Hi Som,

> -----Original Message-----
> From: linux-rdma-ow...@vger.kernel.org [mailto:linux-rdma-
> ow...@vger.kernel.org] On Behalf Of Somnath Kotur
> Sent: Friday, February 20, 2015 3:33 AM
> To: rol...@kernel.org
> Cc: linux-rdma@vger.kernel.org; Devesh Sharma; Somnath Kotur
> Subject: [PATCH 15/30] RDMA/ocrdma: changes to support RoCE-v2 in UD path
> 
> From: Devesh Sharma <devesh.sha...@emulex.com>
> 
> To support UD protocol this patch adds following changes to existing UD
> implementation.
> 
> 1. AH creation resolves gid-type for a given index.
> 2. Based on GID-type protocol header is built.
> 3. Work completion reports l3-type if f/w supports RoCE-v2
>    and sets IB_WC_WITH_NETWORK_HDR_TYPE flag in wc->wc_flags.
> 
> Signed-off-by: Somnath Kotur <somnath.ko...@emulex.com>
> Signed-off-by: Devesh Sharma <devesh.sha...@emulex.com>
> ---
>  drivers/infiniband/hw/ocrdma/ocrdma.h       |    1 +
>  drivers/infiniband/hw/ocrdma/ocrdma_ah.c    |   68
> ++++++++++++++++++++++-----
>  drivers/infiniband/hw/ocrdma/ocrdma_sli.h   |    5 ++-
>  drivers/infiniband/hw/ocrdma/ocrdma_verbs.c |   23 +++++++--
>  4 files changed, 80 insertions(+), 17 deletions(-)
> 
> diff --git a/drivers/infiniband/hw/ocrdma/ocrdma.h
> b/drivers/infiniband/hw/ocrdma/ocrdma.h
> index 97f971a..302fd0e 100644
> --- a/drivers/infiniband/hw/ocrdma/ocrdma.h
> +++ b/drivers/infiniband/hw/ocrdma/ocrdma.h
> @@ -341,6 +341,7 @@ 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 7ecd230..70a885b 100644
> --- a/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
> +++ b/drivers/infiniband/hw/ocrdma/ocrdma_ah.c
> @@ -39,6 +39,20 @@
> 
>  #define OCRDMA_VID_PCP_SHIFT 0xD
> 
> +static u16 ocrdma_hdr_type_to_proto_num(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:
> +             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) @@ -47,22 +61,32
> @@ 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;
> +     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(ah->hdr_type);
> +
> 
>       /* VLAN */
>       if (!vlan_tag || (vlan_tag > 0xFFF))
>               vlan_tag = dev->pvid;
>       if (vlan_tag && (vlan_tag < 0x1000)) {
>               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 */
> @@ -71,18 +95,34 @@ 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 = 0x11;
> +             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));
> +             /* 0x1b is next header value in GRH */
> +             grh.pdid_hoplimit = cpu_to_be32((pdid << 16) |
> +                                             (0x11 << 8) |

We should decide next-header = 0x1b/0x11 based on the proto_num


> +                                             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); @@ -106,6 +146,7 @@
> 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);
> @@ -126,6 +167,9 @@ struct ib_ah *ocrdma_create_ah(struct ib_pd *ibpd,
> struct ib_ah_attr *attr)
>               vlan_tag = vlan_dev_vlan_id(sgid_attr.ndev);
>       rcu_read_unlock();
> 
> +     /* Get network header type for this GID */
> +     ah->hdr_type = ib_gid_to_network_type(sgid_attr.gid_type, &sgid);
> +
>       if (pd->uctx) {
>               status = rdma_addr_find_dmac_by_grh(&sgid, &attr->grh.dgid,
>                                                   attr->dmac, &vlan_tag,
> diff --git a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
> b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
> index 6b74eb9..4fb68ee 100644
> --- a/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
> +++ b/drivers/infiniband/hw/ocrdma/ocrdma_sli.h
> @@ -1681,8 +1681,11 @@ 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,
> @@ -1807,7 +1810,7 @@ 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 17c80d5..90a6351 100644
> --- a/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
> +++ b/drivers/infiniband/hw/ocrdma/ocrdma_verbs.c
> @@ -31,7 +31,6 @@
>  #include <rdma/iw_cm.h>
>  #include <rdma/ib_umem.h>
>  #include <rdma/ib_addr.h>
> -#include <rdma/ib_cache.h>
> 
>  #include "ocrdma.h"
>  #include "ocrdma_hw.h"
> @@ -1963,6 +1962,7 @@ static void ocrdma_build_ud_hdr(struct ocrdma_qp
> *qp,
>       else
>               ud_hdr->qkey = wr->wr.ud.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);  } @@ -2698,9 +2698,11 @@ 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; @@ -2710,7 +2712,17 @@ 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_rocev2_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;
>  }
> 
> @@ -2773,12 +2785,15 @@ 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);
> 
> --
> 1.7.1
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the 
> body
> of a message to majord...@vger.kernel.org More majordomo info at
> http://vger.kernel.org/majordomo-info.html
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majord...@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Reply via email to