#define NVME_RDMA_MAX_SEGMENTS 256
-#define NVME_RDMA_MAX_INLINE_SEGMENTS 1
+#define NVME_RDMA_MAX_INLINE_SEGMENTS 4
struct nvme_rdma_device {
struct ib_device *dev;
struct ib_pd *pd;
struct kref ref;
struct list_head entry;
+ unsigned int num_inline_segments;
};
struct nvme_rdma_qe {
struct sockaddr_storage src_addr;
struct nvme_ctrl ctrl;
+ bool use_inline_data;
};
static inline struct nvme_rdma_ctrl *to_rdma_ctrl(struct nvme_ctrl *ctrl)
static int nvme_rdma_wait_for_cm(struct nvme_rdma_queue *queue)
{
- wait_for_completion_interruptible_timeout(&queue->cm_done,
+ int ret;
+
+ ret = wait_for_completion_interruptible_timeout(&queue->cm_done,
msecs_to_jiffies(NVME_RDMA_CONNECT_TIMEOUT_MS) + 1);
+ if (ret < 0)
+ return ret;
+ if (ret == 0)
+ return -ETIMEDOUT;
+ WARN_ON_ONCE(queue->cm_error > 0);
return queue->cm_error;
}
/* +1 for drain */
init_attr.cap.max_recv_wr = queue->queue_size + 1;
init_attr.cap.max_recv_sge = 1;
- init_attr.cap.max_send_sge = 1 + NVME_RDMA_MAX_INLINE_SEGMENTS;
+ init_attr.cap.max_send_sge = 1 + dev->num_inline_segments;
init_attr.sq_sig_type = IB_SIGNAL_REQ_WR;
init_attr.qp_type = IB_QPT_RC;
init_attr.send_cq = queue->ib_cq;
struct ib_device *ibdev = dev->dev;
int ret;
+ nvme_req(rq)->ctrl = &ctrl->ctrl;
ret = nvme_rdma_alloc_qe(ibdev, &req->sqe, sizeof(struct nvme_command),
DMA_TO_DEVICE);
if (ret)
goto out_free_pd;
}
+ ndev->num_inline_segments = min(NVME_RDMA_MAX_INLINE_SEGMENTS,
+ ndev->dev->attrs.max_send_sge - 1);
list_add(&ndev->entry, &device_list);
out_unlock:
mutex_unlock(&device_list_mutex);
return ret;
}
+static void nvme_rdma_teardown_admin_queue(struct nvme_rdma_ctrl *ctrl,
+ bool remove)
+{
+ blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
+ nvme_rdma_stop_queue(&ctrl->queues[0]);
+ blk_mq_tagset_busy_iter(&ctrl->admin_tag_set, nvme_cancel_request,
+ &ctrl->ctrl);
+ blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
+ nvme_rdma_destroy_admin_queue(ctrl, remove);
+}
+
+static void nvme_rdma_teardown_io_queues(struct nvme_rdma_ctrl *ctrl,
+ bool remove)
+{
+ if (ctrl->ctrl.queue_count > 1) {
+ nvme_stop_queues(&ctrl->ctrl);
+ nvme_rdma_stop_io_queues(ctrl);
+ blk_mq_tagset_busy_iter(&ctrl->tag_set, nvme_cancel_request,
+ &ctrl->ctrl);
+ if (remove)
+ nvme_start_queues(&ctrl->ctrl);
+ nvme_rdma_destroy_io_queues(ctrl, remove);
+ }
+}
+
static void nvme_rdma_stop_ctrl(struct nvme_ctrl *nctrl)
{
struct nvme_rdma_ctrl *ctrl = to_rdma_ctrl(nctrl);
}
}
-static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
+static int nvme_rdma_setup_ctrl(struct nvme_rdma_ctrl *ctrl, bool new)
{
- struct nvme_rdma_ctrl *ctrl = container_of(to_delayed_work(work),
- struct nvme_rdma_ctrl, reconnect_work);
+ int ret = -EINVAL;
bool changed;
- int ret;
- ++ctrl->ctrl.nr_reconnects;
-
- ret = nvme_rdma_configure_admin_queue(ctrl, false);
+ ret = nvme_rdma_configure_admin_queue(ctrl, new);
if (ret)
- goto requeue;
+ return ret;
+
+ if (ctrl->ctrl.icdoff) {
+ dev_err(ctrl->ctrl.device, "icdoff is not supported!\n");
+ goto destroy_admin;
+ }
+
+ if (!(ctrl->ctrl.sgls & (1 << 2))) {
+ dev_err(ctrl->ctrl.device,
+ "Mandatory keyed sgls are not supported!\n");
+ goto destroy_admin;
+ }
+
+ if (ctrl->ctrl.opts->queue_size > ctrl->ctrl.sqsize + 1) {
+ dev_warn(ctrl->ctrl.device,
+ "queue_size %zu > ctrl sqsize %u, clamping down\n",
+ ctrl->ctrl.opts->queue_size, ctrl->ctrl.sqsize + 1);
+ }
+
+ if (ctrl->ctrl.sqsize + 1 > ctrl->ctrl.maxcmd) {
+ dev_warn(ctrl->ctrl.device,
+ "sqsize %u > ctrl maxcmd %u, clamping down\n",
+ ctrl->ctrl.sqsize + 1, ctrl->ctrl.maxcmd);
+ ctrl->ctrl.sqsize = ctrl->ctrl.maxcmd - 1;
+ }
+
+ if (ctrl->ctrl.sgls & (1 << 20))
+ ctrl->use_inline_data = true;
if (ctrl->ctrl.queue_count > 1) {
- ret = nvme_rdma_configure_io_queues(ctrl, false);
+ ret = nvme_rdma_configure_io_queues(ctrl, new);
if (ret)
goto destroy_admin;
}
if (!changed) {
/* state change failure is ok if we're in DELETING state */
WARN_ON_ONCE(ctrl->ctrl.state != NVME_CTRL_DELETING);
- return;
+ ret = -EINVAL;
+ goto destroy_io;
}
nvme_start_ctrl(&ctrl->ctrl);
+ return 0;
+
+destroy_io:
+ if (ctrl->ctrl.queue_count > 1)
+ nvme_rdma_destroy_io_queues(ctrl, new);
+destroy_admin:
+ nvme_rdma_stop_queue(&ctrl->queues[0]);
+ nvme_rdma_destroy_admin_queue(ctrl, new);
+ return ret;
+}
+
+static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
+{
+ struct nvme_rdma_ctrl *ctrl = container_of(to_delayed_work(work),
+ struct nvme_rdma_ctrl, reconnect_work);
+
+ ++ctrl->ctrl.nr_reconnects;
+
+ if (nvme_rdma_setup_ctrl(ctrl, false))
+ goto requeue;
dev_info(ctrl->ctrl.device, "Successfully reconnected (%d attempts)\n",
ctrl->ctrl.nr_reconnects);
return;
-destroy_admin:
- nvme_rdma_stop_queue(&ctrl->queues[0]);
- nvme_rdma_destroy_admin_queue(ctrl, false);
requeue:
dev_info(ctrl->ctrl.device, "Failed reconnect attempt %d\n",
ctrl->ctrl.nr_reconnects);
struct nvme_rdma_ctrl, err_work);
nvme_stop_keep_alive(&ctrl->ctrl);
-
- if (ctrl->ctrl.queue_count > 1) {
- nvme_stop_queues(&ctrl->ctrl);
- nvme_rdma_stop_io_queues(ctrl);
- blk_mq_tagset_busy_iter(&ctrl->tag_set,
- nvme_cancel_request, &ctrl->ctrl);
- nvme_rdma_destroy_io_queues(ctrl, false);
- }
-
- blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
- nvme_rdma_stop_queue(&ctrl->queues[0]);
- blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
- nvme_cancel_request, &ctrl->ctrl);
- nvme_rdma_destroy_admin_queue(ctrl, false);
-
- /*
- * queues are not a live anymore, so restart the queues to fail fast
- * new IO
- */
- blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
+ nvme_rdma_teardown_io_queues(ctrl, false);
nvme_start_queues(&ctrl->ctrl);
+ nvme_rdma_teardown_admin_queue(ctrl, false);
if (!nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_CONNECTING)) {
/* state change failure is ok if we're in DELETING state */
static int nvme_rdma_inv_rkey(struct nvme_rdma_queue *queue,
struct nvme_rdma_request *req)
{
- struct ib_send_wr *bad_wr;
struct ib_send_wr wr = {
.opcode = IB_WR_LOCAL_INV,
.next = NULL,
req->reg_cqe.done = nvme_rdma_inv_rkey_done;
wr.wr_cqe = &req->reg_cqe;
- return ib_post_send(queue->qp, &wr, &bad_wr);
+ return ib_post_send(queue->qp, &wr, NULL);
}
static void nvme_rdma_unmap_data(struct nvme_rdma_queue *queue,
}
static int nvme_rdma_map_sg_inline(struct nvme_rdma_queue *queue,
- struct nvme_rdma_request *req, struct nvme_command *c)
+ struct nvme_rdma_request *req, struct nvme_command *c,
+ int count)
{
struct nvme_sgl_desc *sg = &c->common.dptr.sgl;
+ struct scatterlist *sgl = req->sg_table.sgl;
+ struct ib_sge *sge = &req->sge[1];
+ u32 len = 0;
+ int i;
- req->sge[1].addr = sg_dma_address(req->sg_table.sgl);
- req->sge[1].length = sg_dma_len(req->sg_table.sgl);
- req->sge[1].lkey = queue->device->pd->local_dma_lkey;
+ for (i = 0; i < count; i++, sgl++, sge++) {
+ sge->addr = sg_dma_address(sgl);
+ sge->length = sg_dma_len(sgl);
+ sge->lkey = queue->device->pd->local_dma_lkey;
+ len += sge->length;
+ }
sg->addr = cpu_to_le64(queue->ctrl->ctrl.icdoff);
- sg->length = cpu_to_le32(sg_dma_len(req->sg_table.sgl));
+ sg->length = cpu_to_le32(len);
sg->type = (NVME_SGL_FMT_DATA_DESC << 4) | NVME_SGL_FMT_OFFSET;
- req->num_sge++;
+ req->num_sge += count;
return 0;
}
goto out_free_table;
}
- if (count == 1) {
+ if (count <= dev->num_inline_segments) {
if (rq_data_dir(rq) == WRITE && nvme_rdma_queue_idx(queue) &&
+ queue->ctrl->use_inline_data &&
blk_rq_payload_bytes(rq) <=
nvme_rdma_inline_data_size(queue)) {
- ret = nvme_rdma_map_sg_inline(queue, req, c);
+ ret = nvme_rdma_map_sg_inline(queue, req, c, count);
goto out;
}
- if (dev->pd->flags & IB_PD_UNSAFE_GLOBAL_RKEY) {
+ if (count == 1 && dev->pd->flags & IB_PD_UNSAFE_GLOBAL_RKEY) {
ret = nvme_rdma_map_sg_single(queue, req, c);
goto out;
}
struct nvme_rdma_qe *qe, struct ib_sge *sge, u32 num_sge,
struct ib_send_wr *first)
{
- struct ib_send_wr wr, *bad_wr;
+ struct ib_send_wr wr;
int ret;
sge->addr = qe->dma;
else
first = ≀
- ret = ib_post_send(queue->qp, first, &bad_wr);
+ ret = ib_post_send(queue->qp, first, NULL);
if (unlikely(ret)) {
dev_err(queue->ctrl->ctrl.device,
"%s failed with error code %d\n", __func__, ret);
static int nvme_rdma_post_recv(struct nvme_rdma_queue *queue,
struct nvme_rdma_qe *qe)
{
- struct ib_recv_wr wr, *bad_wr;
+ struct ib_recv_wr wr;
struct ib_sge list;
int ret;
wr.sg_list = &list;
wr.num_sge = 1;
- ret = ib_post_recv(queue->qp, &wr, &bad_wr);
+ ret = ib_post_recv(queue->qp, &wr, NULL);
if (unlikely(ret)) {
dev_err(queue->ctrl->ctrl.device,
"%s failed with error code %d\n", __func__, ret);
case RDMA_CM_EVENT_CONNECT_ERROR:
case RDMA_CM_EVENT_UNREACHABLE:
nvme_rdma_destroy_queue_ib(queue);
+ /* fall through */
case RDMA_CM_EVENT_ADDR_ERROR:
dev_dbg(queue->ctrl->ctrl.device,
"CM error event %d\n", ev->event);
WARN_ON_ONCE(rq->tag < 0);
if (!nvmf_check_ready(&queue->ctrl->ctrl, rq, queue_ready))
- return nvmf_fail_nonready_command(rq);
+ return nvmf_fail_nonready_command(&queue->ctrl->ctrl, rq);
dev = queue->device->dev;
ib_dma_sync_single_for_cpu(dev, sqe->dma,
static void nvme_rdma_shutdown_ctrl(struct nvme_rdma_ctrl *ctrl, bool shutdown)
{
- if (ctrl->ctrl.queue_count > 1) {
- nvme_stop_queues(&ctrl->ctrl);
- nvme_rdma_stop_io_queues(ctrl);
- blk_mq_tagset_busy_iter(&ctrl->tag_set,
- nvme_cancel_request, &ctrl->ctrl);
- nvme_rdma_destroy_io_queues(ctrl, shutdown);
- }
-
+ nvme_rdma_teardown_io_queues(ctrl, shutdown);
if (shutdown)
nvme_shutdown_ctrl(&ctrl->ctrl);
else
nvme_disable_ctrl(&ctrl->ctrl, ctrl->ctrl.cap);
-
- blk_mq_quiesce_queue(ctrl->ctrl.admin_q);
- nvme_rdma_stop_queue(&ctrl->queues[0]);
- blk_mq_tagset_busy_iter(&ctrl->admin_tag_set,
- nvme_cancel_request, &ctrl->ctrl);
- blk_mq_unquiesce_queue(ctrl->ctrl.admin_q);
- nvme_rdma_destroy_admin_queue(ctrl, shutdown);
+ nvme_rdma_teardown_admin_queue(ctrl, shutdown);
}
static void nvme_rdma_delete_ctrl(struct nvme_ctrl *ctrl)
{
struct nvme_rdma_ctrl *ctrl =
container_of(work, struct nvme_rdma_ctrl, ctrl.reset_work);
- int ret;
- bool changed;
nvme_stop_ctrl(&ctrl->ctrl);
nvme_rdma_shutdown_ctrl(ctrl, false);
return;
}
- ret = nvme_rdma_configure_admin_queue(ctrl, false);
- if (ret)
+ if (nvme_rdma_setup_ctrl(ctrl, false))
goto out_fail;
- if (ctrl->ctrl.queue_count > 1) {
- ret = nvme_rdma_configure_io_queues(ctrl, false);
- if (ret)
- goto out_fail;
- }
-
- changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
- if (!changed) {
- /* state change failure is ok if we're in DELETING state */
- WARN_ON_ONCE(ctrl->ctrl.state != NVME_CTRL_DELETING);
- return;
- }
-
- nvme_start_ctrl(&ctrl->ctrl);
-
return;
out_fail:
changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_CONNECTING);
WARN_ON_ONCE(!changed);
- ret = nvme_rdma_configure_admin_queue(ctrl, true);
+ ret = nvme_rdma_setup_ctrl(ctrl, true);
if (ret)
goto out_uninit_ctrl;
- /* sanity check icdoff */
- if (ctrl->ctrl.icdoff) {
- dev_err(ctrl->ctrl.device, "icdoff is not supported!\n");
- ret = -EINVAL;
- goto out_remove_admin_queue;
- }
-
- /* sanity check keyed sgls */
- if (!(ctrl->ctrl.sgls & (1 << 2))) {
- dev_err(ctrl->ctrl.device,
- "Mandatory keyed sgls are not supported!\n");
- ret = -EINVAL;
- goto out_remove_admin_queue;
- }
-
- /* only warn if argument is too large here, will clamp later */
- if (opts->queue_size > ctrl->ctrl.sqsize + 1) {
- dev_warn(ctrl->ctrl.device,
- "queue_size %zu > ctrl sqsize %u, clamping down\n",
- opts->queue_size, ctrl->ctrl.sqsize + 1);
- }
-
- /* warn if maxcmd is lower than sqsize+1 */
- if (ctrl->ctrl.sqsize + 1 > ctrl->ctrl.maxcmd) {
- dev_warn(ctrl->ctrl.device,
- "sqsize %u > ctrl maxcmd %u, clamping down\n",
- ctrl->ctrl.sqsize + 1, ctrl->ctrl.maxcmd);
- ctrl->ctrl.sqsize = ctrl->ctrl.maxcmd - 1;
- }
-
- if (opts->nr_io_queues) {
- ret = nvme_rdma_configure_io_queues(ctrl, true);
- if (ret)
- goto out_remove_admin_queue;
- }
-
- changed = nvme_change_ctrl_state(&ctrl->ctrl, NVME_CTRL_LIVE);
- WARN_ON_ONCE(!changed);
-
dev_info(ctrl->ctrl.device, "new ctrl: NQN \"%s\", addr %pISpcs\n",
ctrl->ctrl.opts->subsysnqn, &ctrl->addr);
list_add_tail(&ctrl->list, &nvme_rdma_ctrl_list);
mutex_unlock(&nvme_rdma_ctrl_mutex);
- nvme_start_ctrl(&ctrl->ctrl);
-
return &ctrl->ctrl;
-out_remove_admin_queue:
- nvme_rdma_stop_queue(&ctrl->queues[0]);
- nvme_rdma_destroy_admin_queue(ctrl, true);
out_uninit_ctrl:
nvme_uninit_ctrl(&ctrl->ctrl);
nvme_put_ctrl(&ctrl->ctrl);