Commit 86a6ac99 authored by Seth Howell's avatar Seth Howell Committed by Tomasz Zawadzki
Browse files

nvmf/rdma: always go through spdk_nvmf_qpair_disconnect. 



We should use this function as the synchronization point
for all qpair disconnects.

Signed-off-by: default avatarSeth Howell <seth.howell@intel.com>
Change-Id: Ic685ac3481765190cc56eeec3ee24dad52e336c9
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/3675


Community-CI: Mellanox Build Bot
Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: default avatarZiye Yang <ziye.yang@intel.com>
Reviewed-by: default avatarChangpeng Liu <changpeng.liu@intel.com>
Reviewed-by: default avatarAleksey Marchuk <alexeymar@mellanox.com>
Reviewed-by: default avatarBen Walker <benjamin.walker@intel.com>
parent 4bdfe13d

lib/nvmf/rdma.c

+10 −71
Original line number Diff line number Diff line
@@ -266,12 +266,6 @@ struct spdk_nvmf_rdma_request { 
	STAILQ_ENTRY(spdk_nvmf_rdma_request)	state_link;

};



enum spdk_nvmf_rdma_qpair_disconnect_flags {

	RDMA_QP_DISCONNECTING		= 1,

	RDMA_QP_RECV_DRAINED		= 1 << 1,

	RDMA_QP_SEND_DRAINED		= 1 << 2

};



struct spdk_nvmf_rdma_resource_opts {

	struct spdk_nvmf_rdma_qpair	*qpair;

	/* qp points either to an ibv_qp object or an ibv_srq object depending on the value of shared. */
@@ -395,8 +389,6 @@ struct spdk_nvmf_rdma_qpair { 
	 */

	enum ibv_qp_state			ibv_state;



	uint32_t				disconnect_flags;



	/* Poller registered in case the qpair doesn't properly

	 * complete the qpair destruct process and becomes defunct.

	 */
@@ -411,11 +403,6 @@ struct spdk_nvmf_rdma_qpair { 
	/* List of ibv async events */

	STAILQ_HEAD(, spdk_nvmf_rdma_ibv_event_ctx)	ibv_events;



	/* There are several ways a disconnect can start on a qpair

	 * and they are not all mutually exclusive. It is important

	 * that we only initialize one of these paths.

	 */

	bool					disconnect_started;

	/* Lets us know that we have received the last_wqe event. */

	bool					last_wqe_reached;

};
@@ -517,9 +504,6 @@ struct spdk_nvmf_rdma_transport { 
	TAILQ_HEAD(, spdk_nvmf_rdma_poll_group)	poll_groups;

};



static inline void

nvmf_rdma_start_disconnect(struct spdk_nvmf_rdma_qpair *rqpair);



static bool

nvmf_rdma_request_process(struct spdk_nvmf_rdma_transport *rtransport,

			  struct spdk_nvmf_rdma_request *rdma_req);
@@ -2104,7 +2088,7 @@ nvmf_rdma_request_process(struct spdk_nvmf_rdma_transport *rtransport, 
					if (rc != 0) {

						SPDK_ERRLOG("DIF generation failed\n");

						rdma_req->state = RDMA_REQUEST_STATE_COMPLETED;

						nvmf_rdma_start_disconnect(rqpair);

						spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);

						break;

					}

				}
@@ -2767,45 +2751,6 @@ nvmf_rdma_qpair_process_pending(struct spdk_nvmf_rdma_transport *rtransport, 
	}

}



static void

_nvmf_rdma_qpair_disconnect(void *ctx)

{

	struct spdk_nvmf_qpair *qpair = ctx;



	spdk_nvmf_qpair_disconnect(qpair, NULL, NULL);

}



static void

_nvmf_rdma_try_disconnect(void *ctx)

{

	struct spdk_nvmf_qpair *qpair = ctx;

	struct spdk_nvmf_poll_group *group;



	/* Read the group out of the qpair. This is normally set and accessed only from

	 * the thread that created the group. Here, we're not on that thread necessarily.

	 * The data member qpair->group begins it's life as NULL and then is assigned to

	 * a pointer and never changes. So fortunately reading this and checking for

	 * non-NULL is thread safe in the x86_64 memory model. */

	group = qpair->group;



	if (group == NULL) {

		/* The qpair hasn't been assigned to a group yet, so we can't

		 * process a disconnect. Send a message to ourself and try again. */

		spdk_thread_send_msg(spdk_get_thread(), _nvmf_rdma_try_disconnect, qpair);

		return;

	}



	spdk_thread_send_msg(group->thread, _nvmf_rdma_qpair_disconnect, qpair);

}



static inline void

nvmf_rdma_start_disconnect(struct spdk_nvmf_rdma_qpair *rqpair)

{

	if (!__atomic_test_and_set(&rqpair->disconnect_started, __ATOMIC_RELAXED)) {

		_nvmf_rdma_try_disconnect(&rqpair->qpair);

	}

}



static void nvmf_rdma_destroy_drained_qpair(void *ctx)

{

	struct spdk_nvmf_rdma_qpair *rqpair = ctx;
@@ -2857,7 +2802,7 @@ nvmf_rdma_disconnect(struct rdma_cm_event *evt) 


	spdk_trace_record(TRACE_RDMA_QP_DISCONNECT, 0, 0, (uintptr_t)rqpair->cm_id, 0);



	nvmf_rdma_start_disconnect(rqpair);

	spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);



	return 0;

}
@@ -2895,7 +2840,7 @@ nvmf_rdma_disconnect_qpairs_on_port(struct spdk_nvmf_rdma_transport *rtransport, 
		TAILQ_FOREACH(rpoller, &rgroup->pollers, link) {

			TAILQ_FOREACH(rqpair, &rpoller->qpairs, link) {

				if (rqpair->listen_id == port->id) {

					nvmf_rdma_start_disconnect(rqpair);

					spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);

				}

			}

		}
@@ -3059,7 +3004,7 @@ static void 
nvmf_rdma_handle_qp_fatal(struct spdk_nvmf_rdma_qpair *rqpair)

{

	nvmf_rdma_update_ibv_state(rqpair);

	nvmf_rdma_start_disconnect(rqpair);

	spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);

}



static void
@@ -3072,7 +3017,7 @@ nvmf_rdma_handle_last_wqe_reached(struct spdk_nvmf_rdma_qpair *rqpair) 
static void

nvmf_rdma_handle_sq_drained(struct spdk_nvmf_rdma_qpair *rqpair)

{

	nvmf_rdma_start_disconnect(rqpair);

	spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);

}



static void
@@ -3641,12 +3586,6 @@ nvmf_rdma_close_qpair(struct spdk_nvmf_qpair *qpair) 
{

	struct spdk_nvmf_rdma_qpair *rqpair = SPDK_CONTAINEROF(qpair, struct spdk_nvmf_rdma_qpair, qpair);



	if (rqpair->disconnect_flags & RDMA_QP_DISCONNECTING) {

		return;

	}



	rqpair->disconnect_flags |= RDMA_QP_DISCONNECTING;



	/* This happens only when the qpair is disconnected before

	 * it is added to the poll group. Since there is no poll group,

	 * the RDMA qp has not been initialized yet and the RDMA CM
@@ -3703,7 +3642,7 @@ _poller_reset_failed_recvs(struct spdk_nvmf_rdma_poller *rpoller, struct ibv_rec 
		rdma_recv->qpair->current_recv_depth++;

		bad_recv_wr = bad_recv_wr->next;

		SPDK_ERRLOG("Failed to post a recv for the qpair %p with errno %d\n", rdma_recv->qpair, -rc);

		nvmf_rdma_start_disconnect(rdma_recv->qpair);

		spdk_nvmf_qpair_disconnect(&rdma_recv->qpair->qpair, NULL, NULL);

	}

}
@@ -3715,7 +3654,7 @@ _qp_reset_failed_recvs(struct spdk_nvmf_rdma_qpair *rqpair, struct ibv_recv_wr * 
		bad_recv_wr = bad_recv_wr->next;

		rqpair->current_recv_depth++;

	}

	nvmf_rdma_start_disconnect(rqpair);

	spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);

}



static void
@@ -3806,7 +3745,7 @@ _qp_reset_failed_sends(struct spdk_nvmf_rdma_transport *rtransport, 


	if (rqpair->qpair.state == SPDK_NVMF_QPAIR_ACTIVE) {

		/* Disconnect the connection. */

		nvmf_rdma_start_disconnect(rqpair);

		spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);

	}



}
@@ -3907,7 +3846,7 @@ nvmf_rdma_poller_poll(struct spdk_nvmf_rdma_transport *rtransport, 
			if (!wc[i].status) {

				assert(wc[i].opcode == IBV_WC_RECV);

				if (rqpair->current_recv_depth >= rqpair->max_queue_depth) {

					nvmf_rdma_start_disconnect(rqpair);

					spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);

					break;

				}

			}
@@ -3969,7 +3908,7 @@ nvmf_rdma_poller_poll(struct spdk_nvmf_rdma_transport *rtransport, 


			if (rqpair->qpair.state == SPDK_NVMF_QPAIR_ACTIVE) {

				/* Disconnect the connection. */

				nvmf_rdma_start_disconnect(rqpair);

				spdk_nvmf_qpair_disconnect(&rqpair->qpair, NULL, NULL);

			} else {

				nvmf_rdma_destroy_drained_qpair(rqpair);

			}