[PATCH 0/4] RDMA/ocrdma: Add RoCE-v2 support

[PATCH 0/4] RDMA/ocrdma: Add RoCE-v2 support

[Devesh Sharma]

This patch series adds support for RoCE-v2 protocol
on Avago's RoCE enabled adapters.

Devesh Sharma (3):
  RDMA/ocrdma: Support RoCE-v2 in the UD path
  RDMA/ocrdma: Support RoCE-v2 in the RC path
  RDMA/ocrdma: Support user AH creation for RoCE-v2

Somnath Kotur (1):
  RDMA/ocrdma: Export udp encapsulation capability

 drivers/infiniband/hw/ocrdma/ocrdma.h       |  8 +++
 drivers/infiniband/hw/ocrdma/ocrdma_ah.c    | 75 ++++++++++++++++++++++++-----
 drivers/infiniband/hw/ocrdma/ocrdma_ah.h    |  5 +-
 drivers/infiniband/hw/ocrdma/ocrdma_hw.c    | 26 +++++++++-
 drivers/infiniband/hw/ocrdma/ocrdma_main.c  |  4 ++
 drivers/infiniband/hw/ocrdma/ocrdma_sli.h   | 16 ++++--
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 22 +++++++--
 7 files changed, 134 insertions(+), 22 deletions(-)

-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 1/4] RDMA/ocrdma: Export udp encapsulation capability

[Devesh Sharma]

From: Somnath Kotur <somnath.kotur-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>

Add support to read device configuration and initialize port-immutables
to report UDP-Encap flag during port query.

Signed-off-by: Devesh Sharma <devesh.sharma-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>
Signed-off-by: Somnath Kotur <somnath.kotur-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>
---
 drivers/infiniband/hw/ocrdma/ocrdma.h      | 7 +++++++
 drivers/infiniband/hw/ocrdma/ocrdma_hw.c   | 3 +++
 drivers/infiniband/hw/ocrdma/ocrdma_main.c | 4 ++++
 drivers/infiniband/hw/ocrdma/ocrdma_sli.h  | 9 +++++++--
 4 files changed, 21 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/ocrdma/ocrdma.h b/drivers/infiniband/hw/ocrdma/ocrdma.h
index 040bb8b..3ef0b52 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma.h
@@ -114,6 +114,7 @@ struct ocrdma_dev_attr {
 	u8 local_ca_ack_delay;
 	u8 ird;
 	u8 num_ird_pages;
+	u8 udp_encap;
 };
 
 struct ocrdma_dma_mem {
@@ -601,4 +602,10 @@ static inline u8 ocrdma_get_ae_link_state(u32 ae_state)
 	return ((ae_state & OCRDMA_AE_LSC_LS_MASK) >> OCRDMA_AE_LSC_LS_SHIFT);
 }
 
+static inline bool ocrdma_is_udp_encap_supported(struct ocrdma_dev *dev)
+{
+	return (dev->attr.udp_encap & OCRDMA_L3_TYPE_IPV4) ||
+	       (dev->attr.udp_encap & OCRDMA_L3_TYPE_IPV6);
+}
+
 #endif
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
index 283ca84..aea7f17 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
@@ -1144,6 +1144,9 @@ static void ocrdma_get_attr(struct ocrdma_dev *dev,
 	attr->max_pd =
 	    (rsp->max_pd_ca_ack_delay & OCRDMA_MBX_QUERY_CFG_MAX_PD_MASK) >>
 	    OCRDMA_MBX_QUERY_CFG_MAX_PD_SHIFT;
+	attr->udp_encap = (rsp->max_pd_ca_ack_delay &
+			   OCRDMA_MBX_QUERY_CFG_L3_TYPE_MASK) >>
+			   OCRDMA_MBX_QUERY_CFG_L3_TYPE_SHIFT;
 	attr->max_dpp_pds =
 	   (rsp->max_dpp_pds_credits & OCRDMA_MBX_QUERY_CFG_MAX_DPP_PDS_MASK) >>
 	    OCRDMA_MBX_QUERY_CFG_MAX_DPP_PDS_OFFSET;
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_main.c b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
index 5738493..f1f610b 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_main.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_main.c
@@ -89,8 +89,10 @@ static int ocrdma_port_immutable(struct ib_device *ibdev, u8 port_num,
 			         struct ib_port_immutable *immutable)
 {
 	struct ib_port_attr attr;
+	struct ocrdma_dev *dev;
 	int err;
 
+	dev = get_ocrdma_dev(ibdev);
 	err = ocrdma_query_port(ibdev, port_num, &attr);
 	if (err)
 		return err;
@@ -98,6 +100,8 @@ static int ocrdma_port_immutable(struct ib_device *ibdev, u8 port_num,
 	immutable->pkey_tbl_len = attr.pkey_tbl_len;
 	immutable->gid_tbl_len = attr.gid_tbl_len;
 	immutable->core_cap_flags = RDMA_CORE_PORT_IBA_ROCE;
+	if (ocrdma_is_udp_encap_supported(dev))
+		immutable->core_cap_flags |= RDMA_CORE_CAP_PROT_ROCE_UDP_ENCAP;
 	immutable->max_mad_size = IB_MGMT_MAD_SIZE;
 
 	return 0;
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
index 99dd6fd..8d75bd4 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
@@ -140,7 +140,11 @@ enum {
 	OCRDMA_DB_RQ_SHIFT		= 24
 };
 
-#define OCRDMA_ROUDP_FLAGS_SHIFT	0x03
+enum {
+	OCRDMA_L3_TYPE_IB_GRH   = 0x00,
+	OCRDMA_L3_TYPE_IPV4     = 0x01,
+	OCRDMA_L3_TYPE_IPV6     = 0x02
+};
 
 #define OCRDMA_DB_CQ_RING_ID_MASK       0x3FF	/* bits 0 - 9 */
 #define OCRDMA_DB_CQ_RING_ID_EXT_MASK  0x0C00	/* bits 10-11 of qid at 12-11 */
@@ -546,7 +550,8 @@ enum {
 	OCRDMA_MBX_QUERY_CFG_CA_ACK_DELAY_SHIFT		= 8,
 	OCRDMA_MBX_QUERY_CFG_CA_ACK_DELAY_MASK		= 0xFF <<
 				OCRDMA_MBX_QUERY_CFG_CA_ACK_DELAY_SHIFT,
-
+	OCRDMA_MBX_QUERY_CFG_L3_TYPE_SHIFT		= 3,
+	OCRDMA_MBX_QUERY_CFG_L3_TYPE_MASK		= 0x18,
 	OCRDMA_MBX_QUERY_CFG_MAX_SEND_SGE_SHIFT		= 0,
 	OCRDMA_MBX_QUERY_CFG_MAX_SEND_SGE_MASK		= 0xFFFF,
 	OCRDMA_MBX_QUERY_CFG_MAX_WRITE_SGE_SHIFT	= 16,
-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 2/4] RDMA/ocrdma: Support RoCE-v2 in the UD path

[Devesh Sharma]

This patch adds following changes to support RoCE-v2
in the UD path.

 * During AH creation GID-type is resolved for a given gid-index.
 * Based on GID-type protocol header is built.
 * Work completion reports network header type and set
   IB_WC_WITH_NETWORK_HDR_TYPE flag in wc->wc_flags to indicate
   that the network header type is valid.

Signed-off-by: Somnath Kotur <somnath.kotur-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>
Signed-off-by: Devesh Sharma <devesh.sharma-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>
---
 drivers/infiniband/hw/ocrdma/ocrdma.h       |  1 +
 drivers/infiniband/hw/ocrdma/ocrdma_ah.c    | 70 ++++++++++++++++++++++++-----
 drivers/infiniband/hw/ocrdma/ocrdma_sli.h   |  5 ++-
 drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 22 +++++++--
 4 files changed, 82 insertions(+), 16 deletions(-)

diff --git a/drivers/infiniband/hw/ocrdma/ocrdma.h b/drivers/infiniband/hw/ocrdma/ocrdma.h
index 3ef0b52..b716259 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma.h
@@ -360,6 +360,7 @@ struct ocrdma_ah {
 	struct ocrdma_av *av;
 	u16 sgid_index;
 	u32 id;
+	u8 hdr_type;
 };
 
 struct ocrdma_qp_hwq_info {
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
index 3790771..4aed1db 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
@@ -55,6 +55,21 @@
 
 #define OCRDMA_VID_PCP_SHIFT	0xD
 
+static u16 ocrdma_hdr_type_to_proto_num(int devid, u8 hdr_type)
+{
+	switch (hdr_type) {
+	case OCRDMA_L3_TYPE_IB_GRH:
+		return (u16)0x8915;
+	case OCRDMA_L3_TYPE_IPV4:
+		return (u16)0x0800;
+	case OCRDMA_L3_TYPE_IPV6:
+		return (u16)0x86dd;
+	default:
+		pr_err("ocrdma%d: Invalid network header\n", devid);
+		return 0;
+	}
+}
+
 static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah,
 			struct ib_ah_attr *attr, union ib_gid *sgid,
 			int pdid, bool *isvlan, u16 vlan_tag)
@@ -63,10 +78,23 @@ static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah,
 	struct ocrdma_eth_vlan eth;
 	struct ocrdma_grh grh;
 	int eth_sz;
+	u16 proto_num = 0;
+	u8 nxthdr = 0x11;
+	struct iphdr ipv4;
+	union {
+		struct sockaddr     _sockaddr;
+		struct sockaddr_in  _sockaddr_in;
+		struct sockaddr_in6 _sockaddr_in6;
+	} sgid_addr, dgid_addr;
 
 	memset(&eth, 0, sizeof(eth));
 	memset(&grh, 0, sizeof(grh));
 
+	/* Protocol Number */
+	proto_num = ocrdma_hdr_type_to_proto_num(dev->id, ah->hdr_type);
+	if (!proto_num)
+		return -EINVAL;
+	nxthdr = (proto_num == 0x8915) ? 0x1b : 0x11;
 	/* VLAN */
 	if (!vlan_tag || (vlan_tag > 0xFFF))
 		vlan_tag = dev->pvid;
@@ -78,13 +106,13 @@ static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah,
 				dev->id);
 		}
 		eth.eth_type = cpu_to_be16(0x8100);
-		eth.roce_eth_type = cpu_to_be16(OCRDMA_ROCE_ETH_TYPE);
+		eth.roce_eth_type = cpu_to_be16(proto_num);
 		vlan_tag |= (dev->sl & 0x07) << OCRDMA_VID_PCP_SHIFT;
 		eth.vlan_tag = cpu_to_be16(vlan_tag);
 		eth_sz = sizeof(struct ocrdma_eth_vlan);
 		*isvlan = true;
 	} else {
-		eth.eth_type = cpu_to_be16(OCRDMA_ROCE_ETH_TYPE);
+		eth.eth_type = cpu_to_be16(proto_num);
 		eth_sz = sizeof(struct ocrdma_eth_basic);
 	}
 	/* MAC */
@@ -93,18 +121,33 @@ static inline int set_av_attr(struct ocrdma_dev *dev, struct ocrdma_ah *ah,
 	if (status)
 		return status;
 	ah->sgid_index = attr->grh.sgid_index;
-	memcpy(&grh.sgid[0], sgid->raw, sizeof(union ib_gid));
-	memcpy(&grh.dgid[0], attr->grh.dgid.raw, sizeof(attr->grh.dgid.raw));
-
-	grh.tclass_flow = cpu_to_be32((6 << 28) |
-			(attr->grh.traffic_class << 24) |
-			attr->grh.flow_label);
-	/* 0x1b is next header value in GRH */
-	grh.pdid_hoplimit = cpu_to_be32((pdid << 16) |
-			(0x1b << 8) | attr->grh.hop_limit);
 	/* Eth HDR */
 	memcpy(&ah->av->eth_hdr, &eth, eth_sz);
-	memcpy((u8 *)ah->av + eth_sz, &grh, sizeof(struct ocrdma_grh));
+	if (ah->hdr_type == RDMA_NETWORK_IPV4) {
+		*((__be16 *)&ipv4) = htons((4 << 12) | (5 << 8) |
+					   attr->grh.traffic_class);
+		ipv4.id = cpu_to_be16(pdid);
+		ipv4.frag_off = htons(IP_DF);
+		ipv4.tot_len = htons(0);
+		ipv4.ttl = attr->grh.hop_limit;
+		ipv4.protocol = nxthdr;
+		rdma_gid2ip(&sgid_addr._sockaddr, sgid);
+		ipv4.saddr = sgid_addr._sockaddr_in.sin_addr.s_addr;
+		rdma_gid2ip(&dgid_addr._sockaddr, &attr->grh.dgid);
+		ipv4.daddr = dgid_addr._sockaddr_in.sin_addr.s_addr;
+		memcpy((u8 *)ah->av + eth_sz, &ipv4, sizeof(struct iphdr));
+	} else {
+		memcpy(&grh.sgid[0], sgid->raw, sizeof(union ib_gid));
+		grh.tclass_flow = cpu_to_be32((6 << 28) |
+					      (attr->grh.traffic_class << 24) |
+					      attr->grh.flow_label);
+		memcpy(&grh.dgid[0], attr->grh.dgid.raw,
+		       sizeof(attr->grh.dgid.raw));
+		grh.pdid_hoplimit = cpu_to_be32((pdid << 16) |
+						(nxthdr << 8) |
+						attr->grh.hop_limit);
+		memcpy((u8 *)ah->av + eth_sz, &grh, sizeof(struct ocrdma_grh));
+	}
 	if (*isvlan)
 		ah->av->valid |= OCRDMA_AV_VLAN_VALID;
 	ah->av->valid = cpu_to_le32(ah->av->valid);
@@ -128,6 +171,7 @@ struct ib_ah *ocrdma_create_ah(struct ib_pd *ibpd, struct ib_ah_attr *attr)
 
 	if (atomic_cmpxchg(&dev->update_sl, 1, 0))
 		ocrdma_init_service_level(dev);
+
 	ah = kzalloc(sizeof(*ah), GFP_ATOMIC);
 	if (!ah)
 		return ERR_PTR(-ENOMEM);
@@ -148,6 +192,8 @@ struct ib_ah *ocrdma_create_ah(struct ib_pd *ibpd, struct ib_ah_attr *attr)
 			vlan_tag = vlan_dev_vlan_id(sgid_attr.ndev);
 		dev_put(sgid_attr.ndev);
 	}
+	/* Get network header type for this GID */
+	ah->hdr_type = ib_gid_to_network_type(sgid_attr.gid_type, &sgid);
 
 	if ((pd->uctx) &&
 	    (!rdma_is_multicast_addr((struct in6_addr *)attr->grh.dgid.raw)) &&
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
index 8d75bd4..3d15948 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
@@ -1740,8 +1740,11 @@ enum {
 
 	/* w1 */
 	OCRDMA_CQE_UD_XFER_LEN_SHIFT	= 16,
+	OCRDMA_CQE_UD_XFER_LEN_MASK     = 0x1FFF,
 	OCRDMA_CQE_PKEY_SHIFT		= 0,
 	OCRDMA_CQE_PKEY_MASK		= 0xFFFF,
+	OCRDMA_CQE_UD_L3TYPE_SHIFT      = 29,
+	OCRDMA_CQE_UD_L3TYPE_MASK       = 0x07,
 
 	/* w2 */
 	OCRDMA_CQE_QPN_SHIFT		= 0,
@@ -1866,7 +1869,7 @@ struct ocrdma_ewqe_ud_hdr {
 	u32 rsvd_dest_qpn;
 	u32 qkey;
 	u32 rsvd_ahid;
-	u32 rsvd;
+	u32 hdr_type;
 };
 
 /* extended wqe followed by hdr_wqe for Fast Memory register */
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
index d4c687b..a8e7391 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
@@ -2006,6 +2006,7 @@ static void ocrdma_build_ud_hdr(struct ocrdma_qp *qp,
 	else
 		ud_hdr->qkey = ud_wr(wr)->remote_qkey;
 	ud_hdr->rsvd_ahid = ah->id;
+	ud_hdr->hdr_type = ah->hdr_type;
 	if (ah->av->valid & OCRDMA_AV_VLAN_VALID)
 		hdr->cw |= (OCRDMA_FLAG_AH_VLAN_PR << OCRDMA_WQE_FLAGS_SHIFT);
 }
@@ -2718,9 +2719,11 @@ static bool ocrdma_poll_scqe(struct ocrdma_qp *qp, struct ocrdma_cqe *cqe,
 	return expand;
 }
 
-static int ocrdma_update_ud_rcqe(struct ib_wc *ibwc, struct ocrdma_cqe *cqe)
+static int ocrdma_update_ud_rcqe(struct ocrdma_dev *dev, struct ib_wc *ibwc,
+				 struct ocrdma_cqe *cqe)
 {
 	int status;
+	u16 hdr_type = 0;
 
 	status = (le32_to_cpu(cqe->flags_status_srcqpn) &
 		OCRDMA_CQE_UD_STATUS_MASK) >> OCRDMA_CQE_UD_STATUS_SHIFT;
@@ -2730,7 +2733,17 @@ static int ocrdma_update_ud_rcqe(struct ib_wc *ibwc, struct ocrdma_cqe *cqe)
 						OCRDMA_CQE_PKEY_MASK;
 	ibwc->wc_flags = IB_WC_GRH;
 	ibwc->byte_len = (le32_to_cpu(cqe->ud.rxlen_pkey) >>
-					OCRDMA_CQE_UD_XFER_LEN_SHIFT);
+			  OCRDMA_CQE_UD_XFER_LEN_SHIFT) &
+			  OCRDMA_CQE_UD_XFER_LEN_MASK;
+
+	if (ocrdma_is_udp_encap_supported(dev)) {
+		hdr_type = (le32_to_cpu(cqe->ud.rxlen_pkey) >>
+			    OCRDMA_CQE_UD_L3TYPE_SHIFT) &
+			    OCRDMA_CQE_UD_L3TYPE_MASK;
+		ibwc->wc_flags |= IB_WC_WITH_NETWORK_HDR_TYPE;
+		ibwc->network_hdr_type = hdr_type;
+	}
+
 	return status;
 }
 
@@ -2793,12 +2806,15 @@ static bool ocrdma_poll_err_rcqe(struct ocrdma_qp *qp, struct ocrdma_cqe *cqe,
 static void ocrdma_poll_success_rcqe(struct ocrdma_qp *qp,
 				     struct ocrdma_cqe *cqe, struct ib_wc *ibwc)
 {
+	struct ocrdma_dev *dev;
+
+	dev = get_ocrdma_dev(qp->ibqp.device);
 	ibwc->opcode = IB_WC_RECV;
 	ibwc->qp = &qp->ibqp;
 	ibwc->status = IB_WC_SUCCESS;
 
 	if (qp->qp_type == IB_QPT_UD || qp->qp_type == IB_QPT_GSI)
-		ocrdma_update_ud_rcqe(ibwc, cqe);
+		ocrdma_update_ud_rcqe(dev, ibwc, cqe);
 	else
 		ibwc->byte_len = le32_to_cpu(cqe->rq.rxlen);
 
-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 3/4] RDMA/ocrdma: Support RoCE-v2 in the RC path

[Devesh Sharma]

This patch implements following changes to support RoCE-v2
in the RC path:

* Get the GID-type for a given sgid.
* Based on the GID-type get IPv4/IPv6 L3-address
  and give those to underlying device.
* Resolve and provide network header type to device.

Signed-off-by: Somnath Kotur <somnath.kotur-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>
Signed-off-by: Devesh Sharma <devesh.sharma-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>
---
 drivers/infiniband/hw/ocrdma/ocrdma_hw.c  | 23 +++++++++++++++++++++--
 drivers/infiniband/hw/ocrdma/ocrdma_sli.h |  2 ++
 2 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
index aea7f17..2cfbf15 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_hw.c
@@ -2504,7 +2504,12 @@ static int ocrdma_set_av_params(struct ocrdma_qp *qp,
 	union ib_gid sgid, zgid;
 	struct ib_gid_attr sgid_attr;
 	u32 vlan_id = 0xFFFF;
-	u8 mac_addr[6];
+	u8 mac_addr[6], hdr_type;
+	union {
+		struct sockaddr     _sockaddr;
+		struct sockaddr_in  _sockaddr_in;
+		struct sockaddr_in6 _sockaddr_in6;
+	} sgid_addr, dgid_addr;
 	struct ocrdma_dev *dev = get_ocrdma_dev(qp->ibqp.device);
 
 	if ((ah_attr->ah_flags & IB_AH_GRH) == 0)
@@ -2519,6 +2524,8 @@ static int ocrdma_set_av_params(struct ocrdma_qp *qp,
 	cmd->params.hop_lmt_rq_psn |=
 	    (ah_attr->grh.hop_limit << OCRDMA_QP_PARAMS_HOP_LMT_SHIFT);
 	cmd->flags |= OCRDMA_QP_PARA_FLOW_LBL_VALID;
+
+	/* GIDs */
 	memcpy(&cmd->params.dgid[0], &ah_attr->grh.dgid.raw[0],
 	       sizeof(cmd->params.dgid));
 
@@ -2541,6 +2548,16 @@ static int ocrdma_set_av_params(struct ocrdma_qp *qp,
 		return status;
 	cmd->params.dmac_b0_to_b3 = mac_addr[0] | (mac_addr[1] << 8) |
 				(mac_addr[2] << 16) | (mac_addr[3] << 24);
+
+	hdr_type = ib_gid_to_network_type(sgid_attr.gid_type, &sgid);
+	if (hdr_type == RDMA_NETWORK_IPV4) {
+		rdma_gid2ip(&sgid_addr._sockaddr, &sgid);
+		rdma_gid2ip(&dgid_addr._sockaddr, &ah_attr->grh.dgid);
+		memcpy(&cmd->params.dgid[0],
+		       &dgid_addr._sockaddr_in.sin_addr.s_addr, 4);
+		memcpy(&cmd->params.sgid[0],
+		       &sgid_addr._sockaddr_in.sin_addr.s_addr, 4);
+	}
 	/* convert them to LE format. */
 	ocrdma_cpu_to_le32(&cmd->params.dgid[0], sizeof(cmd->params.dgid));
 	ocrdma_cpu_to_le32(&cmd->params.sgid[0], sizeof(cmd->params.sgid));
@@ -2561,7 +2578,9 @@ static int ocrdma_set_av_params(struct ocrdma_qp *qp,
 		cmd->params.rnt_rc_sl_fl |=
 			(dev->sl & 0x07) << OCRDMA_QP_PARAMS_SL_SHIFT;
 	}
-
+	cmd->params.max_sge_recv_flags |= ((hdr_type <<
+					OCRDMA_QP_PARAMS_FLAGS_L3_TYPE_SHIFT) &
+					OCRDMA_QP_PARAMS_FLAGS_L3_TYPE_MASK);
 	return 0;
 }
 
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
index 3d15948..0efc966 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
@@ -1112,6 +1112,8 @@ enum {
 	OCRDMA_QP_PARAMS_STATE_MASK		= BIT(5) | BIT(6) | BIT(7),
 	OCRDMA_QP_PARAMS_FLAGS_SQD_ASYNC	= BIT(8),
 	OCRDMA_QP_PARAMS_FLAGS_INB_ATEN		= BIT(9),
+	OCRDMA_QP_PARAMS_FLAGS_L3_TYPE_SHIFT    = 11,
+	OCRDMA_QP_PARAMS_FLAGS_L3_TYPE_MASK     = BIT(11) | BIT(12) | BIT(13),
 	OCRDMA_QP_PARAMS_MAX_SGE_RECV_SHIFT	= 16,
 	OCRDMA_QP_PARAMS_MAX_SGE_RECV_MASK	= 0xFFFF <<
 					OCRDMA_QP_PARAMS_MAX_SGE_RECV_SHIFT,
-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

[PATCH 4/4] RDMA/ocrdma: Support user AH creation for RoCE-v2

[Devesh Sharma]

This patch adds support to create RoCE-v2 compatible AH. It uses ahid
field to tell network-header-type to user space library. The library
has to decode network-header-type from ahid field.

Signed-off-by: Somnath Kotur <somnath.kotur-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>
Signed-off-by: Devesh Sharma <devesh.sharma-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org>
---
 drivers/infiniband/hw/ocrdma/ocrdma_ah.c | 5 +++++
 drivers/infiniband/hw/ocrdma/ocrdma_ah.h | 5 +++--
 2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
index 4aed1db..e3c4f17 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
@@ -218,6 +218,11 @@ struct ib_ah *ocrdma_create_ah(struct ib_pd *ibpd, struct ib_ah_attr *attr)
 		ahid_addr = pd->uctx->ah_tbl.va + attr->dlid;
 		*ahid_addr = 0;
 		*ahid_addr |= ah->id & OCRDMA_AH_ID_MASK;
+		if (ocrdma_is_udp_encap_supported(dev)) {
+			*ahid_addr |= ((u32)ah->hdr_type &
+				       OCRDMA_AH_L3_TYPE_MASK) <<
+				       OCRDMA_AH_L3_TYPE_SHIFT;
+		}
 		if (isvlan)
 			*ahid_addr |= (OCRDMA_AH_VLAN_VALID_MASK <<
 				       OCRDMA_AH_VLAN_VALID_SHIFT);
diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_ah.h b/drivers/infiniband/hw/ocrdma/ocrdma_ah.h
index 04a30ae..3856dd4 100644
--- a/drivers/infiniband/hw/ocrdma/ocrdma_ah.h
+++ b/drivers/infiniband/hw/ocrdma/ocrdma_ah.h
@@ -46,9 +46,10 @@
 enum {
 	OCRDMA_AH_ID_MASK		= 0x3FF,
 	OCRDMA_AH_VLAN_VALID_MASK	= 0x01,
-	OCRDMA_AH_VLAN_VALID_SHIFT	= 0x1F
+	OCRDMA_AH_VLAN_VALID_SHIFT	= 0x1F,
+	OCRDMA_AH_L3_TYPE_MASK		= 0x03,
+	OCRDMA_AH_L3_TYPE_SHIFT		= 0x1D /* 29 bits */
 };
-
 struct ib_ah *ocrdma_create_ah(struct ib_pd *, struct ib_ah_attr *);
 int ocrdma_destroy_ah(struct ib_ah *);
 int ocrdma_query_ah(struct ib_ah *, struct ib_ah_attr *);
-- 
1.8.3.1

--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/4] RDMA/ocrdma: Add RoCE-v2 support

[Devesh Sharma]

Hi Dough,

This is a gentle reminder to pull this patch series into upstream.
Please let me know if you are targeting a specific kernel version.

-Regards
Devesh

On Thu, Jan 28, 2016 at 7:29 PM, Devesh Sharma
<devesh.sharma-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org> wrote:
> This patch series adds support for RoCE-v2 protocol
> on Avago's RoCE enabled adapters.
>
> Devesh Sharma (3):
>   RDMA/ocrdma: Support RoCE-v2 in the UD path
>   RDMA/ocrdma: Support RoCE-v2 in the RC path
>   RDMA/ocrdma: Support user AH creation for RoCE-v2
>
> Somnath Kotur (1):
>   RDMA/ocrdma: Export udp encapsulation capability
>
>  drivers/infiniband/hw/ocrdma/ocrdma.h       |  8 +++
>  drivers/infiniband/hw/ocrdma/ocrdma_ah.c    | 75 ++++++++++++++++++++++++-----
>  drivers/infiniband/hw/ocrdma/ocrdma_ah.h    |  5 +-
>  drivers/infiniband/hw/ocrdma/ocrdma_hw.c    | 26 +++++++++-
>  drivers/infiniband/hw/ocrdma/ocrdma_main.c  |  4 ++
>  drivers/infiniband/hw/ocrdma/ocrdma_sli.h   | 16 ++++--
>  drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 22 +++++++--
>  7 files changed, 134 insertions(+), 22 deletions(-)
>
> --
> 1.8.3.1
>
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/4] RDMA/ocrdma: Add RoCE-v2 support

[Devesh Sharma]

Hi Doug,

Ping, any update about the timeline of this patch series?

-Regards
Devesh

On Mon, Feb 8, 2016 at 4:18 PM, Devesh Sharma
<devesh.sharma-dY08KVG/lbpWk0Htik3J/w@public.gmane.org> wrote:
> Hi Dough,
>
> This is a gentle reminder to pull this patch series into upstream.
> Please let me know if you are targeting a specific kernel version.
>
> -Regards
> Devesh
>
> On Thu, Jan 28, 2016 at 7:29 PM, Devesh Sharma
> <devesh.sharma-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org> wrote:
>> This patch series adds support for RoCE-v2 protocol
>> on Avago's RoCE enabled adapters.
>>
>> Devesh Sharma (3):
>>   RDMA/ocrdma: Support RoCE-v2 in the UD path
>>   RDMA/ocrdma: Support RoCE-v2 in the RC path
>>   RDMA/ocrdma: Support user AH creation for RoCE-v2
>>
>> Somnath Kotur (1):
>>   RDMA/ocrdma: Export udp encapsulation capability
>>
>>  drivers/infiniband/hw/ocrdma/ocrdma.h       |  8 +++
>>  drivers/infiniband/hw/ocrdma/ocrdma_ah.c    | 75 ++++++++++++++++++++++++-----
>>  drivers/infiniband/hw/ocrdma/ocrdma_ah.h    |  5 +-
>>  drivers/infiniband/hw/ocrdma/ocrdma_hw.c    | 26 +++++++++-
>>  drivers/infiniband/hw/ocrdma/ocrdma_main.c  |  4 ++
>>  drivers/infiniband/hw/ocrdma/ocrdma_sli.h   | 16 ++++--
>>  drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 22 +++++++--
>>  7 files changed, 134 insertions(+), 22 deletions(-)
>>
>> --
>> 1.8.3.1
>>
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 0/4] RDMA/ocrdma: Add RoCE-v2 support

[Doug Ledford]

[-- Attachment #1: Type: text/plain, Size: 1989 bytes --]

On 03/02/2016 10:04 PM, Devesh Sharma wrote:
> Hi Doug,
> 
> Ping, any update about the timeline of this patch series?

I already have these pulled in.  I have an ocrdma branch in my github
repo that is going to be part of the overall 4.6 merge and these are in it.

> -Regards
> Devesh
> 
> On Mon, Feb 8, 2016 at 4:18 PM, Devesh Sharma
> <devesh.sharma-dY08KVG/lbpWk0Htik3J/w@public.gmane.org> wrote:
>> Hi Dough,
>>
>> This is a gentle reminder to pull this patch series into upstream.
>> Please let me know if you are targeting a specific kernel version.
>>
>> -Regards
>> Devesh
>>
>> On Thu, Jan 28, 2016 at 7:29 PM, Devesh Sharma
>> <devesh.sharma-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org> wrote:
>>> This patch series adds support for RoCE-v2 protocol
>>> on Avago's RoCE enabled adapters.
>>>
>>> Devesh Sharma (3):
>>>   RDMA/ocrdma: Support RoCE-v2 in the UD path
>>>   RDMA/ocrdma: Support RoCE-v2 in the RC path
>>>   RDMA/ocrdma: Support user AH creation for RoCE-v2
>>>
>>> Somnath Kotur (1):
>>>   RDMA/ocrdma: Export udp encapsulation capability
>>>
>>>  drivers/infiniband/hw/ocrdma/ocrdma.h       |  8 +++
>>>  drivers/infiniband/hw/ocrdma/ocrdma_ah.c    | 75 ++++++++++++++++++++++++-----
>>>  drivers/infiniband/hw/ocrdma/ocrdma_ah.h    |  5 +-
>>>  drivers/infiniband/hw/ocrdma/ocrdma_hw.c    | 26 +++++++++-
>>>  drivers/infiniband/hw/ocrdma/ocrdma_main.c  |  4 ++
>>>  drivers/infiniband/hw/ocrdma/ocrdma_sli.h   | 16 ++++--
>>>  drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 22 +++++++--
>>>  7 files changed, 134 insertions(+), 22 deletions(-)
>>>
>>> --
>>> 1.8.3.1
>>>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 


-- 
Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
              GPG KeyID: 0E572FDD



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 884 bytes --]

Re: [PATCH 0/4] RDMA/ocrdma: Add RoCE-v2 support

[Devesh Sharma]

Thanks Doug for a quick update.

-Regards
Devesh

On Thu, Mar 3, 2016 at 8:21 PM, Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> On 03/02/2016 10:04 PM, Devesh Sharma wrote:
>> Hi Doug,
>>
>> Ping, any update about the timeline of this patch series?
>
> I already have these pulled in.  I have an ocrdma branch in my github
> repo that is going to be part of the overall 4.6 merge and these are in it.
>
>> -Regards
>> Devesh
>>
>> On Mon, Feb 8, 2016 at 4:18 PM, Devesh Sharma
>> <devesh.sharma-dY08KVG/lbpWk0Htik3J/w@public.gmane.org> wrote:
>>> Hi Dough,
>>>
>>> This is a gentle reminder to pull this patch series into upstream.
>>> Please let me know if you are targeting a specific kernel version.
>>>
>>> -Regards
>>> Devesh
>>>
>>> On Thu, Jan 28, 2016 at 7:29 PM, Devesh Sharma
>>> <devesh.sharma-1wcpHE2jlwO1Z/+hSey0Gg@public.gmane.org> wrote:
>>>> This patch series adds support for RoCE-v2 protocol
>>>> on Avago's RoCE enabled adapters.
>>>>
>>>> Devesh Sharma (3):
>>>>   RDMA/ocrdma: Support RoCE-v2 in the UD path
>>>>   RDMA/ocrdma: Support RoCE-v2 in the RC path
>>>>   RDMA/ocrdma: Support user AH creation for RoCE-v2
>>>>
>>>> Somnath Kotur (1):
>>>>   RDMA/ocrdma: Export udp encapsulation capability
>>>>
>>>>  drivers/infiniband/hw/ocrdma/ocrdma.h       |  8 +++
>>>>  drivers/infiniband/hw/ocrdma/ocrdma_ah.c    | 75 ++++++++++++++++++++++++-----
>>>>  drivers/infiniband/hw/ocrdma/ocrdma_ah.h    |  5 +-
>>>>  drivers/infiniband/hw/ocrdma/ocrdma_hw.c    | 26 +++++++++-
>>>>  drivers/infiniband/hw/ocrdma/ocrdma_main.c  |  4 ++
>>>>  drivers/infiniband/hw/ocrdma/ocrdma_sli.h   | 16 ++++--
>>>>  drivers/infiniband/hw/ocrdma/ocrdma_verbs.c | 22 +++++++--
>>>>  7 files changed, 134 insertions(+), 22 deletions(-)
>>>>
>>>> --
>>>> 1.8.3.1
>>>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
>> the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>
>
> --
> Doug Ledford <dledford-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
>               GPG KeyID: 0E572FDD
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html