Commit b4e060b5 authored by Seth Howell's avatar Seth Howell Committed by Tomasz Zawadzki
Browse files

lib/nvme: check that req is not null in RDMA. 



When a request has been aborted, it's possible to get a completion
for an rdma request but the rdma_req->req object has already been
cleared to NULL.

Signed-off-by: default avatarSeth Howell <seth.howell@intel.com>
Change-Id: I5f7b1b96ff4be8c436aae9a7e2a7c9927d04e627
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/1960


Community-CI: Mellanox Build Bot
Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: default avatarBen Walker <benjamin.walker@intel.com>
Reviewed-by: default avatarAleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: default avatarShuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com>
parent c62887ad

lib/nvme/nvme_rdma.c

+9 −3
Original line number Diff line number Diff line
@@ -2143,17 +2143,23 @@ nvme_rdma_cq_process_completions(struct ibv_cq *cq, uint32_t batch_size) 


		case RDMA_WR_TYPE_SEND:

			rdma_req = SPDK_CONTAINEROF(rdma_wr, struct spdk_nvme_rdma_req, rdma_wr);

			rqpair = nvme_rdma_qpair(rdma_req->req->qpair);

			rdma_req->completion_flags |= NVME_RDMA_SEND_COMPLETED;



			/* If we are flushing I/O */

			if (wc[i].status) {

				rqpair = rdma_req->req ? nvme_rdma_qpair(rdma_req->req->qpair) : NULL;

				if (rqpair) {

					nvme_rdma_fail_qpair(&rqpair->qpair, 0);

				}

				SPDK_ERRLOG("CQ error on Queue Pair %p, Response Index %lu (%d): %s\n",

					    rqpair, wc[i].wr_id, wc[i].status, ibv_wc_status_str(wc[i].status));

				nvme_rdma_fail_qpair(&rqpair->qpair, 0);

				completion_rc = -ENXIO;

				continue;

			}



			rqpair = nvme_rdma_qpair(rdma_req->req->qpair);

			rdma_req->completion_flags |= NVME_RDMA_SEND_COMPLETED;





			if ((rdma_req->completion_flags & NVME_RDMA_RECV_COMPLETED) != 0) {

				if (spdk_unlikely(nvme_rdma_request_ready(rqpair, rdma_req))) {

					SPDK_ERRLOG("Unable to re-post rx descriptor\n");