RDMA/hns: Configure DCA mode for the userspace QP

driver inclusion
category: feature
bugzilla: https://gitee.com/openeuler/kernel/issues/I63KVU

----------------------------------------------------------

If the userspace driver assign a NULL to the field of 'buf_addr' in
'struct hns_roce_ib_create_qp' when creating QP, this means the kernel
driver need setup the QP as DCA mode. So add a QP capability bit in
response to indicate the userspace driver that the DCA mode has been
enabled.
Signed-off-by: NChengchang Tang <tangchengchang@huawei.com>
Reviewed-by: NYangyang Li <liyangyang20@huawei.com>
Reviewed-by: NYueHaibing <yuehaibing@huawei.com>
Signed-off-by: NZheng Zengkai <zhengzengkai@huawei.com>

drivers/infiniband/hw/hns/hns_roce_dca.c

@@ -239,7 +239,7 @@ static int shrink_dca_page_proc(struct dca_mem *mem, int index, void *param)
 		return DCA_MEM_NEXT_ITERATE;
 }
 
-static int shrink_dca_mem(struct hns_roce_dev *hr_dev,
+static void shrink_dca_mem(struct hns_roce_dev *hr_dev,
 			  struct hns_roce_ucontext *uctx, u64 reserved_size,
 			  struct hns_dca_shrink_resp *resp)
 {
@@ -252,13 +252,11 @@ static int shrink_dca_mem(struct hns_roce_dev *hr_dev,
 	need_shink = ctx->free_mems > 0 && ctx->free_size > reserved_size;
 	spin_unlock_irqrestore(&ctx->pool_lock, flags);
 	if (!need_shink)
-		return 0;
+		return;
 
 	travel_dca_pages(ctx, &attr, shrink_dca_page_proc);
 	resp->free_mems = attr.shrink_mems;
 	resp->free_key = attr.shrink_key;
-
-	return 0;
 }
 
 static void init_dca_context(struct hns_roce_dca_ctx *ctx)
@@ -356,6 +354,21 @@ static void free_dca_mem(struct dca_mem *mem)
 	spin_unlock(&mem->lock);
 }
 
+void hns_roce_enable_dca(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp)
+{
+	struct hns_roce_dca_cfg *cfg = &hr_qp->dca_cfg;
+
+	cfg->buf_id = HNS_DCA_INVALID_BUF_ID;
+}
+
+void hns_roce_disable_dca(struct hns_roce_dev *hr_dev,
+			  struct hns_roce_qp *hr_qp)
+{
+	struct hns_roce_dca_cfg *cfg = &hr_qp->dca_cfg;
+
+	cfg->buf_id = HNS_DCA_INVALID_BUF_ID;
+}
+
 static inline struct hns_roce_ucontext *
 uverbs_attr_to_hr_uctx(struct uverbs_attr_bundle *attrs)
 {
@@ -449,10 +462,8 @@ static int UVERBS_HANDLER(HNS_IB_METHOD_DCA_MEM_SHRINK)(
 	if (ret)
 		return ret;
 
-	ret = shrink_dca_mem(to_hr_dev(uctx->ibucontext.device), uctx,
-			     reserved_size, &resp);
-	if (ret)
-		return ret;
+	shrink_dca_mem(to_hr_dev(uctx->ibucontext.device), uctx,
+		       reserved_size, &resp);
 
 	ret = uverbs_copy_to(attrs, HNS_IB_ATTR_DCA_MEM_SHRINK_OUT_FREE_KEY,
 			     &resp.free_key, sizeof(resp.free_key));

drivers/infiniband/hw/hns/hns_roce_dca.h

@@ -30,4 +30,8 @@ void hns_roce_register_udca(struct hns_roce_dev *hr_dev,
 void hns_roce_unregister_udca(struct hns_roce_dev *hr_dev,
 			      struct hns_roce_ucontext *uctx);
 
+void hns_roce_enable_dca(struct hns_roce_dev *hr_dev,
+			 struct hns_roce_qp *hr_qp);
+void hns_roce_disable_dca(struct hns_roce_dev *hr_dev,
+			  struct hns_roce_qp *hr_qp);
 #endif

drivers/infiniband/hw/hns/hns_roce_device.h

@@ -314,6 +314,10 @@ struct hns_roce_mtr {
 	struct hns_roce_hem_cfg  hem_cfg; /* config for hardware addressing */
 };
 
+struct hns_roce_dca_cfg {
+	u32 buf_id;
+};
+
 struct hns_roce_mw {
 	struct ib_mw		ibmw;
 	u32			pdn;
@@ -610,6 +614,7 @@ struct hns_roce_qp {
 	struct hns_roce_wq	sq;
 
 	struct hns_roce_mtr	mtr;
+	struct hns_roce_dca_cfg	dca_cfg;
 
 	u32			buff_size;
 	struct mutex		mutex;

drivers/infiniband/hw/hns/hns_roce_hw_v2.c

@@ -4707,6 +4707,16 @@ static int modify_qp_init_to_rtr(struct ib_qp *ibqp,
 	hr_reg_write(context, QPC_TRRL_BA_H, trrl_ba >> (32 + 16 + 4));
 	hr_reg_clear(qpc_mask, QPC_TRRL_BA_H);
 
+	if (hr_dev->pci_dev->revision >= PCI_REVISION_ID_HIP09) {
+		if (hr_qp->en_flags & HNS_ROCE_QP_CAP_DCA) {
+			hr_reg_enable(context, QPC_DCA_MODE);
+			hr_reg_clear(qpc_mask, QPC_DCA_MODE);
+		}
+	} else {
+		/* reset IRRL_HEAD */
+		hr_reg_clear(qpc_mask, QPC_V2_IRRL_HEAD);
+	}
+
 	context->irrl_ba = cpu_to_le32(irrl_ba >> 6);
 	qpc_mask->irrl_ba = 0;
 	hr_reg_write(context, QPC_IRRL_BA_H, irrl_ba >> (32 + 6));
@@ -4843,8 +4853,6 @@ static int modify_qp_rtr_to_rts(struct ib_qp *ibqp,
 
 	hr_reg_clear(qpc_mask, QPC_CHECK_FLG);
 
-	hr_reg_clear(qpc_mask, QPC_V2_IRRL_HEAD);
-
 	return 0;
 }
 

drivers/infiniband/hw/hns/hns_roce_qp.c

@@ -38,6 +38,7 @@
 #include "hns_roce_common.h"
 #include "hns_roce_device.h"
 #include "hns_roce_hem.h"
+#include "hns_roce_dca.h"
 
 static void flush_work_handle(struct work_struct *work)
 {
@@ -638,8 +639,21 @@ static int set_user_sq_size(struct hns_roce_dev *hr_dev,
 	return 0;
 }
 
+static bool check_dca_is_enable(struct hns_roce_dev *hr_dev, bool is_user,
+				unsigned long addr)
+{
+	if (!(hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_DCA_MODE))
+		return false;
+
+	/* If the user QP's buffer addr is 0, the DCA mode should be enabled */
+	if (is_user)
+		return !addr;
+
+	return false;
+}
+
 static int set_wqe_buf_attr(struct hns_roce_dev *hr_dev,
-			    struct hns_roce_qp *hr_qp,
+			    struct hns_roce_qp *hr_qp, bool dca_en,
 			    struct hns_roce_buf_attr *buf_attr)
 {
 	int buf_size;
@@ -683,9 +697,21 @@ static int set_wqe_buf_attr(struct hns_roce_dev *hr_dev,
 	if (hr_qp->buff_size < 1)
 		return -EINVAL;
 
-	buf_attr->page_shift = HNS_HW_PAGE_SHIFT + hr_dev->caps.mtt_buf_pg_sz;
 	buf_attr->region_count = idx;
 
+	if (dca_en) {
+		/*
+		 * When enable DCA, there's no need to alloc buffer now, and
+		 * the page shift should be fixed to 4K.
+		 */
+		buf_attr->mtt_only = true;
+		buf_attr->page_shift = HNS_HW_PAGE_SHIFT;
+	} else {
+		buf_attr->mtt_only = false;
+		buf_attr->page_shift = HNS_HW_PAGE_SHIFT +
+				       hr_dev->caps.mtt_buf_pg_sz;
+	}
+
 	return 0;
 }
 
@@ -738,39 +764,75 @@ static int hns_roce_qp_has_rq(struct ib_qp_init_attr *attr)
 	return 1;
 }
 
-static int alloc_qp_buf(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
-			struct ib_qp_init_attr *init_attr,
-			struct ib_udata *udata, unsigned long addr)
+static int alloc_wqe_buf(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
+			 bool dca_en, struct hns_roce_buf_attr *buf_attr,
+			 struct ib_udata *udata, unsigned long addr)
 {
 	struct ib_device *ibdev = &hr_dev->ib_dev;
-	struct hns_roce_buf_attr buf_attr = {};
 	int ret;
 
-	ret = set_wqe_buf_attr(hr_dev, hr_qp, &buf_attr);
-	if (ret) {
-		ibdev_err(ibdev, "failed to split WQE buf, ret = %d.\n", ret);
-		goto err_inline;
+	if (dca_en) {
+		/* DCA must be enabled after the buffer size is configured. */
+		hns_roce_enable_dca(hr_dev, hr_qp);
+
+		hr_qp->en_flags |= HNS_ROCE_QP_CAP_DCA;
+	} else {
+		/*
+		 * Because DCA and DWQE share the same fileds in RCWQE buffer,
+		 * so DWQE only supported when DCA is disable.
+		 */
+		if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_DIRECT_WQE)
+			hr_qp->en_flags |= HNS_ROCE_QP_CAP_DIRECT_WQE;
 	}
-	ret = hns_roce_mtr_create(hr_dev, &hr_qp->mtr, &buf_attr,
+
+	ret = hns_roce_mtr_create(hr_dev, &hr_qp->mtr, buf_attr,
 				  PAGE_SHIFT + hr_dev->caps.mtt_ba_pg_sz,
 				  udata, addr);
 	if (ret) {
 		ibdev_err(ibdev, "failed to create WQE mtr, ret = %d.\n", ret);
-		goto err_inline;
+		if (dca_en)
+			hns_roce_disable_dca(hr_dev, hr_qp);
 	}
 
-	if (hr_dev->caps.flags & HNS_ROCE_CAP_FLAG_DIRECT_WQE)
-		hr_qp->en_flags |= HNS_ROCE_QP_CAP_DIRECT_WQE;
+	return ret;
+}
 
-	return 0;
-err_inline:
+static void free_wqe_buf(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
+			 struct ib_udata *udata)
+{
+	hns_roce_mtr_destroy(hr_dev, &hr_qp->mtr);
+
+	if (hr_qp->en_flags & HNS_ROCE_QP_CAP_DCA)
+		hns_roce_disable_dca(hr_dev, hr_qp);
+}
+
+static int alloc_qp_wqe(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
+			struct ib_qp_init_attr *init_attr,
+			struct ib_udata *udata, unsigned long addr)
+{
+	struct ib_device *ibdev = &hr_dev->ib_dev;
+	struct hns_roce_buf_attr buf_attr = {};
+	bool dca_en;
+	int ret;
+
+	dca_en = check_dca_is_enable(hr_dev, !!udata, addr);
+	ret = set_wqe_buf_attr(hr_dev, hr_qp, dca_en, &buf_attr);
+	if (ret) {
+		ibdev_err(ibdev, "failed to split WQE buf, ret = %d.\n", ret);
+		return ret;
+	}
+
+	ret = alloc_wqe_buf(hr_dev, hr_qp, dca_en, &buf_attr, udata, addr);
+	if (ret)
+		ibdev_err(ibdev, "failed to alloc WQE buf, ret = %d.\n", ret);
 
 	return ret;
 }
 
-static void free_qp_buf(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp)
+static void free_qp_wqe(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
+			struct ib_udata *udata)
 {
-	hns_roce_mtr_destroy(hr_dev, &hr_qp->mtr);
+	free_wqe_buf(hr_dev, hr_qp, udata);
 }
 
 static inline bool user_qp_has_sdb(struct hns_roce_dev *hr_dev,
@@ -1097,18 +1159,18 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
 		}
 	}
 
-	ret = alloc_qp_buf(hr_dev, hr_qp, init_attr, udata, ucmd.buf_addr);
-	if (ret) {
-		ibdev_err(ibdev, "failed to alloc QP buffer, ret = %d.\n", ret);
-		goto err_buf;
-	}
-
 	ret = alloc_qpn(hr_dev, hr_qp);
 	if (ret) {
 		ibdev_err(ibdev, "failed to alloc QPN, ret = %d.\n", ret);
 		goto err_qpn;
 	}
 
+	ret = alloc_qp_wqe(hr_dev, hr_qp, init_attr, udata, ucmd.buf_addr);
+	if (ret) {
+		ibdev_err(ibdev, "failed to alloc QP buffer, ret = %d.\n", ret);
+		goto err_buf;
+	}
+
 	ret = alloc_qp_db(hr_dev, hr_qp, init_attr, udata, &ucmd, &resp);
 	if (ret) {
 		ibdev_err(ibdev, "failed to alloc QP doorbell, ret = %d.\n",
@@ -1159,10 +1221,10 @@ static int hns_roce_create_qp_common(struct hns_roce_dev *hr_dev,
 err_qpc:
 	free_qp_db(hr_dev, hr_qp, udata);
 err_db:
+	free_qp_wqe(hr_dev, hr_qp, udata);
+err_buf:
 	free_qpn(hr_dev, hr_qp);
 err_qpn:
-	free_qp_buf(hr_dev, hr_qp);
-err_buf:
 	free_kernel_wrid(hr_qp);
 	return ret;
 }
@@ -1176,7 +1238,7 @@ void hns_roce_qp_destroy(struct hns_roce_dev *hr_dev, struct hns_roce_qp *hr_qp,
 
 	free_qpc(hr_dev, hr_qp);
 	free_qpn(hr_dev, hr_qp);
-	free_qp_buf(hr_dev, hr_qp);
+	free_qp_wqe(hr_dev, hr_qp, udata);
 	free_kernel_wrid(hr_qp);
 	free_qp_db(hr_dev, hr_qp, udata);
 

include/uapi/rdma/hns-abi.h

@@ -77,6 +77,7 @@ enum hns_roce_qp_cap_flags {
 	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_DCA = 1 << 4,
 	HNS_ROCE_QP_CAP_DIRECT_WQE = 1 << 5,
 };