Commit 64a741c1 authored by Steve Wise's avatar Steve Wise Committed by Christoph Hellwig

nvme-rdma: support up to 4 segments of inline data

Allow up to 4 segments of inline data for NVMF WRITE operations. This
reduces latency for small WRITEs by removing the need for the target to
issue a READ WR for IB, or a REG_MR + READ WR chain for iWarp.

Also cap the inline segments used based on the limitations of the
device.
Reviewed-by: default avatarSagi Grimberg <sagi@grimberg.me>
Reviewed-by: default avatarMax Gurtovoy <maxg@mellanox.com>
Signed-off-by: default avatarSteve Wise <swise@opengridcomputing.com>
Signed-off-by: default avatarChristoph Hellwig <hch@lst.de>
parent 55eb942e
...@@ -40,13 +40,14 @@ ...@@ -40,13 +40,14 @@
#define NVME_RDMA_MAX_SEGMENTS 256 #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 nvme_rdma_device {
struct ib_device *dev; struct ib_device *dev;
struct ib_pd *pd; struct ib_pd *pd;
struct kref ref; struct kref ref;
struct list_head entry; struct list_head entry;
unsigned int num_inline_segments;
}; };
struct nvme_rdma_qe { struct nvme_rdma_qe {
...@@ -117,6 +118,7 @@ struct nvme_rdma_ctrl { ...@@ -117,6 +118,7 @@ struct nvme_rdma_ctrl {
struct sockaddr_storage src_addr; struct sockaddr_storage src_addr;
struct nvme_ctrl ctrl; struct nvme_ctrl ctrl;
bool use_inline_data;
}; };
static inline struct nvme_rdma_ctrl *to_rdma_ctrl(struct nvme_ctrl *ctrl) static inline struct nvme_rdma_ctrl *to_rdma_ctrl(struct nvme_ctrl *ctrl)
...@@ -249,7 +251,7 @@ static int nvme_rdma_create_qp(struct nvme_rdma_queue *queue, const int factor) ...@@ -249,7 +251,7 @@ static int nvme_rdma_create_qp(struct nvme_rdma_queue *queue, const int factor)
/* +1 for drain */ /* +1 for drain */
init_attr.cap.max_recv_wr = queue->queue_size + 1; init_attr.cap.max_recv_wr = queue->queue_size + 1;
init_attr.cap.max_recv_sge = 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.sq_sig_type = IB_SIGNAL_REQ_WR;
init_attr.qp_type = IB_QPT_RC; init_attr.qp_type = IB_QPT_RC;
init_attr.send_cq = queue->ib_cq; init_attr.send_cq = queue->ib_cq;
...@@ -374,6 +376,8 @@ nvme_rdma_find_get_device(struct rdma_cm_id *cm_id) ...@@ -374,6 +376,8 @@ nvme_rdma_find_get_device(struct rdma_cm_id *cm_id)
goto out_free_pd; goto out_free_pd;
} }
ndev->num_inline_segments = min(NVME_RDMA_MAX_INLINE_SEGMENTS,
ndev->dev->attrs.max_sge - 1);
list_add(&ndev->entry, &device_list); list_add(&ndev->entry, &device_list);
out_unlock: out_unlock:
mutex_unlock(&device_list_mutex); mutex_unlock(&device_list_mutex);
...@@ -925,6 +929,9 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work) ...@@ -925,6 +929,9 @@ static void nvme_rdma_reconnect_ctrl_work(struct work_struct *work)
if (ret) if (ret)
goto requeue; goto requeue;
if (ctrl->ctrl.sgls & (1 << 20))
ctrl->use_inline_data = true;
if (ctrl->ctrl.queue_count > 1) { if (ctrl->ctrl.queue_count > 1) {
ret = nvme_rdma_configure_io_queues(ctrl, false); ret = nvme_rdma_configure_io_queues(ctrl, false);
if (ret) if (ret)
...@@ -1090,19 +1097,27 @@ static int nvme_rdma_set_sg_null(struct nvme_command *c) ...@@ -1090,19 +1097,27 @@ static int nvme_rdma_set_sg_null(struct nvme_command *c)
} }
static int nvme_rdma_map_sg_inline(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 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); for (i = 0; i < count; i++, sgl++, sge++) {
req->sge[1].length = sg_dma_len(req->sg_table.sgl); sge->addr = sg_dma_address(sgl);
req->sge[1].lkey = queue->device->pd->local_dma_lkey; 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->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; sg->type = (NVME_SGL_FMT_DATA_DESC << 4) | NVME_SGL_FMT_OFFSET;
req->num_sge++; req->num_sge += count;
return 0; return 0;
} }
...@@ -1195,15 +1210,16 @@ static int nvme_rdma_map_data(struct nvme_rdma_queue *queue, ...@@ -1195,15 +1210,16 @@ static int nvme_rdma_map_data(struct nvme_rdma_queue *queue,
goto out_free_table; goto out_free_table;
} }
if (count == 1) { if (count <= dev->num_inline_segments) {
if (rq_data_dir(rq) == WRITE && nvme_rdma_queue_idx(queue) && if (rq_data_dir(rq) == WRITE && nvme_rdma_queue_idx(queue) &&
queue->ctrl->use_inline_data &&
blk_rq_payload_bytes(rq) <= blk_rq_payload_bytes(rq) <=
nvme_rdma_inline_data_size(queue)) { 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; 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); ret = nvme_rdma_map_sg_single(queue, req, c);
goto out; goto out;
} }
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment