Commit 6375b60c authored by Alexey Marchuk's avatar Alexey Marchuk Committed by Tomasz Zawadzki
Browse files

nvmf/rdma: Use put_bulk when releasing data WRs 



That is more efficient

Signed-off-by: default avatarAlexey Marchuk <alexeymar@nvidia.com>
Change-Id: I384a67aaa9cb4b160406e14554ab5e1555dd284c
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/19122


Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Community-CI: Mellanox Build Bot
Reviewed-by: default avatarMichael Haeuptle <michaelhaeuptle@gmail.com>
Reviewed-by: default avatarShuhei Matsumoto <smatsumoto@nvidia.com>
Reviewed-by: default avatarJim Harris <jim.harris@gmail.com>
parent 90d91cd3

lib/nvmf/rdma.c

+11 −3
Original line number Diff line number Diff line
@@ -627,8 +627,10 @@ _nvmf_rdma_request_free_data(struct spdk_nvmf_rdma_request *rdma_req, 
			     struct spdk_nvmf_rdma_request_data	*data_wr,

			     struct spdk_mempool *pool)

{

	struct spdk_nvmf_rdma_request_data	*work_requests[SPDK_NVMF_MAX_SGL_ENTRIES];

	struct ibv_send_wr			*next_send_wr;

	uint64_t				req_wrid = data_wr->wr.wr_id;

	uint32_t				num_wrs = 0;



	while (data_wr && data_wr->wr.wr_id == req_wrid) {

		memset(data_wr->sgl, 0, sizeof(data_wr->wr.sg_list[0]) * data_wr->wr.num_sge);
@@ -636,11 +638,17 @@ _nvmf_rdma_request_free_data(struct spdk_nvmf_rdma_request *rdma_req, 
		next_send_wr = data_wr->wr.next;

		if (data_wr != &rdma_req->data) {

			data_wr->wr.next = NULL;

			spdk_mempool_put(pool, data_wr);

			assert(num_wrs < SPDK_NVMF_MAX_SGL_ENTRIES);

			work_requests[num_wrs] = data_wr;

			num_wrs++;

		}

		data_wr = (!next_send_wr || next_send_wr == &rdma_req->rsp.wr) ? NULL :

			  SPDK_CONTAINEROF(next_send_wr, struct spdk_nvmf_rdma_request_data, wr);

	}



	if (num_wrs) {

		spdk_mempool_put_bulk(pool, (void **) work_requests, num_wrs);

	}

}



static void