RDMA/bnxt_re: Enable GSI QP support for 57500 series

In the new 57500 series of adapters the GSI qp is a UD type QP unlike the
previous generation where it was a Raw Eth QP. Changing the control and
data path to support the same. Listing all the significant diffs:

 - AH creation resolve network type unconditionally
 - Add check at relevant places to distinguish from Raw Eth
   processing flow.
 - bnxt_re_process_res_ud_wc report completion with GRH flag
   when qp is GSI.
 - Change length, cfa_meta and smac to match new driver/hardware
   interface.
 - Add new driver/hardware interface.
Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com>
Signed-off-by: Devesh Sharma <devesh.sharma@broadcom.com>
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>

drivers/infiniband/hw/bnxt_re/ib_verbs.c

@@ -663,17 +663,36 @@ int bnxt_re_destroy_ah(struct ib_ah *ib_ah, u32 flags)
 	return 0;
 }
 
+static u8 bnxt_re_stack_to_dev_nw_type(enum rdma_network_type ntype)
+{
+	u8 nw_type;
+
+	switch (ntype) {
+	case RDMA_NETWORK_IPV4:
+		nw_type = CMDQ_CREATE_AH_TYPE_V2IPV4;
+		break;
+	case RDMA_NETWORK_IPV6:
+		nw_type = CMDQ_CREATE_AH_TYPE_V2IPV6;
+		break;
+	default:
+		nw_type = CMDQ_CREATE_AH_TYPE_V1;
+		break;
+	}
+	return nw_type;
+}
+
 struct ib_ah *bnxt_re_create_ah(struct ib_pd *ib_pd,
 				struct rdma_ah_attr *ah_attr,
 				u32 flags,
 				struct ib_udata *udata)
 {
 	struct bnxt_re_pd *pd = container_of(ib_pd, struct bnxt_re_pd, ib_pd);
+	const struct ib_global_route *grh = rdma_ah_read_grh(ah_attr);
 	struct bnxt_re_dev *rdev = pd->rdev;
+	const struct ib_gid_attr *sgid_attr;
 	struct bnxt_re_ah *ah;
-	const struct ib_global_route *grh = rdma_ah_read_grh(ah_attr);
-	int rc;
 	u8 nw_type;
+	int rc;
 
 	if (!(rdma_ah_get_ah_flags(ah_attr) & IB_AH_GRH)) {
 		dev_err(rdev_to_dev(rdev), "Failed to alloc AH: GRH not set");
@@ -700,28 +719,11 @@ struct ib_ah *bnxt_re_create_ah(struct ib_pd *ib_pd,
 	ah->qplib_ah.flow_label = grh->flow_label;
 	ah->qplib_ah.hop_limit = grh->hop_limit;
 	ah->qplib_ah.sl = rdma_ah_get_sl(ah_attr);
-	if (udata &&
-	    !rdma_is_multicast_addr((struct in6_addr *)
-				    grh->dgid.raw) &&
-	    !rdma_link_local_addr((struct in6_addr *)
-				  grh->dgid.raw)) {
-		const struct ib_gid_attr *sgid_attr;
 
-		sgid_attr = grh->sgid_attr;
-		/* Get network header type for this GID */
-		nw_type = rdma_gid_attr_network_type(sgid_attr);
-		switch (nw_type) {
-		case RDMA_NETWORK_IPV4:
-			ah->qplib_ah.nw_type = CMDQ_CREATE_AH_TYPE_V2IPV4;
-			break;
-		case RDMA_NETWORK_IPV6:
-			ah->qplib_ah.nw_type = CMDQ_CREATE_AH_TYPE_V2IPV6;
-			break;
-		default:
-			ah->qplib_ah.nw_type = CMDQ_CREATE_AH_TYPE_V1;
-			break;
-		}
-	}
+	sgid_attr = grh->sgid_attr;
+	/* Get network header type for this GID */
+	nw_type = rdma_gid_attr_network_type(sgid_attr);
+	ah->qplib_ah.nw_type = bnxt_re_stack_to_dev_nw_type(nw_type);
 
 	memcpy(ah->qplib_ah.dmac, ah_attr->roce.dmac, ETH_ALEN);
 	rc = bnxt_qplib_create_ah(&rdev->qplib_res, &ah->qplib_ah,
@@ -1065,12 +1067,17 @@ struct ib_qp *bnxt_re_create_qp(struct ib_pd *ib_pd,
 	qp->qplib_qp.pd = &pd->qplib_pd;
 	qp->qplib_qp.qp_handle = (u64)(unsigned long)(&qp->qplib_qp);
 	qp->qplib_qp.type = __from_ib_qp_type(qp_init_attr->qp_type);
+
+	if (qp_init_attr->qp_type == IB_QPT_GSI &&
+	    bnxt_qplib_is_chip_gen_p5(&rdev->chip_ctx))
+		qp->qplib_qp.type = CMDQ_CREATE_QP_TYPE_GSI;
 	if (qp->qplib_qp.type == IB_QPT_MAX) {
 		dev_err(rdev_to_dev(rdev), "QP type 0x%x not supported",
 			qp->qplib_qp.type);
 		rc = -EINVAL;
 		goto fail;
 	}
+
 	qp->qplib_qp.max_inline_data = qp_init_attr->cap.max_inline_data;
 	qp->qplib_qp.sig_type = ((qp_init_attr->sq_sig_type ==
 				  IB_SIGNAL_ALL_WR) ? true : false);
@@ -1131,7 +1138,8 @@ struct ib_qp *bnxt_re_create_qp(struct ib_pd *ib_pd,
 
 	qp->qplib_qp.mtu = ib_mtu_enum_to_int(iboe_get_mtu(rdev->netdev->mtu));
 
-	if (qp_init_attr->qp_type == IB_QPT_GSI) {
+	if (qp_init_attr->qp_type == IB_QPT_GSI &&
+	    !(bnxt_qplib_is_chip_gen_p5(&rdev->chip_ctx))) {
 		/* Allocate 1 more than what's provided */
 		entries = roundup_pow_of_two(qp_init_attr->cap.max_send_wr + 1);
 		qp->qplib_qp.sq.max_wqe = min_t(u32, entries,
@@ -2091,7 +2099,8 @@ static int bnxt_re_build_qp1_shadow_qp_recv(struct bnxt_re_qp *qp,
 
 static int is_ud_qp(struct bnxt_re_qp *qp)
 {
-	return qp->qplib_qp.type == CMDQ_CREATE_QP_TYPE_UD;
+	return (qp->qplib_qp.type == CMDQ_CREATE_QP_TYPE_UD ||
+		qp->qplib_qp.type == CMDQ_CREATE_QP_TYPE_GSI);
 }
 
 static int bnxt_re_build_send_wqe(struct bnxt_re_qp *qp,
@@ -2395,7 +2404,7 @@ int bnxt_re_post_send(struct ib_qp *ib_qp, const struct ib_send_wr *wr,
 		switch (wr->opcode) {
 		case IB_WR_SEND:
 		case IB_WR_SEND_WITH_IMM:
-			if (ib_qp->qp_type == IB_QPT_GSI) {
+			if (qp->qplib_qp.type == CMDQ_CREATE_QP1_TYPE_GSI) {
 				rc = bnxt_re_build_qp1_send_v2(qp, wr, &wqe,
 							       payload_sz);
 				if (rc)
@@ -2525,7 +2534,8 @@ int bnxt_re_post_recv(struct ib_qp *ib_qp, const struct ib_recv_wr *wr,
 		wqe.wr_id = wr->wr_id;
 		wqe.type = BNXT_QPLIB_SWQE_TYPE_RECV;
 
-		if (ib_qp->qp_type == IB_QPT_GSI)
+		if (ib_qp->qp_type == IB_QPT_GSI &&
+		    qp->qplib_qp.type != CMDQ_CREATE_QP_TYPE_GSI)
 			rc = bnxt_re_build_qp1_shadow_qp_recv(qp, wr, &wqe,
 							      payload_sz);
 		if (!rc)
@@ -3120,19 +3130,33 @@ static void bnxt_re_process_res_shadow_qp_wc(struct bnxt_re_qp *qp,
 	}
 }
 
-static void bnxt_re_process_res_ud_wc(struct ib_wc *wc,
+static void bnxt_re_process_res_ud_wc(struct bnxt_re_qp *qp,
+				      struct ib_wc *wc,
 				      struct bnxt_qplib_cqe *cqe)
 {
+	u8 nw_type;
+
 	wc->opcode = IB_WC_RECV;
 	wc->status = __rc_to_ib_wc_status(cqe->status);
 
-	if (cqe->flags & CQ_RES_RC_FLAGS_IMM)
+	if (cqe->flags & CQ_RES_UD_FLAGS_IMM)
 		wc->wc_flags |= IB_WC_WITH_IMM;
-	if (cqe->flags & CQ_RES_RC_FLAGS_INV)
-		wc->wc_flags |= IB_WC_WITH_INVALIDATE;
-	if ((cqe->flags & (CQ_RES_RC_FLAGS_RDMA | CQ_RES_RC_FLAGS_IMM)) ==
-	    (CQ_RES_RC_FLAGS_RDMA | CQ_RES_RC_FLAGS_IMM))
-		wc->opcode = IB_WC_RECV_RDMA_WITH_IMM;
+	/* report only on GSI QP for Thor */
+	if (qp->qplib_qp.type == CMDQ_CREATE_QP_TYPE_GSI) {
+		wc->wc_flags |= IB_WC_GRH;
+		memcpy(wc->smac, cqe->smac, ETH_ALEN);
+		wc->wc_flags |= IB_WC_WITH_SMAC;
+		if (cqe->flags & CQ_RES_UD_FLAGS_META_FORMAT_VLAN) {
+			wc->vlan_id = (cqe->cfa_meta & 0xFFF);
+			if (wc->vlan_id < 0x1000)
+				wc->wc_flags |= IB_WC_WITH_VLAN;
+		}
+		nw_type = (cqe->flags & CQ_RES_UD_FLAGS_ROCE_IP_VER_MASK) >>
+			   CQ_RES_UD_FLAGS_ROCE_IP_VER_SFT;
+		wc->network_hdr_type = bnxt_re_to_ib_nw_type(nw_type);
+		wc->wc_flags |= IB_WC_WITH_NETWORK_HDR_TYPE;
+	}
+
 }
 
 static int send_phantom_wqe(struct bnxt_re_qp *qp)
@@ -3224,7 +3248,7 @@ int bnxt_re_poll_cq(struct ib_cq *ib_cq, int num_entries, struct ib_wc *wc)
 
 			switch (cqe->opcode) {
 			case CQ_BASE_CQE_TYPE_REQ:
-				if (qp->qplib_qp.id ==
+				if (qp->rdev->qp1_sqp && qp->qplib_qp.id ==
 				    qp->rdev->qp1_sqp->qplib_qp.id) {
 					/* Handle this completion with
 					 * the stored completion
@@ -3259,7 +3283,7 @@ int bnxt_re_poll_cq(struct ib_cq *ib_cq, int num_entries, struct ib_wc *wc)
 				bnxt_re_process_res_rc_wc(wc, cqe);
 				break;
 			case CQ_BASE_CQE_TYPE_RES_UD:
-				if (qp->qplib_qp.id ==
+				if (qp->rdev->qp1_sqp && qp->qplib_qp.id ==
 				    qp->rdev->qp1_sqp->qplib_qp.id) {
 					/* Handle this completion with
 					 * the stored completion
@@ -3272,7 +3296,7 @@ int bnxt_re_poll_cq(struct ib_cq *ib_cq, int num_entries, struct ib_wc *wc)
 						break;
 					}
 				}
-				bnxt_re_process_res_ud_wc(wc, cqe);
+				bnxt_re_process_res_ud_wc(qp, wc, cqe);
 				break;
 			default:
 				dev_err(rdev_to_dev(cq->rdev),

drivers/infiniband/hw/bnxt_re/qplib_fp.c

@@ -44,6 +44,7 @@
 #include <linux/slab.h>
 #include <linux/pci.h>
 #include <linux/prefetch.h>
+#include <linux/if_ether.h>
 
 #include "roce_hsi.h"
 
@@ -1622,7 +1623,8 @@ int bnxt_qplib_post_send(struct bnxt_qplib_qp *qp,
 				((offsetof(typeof(*sqe), data) + 15) >> 4);
 		sqe->inv_key_or_imm_data = cpu_to_le32(
 						wqe->send.inv_key);
-		if (qp->type == CMDQ_CREATE_QP_TYPE_UD) {
+		if (qp->type == CMDQ_CREATE_QP_TYPE_UD ||
+		    qp->type == CMDQ_CREATE_QP_TYPE_GSI) {
 			sqe->q_key = cpu_to_le32(wqe->send.q_key);
 			sqe->dst_qp = cpu_to_le32(
 					wqe->send.dst_qp & SQ_SEND_DST_QP_MASK);
@@ -2408,12 +2410,14 @@ static int bnxt_qplib_cq_process_res_ud(struct bnxt_qplib_cq *cq,
 	}
 	cqe = *pcqe;
 	cqe->opcode = hwcqe->cqe_type_toggle & CQ_BASE_CQE_TYPE_MASK;
-	cqe->length = le32_to_cpu(hwcqe->length);
+	cqe->length = (u32)le16_to_cpu(hwcqe->length);
+	cqe->cfa_meta = le16_to_cpu(hwcqe->cfa_metadata);
 	cqe->invrkey = le32_to_cpu(hwcqe->imm_data);
 	cqe->flags = le16_to_cpu(hwcqe->flags);
 	cqe->status = hwcqe->status;
 	cqe->qp_handle = (u64)(unsigned long)qp;
-	memcpy(cqe->smac, hwcqe->src_mac, 6);
+	/*FIXME: Endianness fix needed for smace */
+	memcpy(cqe->smac, hwcqe->src_mac, ETH_ALEN);
 	wr_id_idx = le32_to_cpu(hwcqe->src_qp_high_srq_or_rq_wr_id)
 				& CQ_RES_UD_SRQ_OR_RQ_WR_ID_MASK;
 	cqe->src_qp = le16_to_cpu(hwcqe->src_qp_low) |

drivers/infiniband/hw/bnxt_re/qplib_fp.h

@@ -347,6 +347,7 @@ struct bnxt_qplib_cqe {
 	u8				type;
 	u8				opcode;
 	u32				length;
+	u16				cfa_meta;
 	u64				wr_id;
 	union {
 		__be32			immdata;

drivers/infiniband/hw/bnxt_re/roce_hsi.h

@@ -515,22 +515,24 @@ struct cq_res_rc {
 
 /* Responder UD CQE (32 bytes) */
 struct cq_res_ud {
-	__le32 length;
+	__le16 length;
 	#define CQ_RES_UD_LENGTH_MASK				    0x3fffUL
 	#define CQ_RES_UD_LENGTH_SFT				    0
-	#define CQ_RES_UD_RESERVED18_MASK			    0xffffc000UL
-	#define CQ_RES_UD_RESERVED18_SFT			    14
+	__le16 cfa_metadata;
+	#define CQ_RES_UD_CFA_METADATA_VID_MASK			0xfffUL
+	#define CQ_RES_UD_CFA_METADATA_VID_SFT			0
+	#define CQ_RES_UD_CFA_METADATA_DE			0x1000UL
+	#define CQ_RES_UD_CFA_METADATA_PRI_MASK			0xe000UL
+	#define CQ_RES_UD_CFA_METADATA_PRI_SFT			13
 	__le32 imm_data;
 	__le64 qp_handle;
 	__le16 src_mac[3];
 	__le16 src_qp_low;
 	u8 cqe_type_toggle;
-	#define CQ_RES_UD_TOGGLE				    0x1UL
-	#define CQ_RES_UD_CQE_TYPE_MASK			    0x1eUL
-	#define CQ_RES_UD_CQE_TYPE_SFT				    1
+	#define CQ_RES_UD_TOGGLE				   0x1UL
+	#define CQ_RES_UD_CQE_TYPE_MASK				   0x1eUL
+	#define CQ_RES_UD_CQE_TYPE_SFT				   1
 	#define CQ_RES_UD_CQE_TYPE_RES_UD			   (0x2UL << 1)
-	#define CQ_RES_UD_RESERVED3_MASK			    0xe0UL
-	#define CQ_RES_UD_RESERVED3_SFT			    5
 	u8 status;
 	#define CQ_RES_UD_STATUS_OK				   0x0UL
 	#define CQ_RES_UD_STATUS_LOCAL_ACCESS_ERROR		   0x1UL
@@ -546,18 +548,30 @@ struct cq_res_ud {
 	#define CQ_RES_UD_FLAGS_SRQ_SRQ			   (0x1UL << 0)
 	#define CQ_RES_UD_FLAGS_SRQ_LAST    CQ_RES_UD_FLAGS_SRQ_SRQ
 	#define CQ_RES_UD_FLAGS_IMM				    0x2UL
-	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_MASK		    0xcUL
-	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_SFT		    2
-	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_V1			   (0x0UL << 2)
-	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_V2IPV4		   (0x2UL << 2)
-	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_V2IPV6		   (0x3UL << 2)
+	#define CQ_RES_UD_FLAGS_UNUSED_MASK			0xcUL
+	#define CQ_RES_UD_FLAGS_UNUSED_SFT			2
+	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_MASK		0x30UL
+	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_SFT			4
+	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_V1			(0x0UL << 4)
+	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_V2IPV4		(0x2UL << 4)
+	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_V2IPV6		(0x3UL << 4)
 	#define CQ_RES_UD_FLAGS_ROCE_IP_VER_LAST		\
 					CQ_RES_UD_FLAGS_ROCE_IP_VER_V2IPV6
+	#define CQ_RES_UD_FLAGS_META_FORMAT_MASK		0x3c0UL
+	#define CQ_RES_UD_FLAGS_META_FORMAT_SFT			6
+	#define CQ_RES_UD_FLAGS_META_FORMAT_NONE		(0x0UL << 6)
+	#define CQ_RES_UD_FLAGS_META_FORMAT_VLAN		(0x1UL << 6)
+	#define CQ_RES_UD_FLAGS_META_FORMAT_TUNNEL_ID		(0x2UL << 6)
+	#define CQ_RES_UD_FLAGS_META_FORMAT_CHDR_DATA		(0x3UL << 6)
+	#define CQ_RES_UD_FLAGS_META_FORMAT_HDR_OFFSET		(0x4UL << 6)
+	#define CQ_RES_UD_FLAGS_META_FORMAT_LAST		\
+					CQ_RES_UD_FLAGS_META_FORMAT_HDR_OFFSET
+	#define CQ_RES_UD_FLAGS_EXT_META_FORMAT_MASK		0xc00UL
+	#define CQ_RES_UD_FLAGS_EXT_META_FORMAT_SFT		10
+
 	__le32 src_qp_high_srq_or_rq_wr_id;
 	#define CQ_RES_UD_SRQ_OR_RQ_WR_ID_MASK			    0xfffffUL
 	#define CQ_RES_UD_SRQ_OR_RQ_WR_ID_SFT			    0
-	#define CQ_RES_UD_RESERVED4_MASK			    0xf00000UL
-	#define CQ_RES_UD_RESERVED4_SFT			    20
 	#define CQ_RES_UD_SRC_QP_HIGH_MASK			    0xff000000UL
 	#define CQ_RES_UD_SRC_QP_HIGH_SFT			    24
 };
@@ -993,6 +1007,7 @@ struct cmdq_create_qp {
 	#define CMDQ_CREATE_QP_TYPE_RC				   0x2UL
 	#define CMDQ_CREATE_QP_TYPE_UD				   0x4UL
 	#define CMDQ_CREATE_QP_TYPE_RAW_ETHERTYPE		   0x6UL
+	#define CMDQ_CREATE_QP_TYPE_GSI				   0x7UL
 	u8 sq_pg_size_sq_lvl;
 	#define CMDQ_CREATE_QP_SQ_LVL_MASK			    0xfUL
 	#define CMDQ_CREATE_QP_SQ_LVL_SFT			    0