5 files changed, 43 insertions, 55 deletions

diff --git a/drivers/infiniband/hw/mlx4/cq.c b/drivers/infiniband/hw/mlx4/cq.c
index 4521319b1406..299f20832ab6 100644
--- a/drivers/infiniband/hw/mlx4/cq.c
+++ b/drivers/infiniband/hw/mlx4/cq.c
@@ -663,18 +663,18 @@ repoll:
 
                 switch (cqe->owner_sr_opcode & MLX4_CQE_OPCODE_MASK) {
                 case MLX4_RECV_OPCODE_RDMA_WRITE_IMM:
-                        wc->opcode   = IB_WC_RECV_RDMA_WITH_IMM;
-                        wc->wc_flags = IB_WC_WITH_IMM;
-                        wc->imm_data = cqe->immed_rss_invalid;
+                        wc->opcode      = IB_WC_RECV_RDMA_WITH_IMM;
+                        wc->wc_flags    = IB_WC_WITH_IMM;
+                        wc->ex.imm_data = cqe->immed_rss_invalid;
                         break;
                 case MLX4_RECV_OPCODE_SEND:
                         wc->opcode   = IB_WC_RECV;
                         wc->wc_flags = 0;
                         break;
                 case MLX4_RECV_OPCODE_SEND_IMM:
-                        wc->opcode   = IB_WC_RECV;
-                        wc->wc_flags = IB_WC_WITH_IMM;
-                        wc->imm_data = cqe->immed_rss_invalid;
+                        wc->opcode      = IB_WC_RECV;
+                        wc->wc_flags    = IB_WC_WITH_IMM;
+                        wc->ex.imm_data = cqe->immed_rss_invalid;
                         break;
                 }
 
diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c
index 4c1e72fc8f57..cdca3a511e1c 100644
--- a/drivers/infiniband/hw/mlx4/mad.c
+++ b/drivers/infiniband/hw/mlx4/mad.c
@@ -255,7 +255,8 @@ int mlx4_ib_process_mad(struct ib_device *ibdev, int mad_flags,	u8 port_num,
                         return IB_MAD_RESULT_SUCCESS;
         } else if (in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_PERF_MGMT ||
                    in_mad->mad_hdr.mgmt_class == MLX4_IB_VENDOR_CLASS1   ||
-                   in_mad->mad_hdr.mgmt_class == MLX4_IB_VENDOR_CLASS2) {
+                   in_mad->mad_hdr.mgmt_class == MLX4_IB_VENDOR_CLASS2   ||
+                   in_mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_CONG_MGMT) {
                 if (in_mad->mad_hdr.method  != IB_MGMT_METHOD_GET &&
                     in_mad->mad_hdr.method  != IB_MGMT_METHOD_SET)
                         return IB_MAD_RESULT_SUCCESS;
diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c
index 4d61e32866c6..bcf50648fa18 100644
--- a/drivers/infiniband/hw/mlx4/main.c
+++ b/drivers/infiniband/hw/mlx4/main.c
@@ -90,7 +90,8 @@ static int mlx4_ib_query_device(struct ib_device *ibdev,
         props->device_cap_flags    = IB_DEVICE_CHANGE_PHY_PORT |
                 IB_DEVICE_PORT_ACTIVE_EVENT             |
                 IB_DEVICE_SYS_IMAGE_GUID                |
-                IB_DEVICE_RC_RNR_NAK_GEN;
+                IB_DEVICE_RC_RNR_NAK_GEN                |
+                IB_DEVICE_BLOCK_MULTICAST_LOOPBACK;
         if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BAD_PKEY_CNTR)
                 props->device_cap_flags |= IB_DEVICE_BAD_PKEY_CNTR;
         if (dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_BAD_QKEY_CNTR)
@@ -437,7 +438,9 @@ static int mlx4_ib_dealloc_pd(struct ib_pd *pd)
 static int mlx4_ib_mcg_attach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
 {
         return mlx4_multicast_attach(to_mdev(ibqp->device)->dev,
-                                     &to_mqp(ibqp)->mqp, gid->raw);
+                                     &to_mqp(ibqp)->mqp, gid->raw,
+                                     !!(to_mqp(ibqp)->flags &
+                                        MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK));
 }
 
 static int mlx4_ib_mcg_detach(struct ib_qp *ibqp, union ib_gid *gid, u16 lid)
diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h
index 5cf994794d25..c4cf5b69eefa 100644
--- a/drivers/infiniband/hw/mlx4/mlx4_ib.h
+++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h
@@ -101,7 +101,8 @@ struct mlx4_ib_wq {
 };
 
 enum mlx4_ib_qp_flags {
-        MLX4_IB_QP_LSO          = 1 << 0
+        MLX4_IB_QP_LSO                          = 1 << 0,
+        MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK     = 1 << 1,
 };
 
 struct mlx4_ib_qp {
diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c
index a80df22deae8..89eb6cbe592e 100644
--- a/drivers/infiniband/hw/mlx4/qp.c
+++ b/drivers/infiniband/hw/mlx4/qp.c
@@ -129,9 +129,10 @@ static void stamp_send_wqe(struct mlx4_ib_qp *qp, int n, int size)
         int ind;
         void *buf;
         __be32 stamp;
+        struct mlx4_wqe_ctrl_seg *ctrl;
 
-        s = roundup(size, 1U << qp->sq.wqe_shift);
         if (qp->sq_max_wqes_per_wr > 1) {
+                s = roundup(size, 1U << qp->sq.wqe_shift);
                 for (i = 0; i < s; i += 64) {
                         ind = (i >> qp->sq.wqe_shift) + n;
                         stamp = ind & qp->sq.wqe_cnt ? cpu_to_be32(0x7fffffff) :
@@ -141,7 +142,8 @@ static void stamp_send_wqe(struct mlx4_ib_qp *qp, int n, int size)
                         *wqe = stamp;
                 }
         } else {
-                buf = get_send_wqe(qp, n & (qp->sq.wqe_cnt - 1));
+                ctrl = buf = get_send_wqe(qp, n & (qp->sq.wqe_cnt - 1));
+                s = (ctrl->fence_size & 0x3f) << 4;
                 for (i = 64; i < s; i += 64) {
                         wqe = buf + i;
                         *wqe = cpu_to_be32(0xffffffff);
@@ -452,19 +454,8 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd,
         spin_lock_init(&qp->rq.lock);
 
         qp->state        = IB_QPS_RESET;
-        qp->atomic_rd_en = 0;
-        qp->resp_depth   = 0;
-
-        qp->rq.head         = 0;
-        qp->rq.tail         = 0;
-        qp->sq.head         = 0;
-        qp->sq.tail         = 0;
-        qp->sq_next_wqe     = 0;
-
         if (init_attr->sq_sig_type == IB_SIGNAL_ALL_WR)
                 qp->sq_signal_bits = cpu_to_be32(MLX4_WQE_CTRL_CQ_UPDATE);
-        else
-                qp->sq_signal_bits = 0;
 
         err = set_rq_size(dev, &init_attr->cap, !!pd->uobject, !!init_attr->srq, qp);
         if (err)
@@ -509,6 +500,9 @@ static int create_qp_common(struct mlx4_ib_dev *dev, struct ib_pd *pd,
         } else {
                 qp->sq_no_prefetch = 0;
 
+                if (init_attr->create_flags & IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK)
+                        qp->flags |= MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK;
+
                 if (init_attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO)
                         qp->flags |= MLX4_IB_QP_LSO;
 
@@ -682,10 +676,15 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
         struct mlx4_ib_qp *qp;
         int err;
 
-        /* We only support LSO, and only for kernel UD QPs. */
-        if (init_attr->create_flags & ~IB_QP_CREATE_IPOIB_UD_LSO)
+        /*
+         * We only support LSO and multicast loopback blocking, and
+         * only for kernel UD QPs.
+         */
+        if (init_attr->create_flags & ~(IB_QP_CREATE_IPOIB_UD_LSO |
+                                        IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK))
                 return ERR_PTR(-EINVAL);
-        if (init_attr->create_flags & IB_QP_CREATE_IPOIB_UD_LSO &&
+
+        if (init_attr->create_flags &&
             (pd->uobject || init_attr->qp_type != IB_QPT_UD))
                 return ERR_PTR(-EINVAL);
 
@@ -694,7 +693,7 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
         case IB_QPT_UC:
         case IB_QPT_UD:
         {
-                qp = kmalloc(sizeof *qp, GFP_KERNEL);
+                qp = kzalloc(sizeof *qp, GFP_KERNEL);
                 if (!qp)
                         return ERR_PTR(-ENOMEM);
 
@@ -715,7 +714,7 @@ struct ib_qp *mlx4_ib_create_qp(struct ib_pd *pd,
                 if (pd->uobject)
                         return ERR_PTR(-EINVAL);
 
-                sqp = kmalloc(sizeof *sqp, GFP_KERNEL);
+                sqp = kzalloc(sizeof *sqp, GFP_KERNEL);
                 if (!sqp)
                         return ERR_PTR(-ENOMEM);
 
@@ -906,7 +905,8 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
                                attr->path_mtu);
                         goto out;
                 }
-                context->mtu_msgmax = (attr->path_mtu << 5) | 31;
+                context->mtu_msgmax = (attr->path_mtu << 5) |
+                        ilog2(dev->dev->caps.max_msg_sz);
         }
 
         if (qp->rq.wqe_cnt)
@@ -1063,6 +1063,8 @@ static int __mlx4_ib_modify_qp(struct ib_qp *ibqp,
                 for (i = 0; i < qp->sq.wqe_cnt; ++i) {
                         ctrl = get_send_wqe(qp, i);
                         ctrl->owner_opcode = cpu_to_be32(1 << 31);
+                        if (qp->sq_max_wqes_per_wr == 1)
+                                ctrl->fence_size = 1 << (qp->sq.wqe_shift - 4);
 
                         stamp_send_wqe(qp, i, 1 << qp->sq.wqe_shift);
                 }
@@ -1127,23 +1129,6 @@ out:
         return err;
 }
 
-static const struct ib_qp_attr mlx4_ib_qp_attr = { .port_num = 1 };
-static const int mlx4_ib_qp_attr_mask_table[IB_QPT_UD + 1] = {
-                [IB_QPT_UD]  = (IB_QP_PKEY_INDEX                |
-                                IB_QP_PORT                      |
-                                IB_QP_QKEY),
-                [IB_QPT_UC]  = (IB_QP_PKEY_INDEX                |
-                                IB_QP_PORT                      |
-                                IB_QP_ACCESS_FLAGS),
-                [IB_QPT_RC]  = (IB_QP_PKEY_INDEX                |
-                                IB_QP_PORT                      |
-                                IB_QP_ACCESS_FLAGS),
-                [IB_QPT_SMI] = (IB_QP_PKEY_INDEX                |
-                                IB_QP_QKEY),
-                [IB_QPT_GSI] = (IB_QP_PKEY_INDEX                |
-                                IB_QP_QKEY),
-};
-
 int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
                       int attr_mask, struct ib_udata *udata)
 {
@@ -1186,15 +1171,6 @@ int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr,
                 goto out;
         }
 
-        if (cur_state == IB_QPS_RESET && new_state == IB_QPS_ERR) {
-                err = __mlx4_ib_modify_qp(ibqp, &mlx4_ib_qp_attr,
-                                          mlx4_ib_qp_attr_mask_table[ibqp->qp_type],
-                                          IB_QPS_RESET, IB_QPS_INIT);
-                if (err)
-                        goto out;
-                cur_state = IB_QPS_INIT;
-        }
-
         err = __mlx4_ib_modify_qp(ibqp, attr, attr_mask, cur_state, new_state);
 
 out:
@@ -1865,6 +1841,13 @@ done:
 
         qp_init_attr->cap            = qp_attr->cap;
 
+        qp_init_attr->create_flags = 0;
+        if (qp->flags & MLX4_IB_QP_BLOCK_MULTICAST_LOOPBACK)
+                qp_init_attr->create_flags |= IB_QP_CREATE_BLOCK_MULTICAST_LOOPBACK;
+
+        if (qp->flags & MLX4_IB_QP_LSO)
+                qp_init_attr->create_flags |= IB_QP_CREATE_IPOIB_UD_LSO;
+
 out:
         mutex_unlock(&qp->mutex);
         return err;