RDMA/hns: Support direct wqe of userspace

The current write wqe mechanism is to write to DDR first, and then notify
the hardware through doorbell to read the data. Direct wqe is a mechanism
to fill wqe directly into the hardware. In the case of light load, the wqe
will be filled into pcie bar space of the hardware, this will reduce one
memory access operation and therefore reduce the latency. SIMD
instructions allows cpu to write the 512 bits at one time to device
memory, thus it can be used for posting direct wqe.

Add direct wqe enable switch and address mapping.

Link: https://lore.kernel.org/r/[e-mail]
Signed-off-by: Yixing Liu <[e-mail]>
Signed-off-by: Wenpeng Liang <[e-mail]>
Signed-off-by: Jason Gunthorpe <[e-mail]>

drivers/infiniband/hw/hns/hns_roce_device.h

···
 	HNS_ROCE_CAP_FLAG_FRMR                  = BIT(8),
 	HNS_ROCE_CAP_FLAG_QP_FLOW_CTRL		= BIT(9),
 	HNS_ROCE_CAP_FLAG_ATOMIC		= BIT(10),
+	HNS_ROCE_CAP_FLAG_DIRECT_WQE		= BIT(12),
 	HNS_ROCE_CAP_FLAG_SDI_MODE		= BIT(14),
 	HNS_ROCE_CAP_FLAG_STASH			= BIT(17),
 };
···
 enum hns_roce_mmap_type {
 	HNS_ROCE_MMAP_TYPE_DB = 1,
 	HNS_ROCE_MMAP_TYPE_TPTR,
+	HNS_ROCE_MMAP_TYPE_DWQE,
 };
 
 struct hns_user_mmap_entry {
···
 	u32 queue_num;
 };
 
-enum {
-	HNS_ROCE_QP_CAP_DIRECT_WQE = BIT(5),
-};
-
 struct hns_roce_qp {
 	struct ib_qp		ibqp;
 	struct hns_roce_wq	rq;
···
 	struct list_head	node; /* all qps are on a list */
 	struct list_head	rq_node; /* all recv qps are on a list */
 	struct list_head	sq_node; /* all send qps are on a list */
+	struct hns_user_mmap_entry *dwqe_mmap_entry;
 };
 
 struct hns_roce_ib_iboe {
···
 	u32 func_num;
 	u32 is_vf;
 	u32 cong_algo_tmpl_id;
+	u64 dwqe_page;
 };
 
 static inline struct hns_roce_dev *to_hr_dev(struct ib_device *ib_dev)

drivers/infiniband/hw/hns/hns_roce_hw_v2.c

···
 	caps->gid_table_len[0] = HNS_ROCE_V2_GID_INDEX_NUM;
 
 	if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) {
-		caps->flags |= HNS_ROCE_CAP_FLAG_STASH;
+		caps->flags |= HNS_ROCE_CAP_FLAG_STASH |
+			       HNS_ROCE_CAP_FLAG_DIRECT_WQE;
 		caps->max_sq_inline = HNS_ROCE_V3_MAX_SQ_INLINE;
 	} else {
 		caps->max_sq_inline = HNS_ROCE_V2_MAX_SQ_INLINE;

drivers/infiniband/hw/hns/hns_roce_main.c

···
 	entry->address = address;
 	entry->mmap_type = mmap_type;
 
-	ret = rdma_user_mmap_entry_insert_exact(
-		ucontext, &entry->rdma_entry, length,
-		mmap_type == HNS_ROCE_MMAP_TYPE_DB ? 0 : 1);
+	switch (mmap_type) {
+	case HNS_ROCE_MMAP_TYPE_DB:
+		ret = rdma_user_mmap_entry_insert_exact(
+				ucontext, &entry->rdma_entry, length, 0);
+		break;
+	case HNS_ROCE_MMAP_TYPE_TPTR:
+		ret = rdma_user_mmap_entry_insert_exact(
+				ucontext, &entry->rdma_entry, length, 1);
+		break;
+	case HNS_ROCE_MMAP_TYPE_DWQE:
+		ret = rdma_user_mmap_entry_insert_range(
+				ucontext, &entry->rdma_entry, length, 2,
+				U32_MAX);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+
 	if (ret) {
 		kfree(entry);
 		return NULL;
···
 
 	entry = to_hns_mmap(rdma_entry);
 	pfn = entry->address >> PAGE_SHIFT;
-	prot = vma->vm_page_prot;
 
-	if (entry->mmap_type != HNS_ROCE_MMAP_TYPE_TPTR)
-		prot = pgprot_device(prot);
+	switch (entry->mmap_type) {
+	case HNS_ROCE_MMAP_TYPE_DB:
+	case HNS_ROCE_MMAP_TYPE_DWQE:
+		prot = pgprot_device(vma->vm_page_prot);
+		break;
+	case HNS_ROCE_MMAP_TYPE_TPTR:
+		prot = vma->vm_page_prot;
+		break;
+	default:
+		return -EINVAL;
+	}
 
 	ret = rdma_user_mmap_io(uctx, vma, pfn, rdma_entry->npages * PAGE_SIZE,
 				prot, rdma_entry);

drivers/infiniband/hw/hns/hns_roce_pd.c

···
 	} else {
 		uar->pfn = ((pci_resource_start(hr_dev->pci_dev, 2))
 			   >> PAGE_SHIFT);
+		if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_DIRECT_WQE)
+			hr_dev->dwqe_page =
+				pci_resource_start(hr_dev->pci_dev, 4);
 	}
 
 	return 0;

drivers/infiniband/hw/hns/hns_roce_qp.c

···
 	return ret;
 }
 
+static void qp_user_mmap_entry_remove(struct hns_roce_qp *hr_qp)
+{
+	rdma_user_mmap_entry_remove(&hr_qp->dwqe_mmap_entry->rdma_entry);
+}
+
 void hns_roce_qp_remove(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp)
 {
 	struct xarray *xa = &hr_dev->qp_table_xa;
···
 		goto err_inline;
 	}
 
+	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_DIRECT_WQE)
+		hr_qp->en_flags |= HNS_ROCE_QP_CAP_DIRECT_WQE;
+
 	return 0;
+
 err_inline:
 	free_rq_inline_buf(hr_qp);
 
···
 {
 	return ((hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_QP_RECORD_DB) &&
 		hns_roce_qp_has_rq(init_attr));
+}
+
+static int qp_mmap_entry(struct hns_roce_qp *hr_qp,
+			 struct hns_roce_dev *hr_dev,
+			 struct ib_udata *udata,
+			 struct hns_roce_ib_create_qp_resp *resp)
+{
+	struct hns_roce_ucontext *uctx =
+		rdma_udata_to_drv_context(udata,
+			struct hns_roce_ucontext, ibucontext);
+	struct rdma_user_mmap_entry *rdma_entry;
+	u64 address;
+
+	address = hr_dev->dwqe_page + hr_qp->qpn * HNS_ROCE_DWQE_SIZE;
+
+	hr_qp->dwqe_mmap_entry =
+		hns_roce_user_mmap_entry_insert(&uctx->ibucontext, address,
+						HNS_ROCE_DWQE_SIZE,
+						HNS_ROCE_MMAP_TYPE_DWQE);
+
+	if (!hr_qp->dwqe_mmap_entry) {
+		ibdev_err(&hr_dev->ib_dev, "failed to get dwqe mmap entry.\n");
+		return -ENOMEM;
+	}
+
+	rdma_entry = &hr_qp->dwqe_mmap_entry->rdma_entry;
+	resp->dwqe_mmap_key = rdma_user_mmap_get_offset(rdma_entry);
+
+	return 0;
 }
 
 static int alloc_user_qp_db(struct hns_roce_dev *hr_dev,
···
 		hr_qp->en_flags |= HNS_ROCE_QP_CAP_OWNER_DB;
 
 	if (udata) {
+		if (hr_qp->en_flags & HNS_ROCE_QP_CAP_DIRECT_WQE) {
+			ret = qp_mmap_entry(hr_qp, hr_dev, udata, resp);
+			if (ret)
+				return ret;
+		}
+
 		ret = alloc_user_qp_db(hr_dev, hr_qp, init_attr, udata, ucmd,
 				       resp);
 		if (ret)
-			return ret;
+			goto err_remove_qp;
 	} else {
 		ret = alloc_kernel_qp_db(hr_dev, hr_qp, init_attr);
 		if (ret)
···
 	}
 
 	return 0;
+
+err_remove_qp:
+	if (hr_qp->en_flags & HNS_ROCE_QP_CAP_DIRECT_WQE)
+		qp_user_mmap_entry_remove(hr_qp);
+
+	return ret;
 }
 
 static void free_qp_db(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
···
 			hns_roce_db_unmap_user(uctx, &hr_qp->rdb);
 		if (hr_qp->en_flags & HNS_ROCE_QP_CAP_SQ_RECORD_DB)
 			hns_roce_db_unmap_user(uctx, &hr_qp->sdb);
+		if (hr_qp->en_flags & HNS_ROCE_QP_CAP_DIRECT_WQE)
+			qp_user_mmap_entry_remove(hr_qp);
 	} else {
 		if (hr_qp->en_flags & HNS_ROCE_QP_CAP_RQ_RECORD_DB)
 			hns_roce_free_db(hr_dev, &hr_qp->rdb);

include/uapi/rdma/hns-abi.h

···
 	HNS_ROCE_QP_CAP_RQ_RECORD_DB = 1 << 0,
 	HNS_ROCE_QP_CAP_SQ_RECORD_DB = 1 << 1,
 	HNS_ROCE_QP_CAP_OWNER_DB = 1 << 2,
+	HNS_ROCE_QP_CAP_DIRECT_WQE = 1 << 5,
 };
 
 struct hns_roce_ib_create_qp_resp {
 	__aligned_u64 cap_flags;
+	__aligned_u64 dwqe_mmap_key;
 };
 
 struct hns_roce_ib_alloc_ucontext_resp {