From patchwork Thu Jul 30 15:33:29 2015 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Matan Barak X-Patchwork-Id: 6903641 Return-Path: X-Original-To: patchwork-linux-rdma@patchwork.kernel.org Delivered-To: patchwork-parsemail@patchwork1.web.kernel.org Received: from mail.kernel.org (mail.kernel.org [198.145.29.136]) by patchwork1.web.kernel.org (Postfix) with ESMTP id B7F169F358 for ; Thu, 30 Jul 2015 15:35:03 +0000 (UTC) Received: from mail.kernel.org (localhost [127.0.0.1]) by mail.kernel.org (Postfix) with ESMTP id 86134205DB for ; Thu, 30 Jul 2015 15:35:02 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 353AF20361 for ; Thu, 30 Jul 2015 15:35:00 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752750AbbG3Pe6 (ORCPT ); Thu, 30 Jul 2015 11:34:58 -0400 Received: from [193.47.165.129] ([193.47.165.129]:45292 "EHLO mellanox.co.il" rhost-flags-FAIL-FAIL-OK-FAIL) by vger.kernel.org with ESMTP id S1752375AbbG3Pe4 (ORCPT ); Thu, 30 Jul 2015 11:34:56 -0400 Received: from Internal Mail-Server by MTLPINE1 (envelope-from matanb@mellanox.com) with ESMTPS (AES256-SHA encrypted); 30 Jul 2015 18:34:05 +0300 Received: from r-vnc06.mtr.labs.mlnx (r-vnc06.mtr.labs.mlnx [10.208.0.117]) by labmailer.mlnx (8.13.8/8.13.8) with ESMTP id t6UFY49u008864; Thu, 30 Jul 2015 18:34:05 +0300 From: Matan Barak To: Doug Ledford Cc: Matan Barak , Or Gerlitz , Jason Gunthorpe , linux-rdma@vger.kernel.org, Sean Hefty , Somnath Kotur , Moni Shoua , talal@mellanox.com, haggaie@mellanox.com Subject: [PATCH for-next V7 08/10] IB/mlx4: Implement ib_device callbacks Date: Thu, 30 Jul 2015 18:33:29 +0300 Message-Id: <1438270411-17648-9-git-send-email-matanb@mellanox.com> X-Mailer: git-send-email 1.7.6.4 In-Reply-To: <1438270411-17648-1-git-send-email-matanb@mellanox.com> References: <1438270411-17648-1-git-send-email-matanb@mellanox.com> Sender: linux-rdma-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-rdma@vger.kernel.org X-Spam-Status: No, score=-8.3 required=5.0 tests=BAYES_00, RCVD_IN_DNSWL_HI, RP_MATCHES_RCVD, UNPARSEABLE_RELAY autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on mail.kernel.org X-Virus-Scanned: ClamAV using ClamSMTP From: Moni Shoua get_netdev: get the net_device on the physical port of the IB transport port. In port aggregation mode it is required to return the netdev of the active port. modify_gid: note for a change in the RoCE gid cache. Handle this by writing to the harsware GID table. It is possible that indexes in cahce and hardware tables won't match so a translation is required when modifying a QP or creating an address handle. Signed-off-by: Moni Shoua --- drivers/infiniband/core/cache.c | 3 +- drivers/infiniband/hw/mlx4/main.c | 236 ++++++++++++++++++++++++++++++++++- drivers/infiniband/hw/mlx4/mlx4_ib.h | 17 +++ include/linux/mlx4/device.h | 3 +- include/rdma/ib_verbs.h | 2 + 5 files changed, 257 insertions(+), 4 deletions(-) diff --git a/drivers/infiniband/core/cache.c b/drivers/infiniband/core/cache.c index 9b4f16b..a9d5c70 100644 --- a/drivers/infiniband/core/cache.c +++ b/drivers/infiniband/core/cache.c @@ -55,7 +55,8 @@ struct ib_update_work { u8 port_num; }; -static union ib_gid zgid; +union ib_gid zgid; +EXPORT_SYMBOL(zgid); static const struct ib_gid_attr zattr; diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 2f81723..61df5c9 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -45,6 +45,9 @@ #include #include #include +#include + +#include #include #include @@ -93,8 +96,6 @@ static void init_query_mad(struct ib_smp *mad) mad->method = IB_MGMT_METHOD_GET; } -static union ib_gid zgid; - static int check_flow_steering_support(struct mlx4_dev *dev) { int eth_num_ports = 0; @@ -131,6 +132,237 @@ static int num_ib_ports(struct mlx4_dev *dev) return ib_ports; } +static struct net_device *mlx4_ib_get_netdev(struct ib_device *device, u8 port_num) +{ + struct mlx4_ib_dev *ibdev = to_mdev(device); + struct net_device *dev; + + rcu_read_lock(); + dev = mlx4_get_protocol_dev(ibdev->dev, MLX4_PROT_ETH, port_num); + + if (dev) { + if (mlx4_is_bonded(ibdev->dev)) { + struct net_device *upper = NULL; + + upper = netdev_master_upper_dev_get_rcu(dev); + if (upper) { + struct net_device *active; + + active = bond_option_active_slave_get_rcu(netdev_priv(upper)); + if (active) + dev = active; + } + } + } + if (dev) + dev_hold(dev); + + rcu_read_unlock(); + return dev; +} + +static int mlx4_ib_update_gids(struct gid_entry *gids, + struct mlx4_ib_dev *ibdev, + u8 port_num) +{ + struct mlx4_cmd_mailbox *mailbox; + int err; + struct mlx4_dev *dev = ibdev->dev; + int i; + union ib_gid *gid_tbl; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) + return -ENOMEM; + + gid_tbl = mailbox->buf; + + for (i = 0; i < MLX4_MAX_PORT_GIDS; ++i) + memcpy(&gid_tbl[i], &gids[i].gid, sizeof(union ib_gid)); + + err = mlx4_cmd(dev, mailbox->dma, + MLX4_SET_PORT_GID_TABLE << 8 | port_num, + 1, MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_WRAPPED); + if (mlx4_is_bonded(dev)) + err += mlx4_cmd(dev, mailbox->dma, + MLX4_SET_PORT_GID_TABLE << 8 | 2, + 1, MLX4_CMD_SET_PORT, MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_WRAPPED); + + mlx4_free_cmd_mailbox(dev, mailbox); + return err; +} + +static int mlx4_ib_add_gid(struct ib_device *device, + u8 port_num, + unsigned int index, + const union ib_gid *gid, + const struct ib_gid_attr *attr, + void **context) +{ + struct mlx4_ib_dev *ibdev = to_mdev(device); + struct mlx4_ib_iboe *iboe = &ibdev->iboe; + struct mlx4_port_gid_table *port_gid_table; + int free = -1, found = -1; + int ret = 0; + int hw_update = 0; + int i; + struct gid_entry *gids = NULL; + + if (!rdma_cap_roce_gid_table(device, port_num)) + return -EINVAL; + + if (port_num > MLX4_MAX_PORTS) + return -EINVAL; + + if (!context) + return -EINVAL; + + port_gid_table = &iboe->gids[port_num - 1]; + spin_lock_bh(&iboe->lock); + for (i = 0; i < MLX4_MAX_PORT_GIDS; ++i) { + if (!memcmp(&port_gid_table->gids[i].gid, gid, sizeof(*gid))) { + found = i; + break; + } + if (free < 0 && !memcmp(&port_gid_table->gids[i].gid, &zgid, sizeof(*gid))) + free = i; /* HW has space */ + } + + if (found < 0) { + if (free < 0) { + ret = -ENOSPC; + } else { + port_gid_table->gids[free].ctx = kmalloc(sizeof(*port_gid_table->gids[free].ctx), GFP_ATOMIC); + if (!port_gid_table->gids[free].ctx) { + ret = -ENOMEM; + } else { + *context = port_gid_table->gids[free].ctx; + memcpy(&port_gid_table->gids[free].gid, gid, sizeof(*gid)); + port_gid_table->gids[free].ctx->real_index = free; + port_gid_table->gids[free].ctx->refcount = 1; + hw_update = 1; + } + } + } else { + struct gid_cache_context *ctx = port_gid_table->gids[found].ctx; + *context = ctx; + ctx->refcount++; + } + if (!ret && hw_update) { + gids = kmalloc(sizeof(*gids) * MLX4_MAX_PORT_GIDS, GFP_ATOMIC); + if (!gids) { + ret = -ENOMEM; + } else { + for (i = 0; i < MLX4_MAX_PORT_GIDS; i++) + memcpy(&gids[i].gid, &port_gid_table->gids[i].gid, sizeof(union ib_gid)); + } + } + spin_unlock_bh(&iboe->lock); + + if (!ret && hw_update) { + ret = mlx4_ib_update_gids(gids, ibdev, port_num); + kfree(gids); + } + + return ret; +} + +static int mlx4_ib_del_gid(struct ib_device *device, + u8 port_num, + unsigned int index, + void **context) +{ + struct gid_cache_context *ctx = *context; + struct mlx4_ib_dev *ibdev = to_mdev(device); + struct mlx4_ib_iboe *iboe = &ibdev->iboe; + struct mlx4_port_gid_table *port_gid_table; + int ret = 0; + int hw_update = 0; + struct gid_entry *gids = NULL; + + if (!rdma_cap_roce_gid_table(device, port_num)) + return -EINVAL; + + if (port_num > MLX4_MAX_PORTS) + return -EINVAL; + + port_gid_table = &iboe->gids[port_num - 1]; + spin_lock_bh(&iboe->lock); + if (ctx) { + ctx->refcount--; + if (!ctx->refcount) { + unsigned int real_index = ctx->real_index; + + memcpy(&port_gid_table->gids[real_index].gid, &zgid, sizeof(zgid)); + kfree(port_gid_table->gids[real_index].ctx); + port_gid_table->gids[real_index].ctx = NULL; + hw_update = 1; + } + } + if (!ret && hw_update) { + int i; + + gids = kmalloc(sizeof(*gids) * MLX4_MAX_PORT_GIDS, GFP_ATOMIC); + if (!gids) { + ret = -ENOMEM; + } else { + for (i = 0; i < MLX4_MAX_PORT_GIDS; i++) + memcpy(&gids[i].gid, &port_gid_table->gids[i].gid, sizeof(union ib_gid)); + } + } + spin_unlock_bh(&iboe->lock); + + if (!ret && hw_update) { + ret = mlx4_ib_update_gids(gids, ibdev, port_num); + kfree(gids); + } + return ret; +} + +int mlx4_ib_gid_index_to_real_index(struct mlx4_ib_dev *ibdev, + u8 port_num, int index) +{ + struct mlx4_ib_iboe *iboe = &ibdev->iboe; + struct gid_cache_context *ctx = NULL; + union ib_gid gid; + struct mlx4_port_gid_table *port_gid_table; + int real_index = -EINVAL; + int i; + int ret; + unsigned long flags; + + if (port_num > MLX4_MAX_PORTS) + return -EINVAL; + + if (mlx4_is_bonded(ibdev->dev)) + port_num = 1; + + if (!rdma_cap_roce_gid_table(&ibdev->ib_dev, port_num)) + return index; + + ret = ib_get_cached_gid(&ibdev->ib_dev, port_num, index, &gid); + if (ret) + return ret; + + if (!memcmp(&gid, &zgid, sizeof(gid))) + return -EINVAL; + + spin_lock_irqsave(&iboe->lock, flags); + port_gid_table = &iboe->gids[port_num - 1]; + + for (i = 0; i < MLX4_MAX_PORT_GIDS; ++i) + if (!memcmp(&port_gid_table->gids[i].gid, &gid, sizeof(gid))) { + ctx = port_gid_table->gids[i].ctx; + break; + } + if (ctx) + real_index = ctx->real_index; + spin_unlock_irqrestore(&iboe->lock, flags); + return real_index; +} + static int mlx4_ib_query_device(struct ib_device *ibdev, struct ib_device_attr *props, struct ib_udata *uhw) diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index 4eb7a85..4e91ca5 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -470,6 +470,20 @@ struct mlx4_ib_sriov { struct idr pv_id_table; }; +struct gid_cache_context { + int real_index; + int refcount; +}; + +struct gid_entry { + union ib_gid gid; + struct gid_cache_context *ctx; +}; + +struct mlx4_port_gid_table { + struct gid_entry gids[MLX4_MAX_PORT_GIDS]; +}; + struct mlx4_ib_iboe { spinlock_t lock; struct net_device *netdevs[MLX4_MAX_PORTS]; @@ -479,6 +493,7 @@ struct mlx4_ib_iboe { struct notifier_block nb_inet; struct notifier_block nb_inet6; union ib_gid gid_table[MLX4_MAX_PORTS][128]; + struct mlx4_port_gid_table gids[MLX4_MAX_PORTS]; }; struct pkey_mgt { @@ -851,5 +866,7 @@ int mlx4_ib_rereg_user_mr(struct ib_mr *mr, int flags, u64 start, u64 length, u64 virt_addr, int mr_access_flags, struct ib_pd *pd, struct ib_udata *udata); +int mlx4_ib_gid_index_to_real_index(struct mlx4_ib_dev *ibdev, + u8 port_num, int index); #endif /* MLX4_IB_H */ diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h index fd13c1c..5c7687a 100644 --- a/include/linux/mlx4/device.h +++ b/include/linux/mlx4/device.h @@ -79,7 +79,8 @@ enum { enum { MLX4_MAX_PORTS = 2, - MLX4_MAX_PORT_PKEYS = 128 + MLX4_MAX_PORT_PKEYS = 128, + MLX4_MAX_PORT_GIDS = 128 }; /* base qkey for use in sriov tunnel-qp/proxy-qp communication. diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index 74fc94c..c856276 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -64,6 +64,8 @@ union ib_gid { } global; }; +extern union ib_gid zgid; + struct ib_gid_attr { struct net_device *ndev; };