipath.c
> +++ b/drivers/nvme/host/multipath.c
> @@ -446,6 +446,15 @@ int nvme_mpath_alloc_disk(struct nvme_ctrl *ctrl, struct nvme_ns_head *head)
>   		goto out;
>   	blk_queue_flag_set(QUEUE_FLAG_NONROT, q);
>   	blk_queue_flag_set(QUEUE_FLAG_NOWAIT, q);
> +	/*
> +	 * This assumes all controllers that refer to a namespace either
> +	 * support poll queues or not.  That is not a strict guarantee,
> +	 * but if the assumption is wrong the effect is only suboptimal
> +	 * performance but not correctness problem.
> +	 */
> +	if (ctrl->tagset->nr_maps > HCTX_TYPE_POLL &&
> +	    ctrl->tagset->map[HCTX_TYPE_POLL].nr_queues)
> +		blk_queue_flag_set(QUEUE_FLAG_POLL, q);

If one controller does not support polling and the other does, won't
the block layer fail to map a queue for REQ_POLLED requests?

Maybe clear in the else case here?
```
