From b1d8eb5a213640f1be98a90e73a241d15b70045c Mon Sep 17 00:00:00 2001 From: Jack Morgenstein Date: Tue, 19 Jun 2012 11:21:35 +0300 Subject: IB/mlx4: Add debug prints Define pr_fmt and add some pr_debug prints. Signed-off-by: Jack Morgenstein Signed-off-by: Roland Dreier --- drivers/infiniband/hw/mlx4/mad.c | 19 +++++++++++++++++++ drivers/infiniband/hw/mlx4/main.c | 2 +- drivers/infiniband/hw/mlx4/mlx4_ib.h | 10 ++++++++++ drivers/infiniband/hw/mlx4/qp.c | 27 +++++++++++++++++++++++++-- 4 files changed, 55 insertions(+), 3 deletions(-) (limited to 'drivers/infiniband/hw/mlx4') diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c index 259b0670b51c..84786a9fb64f 100644 --- a/drivers/infiniband/hw/mlx4/mad.c +++ b/drivers/infiniband/hw/mlx4/mad.c @@ -242,6 +242,25 @@ static int ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, int err; struct ib_port_attr pattr; + if (in_wc && in_wc->qp->qp_num) { + pr_debug("received MAD: slid:%d sqpn:%d " + "dlid_bits:%d dqpn:%d wc_flags:0x%x, cls %x, mtd %x, atr %x\n", + in_wc->slid, in_wc->src_qp, + in_wc->dlid_path_bits, + in_wc->qp->qp_num, + in_wc->wc_flags, + in_mad->mad_hdr.mgmt_class, in_mad->mad_hdr.method, + be16_to_cpu(in_mad->mad_hdr.attr_id)); + if (in_wc->wc_flags & IB_WC_GRH) { + pr_debug("sgid_hi:0x%016llx sgid_lo:0x%016llx\n", + be64_to_cpu(in_grh->sgid.global.subnet_prefix), + be64_to_cpu(in_grh->sgid.global.interface_id)); + pr_debug("dgid_hi:0x%016llx dgid_lo:0x%016llx\n", + be64_to_cpu(in_grh->dgid.global.subnet_prefix), + be64_to_cpu(in_grh->dgid.global.interface_id)); + } + } + slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE); if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP && slid == 0) { diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 3530c41fcd1f..5266b49c46ee 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -50,7 +50,7 @@ #include "mlx4_ib.h" #include "user.h" -#define DRV_NAME "mlx4_ib" +#define DRV_NAME MLX4_IB_DRV_NAME #define DRV_VERSION "1.0" #define DRV_RELDATE "April 4, 2008" diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index ff36655d23d3..5f298afaa81f 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -44,6 +44,16 @@ #include #include +#define MLX4_IB_DRV_NAME "mlx4_ib" + +#ifdef pr_fmt +#undef pr_fmt +#endif +#define pr_fmt(fmt) "<" MLX4_IB_DRV_NAME "> %s: " fmt, __func__ + +#define mlx4_ib_warn(ibdev, format, arg...) \ + dev_warn((ibdev)->dma_device, MLX4_IB_DRV_NAME ": " format, ## arg) + enum { MLX4_IB_SQ_MIN_WQE_SHIFT = 6, MLX4_IB_MAX_HEADROOM = 2048 diff --git a/drivers/infiniband/hw/mlx4/qp.c b/drivers/infiniband/hw/mlx4/qp.c index 8d4ed24aef93..84b26963c8d4 100644 --- a/drivers/infiniband/hw/mlx4/qp.c +++ b/drivers/infiniband/hw/mlx4/qp.c @@ -1335,11 +1335,21 @@ int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, cur_state = attr_mask & IB_QP_CUR_STATE ? attr->cur_qp_state : qp->state; new_state = attr_mask & IB_QP_STATE ? attr->qp_state : cur_state; - if (!ib_modify_qp_is_ok(cur_state, new_state, ibqp->qp_type, attr_mask)) + if (!ib_modify_qp_is_ok(cur_state, new_state, ibqp->qp_type, attr_mask)) { + pr_debug("qpn 0x%x: invalid attribute mask specified " + "for transition %d to %d. qp_type %d," + " attr_mask 0x%x\n", + ibqp->qp_num, cur_state, new_state, + ibqp->qp_type, attr_mask); goto out; + } if ((attr_mask & IB_QP_PORT) && (attr->port_num == 0 || attr->port_num > dev->dev->caps.num_ports)) { + pr_debug("qpn 0x%x: invalid port number (%d) specified " + "for transition %d to %d. qp_type %d\n", + ibqp->qp_num, attr->port_num, cur_state, + new_state, ibqp->qp_type); goto out; } @@ -1350,17 +1360,30 @@ int mlx4_ib_modify_qp(struct ib_qp *ibqp, struct ib_qp_attr *attr, if (attr_mask & IB_QP_PKEY_INDEX) { int p = attr_mask & IB_QP_PORT ? attr->port_num : qp->port; - if (attr->pkey_index >= dev->dev->caps.pkey_table_len[p]) + if (attr->pkey_index >= dev->dev->caps.pkey_table_len[p]) { + pr_debug("qpn 0x%x: invalid pkey index (%d) specified " + "for transition %d to %d. qp_type %d\n", + ibqp->qp_num, attr->pkey_index, cur_state, + new_state, ibqp->qp_type); goto out; + } } if (attr_mask & IB_QP_MAX_QP_RD_ATOMIC && attr->max_rd_atomic > dev->dev->caps.max_qp_init_rdma) { + pr_debug("qpn 0x%x: max_rd_atomic (%d) too large. " + "Transition %d to %d. qp_type %d\n", + ibqp->qp_num, attr->max_rd_atomic, cur_state, + new_state, ibqp->qp_type); goto out; } if (attr_mask & IB_QP_MAX_DEST_RD_ATOMIC && attr->max_dest_rd_atomic > dev->dev->caps.max_qp_dest_rdma) { + pr_debug("qpn 0x%x: max_dest_rd_atomic (%d) too large. " + "Transition %d to %d. qp_type %d\n", + ibqp->qp_num, attr->max_dest_rd_atomic, cur_state, + new_state, ibqp->qp_type); goto out; } -- cgit v1.2.1 From 00f5ce99dc6ee46c3113393cc8fa12173f9bbcd7 Mon Sep 17 00:00:00 2001 From: Jack Morgenstein Date: Tue, 19 Jun 2012 11:21:40 +0300 Subject: mlx4: Use port management change event instead of smp_snoop The port management change event can replace smp_snoop. If the capability bit for this event is set in dev-caps, the event is used (by the driver setting the PORT_MNG_CHG_EVENT bit in the async event mask in the MAP_EQ fw command). In this case, when the driver passes incoming SMP PORT_INFO SET mads to the FW, the FW generates port management change events to signal any changes to the driver. If the FW generates these events, smp_snoop shouldn't be invoked in ib_process_mad(), or duplicate events will occur (once from the FW-generated event, and once from smp_snoop). In the case where the FW does not generate port management change events smp_snoop needs to be invoked to create these events. The flow in smp_snoop has been modified to make use of the same procedures as in the fw-generated-event event case to generate the port management events (LID change, Client-rereg, Pkey change, and/or GID change). Port management change event handling required changing the mlx4_ib_event and mlx4_dispatch_event prototypes; the "param" argument (last argument) had to be changed to unsigned long in order to accomodate passing the EQE pointer. We also needed to move the definition of struct mlx4_eqe from net/mlx4.h to file device.h -- to make it available to the IB driver, to handle port management change events. Signed-off-by: Jack Morgenstein Signed-off-by: Or Gerlitz Signed-off-by: Roland Dreier --- drivers/infiniband/hw/mlx4/mad.c | 118 ++++++++++++++++++++++++++--------- drivers/infiniband/hw/mlx4/main.c | 29 +++++++-- drivers/infiniband/hw/mlx4/mlx4_ib.h | 9 +++ 3 files changed, 122 insertions(+), 34 deletions(-) (limited to 'drivers/infiniband/hw/mlx4') diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c index 84786a9fb64f..58c45fb5bd31 100644 --- a/drivers/infiniband/hw/mlx4/mad.c +++ b/drivers/infiniband/hw/mlx4/mad.c @@ -147,47 +147,49 @@ static void update_sm_ah(struct mlx4_ib_dev *dev, u8 port_num, u16 lid, u8 sl) } /* - * Snoop SM MADs for port info and P_Key table sets, so we can - * synthesize LID change and P_Key change events. + * Snoop SM MADs for port info, GUID info, and P_Key table sets, so we can + * synthesize LID change, Client-Rereg, GID change, and P_Key change events. */ static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad, - u16 prev_lid) + u16 prev_lid) { - struct ib_event event; + struct ib_port_info *pinfo; + u16 lid; + struct mlx4_ib_dev *dev = to_mdev(ibdev); if ((mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED || mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) && - mad->mad_hdr.method == IB_MGMT_METHOD_SET) { - if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PORT_INFO) { - struct ib_port_info *pinfo = - (struct ib_port_info *) ((struct ib_smp *) mad)->data; - u16 lid = be16_to_cpu(pinfo->lid); + mad->mad_hdr.method == IB_MGMT_METHOD_SET) + switch (mad->mad_hdr.attr_id) { + case IB_SMP_ATTR_PORT_INFO: + pinfo = (struct ib_port_info *) ((struct ib_smp *) mad)->data; + lid = be16_to_cpu(pinfo->lid); - update_sm_ah(to_mdev(ibdev), port_num, + update_sm_ah(dev, port_num, be16_to_cpu(pinfo->sm_lid), pinfo->neighbormtu_mastersmsl & 0xf); - event.device = ibdev; - event.element.port_num = port_num; + if (pinfo->clientrereg_resv_subnetto & 0x80) + mlx4_ib_dispatch_event(dev, port_num, + IB_EVENT_CLIENT_REREGISTER); - if (pinfo->clientrereg_resv_subnetto & 0x80) { - event.event = IB_EVENT_CLIENT_REREGISTER; - ib_dispatch_event(&event); - } + if (prev_lid != lid) + mlx4_ib_dispatch_event(dev, port_num, + IB_EVENT_LID_CHANGE); + break; - if (prev_lid != lid) { - event.event = IB_EVENT_LID_CHANGE; - ib_dispatch_event(&event); - } - } + case IB_SMP_ATTR_PKEY_TABLE: + mlx4_ib_dispatch_event(dev, port_num, + IB_EVENT_PKEY_CHANGE); + break; - if (mad->mad_hdr.attr_id == IB_SMP_ATTR_PKEY_TABLE) { - event.device = ibdev; - event.event = IB_EVENT_PKEY_CHANGE; - event.element.port_num = port_num; - ib_dispatch_event(&event); + case IB_SMP_ATTR_GUID_INFO: + mlx4_ib_dispatch_event(dev, port_num, + IB_EVENT_GID_CHANGE); + break; + default: + break; } - } } static void node_desc_override(struct ib_device *dev, @@ -305,7 +307,8 @@ static int ib_process_mad(struct ib_device *ibdev, int mad_flags, u8 port_num, return IB_MAD_RESULT_FAILURE; if (!out_mad->mad_hdr.status) { - smp_snoop(ibdev, port_num, in_mad, prev_lid); + if (!(to_mdev(ibdev)->dev->caps.flags & MLX4_DEV_CAP_FLAG_PORT_MNG_CHG_EV)) + smp_snoop(ibdev, port_num, in_mad, prev_lid); node_desc_override(ibdev, out_mad); } @@ -446,3 +449,62 @@ void mlx4_ib_mad_cleanup(struct mlx4_ib_dev *dev) ib_destroy_ah(dev->sm_ah[p]); } } + +void handle_port_mgmt_change_event(struct work_struct *work) +{ + struct ib_event_work *ew = container_of(work, struct ib_event_work, work); + struct mlx4_ib_dev *dev = ew->ib_dev; + struct mlx4_eqe *eqe = &(ew->ib_eqe); + u8 port = eqe->event.port_mgmt_change.port; + u32 changed_attr; + + switch (eqe->subtype) { + case MLX4_DEV_PMC_SUBTYPE_PORT_INFO: + changed_attr = be32_to_cpu(eqe->event.port_mgmt_change.params.port_info.changed_attr); + + /* Update the SM ah - This should be done before handling + the other changed attributes so that MADs can be sent to the SM */ + if (changed_attr & MSTR_SM_CHANGE_MASK) { + u16 lid = be16_to_cpu(eqe->event.port_mgmt_change.params.port_info.mstr_sm_lid); + u8 sl = eqe->event.port_mgmt_change.params.port_info.mstr_sm_sl & 0xf; + update_sm_ah(dev, port, lid, sl); + } + + /* Check if it is a lid change event */ + if (changed_attr & MLX4_EQ_PORT_INFO_LID_CHANGE_MASK) + mlx4_ib_dispatch_event(dev, port, IB_EVENT_LID_CHANGE); + + /* Generate GUID changed event */ + if (changed_attr & MLX4_EQ_PORT_INFO_GID_PFX_CHANGE_MASK) + mlx4_ib_dispatch_event(dev, port, IB_EVENT_GID_CHANGE); + + if (changed_attr & MLX4_EQ_PORT_INFO_CLIENT_REREG_MASK) + mlx4_ib_dispatch_event(dev, port, + IB_EVENT_CLIENT_REREGISTER); + break; + + case MLX4_DEV_PMC_SUBTYPE_PKEY_TABLE: + mlx4_ib_dispatch_event(dev, port, IB_EVENT_PKEY_CHANGE); + break; + case MLX4_DEV_PMC_SUBTYPE_GUID_INFO: + mlx4_ib_dispatch_event(dev, port, IB_EVENT_GID_CHANGE); + break; + default: + pr_warn("Unsupported subtype 0x%x for " + "Port Management Change event\n", eqe->subtype); + } + + kfree(ew); +} + +void mlx4_ib_dispatch_event(struct mlx4_ib_dev *dev, u8 port_num, + enum ib_event_type type) +{ + struct ib_event event; + + event.device = &dev->ib_dev; + event.element.port_num = port_num; + event.event = type; + + ib_dispatch_event(&event); +} diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 5266b49c46ee..4f230c26622d 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -898,7 +898,6 @@ static void update_gids_task(struct work_struct *work) union ib_gid *gids; int err; struct mlx4_dev *dev = gw->dev->dev; - struct ib_event event; mailbox = mlx4_alloc_cmd_mailbox(dev); if (IS_ERR(mailbox)) { @@ -916,10 +915,7 @@ static void update_gids_task(struct work_struct *work) pr_warn("set port command failed\n"); else { memcpy(gw->dev->iboe.gid_table[gw->port - 1], gw->gids, sizeof gw->gids); - event.device = &gw->dev->ib_dev; - event.element.port_num = gw->port; - event.event = IB_EVENT_GID_CHANGE; - ib_dispatch_event(&event); + mlx4_ib_dispatch_event(gw->dev, gw->port, IB_EVENT_GID_CHANGE); } mlx4_free_cmd_mailbox(dev, mailbox); @@ -1383,10 +1379,18 @@ static void mlx4_ib_remove(struct mlx4_dev *dev, void *ibdev_ptr) } static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr, - enum mlx4_dev_event event, int port) + enum mlx4_dev_event event, unsigned long param) { struct ib_event ibev; struct mlx4_ib_dev *ibdev = to_mdev((struct ib_device *) ibdev_ptr); + struct mlx4_eqe *eqe = NULL; + struct ib_event_work *ew; + int port = 0; + + if (event == MLX4_DEV_EVENT_PORT_MGMT_CHANGE) + eqe = (struct mlx4_eqe *)param; + else + port = (u8)param; if (port > ibdev->num_ports) return; @@ -1405,6 +1409,19 @@ static void mlx4_ib_event(struct mlx4_dev *dev, void *ibdev_ptr, ibev.event = IB_EVENT_DEVICE_FATAL; break; + case MLX4_DEV_EVENT_PORT_MGMT_CHANGE: + ew = kmalloc(sizeof *ew, GFP_ATOMIC); + if (!ew) { + pr_err("failed to allocate memory for events work\n"); + break; + } + + INIT_WORK(&ew->work, handle_port_mgmt_change_event); + memcpy(&ew->ib_eqe, eqe, sizeof *eqe); + ew->ib_dev = ibdev; + handle_port_mgmt_change_event(&ew->work); + return; + default: return; } diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index 5f298afaa81f..23bfbf9ee0e0 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -224,6 +224,12 @@ struct mlx4_ib_dev { int eq_added; }; +struct ib_event_work { + struct work_struct work; + struct mlx4_ib_dev *ib_dev; + struct mlx4_eqe ib_eqe; +}; + static inline struct mlx4_ib_dev *to_mdev(struct ib_device *ibdev) { return container_of(ibdev, struct mlx4_ib_dev, ib_dev); @@ -381,4 +387,7 @@ static inline int mlx4_ib_ah_grh_present(struct mlx4_ib_ah *ah) int mlx4_ib_add_mc(struct mlx4_ib_dev *mdev, struct mlx4_ib_qp *mqp, union ib_gid *gid); +void mlx4_ib_dispatch_event(struct mlx4_ib_dev *dev, u8 port_num, + enum ib_event_type type); + #endif /* MLX4_IB_H */ -- cgit v1.2.1 From 47e956b2a6a4539f948ff2b85dfd259704dc5021 Mon Sep 17 00:00:00 2001 From: Dotan Barak Date: Wed, 11 Jul 2012 15:39:29 +0000 Subject: IB/mlx4: Fill the masked_atomic_cap attribute in query device When the user queries for device capabilities, fill in the masked_atomic_cap attribute with the real support level of atomic capabilities instead of using a hard coded value. Signed-off-by: Dotan Barak Reviewed-by: Eli Cohen Signed-off-by: Or Gerlitz Signed-off-by: Roland Dreier --- drivers/infiniband/hw/mlx4/main.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/infiniband/hw/mlx4') diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 3530c41fcd1f..07efe4b84a74 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -157,7 +157,7 @@ static int mlx4_ib_query_device(struct ib_device *ibdev, props->local_ca_ack_delay = dev->dev->caps.local_ca_ack_delay; props->atomic_cap = dev->dev->caps.flags & MLX4_DEV_CAP_FLAG_ATOMIC ? IB_ATOMIC_HCA : IB_ATOMIC_NONE; - props->masked_atomic_cap = IB_ATOMIC_HCA; + props->masked_atomic_cap = props->atomic_cap; props->max_pkeys = dev->dev->caps.pkey_table_len[1]; props->max_mcast_grp = dev->dev->caps.num_mgms + dev->dev->caps.num_amgms; props->max_mcast_qp_attach = dev->dev->caps.num_qp_per_mgm; -- cgit v1.2.1 From 6634961c14d38ef64ec284c07aecb03d3dd03b4a Mon Sep 17 00:00:00 2001 From: Jack Morgenstein Date: Tue, 19 Jun 2012 11:21:44 +0300 Subject: mlx4: Put physical GID and P_Key table sizes in mlx4_phys_caps struct and paravirtualize them To allow easy paravirtualization of P_Key and GID table sizes, keep paravirtualized sizes in mlx4_dev->caps, but save the actual physical sizes from FW in struct: mlx4_dev->phys_cap. In addition, in SR-IOV mode, do the following: 1. Reduce reported P_Key table size by 1. This is done to reserve the highest P_Key index for internal use, for declaring an invalid P_Key in P_Key paravirtualization. We require a P_Key index which always contain an invalid P_Key value for this purpose (i.e., one which cannot be modified by the subnet manager). The way to do this is to reduce the P_Key table size reported to the subnet manager by 1, so that it will not attempt to access the P_Key at index #127. 2. Paravirtualize the GID table size to 1. Thus, each guest sees only a single GID (at its paravirtualized index 0). In addition, since we are paravirtualizing the GID table size to 1, we add paravirtualization of the master GID event here (i.e., we do not do ib_dispatch_event() for the GUID change event on the master, since its (only) GUID never changes). Signed-off-by: Jack Morgenstein Signed-off-by: Roland Dreier --- drivers/infiniband/hw/mlx4/mad.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) (limited to 'drivers/infiniband/hw/mlx4') diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c index 58c45fb5bd31..c27141fef1ab 100644 --- a/drivers/infiniband/hw/mlx4/mad.c +++ b/drivers/infiniband/hw/mlx4/mad.c @@ -184,8 +184,10 @@ static void smp_snoop(struct ib_device *ibdev, u8 port_num, struct ib_mad *mad, break; case IB_SMP_ATTR_GUID_INFO: - mlx4_ib_dispatch_event(dev, port_num, - IB_EVENT_GID_CHANGE); + /* paravirtualized master's guid is guid 0 -- does not change */ + if (!mlx4_is_master(dev->dev)) + mlx4_ib_dispatch_event(dev, port_num, + IB_EVENT_GID_CHANGE); break; default: break; @@ -487,7 +489,9 @@ void handle_port_mgmt_change_event(struct work_struct *work) mlx4_ib_dispatch_event(dev, port, IB_EVENT_PKEY_CHANGE); break; case MLX4_DEV_PMC_SUBTYPE_GUID_INFO: - mlx4_ib_dispatch_event(dev, port, IB_EVENT_GID_CHANGE); + /* paravirtualized master's guid is guid 0 -- does not change */ + if (!mlx4_is_master(dev->dev)) + mlx4_ib_dispatch_event(dev, port, IB_EVENT_GID_CHANGE); break; default: pr_warn("Unsupported subtype 0x%x for " -- cgit v1.2.1