Commit 50569293 authored by Alexey Marchuk's avatar Alexey Marchuk Committed by Tomasz Zawadzki
Browse files

nvme/rdma: Add poller statistics 



New statistics include number of poller calls,
number of idle polls and total number of completions.
These statistics allow to estimate % of idle polls
and the number of completions per poll.

Since nvme_rdma_cq_process_completions function
returns number of completed NVMF requests and each
NVMF request consumes 2 RDMA completions (send+recv),
this function was extended to return the number of
RDMA completions.

Signed-off-by: default avatarAlexey Marchuk <alexeymar@mellanox.com>
Change-Id: Ifdc1e2e467f645adb5d66d39ff2a379e161fbd77
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/6298


Community-CI: Broadcom CI
Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: default avatarShuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
Reviewed-by: default avatarJim Harris <james.r.harris@intel.com>
Reviewed-by: default avatarBen Walker <benjamin.walker@intel.com>
parent 527f406b

lib/nvme/nvme_rdma.c

+21 −4
Original line number Diff line number Diff line
@@ -149,6 +149,10 @@ struct nvme_rdma_destroyed_qpair { 
};



struct nvme_rdma_poller_stats {

	uint64_t polls;

	uint64_t idle_polls;

	uint64_t queued_requests;

	uint64_t completions;

	struct spdk_rdma_qp_stats rdma_stats;

};
@@ -1897,7 +1901,10 @@ nvme_rdma_qpair_submit_request(struct spdk_nvme_qpair *qpair, 
	assert(req != NULL);



	rdma_req = nvme_rdma_req_get(rqpair);

	if (!rdma_req) {

	if (spdk_unlikely(!rdma_req)) {

		if (rqpair->poller) {

			rqpair->poller->stats.queued_requests++;

		}

		/* Inform the upper layer to try again later. */

		return -EAGAIN;

	}
@@ -2029,7 +2036,8 @@ nvme_rdma_conditional_fail_qpair(struct nvme_rdma_qpair *rqpair, struct nvme_rdm 
static int

nvme_rdma_cq_process_completions(struct ibv_cq *cq, uint32_t batch_size,

				 struct nvme_rdma_poll_group *group,

				 struct nvme_rdma_qpair *rdma_qpair)

				 struct nvme_rdma_qpair *rdma_qpair,

				 uint64_t *rdma_completions)

{

	struct ibv_wc			wc[MAX_COMPLETIONS_PER_POLL];

	struct nvme_rdma_qpair		*rqpair;
@@ -2132,6 +2140,8 @@ nvme_rdma_cq_process_completions(struct ibv_cq *cq, uint32_t batch_size, 
		}

	}



	*rdma_completions += rc;



	if (completion_rc) {

		return completion_rc;

	}
@@ -2153,6 +2163,7 @@ nvme_rdma_qpair_process_completions(struct spdk_nvme_qpair *qpair, 
	int				rc = 0, batch_size;

	struct ibv_cq			*cq;

	struct nvme_rdma_ctrlr		*rctrlr;

	uint64_t			rdma_completions = 0;



	/*

	 * This is used during the connection phase. It's possible that we are still reaping error completions
@@ -2186,7 +2197,7 @@ nvme_rdma_qpair_process_completions(struct spdk_nvme_qpair *qpair, 
	rqpair->num_completions = 0;

	do {

		batch_size = spdk_min((max_completions - rqpair->num_completions), MAX_COMPLETIONS_PER_POLL);

		rc = nvme_rdma_cq_process_completions(cq, batch_size, NULL, rqpair);

		rc = nvme_rdma_cq_process_completions(cq, batch_size, NULL, rqpair, &rdma_completions);



		if (rc == 0) {

			break;
@@ -2530,6 +2541,7 @@ nvme_rdma_poll_group_process_completions(struct spdk_nvme_transport_poll_group * 
	uint64_t				completions_allowed = 0;

	uint64_t				completions_per_poller = 0;

	uint64_t				poller_completions = 0;

	uint64_t				rdma_completions;





	if (completions_per_qpair == 0) {
@@ -2559,12 +2571,16 @@ nvme_rdma_poll_group_process_completions(struct spdk_nvme_transport_poll_group * 


	STAILQ_FOREACH(poller, &group->pollers, link) {

		poller_completions = 0;

		rdma_completions = 0;

		do {

			poller->stats.polls++;

			batch_size = spdk_min((completions_per_poller - poller_completions), MAX_COMPLETIONS_PER_POLL);

			rc = nvme_rdma_cq_process_completions(poller->cq, batch_size, group, NULL);

			rc = nvme_rdma_cq_process_completions(poller->cq, batch_size, group, NULL, &rdma_completions);

			if (rc <= 0) {

				if (rc == -ECANCELED) {

					return -EIO;

				} else if (rc == 0) {

					poller->stats.idle_polls++;

				}

				break;

			}
@@ -2572,6 +2588,7 @@ nvme_rdma_poll_group_process_completions(struct spdk_nvme_transport_poll_group * 
			poller_completions += rc;

		} while (poller_completions < completions_per_poller);

		total_completions += poller_completions;

		poller->stats.completions += rdma_completions;

	}



	STAILQ_FOREACH_SAFE(qpair, &tgroup->connected_qpairs, poll_group_stailq, tmp_qpair) {