diff options
46 files changed, 1254 insertions, 354 deletions
diff --git a/Documentation/infiniband/user_mad.txt b/Documentation/infiniband/user_mad.txt index 8a366959f5cc..7aca13a54a3a 100644 --- a/Documentation/infiniband/user_mad.txt +++ b/Documentation/infiniband/user_mad.txt @@ -26,6 +26,11 @@ Creating MAD agents ioctl. Also, all agents registered through a file descriptor will be unregistered when the descriptor is closed. + 2014 -- a new registration ioctl is now provided which allows additional + fields to be provided during registration. + Users of this registration call are implicitly setting the use of + pkey_index (see below). + Receiving MADs MADs are received using read(). The receive side now supports @@ -104,10 +109,10 @@ P_Key Index Handling The old ib_umad interface did not allow setting the P_Key index for MADs that are sent and did not provide a way for obtaining the P_Key index of received MADs. A new layout for struct ib_user_mad_hdr - with a pkey_index member has been defined; however, to preserve - binary compatibility with older applications, this new layout will - not be used unless the IB_USER_MAD_ENABLE_PKEY ioctl is called - before a file descriptor is used for anything else. + with a pkey_index member has been defined; however, to preserve binary + compatibility with older applications, this new layout will not be used + unless one of IB_USER_MAD_ENABLE_PKEY or IB_USER_MAD_REGISTER_AGENT2 ioctl's + are called before a file descriptor is used for anything else. In September 2008, the IB_USER_MAD_ABI_VERSION will be incremented to 6, the new layout of struct ib_user_mad_hdr will be used by diff --git a/drivers/infiniband/core/agent.c b/drivers/infiniband/core/agent.c index 2bc7f5af64f4..f6d29614cb01 100644 --- a/drivers/infiniband/core/agent.c +++ b/drivers/infiniband/core/agent.c @@ -94,14 +94,14 @@ void agent_send_response(struct ib_mad *mad, struct ib_grh *grh, port_priv = ib_get_agent_port(device, port_num); if (!port_priv) { - printk(KERN_ERR SPFX "Unable to find port agent\n"); + dev_err(&device->dev, "Unable to find port agent\n"); return; } agent = port_priv->agent[qpn]; ah = ib_create_ah_from_wc(agent->qp->pd, wc, grh, port_num); if (IS_ERR(ah)) { - printk(KERN_ERR SPFX "ib_create_ah_from_wc error %ld\n", + dev_err(&device->dev, "ib_create_ah_from_wc error %ld\n", PTR_ERR(ah)); return; } @@ -110,7 +110,7 @@ void agent_send_response(struct ib_mad *mad, struct ib_grh *grh, IB_MGMT_MAD_HDR, IB_MGMT_MAD_DATA, GFP_KERNEL); if (IS_ERR(send_buf)) { - printk(KERN_ERR SPFX "ib_create_send_mad error\n"); + dev_err(&device->dev, "ib_create_send_mad error\n"); goto err1; } @@ -125,7 +125,7 @@ void agent_send_response(struct ib_mad *mad, struct ib_grh *grh, } if (ib_post_send_mad(send_buf, NULL)) { - printk(KERN_ERR SPFX "ib_post_send_mad error\n"); + dev_err(&device->dev, "ib_post_send_mad error\n"); goto err2; } return; @@ -151,7 +151,7 @@ int ib_agent_port_open(struct ib_device *device, int port_num) /* Create new device info */ port_priv = kzalloc(sizeof *port_priv, GFP_KERNEL); if (!port_priv) { - printk(KERN_ERR SPFX "No memory for ib_agent_port_private\n"); + dev_err(&device->dev, "No memory for ib_agent_port_private\n"); ret = -ENOMEM; goto error1; } @@ -161,7 +161,7 @@ int ib_agent_port_open(struct ib_device *device, int port_num) port_priv->agent[0] = ib_register_mad_agent(device, port_num, IB_QPT_SMI, NULL, 0, &agent_send_handler, - NULL, NULL); + NULL, NULL, 0); if (IS_ERR(port_priv->agent[0])) { ret = PTR_ERR(port_priv->agent[0]); goto error2; @@ -172,7 +172,7 @@ int ib_agent_port_open(struct ib_device *device, int port_num) port_priv->agent[1] = ib_register_mad_agent(device, port_num, IB_QPT_GSI, NULL, 0, &agent_send_handler, - NULL, NULL); + NULL, NULL, 0); if (IS_ERR(port_priv->agent[1])) { ret = PTR_ERR(port_priv->agent[1]); goto error3; @@ -202,7 +202,7 @@ int ib_agent_port_close(struct ib_device *device, int port_num) port_priv = __ib_get_agent_port(device, port_num); if (port_priv == NULL) { spin_unlock_irqrestore(&ib_agent_port_list_lock, flags); - printk(KERN_ERR SPFX "Port %d not found\n", port_num); + dev_err(&device->dev, "Port %d not found\n", port_num); return -ENODEV; } list_del(&port_priv->port_list); diff --git a/drivers/infiniband/core/cm.c b/drivers/infiniband/core/cm.c index c3239170d8b7..e28a494e2a3a 100644 --- a/drivers/infiniband/core/cm.c +++ b/drivers/infiniband/core/cm.c @@ -3753,7 +3753,7 @@ static void cm_add_one(struct ib_device *ib_device) struct cm_port *port; struct ib_mad_reg_req reg_req = { .mgmt_class = IB_MGMT_CLASS_CM, - .mgmt_class_version = IB_CM_CLASS_VERSION + .mgmt_class_version = IB_CM_CLASS_VERSION, }; struct ib_port_modify port_modify = { .set_port_cap_mask = IB_PORT_CM_SUP @@ -3801,7 +3801,8 @@ static void cm_add_one(struct ib_device *ib_device) 0, cm_send_handler, cm_recv_handler, - port); + port, + 0); if (IS_ERR(port->mad_agent)) goto error2; diff --git a/drivers/infiniband/core/iwcm.c b/drivers/infiniband/core/iwcm.c index 3d2e489ab732..ff9163dc1596 100644 --- a/drivers/infiniband/core/iwcm.c +++ b/drivers/infiniband/core/iwcm.c @@ -46,6 +46,7 @@ #include <linux/completion.h> #include <linux/slab.h> #include <linux/module.h> +#include <linux/sysctl.h> #include <rdma/iw_cm.h> #include <rdma/ib_addr.h> @@ -65,6 +66,20 @@ struct iwcm_work { struct list_head free_list; }; +static unsigned int default_backlog = 256; + +static struct ctl_table_header *iwcm_ctl_table_hdr; +static struct ctl_table iwcm_ctl_table[] = { + { + .procname = "default_backlog", + .data = &default_backlog, + .maxlen = sizeof(default_backlog), + .mode = 0644, + .proc_handler = proc_dointvec, + }, + { } +}; + /* * The following services provide a mechanism for pre-allocating iwcm_work * elements. The design pre-allocates them based on the cm_id type: @@ -425,6 +440,9 @@ int iw_cm_listen(struct iw_cm_id *cm_id, int backlog) cm_id_priv = container_of(cm_id, struct iwcm_id_private, id); + if (!backlog) + backlog = default_backlog; + ret = alloc_work_entries(cm_id_priv, backlog); if (ret) return ret; @@ -1030,11 +1048,20 @@ static int __init iw_cm_init(void) if (!iwcm_wq) return -ENOMEM; + iwcm_ctl_table_hdr = register_net_sysctl(&init_net, "net/iw_cm", + iwcm_ctl_table); + if (!iwcm_ctl_table_hdr) { + pr_err("iw_cm: couldn't register sysctl paths\n"); + destroy_workqueue(iwcm_wq); + return -ENOMEM; + } + return 0; } static void __exit iw_cm_cleanup(void) { + unregister_net_sysctl_table(iwcm_ctl_table_hdr); destroy_workqueue(iwcm_wq); } diff --git a/drivers/infiniband/core/mad.c b/drivers/infiniband/core/mad.c index ab31f136d04b..74c30f4c557e 100644 --- a/drivers/infiniband/core/mad.c +++ b/drivers/infiniband/core/mad.c @@ -33,6 +33,9 @@ * SOFTWARE. * */ + +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + #include <linux/dma-mapping.h> #include <linux/slab.h> #include <linux/module.h> @@ -195,7 +198,8 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, u8 rmpp_version, ib_mad_send_handler send_handler, ib_mad_recv_handler recv_handler, - void *context) + void *context, + u32 registration_flags) { struct ib_mad_port_private *port_priv; struct ib_mad_agent *ret = ERR_PTR(-EINVAL); @@ -211,68 +215,109 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, /* Validate parameters */ qpn = get_spl_qp_index(qp_type); - if (qpn == -1) + if (qpn == -1) { + dev_notice(&device->dev, + "ib_register_mad_agent: invalid QP Type %d\n", + qp_type); goto error1; + } - if (rmpp_version && rmpp_version != IB_MGMT_RMPP_VERSION) + if (rmpp_version && rmpp_version != IB_MGMT_RMPP_VERSION) { + dev_notice(&device->dev, + "ib_register_mad_agent: invalid RMPP Version %u\n", + rmpp_version); goto error1; + } /* Validate MAD registration request if supplied */ if (mad_reg_req) { - if (mad_reg_req->mgmt_class_version >= MAX_MGMT_VERSION) + if (mad_reg_req->mgmt_class_version >= MAX_MGMT_VERSION) { + dev_notice(&device->dev, + "ib_register_mad_agent: invalid Class Version %u\n", + mad_reg_req->mgmt_class_version); goto error1; - if (!recv_handler) + } + if (!recv_handler) { + dev_notice(&device->dev, + "ib_register_mad_agent: no recv_handler\n"); goto error1; + } if (mad_reg_req->mgmt_class >= MAX_MGMT_CLASS) { /* * IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE is the only * one in this range currently allowed */ if (mad_reg_req->mgmt_class != - IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) + IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) { + dev_notice(&device->dev, + "ib_register_mad_agent: Invalid Mgmt Class 0x%x\n", + mad_reg_req->mgmt_class); goto error1; + } } else if (mad_reg_req->mgmt_class == 0) { /* * Class 0 is reserved in IBA and is used for * aliasing of IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE */ + dev_notice(&device->dev, + "ib_register_mad_agent: Invalid Mgmt Class 0\n"); goto error1; } else if (is_vendor_class(mad_reg_req->mgmt_class)) { /* * If class is in "new" vendor range, * ensure supplied OUI is not zero */ - if (!is_vendor_oui(mad_reg_req->oui)) + if (!is_vendor_oui(mad_reg_req->oui)) { + dev_notice(&device->dev, + "ib_register_mad_agent: No OUI specified for class 0x%x\n", + mad_reg_req->mgmt_class); goto error1; + } } /* Make sure class supplied is consistent with RMPP */ if (!ib_is_mad_class_rmpp(mad_reg_req->mgmt_class)) { - if (rmpp_version) + if (rmpp_version) { + dev_notice(&device->dev, + "ib_register_mad_agent: RMPP version for non-RMPP class 0x%x\n", + mad_reg_req->mgmt_class); goto error1; + } } + /* Make sure class supplied is consistent with QP type */ if (qp_type == IB_QPT_SMI) { if ((mad_reg_req->mgmt_class != IB_MGMT_CLASS_SUBN_LID_ROUTED) && (mad_reg_req->mgmt_class != - IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE)) + IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE)) { + dev_notice(&device->dev, + "ib_register_mad_agent: Invalid SM QP type: class 0x%x\n", + mad_reg_req->mgmt_class); goto error1; + } } else { if ((mad_reg_req->mgmt_class == IB_MGMT_CLASS_SUBN_LID_ROUTED) || (mad_reg_req->mgmt_class == - IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE)) + IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE)) { + dev_notice(&device->dev, + "ib_register_mad_agent: Invalid GS QP type: class 0x%x\n", + mad_reg_req->mgmt_class); goto error1; + } } } else { /* No registration request supplied */ if (!send_handler) goto error1; + if (registration_flags & IB_MAD_USER_RMPP) + goto error1; } /* Validate device and port */ port_priv = ib_get_mad_port(device, port_num); if (!port_priv) { + dev_notice(&device->dev, "ib_register_mad_agent: Invalid port\n"); ret = ERR_PTR(-ENODEV); goto error1; } @@ -280,6 +325,8 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, /* Verify the QP requested is supported. For example, Ethernet devices * will not have QP0 */ if (!port_priv->qp_info[qpn].qp) { + dev_notice(&device->dev, + "ib_register_mad_agent: QP %d not supported\n", qpn); ret = ERR_PTR(-EPROTONOSUPPORT); goto error1; } @@ -316,6 +363,7 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, mad_agent_priv->agent.context = context; mad_agent_priv->agent.qp = port_priv->qp_info[qpn].qp; mad_agent_priv->agent.port_num = port_num; + mad_agent_priv->agent.flags = registration_flags; spin_lock_init(&mad_agent_priv->lock); INIT_LIST_HEAD(&mad_agent_priv->send_list); INIT_LIST_HEAD(&mad_agent_priv->wait_list); @@ -706,7 +754,7 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, smi_handle_dr_smp_send(smp, device->node_type, port_num) == IB_SMI_DISCARD) { ret = -EINVAL; - printk(KERN_ERR PFX "Invalid directed route\n"); + dev_err(&device->dev, "Invalid directed route\n"); goto out; } @@ -718,7 +766,7 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, local = kmalloc(sizeof *local, GFP_ATOMIC); if (!local) { ret = -ENOMEM; - printk(KERN_ERR PFX "No memory for ib_mad_local_private\n"); + dev_err(&device->dev, "No memory for ib_mad_local_private\n"); goto out; } local->mad_priv = NULL; @@ -726,7 +774,7 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, mad_priv = kmem_cache_alloc(ib_mad_cache, GFP_ATOMIC); if (!mad_priv) { ret = -ENOMEM; - printk(KERN_ERR PFX "No memory for local response MAD\n"); + dev_err(&device->dev, "No memory for local response MAD\n"); kfree(local); goto out; } @@ -837,9 +885,9 @@ static int alloc_send_rmpp_list(struct ib_mad_send_wr_private *send_wr, for (left = send_buf->data_len + pad; left > 0; left -= seg_size) { seg = kmalloc(sizeof (*seg) + seg_size, gfp_mask); if (!seg) { - printk(KERN_ERR "alloc_send_rmpp_segs: RMPP mem " - "alloc failed for len %zd, gfp %#x\n", - sizeof (*seg) + seg_size, gfp_mask); + dev_err(&send_buf->mad_agent->device->dev, + "alloc_send_rmpp_segs: RMPP mem alloc failed for len %zd, gfp %#x\n", + sizeof (*seg) + seg_size, gfp_mask); free_send_rmpp_list(send_wr); return -ENOMEM; } @@ -862,6 +910,12 @@ static int alloc_send_rmpp_list(struct ib_mad_send_wr_private *send_wr, return 0; } +int ib_mad_kernel_rmpp_agent(struct ib_mad_agent *agent) +{ + return agent->rmpp_version && !(agent->flags & IB_MAD_USER_RMPP); +} +EXPORT_SYMBOL(ib_mad_kernel_rmpp_agent); + struct ib_mad_send_buf * ib_create_send_mad(struct ib_mad_agent *mad_agent, u32 remote_qpn, u16 pkey_index, int rmpp_active, @@ -878,10 +932,12 @@ struct ib_mad_send_buf * ib_create_send_mad(struct ib_mad_agent *mad_agent, pad = get_pad_size(hdr_len, data_len); message_size = hdr_len + data_len + pad; - if ((!mad_agent->rmpp_version && - (rmpp_active || message_size > sizeof(struct ib_mad))) || - (!rmpp_active && message_size > sizeof(struct ib_mad))) - return ERR_PTR(-EINVAL); + if (ib_mad_kernel_rmpp_agent(mad_agent)) { + if (!rmpp_active && message_size > sizeof(struct ib_mad)) + return ERR_PTR(-EINVAL); + } else + if (rmpp_active || message_size > sizeof(struct ib_mad)) + return ERR_PTR(-EINVAL); size = rmpp_active ? hdr_len : sizeof(struct ib_mad); buf = kzalloc(sizeof *mad_send_wr + size, gfp_mask); @@ -1135,7 +1191,7 @@ int ib_post_send_mad(struct ib_mad_send_buf *send_buf, &mad_agent_priv->send_list); spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - if (mad_agent_priv->agent.rmpp_version) { + if (ib_mad_kernel_rmpp_agent(&mad_agent_priv->agent)) { ret = ib_send_rmpp_mad(mad_send_wr); if (ret >= 0 && ret != IB_RMPP_RESULT_CONSUMED) ret = ib_send_mad(mad_send_wr); @@ -1199,7 +1255,8 @@ EXPORT_SYMBOL(ib_redirect_mad_qp); int ib_process_mad_wc(struct ib_mad_agent *mad_agent, struct ib_wc *wc) { - printk(KERN_ERR PFX "ib_process_mad_wc() not implemented yet\n"); + dev_err(&mad_agent->device->dev, + "ib_process_mad_wc() not implemented yet\n"); return 0; } EXPORT_SYMBOL(ib_process_mad_wc); @@ -1211,7 +1268,7 @@ static int method_in_use(struct ib_mad_mgmt_method_table **method, for_each_set_bit(i, mad_reg_req->method_mask, IB_MGMT_MAX_METHODS) { if ((*method)->agent[i]) { - printk(KERN_ERR PFX "Method %d already in use\n", i); + pr_err("Method %d already in use\n", i); return -EINVAL; } } @@ -1223,8 +1280,7 @@ static int allocate_method_table(struct ib_mad_mgmt_method_table **method) /* Allocate management method table */ *method = kzalloc(sizeof **method, GFP_ATOMIC); if (!*method) { - printk(KERN_ERR PFX "No memory for " - "ib_mad_mgmt_method_table\n"); + pr_err("No memory for ib_mad_mgmt_method_table\n"); return -ENOMEM; } @@ -1319,8 +1375,8 @@ static int add_nonoui_reg_req(struct ib_mad_reg_req *mad_reg_req, /* Allocate management class table for "new" class version */ *class = kzalloc(sizeof **class, GFP_ATOMIC); if (!*class) { - printk(KERN_ERR PFX "No memory for " - "ib_mad_mgmt_class_table\n"); + dev_err(&agent_priv->agent.device->dev, + "No memory for ib_mad_mgmt_class_table\n"); ret = -ENOMEM; goto error1; } @@ -1386,8 +1442,8 @@ static int add_oui_reg_req(struct ib_mad_reg_req *mad_reg_req, /* Allocate mgmt vendor class table for "new" class version */ vendor = kzalloc(sizeof *vendor, GFP_ATOMIC); if (!vendor) { - printk(KERN_ERR PFX "No memory for " - "ib_mad_mgmt_vendor_class_table\n"); + dev_err(&agent_priv->agent.device->dev, + "No memory for ib_mad_mgmt_vendor_class_table\n"); goto error1; } @@ -1397,8 +1453,8 @@ static int add_oui_reg_req(struct ib_mad_reg_req *mad_reg_req, /* Allocate table for this management vendor class */ vendor_class = kzalloc(sizeof *vendor_class, GFP_ATOMIC); if (!vendor_class) { - printk(KERN_ERR PFX "No memory for " - "ib_mad_mgmt_vendor_class\n"); + dev_err(&agent_priv->agent.device->dev, + "No memory for ib_mad_mgmt_vendor_class\n"); goto error2; } @@ -1429,7 +1485,7 @@ static int add_oui_reg_req(struct ib_mad_reg_req *mad_reg_req, goto check_in_use; } } - printk(KERN_ERR PFX "All OUI slots in use\n"); + dev_err(&agent_priv->agent.device->dev, "All OUI slots in use\n"); goto error3; check_in_use: @@ -1640,9 +1696,9 @@ find_mad_agent(struct ib_mad_port_private *port_priv, if (mad_agent->agent.recv_handler) atomic_inc(&mad_agent->refcount); else { - printk(KERN_NOTICE PFX "No receive handler for client " - "%p on port %d\n", - &mad_agent->agent, port_priv->port_num); + dev_notice(&port_priv->device->dev, + "No receive handler for client %p on port %d\n", + &mad_agent->agent, port_priv->port_num); mad_agent = NULL; } } @@ -1658,8 +1714,8 @@ static int validate_mad(struct ib_mad *mad, u32 qp_num) /* Make sure MAD base version is understood */ if (mad->mad_hdr.base_version != IB_MGMT_BASE_VERSION) { - printk(KERN_ERR PFX "MAD received with unsupported base " - "version %d\n", mad->mad_hdr.base_version); + pr_err("MAD received with unsupported base version %d\n", + mad->mad_hdr.base_version); goto out; } @@ -1685,6 +1741,7 @@ static int is_data_mad(struct ib_mad_agent_private *mad_agent_priv, rmpp_mad = (struct ib_rmpp_mad *)mad_hdr; return !mad_agent_priv->agent.rmpp_version || + !ib_mad_kernel_rmpp_agent(&mad_agent_priv->agent) || !(ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) & IB_MGMT_RMPP_FLAG_ACTIVE) || (rmpp_mad->rmpp_hdr.rmpp_type == IB_MGMT_RMPP_TYPE_DATA); @@ -1812,7 +1869,7 @@ static void ib_mad_complete_recv(struct ib_mad_agent_private *mad_agent_priv, INIT_LIST_HEAD(&mad_recv_wc->rmpp_list); list_add(&mad_recv_wc->recv_buf.list, &mad_recv_wc->rmpp_list); - if (mad_agent_priv->agent.rmpp_version) { + if (ib_mad_kernel_rmpp_agent(&mad_agent_priv->agent)) { mad_recv_wc = ib_process_rmpp_recv_wc(mad_agent_priv, mad_recv_wc); if (!mad_recv_wc) { @@ -1827,23 +1884,39 @@ static void ib_mad_complete_recv(struct ib_mad_agent_private *mad_agent_priv, mad_send_wr = ib_find_send_mad(mad_agent_priv, mad_recv_wc); if (!mad_send_wr) { spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - ib_free_recv_mad(mad_recv_wc); - deref_mad_agent(mad_agent_priv); - return; - } - ib_mark_mad_done(mad_send_wr); - spin_unlock_irqrestore(&mad_agent_priv->lock, flags); + if (!ib_mad_kernel_rmpp_agent(&mad_agent_priv->agent) + && ib_is_mad_class_rmpp(mad_recv_wc->recv_buf.mad->mad_hdr.mgmt_class) + && (ib_get_rmpp_flags(&((struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad)->rmpp_hdr) + & IB_MGMT_RMPP_FLAG_ACTIVE)) { + /* user rmpp is in effect + * and this is an active RMPP MAD + */ + mad_recv_wc->wc->wr_id = 0; + mad_agent_priv->agent.recv_handler(&mad_agent_priv->agent, + mad_recv_wc); + atomic_dec(&mad_agent_priv->refcount); + } else { + /* not user rmpp, revert to normal behavior and + * drop the mad */ + ib_free_recv_mad(mad_recv_wc); + deref_mad_agent(mad_agent_priv); + return; + } + } else { + ib_mark_mad_done(mad_send_wr); + spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - /* Defined behavior is to complete response before request */ - mad_recv_wc->wc->wr_id = (unsigned long) &mad_send_wr->send_buf; - mad_agent_priv->agent.recv_handler(&mad_agent_priv->agent, - mad_recv_wc); - atomic_dec(&mad_agent_priv->refcount); + /* Defined behavior is to complete response before request */ + mad_recv_wc->wc->wr_id = (unsigned long) &mad_send_wr->send_buf; + mad_agent_priv->agent.recv_handler(&mad_agent_priv->agent, + mad_recv_wc); + atomic_dec(&mad_agent_priv->refcount); - mad_send_wc.status = IB_WC_SUCCESS; - mad_send_wc.vendor_err = 0; - mad_send_wc.send_buf = &mad_send_wr->send_buf; - ib_mad_complete_send_wr(mad_send_wr, &mad_send_wc); + mad_send_wc.status = IB_WC_SUCCESS; + mad_send_wc.vendor_err = 0; + mad_send_wc.send_buf = &mad_send_wr->send_buf; + ib_mad_complete_send_wr(mad_send_wr, &mad_send_wc); + } } else { mad_agent_priv->agent.recv_handler(&mad_agent_priv->agent, mad_recv_wc); @@ -1911,8 +1984,8 @@ static void ib_mad_recv_done_handler(struct ib_mad_port_private *port_priv, response = kmem_cache_alloc(ib_mad_cache, GFP_KERNEL); if (!response) { - printk(KERN_ERR PFX "ib_mad_recv_done_handler no memory " - "for response buffer\n"); + dev_err(&port_priv->device->dev, + "ib_mad_recv_done_handler no memory for response buffer\n"); goto out; } @@ -2083,7 +2156,7 @@ void ib_mad_complete_send_wr(struct ib_mad_send_wr_private *mad_send_wr, mad_agent_priv = mad_send_wr->mad_agent_priv; spin_lock_irqsave(&mad_agent_priv->lock, flags); - if (mad_agent_priv->agent.rmpp_version) { + if (ib_mad_kernel_rmpp_agent(&mad_agent_priv->agent)) { ret = ib_process_rmpp_send_wc(mad_send_wr, mad_send_wc); if (ret == IB_RMPP_RESULT_CONSUMED) goto done; @@ -2176,7 +2249,8 @@ retry: ret = ib_post_send(qp_info->qp, &queued_send_wr->send_wr, &bad_send_wr); if (ret) { - printk(KERN_ERR PFX "ib_post_send failed: %d\n", ret); + dev_err(&port_priv->device->dev, + "ib_post_send failed: %d\n", ret); mad_send_wr = queued_send_wr; wc->status = IB_WC_LOC_QP_OP_ERR; goto retry; @@ -2248,8 +2322,9 @@ static void mad_error_handler(struct ib_mad_port_private *port_priv, IB_QP_STATE | IB_QP_CUR_STATE); kfree(attr); if (ret) - printk(KERN_ERR PFX "mad_error_handler - " - "ib_modify_qp to RTS : %d\n", ret); + dev_err(&port_priv->device->dev, + "mad_error_handler - ib_modify_qp to RTS : %d\n", + ret); else mark_sends_for_retry(qp_info); } @@ -2408,7 +2483,8 @@ static void local_completions(struct work_struct *work) if (local->mad_priv) { recv_mad_agent = local->recv_mad_agent; if (!recv_mad_agent) { - printk(KERN_ERR PFX "No receive MAD agent for local completion\n"); + dev_err(&mad_agent_priv->agent.device->dev, + "No receive MAD agent for local completion\n"); free_mad = 1; goto local_send_completion; } @@ -2476,7 +2552,7 @@ static int retry_send(struct ib_mad_send_wr_private *mad_send_wr) mad_send_wr->timeout = msecs_to_jiffies(mad_send_wr->send_buf.timeout_ms); - if (mad_send_wr->mad_agent_priv->agent.rmpp_version) { + if (ib_mad_kernel_rmpp_agent(&mad_send_wr->mad_agent_priv->agent)) { ret = ib_retry_rmpp(mad_send_wr); switch (ret) { case IB_RMPP_RESULT_UNHANDLED: @@ -2589,7 +2665,8 @@ static int ib_mad_post_receive_mads(struct ib_mad_qp_info *qp_info, } else { mad_priv = kmem_cache_alloc(ib_mad_cache, GFP_KERNEL); if (!mad_priv) { - printk(KERN_ERR PFX "No memory for receive buffer\n"); + dev_err(&qp_info->port_priv->device->dev, + "No memory for receive buffer\n"); ret = -ENOMEM; break; } @@ -2625,7 +2702,8 @@ static int ib_mad_post_receive_mads(struct ib_mad_qp_info *qp_info, sizeof mad_priv->header, DMA_FROM_DEVICE); kmem_cache_free(ib_mad_cache, mad_priv); - printk(KERN_ERR PFX "ib_post_recv failed: %d\n", ret); + dev_err(&qp_info->port_priv->device->dev, + "ib_post_recv failed: %d\n", ret); break; } } while (post); @@ -2681,7 +2759,8 @@ static int ib_mad_port_start(struct ib_mad_port_private *port_priv) attr = kmalloc(sizeof *attr, GFP_KERNEL); if (!attr) { - printk(KERN_ERR PFX "Couldn't kmalloc ib_qp_attr\n"); + dev_err(&port_priv->device->dev, + "Couldn't kmalloc ib_qp_attr\n"); return -ENOMEM; } @@ -2705,16 +2784,18 @@ static int ib_mad_port_start(struct ib_mad_port_private *port_priv) ret = ib_modify_qp(qp, attr, IB_QP_STATE | IB_QP_PKEY_INDEX | IB_QP_QKEY); if (ret) { - printk(KERN_ERR PFX "Couldn't change QP%d state to " - "INIT: %d\n", i, ret); + dev_err(&port_priv->device->dev, + "Couldn't change QP%d state to INIT: %d\n", + i, ret); goto out; } attr->qp_state = IB_QPS_RTR; ret = ib_modify_qp(qp, attr, IB_QP_STATE); if (ret) { - printk(KERN_ERR PFX "Couldn't change QP%d state to " - "RTR: %d\n", i, ret); + dev_err(&port_priv->device->dev, + "Couldn't change QP%d state to RTR: %d\n", + i, ret); goto out; } @@ -2722,16 +2803,18 @@ static int ib_mad_port_start(struct ib_mad_port_private *port_priv) attr->sq_psn = IB_MAD_SEND_Q_PSN; ret = ib_modify_qp(qp, attr, IB_QP_STATE | IB_QP_SQ_PSN); if (ret) { - printk(KERN_ERR PFX "Couldn't change QP%d state to " - "RTS: %d\n", i, ret); + dev_err(&port_priv->device->dev, + "Couldn't change QP%d state to RTS: %d\n", + i, ret); goto out; } } ret = ib_req_notify_cq(port_priv->cq, IB_CQ_NEXT_COMP); if (ret) { - printk(KERN_ERR PFX "Failed to request completion " - "notification: %d\n", ret); + dev_err(&port_priv->device->dev, + "Failed to request completion notification: %d\n", + ret); goto out; } @@ -2741,7 +2824,8 @@ static int ib_mad_port_start(struct ib_mad_port_private *port_priv) ret = ib_mad_post_receive_mads(&port_priv->qp_info[i], NULL); if (ret) { - printk(KERN_ERR PFX "Couldn't post receive WRs\n"); + dev_err(&port_priv->device->dev, + "Couldn't post receive WRs\n"); goto out; } } @@ -2755,7 +2839,8 @@ static void qp_event_handler(struct ib_event *event, void *qp_context) struct ib_mad_qp_info *qp_info = qp_context; /* It's worse than that! He's dead, Jim! */ - printk(KERN_ERR PFX "Fatal error (%d) on MAD QP (%d)\n", + dev_err(&qp_info->port_priv->device->dev, + "Fatal error (%d) on MAD QP (%d)\n", event->event, qp_info->qp->qp_num); } @@ -2801,8 +2886,9 @@ static int create_mad_qp(struct ib_mad_qp_info *qp_info, qp_init_attr.event_handler = qp_event_handler; qp_info->qp = ib_create_qp(qp_info->port_priv->pd, &qp_init_attr); if (IS_ERR(qp_info->qp)) { - printk(KERN_ERR PFX "Couldn't create ib_mad QP%d\n", - get_spl_qp_index(qp_type)); + dev_err(&qp_info->port_priv->device->dev, + "Couldn't create ib_mad QP%d\n", + get_spl_qp_index(qp_type)); ret = PTR_ERR(qp_info->qp); goto error; } @@ -2840,7 +2926,7 @@ static int ib_mad_port_open(struct ib_device *device, /* Create new device info */ port_priv = kzalloc(sizeof *port_priv, GFP_KERNEL); if (!port_priv) { - printk(KERN_ERR PFX "No memory for ib_mad_port_private\n"); + dev_err(&device->dev, "No memory for ib_mad_port_private\n"); return -ENOMEM; } @@ -2860,21 +2946,21 @@ static int ib_mad_port_open(struct ib_device *device, ib_mad_thread_completion_handler, NULL, port_priv, cq_size, 0); if (IS_ERR(port_priv->cq)) { - printk(KERN_ERR PFX "Couldn't create ib_mad CQ\n"); + dev_err(&device->dev, "Couldn't create ib_mad CQ\n"); ret = PTR_ERR(port_priv->cq); goto error3; } port_priv->pd = ib_alloc_pd(device); if (IS_ERR(port_priv->pd)) { - printk(KERN_ERR PFX "Couldn't create ib_mad PD\n"); + dev_err(&device->dev, "Couldn't create ib_mad PD\n"); ret = PTR_ERR(port_priv->pd); goto error4; } port_priv->mr = ib_get_dma_mr(port_priv->pd, IB_ACCESS_LOCAL_WRITE); if (IS_ERR(port_priv->mr)) { - printk(KERN_ERR PFX "Couldn't get ib_mad DMA MR\n"); + dev_err(&device->dev, "Couldn't get ib_mad DMA MR\n"); ret = PTR_ERR(port_priv->mr); goto error5; } @@ -2902,7 +2988,7 @@ static int ib_mad_port_open(struct ib_device *device, ret = ib_mad_port_start(port_priv); if (ret) { - printk(KERN_ERR PFX "Couldn't start port\n"); + dev_err(&device->dev, "Couldn't start port\n"); goto error9; } @@ -2946,7 +3032,7 @@ static int ib_mad_port_close(struct ib_device *device, int port_num) port_priv = __ib_get_mad_port(device, port_num); if (port_priv == NULL) { spin_unlock_irqrestore(&ib_mad_port_list_lock, flags); - printk(KERN_ERR PFX "Port %d not found\n", port_num); + dev_err(&device->dev, "Port %d not found\n", port_num); return -ENODEV; } list_del_init(&port_priv->port_list); @@ -2984,14 +3070,12 @@ static void ib_mad_init_device(struct ib_device *device) for (i = start; i <= end; i++) { if (ib_mad_port_open(device, i)) { - printk(KERN_ERR PFX "Couldn't open %s port %d\n", - device->name, i); + dev_err(&device->dev, "Couldn't open port %d\n", i); goto error; } if (ib_agent_port_open(device, i)) { - printk(KERN_ERR PFX "Couldn't open %s port %d " - "for agents\n", - device->name, i); + dev_err(&device->dev, + "Couldn't open port %d for agents\n", i); goto error_agent; } } @@ -2999,20 +3083,17 @@ static void ib_mad_init_device(struct ib_device *device) error_agent: if (ib_mad_port_close(device, i)) - printk(KERN_ERR PFX "Couldn't close %s port %d\n", - device->name, i); + dev_err(&device->dev, "Couldn't close port %d\n", i); error: i--; while (i >= start) { if (ib_agent_port_close(device, i)) - printk(KERN_ERR PFX "Couldn't close %s port %d " - "for agents\n", - device->name, i); + dev_err(&device->dev, + "Couldn't close port %d for agents\n", i); if (ib_mad_port_close(device, i)) - printk(KERN_ERR PFX "Couldn't close %s port %d\n", - device->name, i); + dev_err(&device->dev, "Couldn't close port %d\n", i); i--; } } @@ -3033,12 +3114,12 @@ static void ib_mad_remove_device(struct ib_device *device) } for (i = 0; i < num_ports; i++, cur_port++) { if (ib_agent_port_close(device, cur_port)) - printk(KERN_ERR PFX "Couldn't close %s port %d " - "for agents\n", - device->name, cur_port); + dev_err(&device->dev, + "Couldn't close port %d for agents\n", + cur_port); if (ib_mad_port_close(device, cur_port)) - printk(KERN_ERR PFX "Couldn't close %s port %d\n", - device->name, cur_port); + dev_err(&device->dev, "Couldn't close port %d\n", + cur_port); } } @@ -3064,7 +3145,7 @@ static int __init ib_mad_init_module(void) SLAB_HWCACHE_ALIGN, NULL); if (!ib_mad_cache) { - printk(KERN_ERR PFX "Couldn't create ib_mad cache\n"); + pr_err("Couldn't create ib_mad cache\n"); ret = -ENOMEM; goto error1; } @@ -3072,7 +3153,7 @@ static int __init ib_mad_init_module(void) INIT_LIST_HEAD(&ib_mad_port_list); if (ib_register_client(&mad_client)) { - printk(KERN_ERR PFX "Couldn't register ib_mad client\n"); + pr_err("Couldn't register ib_mad client\n"); ret = -EINVAL; goto error2; } diff --git a/drivers/infiniband/core/mad_priv.h b/drivers/infiniband/core/mad_priv.h index 9430ab4969c5..d1a0b0ee9444 100644 --- a/drivers/infiniband/core/mad_priv.h +++ b/drivers/infiniband/core/mad_priv.h @@ -42,9 +42,6 @@ #include <rdma/ib_mad.h> #include <rdma/ib_smi.h> - -#define PFX "ib_mad: " - #define IB_MAD_QPS_CORE 2 /* Always QP0 and QP1 as a minimum */ /* QP and CQ parameters */ diff --git a/drivers/infiniband/core/sa_query.c b/drivers/infiniband/core/sa_query.c index 233eaf541f55..c38f030f0dc9 100644 --- a/drivers/infiniband/core/sa_query.c +++ b/drivers/infiniband/core/sa_query.c @@ -1184,7 +1184,7 @@ static void ib_sa_add_one(struct ib_device *device) sa_dev->port[i].agent = ib_register_mad_agent(device, i + s, IB_QPT_GSI, NULL, 0, send_handler, - recv_handler, sa_dev); + recv_handler, sa_dev, 0); if (IS_ERR(sa_dev->port[i].agent)) goto err; diff --git a/drivers/infiniband/core/user_mad.c b/drivers/infiniband/core/user_mad.c index 1acb99100556..928cdd20e2d1 100644 --- a/drivers/infiniband/core/user_mad.c +++ b/drivers/infiniband/core/user_mad.c @@ -33,6 +33,8 @@ * SOFTWARE. */ +#define pr_fmt(fmt) "user_mad: " fmt + #include <linux/module.h> #include <linux/init.h> #include <linux/device.h> @@ -504,13 +506,15 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf, rmpp_mad = (struct ib_rmpp_mad *) packet->mad.data; hdr_len = ib_get_mad_data_offset(rmpp_mad->mad_hdr.mgmt_class); - if (!ib_is_mad_class_rmpp(rmpp_mad->mad_hdr.mgmt_class)) { - copy_offset = IB_MGMT_MAD_HDR; - rmpp_active = 0; - } else { + + if (ib_is_mad_class_rmpp(rmpp_mad->mad_hdr.mgmt_class) + && ib_mad_kernel_rmpp_agent(agent)) { copy_offset = IB_MGMT_RMPP_HDR; rmpp_active = ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) & - IB_MGMT_RMPP_FLAG_ACTIVE; + IB_MGMT_RMPP_FLAG_ACTIVE; + } else { + copy_offset = IB_MGMT_MAD_HDR; + rmpp_active = 0; } data_len = count - hdr_size(file) - hdr_len; @@ -556,14 +560,22 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf, rmpp_mad->mad_hdr.tid = *tid; } - spin_lock_irq(&file->send_lock); - ret = is_duplicate(file, packet); - if (!ret) + if (!ib_mad_kernel_rmpp_agent(agent) + && ib_is_mad_class_rmpp(rmpp_mad->mad_hdr.mgmt_class) + && (ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) & IB_MGMT_RMPP_FLAG_ACTIVE)) { + spin_lock_irq(&file->send_lock); list_add_tail(&packet->list, &file->send_list); - spin_unlock_irq(&file->send_lock); - if (ret) { - ret = -EINVAL; - goto err_msg; + spin_unlock_irq(&file->send_lock); + } else { + spin_lock_irq(&file->send_lock); + ret = is_duplicate(file, packet); + if (!ret) + list_add_tail(&packet->list, &file->send_list); + spin_unlock_irq(&file->send_lock); + if (ret) { + ret = -EINVAL; + goto err_msg; + } } ret = ib_post_send_mad(packet->msg, NULL); @@ -614,6 +626,8 @@ static int ib_umad_reg_agent(struct ib_umad_file *file, void __user *arg, mutex_lock(&file->mutex); if (!file->port->ib_dev) { + dev_notice(file->port->dev, + "ib_umad_reg_agent: invalid device\n"); ret = -EPIPE; goto out; } @@ -624,6 +638,9 @@ static int ib_umad_reg_agent(struct ib_umad_file *file, void __user *arg, } if (ureq.qpn != 0 && ureq.qpn != 1) { + dev_notice(file->port->dev, + "ib_umad_reg_agent: invalid QPN %d specified\n", + ureq.qpn); ret = -EINVAL; goto out; } @@ -632,11 +649,15 @@ static int ib_umad_reg_agent(struct ib_umad_file *file, void __user *arg, if (!__get_agent(file, agent_id)) goto found; + dev_notice(file->port->dev, + "ib_umad_reg_agent: Max Agents (%u) reached\n", + IB_UMAD_MAX_AGENTS); ret = -ENOMEM; goto out; found: if (ureq.mgmt_class) { + memset(&req, 0, sizeof(req)); req.mgmt_class = ureq.mgmt_class; req.mgmt_class_version = ureq.mgmt_class_version; memcpy(req.oui, ureq.oui, sizeof req.oui); @@ -657,7 +678,7 @@ found: ureq.qpn ? IB_QPT_GSI : IB_QPT_SMI, ureq.mgmt_class ? &req : NULL, ureq.rmpp_version, - send_handler, recv_handler, file); + send_handler, recv_handler, file, 0); if (IS_ERR(agent)) { ret = PTR_ERR(agent); agent = NULL; @@ -673,10 +694,11 @@ found: if (!file->already_used) { file->already_used = 1; if (!file->use_pkey_index) { - printk(KERN_WARNING "user_mad: process %s did not enable " - "P_Key index support.\n", current->comm); - printk(KERN_WARNING "user_mad: Documentation/infiniband/user_mad.txt " - "has info on the new ABI.\n"); + dev_warn(file->port->dev, + "process %s did not enable P_Key index support.\n", + current->comm); + dev_warn(file->port->dev, + " Documentation/infiniband/user_mad.txt has info on the new ABI.\n"); } } @@ -694,6 +716,119 @@ out: return ret; } +static int ib_umad_reg_agent2(struct ib_umad_file *file, void __user *arg) +{ + struct ib_user_mad_reg_req2 ureq; + struct ib_mad_reg_req req; + struct ib_mad_agent *agent = NULL; + int agent_id; + int ret; + + mutex_lock(&file->port->file_mutex); + mutex_lock(&file->mutex); + + if (!file->port->ib_dev) { + dev_notice(file->port->dev, + "ib_umad_reg_agent2: invalid device\n"); + ret = -EPIPE; + goto out; + } + + if (copy_from_user(&ureq, arg, sizeof(ureq))) { + ret = -EFAULT; + goto out; + } + + if (ureq.qpn != 0 && ureq.qpn != 1) { + dev_notice(file->port->dev, + "ib_umad_reg_agent2: invalid QPN %d specified\n", + ureq.qpn); + ret = -EINVAL; + goto out; + } + + if (ureq.flags & ~IB_USER_MAD_REG_FLAGS_CAP) { + dev_notice(file->port->dev, + "ib_umad_reg_agent2 failed: invalid registration flags specified 0x%x; supported 0x%x\n", + ureq.flags, IB_USER_MAD_REG_FLAGS_CAP); + ret = -EINVAL; + + if (put_user((u32)IB_USER_MAD_REG_FLAGS_CAP, + (u32 __user *) (arg + offsetof(struct + ib_user_mad_reg_req2, flags)))) + ret = -EFAULT; + + goto out; + } + + for (agent_id = 0; agent_id < IB_UMAD_MAX_AGENTS; ++agent_id) + if (!__get_agent(file, agent_id)) + goto found; + + dev_notice(file->port->dev, + "ib_umad_reg_agent2: Max Agents (%u) reached\n", + IB_UMAD_MAX_AGENTS); + ret = -ENOMEM; + goto out; + +found: + if (ureq.mgmt_class) { + memset(&req, 0, sizeof(req)); + req.mgmt_class = ureq.mgmt_class; + req.mgmt_class_version = ureq.mgmt_class_version; + if (ureq.oui & 0xff000000) { + dev_notice(file->port->dev, + "ib_umad_reg_agent2 failed: oui invalid 0x%08x\n", + ureq.oui); + ret = -EINVAL; + goto out; + } + req.oui[2] = ureq.oui & 0x0000ff; + req.oui[1] = (ureq.oui & 0x00ff00) >> 8; + req.oui[0] = (ureq.oui & 0xff0000) >> 16; + memcpy(req.method_mask, ureq.method_mask, + sizeof(req.method_mask)); + } + + agent = ib_register_mad_agent(file->port->ib_dev, file->port->port_num, + ureq.qpn ? IB_QPT_GSI : IB_QPT_SMI, + ureq.mgmt_class ? &req : NULL, + ureq.rmpp_version, + send_handler, recv_handler, file, + ureq.flags); + if (IS_ERR(agent)) { + ret = PTR_ERR(agent); + agent = NULL; + goto out; + } + + if (put_user(agent_id, + (u32 __user *)(arg + + offsetof(struct ib_user_mad_reg_req2, id)))) { + ret = -EFAULT; + goto out; + } + + if (!file->already_used) { + file->already_used = 1; + file->use_pkey_index = 1; + } + + file->agent[agent_id] = agent; + ret = 0; + +out: + mutex_unlock(&file->mutex); + + if (ret && agent) + ib_unregister_mad_agent(agent); + + mutex_unlock(&file->port->file_mutex); + + return ret; +} + + static int ib_umad_unreg_agent(struct ib_umad_file *file, u32 __user *arg) { struct ib_mad_agent *agent = NULL; @@ -749,6 +884,8 @@ static long ib_umad_ioctl(struct file *filp, unsigned int cmd, return ib_umad_unreg_agent(filp->private_data, (__u32 __user *) arg); case IB_USER_MAD_ENABLE_PKEY: return ib_umad_enable_pkey(filp->private_data); + case IB_USER_MAD_REGISTER_AGENT2: + return ib_umad_reg_agent2(filp->private_data, (void __user *) arg); default: return -ENOIOCTLCMD; } @@ -765,6 +902,8 @@ static long ib_umad_compat_ioctl(struct file *filp, unsigned int cmd, return ib_umad_unreg_agent(filp->private_data, compat_ptr(arg)); case IB_USER_MAD_ENABLE_PKEY: return ib_umad_enable_pkey(filp->private_data); + case IB_USER_MAD_REGISTER_AGENT2: + return ib_umad_reg_agent2(filp->private_data, compat_ptr(arg)); default: return -ENOIOCTLCMD; } @@ -983,7 +1122,7 @@ static CLASS_ATTR_STRING(abi_version, S_IRUGO, static dev_t overflow_maj; static DECLARE_BITMAP(overflow_map, IB_UMAD_MAX_PORTS); -static int find_overflow_devnum(void) +static int find_overflow_devnum(struct ib_device *device) { int ret; @@ -991,7 +1130,8 @@ static int find_overflow_devnum(void) ret = alloc_chrdev_region(&overflow_maj, 0, IB_UMAD_MAX_PORTS * 2, "infiniband_mad"); if (ret) { - printk(KERN_ERR "user_mad: couldn't register dynamic device number\n"); + dev_err(&device->dev, + "couldn't register dynamic device number\n"); return ret; } } @@ -1014,7 +1154,7 @@ static int ib_umad_init_port(struct ib_device *device, int port_num, devnum = find_first_zero_bit(dev_map, IB_UMAD_MAX_PORTS); if (devnum >= IB_UMAD_MAX_PORTS) { spin_unlock(&port_lock); - devnum = find_overflow_devnum(); + devnum = find_overflow_devnum(device); if (devnum < 0) return -1; @@ -1200,14 +1340,14 @@ static int __init ib_umad_init(void) ret = register_chrdev_region(base_dev, IB_UMAD_MAX_PORTS * 2, "infiniband_mad"); if (ret) { - printk(KERN_ERR "user_mad: couldn't register device number\n"); + pr_err("couldn't register device number\n"); goto out; } umad_class = class_create(THIS_MODULE, "infiniband_mad"); if (IS_ERR(umad_class)) { ret = PTR_ERR(umad_class); - printk(KERN_ERR "user_mad: couldn't create class infiniband_mad\n"); + pr_err("couldn't create class infiniband_mad\n"); goto out_chrdev; } @@ -1215,13 +1355,13 @@ static int __init ib_umad_init(void) ret = class_create_file(umad_class, &class_attr_abi_version.attr); if (ret) { - printk(KERN_ERR "user_mad: couldn't create abi_version attribute\n"); + pr_err("couldn't create abi_version attribute\n"); goto out_class; } ret = ib_register_client(&umad_client); if (ret) { - printk(KERN_ERR "user_mad: couldn't register ib_umad client\n"); + pr_err("couldn't register ib_umad client\n"); goto out_class; } diff --git a/drivers/infiniband/core/uverbs.h b/drivers/infiniband/core/uverbs.h index a283274a5a09..643c08a025a5 100644 --- a/drivers/infiniband/core/uverbs.h +++ b/drivers/infiniband/core/uverbs.h @@ -221,6 +221,7 @@ IB_UVERBS_DECLARE_CMD(query_port); IB_UVERBS_DECLARE_CMD(alloc_pd); IB_UVERBS_DECLARE_CMD(dealloc_pd); IB_UVERBS_DECLARE_CMD(reg_mr); +IB_UVERBS_DECLARE_CMD(rereg_mr); IB_UVERBS_DECLARE_CMD(dereg_mr); IB_UVERBS_DECLARE_CMD(alloc_mw); IB_UVERBS_DECLARE_CMD(dealloc_mw); diff --git a/drivers/infiniband/core/uverbs_cmd.c b/drivers/infiniband/core/uverbs_cmd.c index ea6203ee7bcc..0600c50e6215 100644 --- a/drivers/infiniband/core/uverbs_cmd.c +++ b/drivers/infiniband/core/uverbs_cmd.c @@ -1002,6 +1002,99 @@ err_free: return ret; } +ssize_t ib_uverbs_rereg_mr(struct ib_uverbs_file *file, + const char __user *buf, int in_len, + int out_len) +{ + struct ib_uverbs_rereg_mr cmd; + struct ib_uverbs_rereg_mr_resp resp; + struct ib_udata udata; + struct ib_pd *pd = NULL; + struct ib_mr *mr; + struct ib_pd *old_pd; + int ret; + struct ib_uobject *uobj; + + if (out_len < sizeof(resp)) + return -ENOSPC; + + if (copy_from_user(&cmd, buf, sizeof(cmd))) + return -EFAULT; + + INIT_UDATA(&udata, buf + sizeof(cmd), + (unsigned long) cmd.response + sizeof(resp), + in_len - sizeof(cmd), out_len - sizeof(resp)); + + if (cmd.flags & ~IB_MR_REREG_SUPPORTED || !cmd.flags) + return -EINVAL; + + if ((cmd.flags & IB_MR_REREG_TRANS) && + (!cmd.start || !cmd.hca_va || 0 >= cmd.length || + (cmd.start & ~PAGE_MASK) != (cmd.hca_va & ~PAGE_MASK))) + return -EINVAL; + + uobj = idr_write_uobj(&ib_uverbs_mr_idr, cmd.mr_handle, + file->ucontext); + + if (!uobj) + return -EINVAL; + + mr = uobj->object; + + if (cmd.flags & IB_MR_REREG_ACCESS) { + ret = ib_check_mr_access(cmd.access_flags); + if (ret) + goto put_uobjs; + } + + if (cmd.flags & IB_MR_REREG_PD) { + pd = idr_read_pd(cmd.pd_handle, file->ucontext); + if (!pd) { + ret = -EINVAL; + goto put_uobjs; + } + } + + if (atomic_read(&mr->usecnt)) { + ret = -EBUSY; + goto put_uobj_pd; + } + + old_pd = mr->pd; + ret = mr->device->rereg_user_mr(mr, cmd.flags, cmd.start, + cmd.length, cmd.hca_va, + cmd.access_flags, pd, &udata); + if (!ret) { + if (cmd.flags & IB_MR_REREG_PD) { + atomic_inc(&pd->usecnt); + mr->pd = pd; + atomic_dec(&old_pd->usecnt); + } + } else { + goto put_uobj_pd; + } + + memset(&resp, 0, sizeof(resp)); + resp.lkey = mr->lkey; + resp.rkey = mr->rkey; + + if (copy_to_user((void __user *)(unsigned long)cmd.response, + &resp, sizeof(resp))) + ret = -EFAULT; + else + ret = in_len; + +put_uobj_pd: + if (cmd.flags & IB_MR_REREG_PD) + put_pd_read(pd); + +put_uobjs: + + put_uobj_write(mr->uobject); + + return ret; +} + ssize_t ib_uverbs_dereg_mr(struct ib_uverbs_file *file, const char __user *buf, int in_len, int out_len) diff --git a/drivers/infiniband/core/uverbs_main.c b/drivers/infiniband/core/uverbs_main.c index 08219fb3338b..c73b22a257fe 100644 --- a/drivers/infiniband/core/uverbs_main.c +++ b/drivers/infiniband/core/uverbs_main.c @@ -87,6 +87,7 @@ static ssize_t (*uverbs_cmd_table[])(struct ib_uverbs_file *file, [IB_USER_VERBS_CMD_ALLOC_PD] = ib_uverbs_alloc_pd, [IB_USER_VERBS_CMD_DEALLOC_PD] = ib_uverbs_dealloc_pd, [IB_USER_VERBS_CMD_REG_MR] = ib_uverbs_reg_mr, + [IB_USER_VERBS_CMD_REREG_MR] = ib_uverbs_rereg_mr, [IB_USER_VERBS_CMD_DEREG_MR] = ib_uverbs_dereg_mr, [IB_USER_VERBS_CMD_ALLOC_MW] = ib_uverbs_alloc_mw, [IB_USER_VERBS_CMD_DEALLOC_MW] = ib_uverbs_dealloc_mw, diff --git a/drivers/infiniband/hw/amso1100/c2_cq.c b/drivers/infiniband/hw/amso1100/c2_cq.c index 49e0e8533f74..1b63185b4ad4 100644 --- a/drivers/infiniband/hw/amso1100/c2_cq.c +++ b/drivers/infiniband/hw/amso1100/c2_cq.c @@ -260,11 +260,14 @@ static void c2_free_cq_buf(struct c2_dev *c2dev, struct c2_mq *mq) mq->msg_pool.host, dma_unmap_addr(mq, mapping)); } -static int c2_alloc_cq_buf(struct c2_dev *c2dev, struct c2_mq *mq, int q_size, - int msg_size) +static int c2_alloc_cq_buf(struct c2_dev *c2dev, struct c2_mq *mq, + size_t q_size, size_t msg_size) { u8 *pool_start; + if (q_size > SIZE_MAX / msg_size) + return -EINVAL; + pool_start = dma_alloc_coherent(&c2dev->pcidev->dev, q_size * msg_size, &mq->host_dma, GFP_KERNEL); if (!pool_start) diff --git a/drivers/infiniband/hw/cxgb4/ev.c b/drivers/infiniband/hw/cxgb4/ev.c index d61d0a18f784..a98426fed9ee 100644 --- a/drivers/infiniband/hw/cxgb4/ev.c +++ b/drivers/infiniband/hw/cxgb4/ev.c @@ -182,6 +182,7 @@ int c4iw_ev_handler(struct c4iw_dev *dev, u32 qid) chp = get_chp(dev, qid); if (chp) { + t4_clear_cq_armed(&chp->cq); spin_lock_irqsave(&chp->comp_handler_lock, flag); (*chp->ibcq.comp_handler)(&chp->ibcq, chp->ibcq.cq_context); spin_unlock_irqrestore(&chp->comp_handler_lock, flag); diff --git a/drivers/infiniband/hw/cxgb4/qp.c b/drivers/infiniband/hw/cxgb4/qp.c index 086f62f5dc9e..60cfc11a66e4 100644 --- a/drivers/infiniband/hw/cxgb4/qp.c +++ b/drivers/infiniband/hw/cxgb4/qp.c @@ -1066,7 +1066,7 @@ static void __flush_qp(struct c4iw_qp *qhp, struct c4iw_cq *rchp, struct c4iw_cq *schp) { int count; - int flushed; + int rq_flushed, sq_flushed; unsigned long flag; PDBG("%s qhp %p rchp %p schp %p\n", __func__, qhp, rchp, schp); @@ -1084,27 +1084,40 @@ static void __flush_qp(struct c4iw_qp *qhp, struct c4iw_cq *rchp, c4iw_flush_hw_cq(rchp); c4iw_count_rcqes(&rchp->cq, &qhp->wq, &count); - flushed = c4iw_flush_rq(&qhp->wq, &rchp->cq, count); + rq_flushed = c4iw_flush_rq(&qhp->wq, &rchp->cq, count); spin_unlock(&qhp->lock); spin_unlock_irqrestore(&rchp->lock, flag); - if (flushed) { - spin_lock_irqsave(&rchp->comp_handler_lock, flag); - (*rchp->ibcq.comp_handler)(&rchp->ibcq, rchp->ibcq.cq_context); - spin_unlock_irqrestore(&rchp->comp_handler_lock, flag); - } /* locking hierarchy: cq lock first, then qp lock. */ spin_lock_irqsave(&schp->lock, flag); spin_lock(&qhp->lock); if (schp != rchp) c4iw_flush_hw_cq(schp); - flushed = c4iw_flush_sq(qhp); + sq_flushed = c4iw_flush_sq(qhp); spin_unlock(&qhp->lock); spin_unlock_irqrestore(&schp->lock, flag); - if (flushed) { - spin_lock_irqsave(&schp->comp_handler_lock, flag); - (*schp->ibcq.comp_handler)(&schp->ibcq, schp->ibcq.cq_context); - spin_unlock_irqrestore(&schp->comp_handler_lock, flag); + + if (schp == rchp) { + if (t4_clear_cq_armed(&rchp->cq) && + (rq_flushed || sq_flushed)) { + spin_lock_irqsave(&rchp->comp_handler_lock, flag); + (*rchp->ibcq.comp_handler)(&rchp->ibcq, + rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, flag); + } + } else { + if (t4_clear_cq_armed(&rchp->cq) && rq_flushed) { + spin_lock_irqsave(&rchp->comp_handler_lock, flag); + (*rchp->ibcq.comp_handler)(&rchp->ibcq, + rchp->ibcq.cq_context); + spin_unlock_irqrestore(&rchp->comp_handler_lock, flag); + } + if (t4_clear_cq_armed(&schp->cq) && sq_flushed) { + spin_lock_irqsave(&schp->comp_handler_lock, flag); + (*schp->ibcq.comp_handler)(&schp->ibcq, + schp->ibcq.cq_context); + spin_unlock_irqrestore(&schp->comp_handler_lock, flag); + } } } diff --git a/drivers/infiniband/hw/cxgb4/t4.h b/drivers/infiniband/hw/cxgb4/t4.h index 68b0a6bf4eb0..d8d7fa3e446d 100644 --- a/drivers/infiniband/hw/cxgb4/t4.h +++ b/drivers/infiniband/hw/cxgb4/t4.h @@ -531,6 +531,10 @@ static inline int t4_wq_db_enabled(struct t4_wq *wq) return !wq->rq.queue[wq->rq.size].status.db_off; } +enum t4_cq_flags { + CQ_ARMED = 1, +}; + struct t4_cq { struct t4_cqe *queue; dma_addr_t dma_addr; @@ -551,12 +555,19 @@ struct t4_cq { u16 cidx_inc; u8 gen; u8 error; + unsigned long flags; }; +static inline int t4_clear_cq_armed(struct t4_cq *cq) +{ + return test_and_clear_bit(CQ_ARMED, &cq->flags); +} + static inline int t4_arm_cq(struct t4_cq *cq, int se) { u32 val; + set_bit(CQ_ARMED, &cq->flags); while (cq->cidx_inc > CIDXINC_MASK) { val = SEINTARM(0) | CIDXINC(CIDXINC_MASK) | TIMERREG(7) | INGRESSQID(cq->cqid); diff --git a/drivers/infiniband/hw/ipath/ipath_mad.c b/drivers/infiniband/hw/ipath/ipath_mad.c index 43f2d0424d4f..e890e5ba0e01 100644 --- a/drivers/infiniband/hw/ipath/ipath_mad.c +++ b/drivers/infiniband/hw/ipath/ipath_mad.c @@ -726,7 +726,7 @@ bail: * @dd: the infinipath device * @pkeys: the PKEY table */ -static int set_pkeys(struct ipath_devdata *dd, u16 *pkeys) +static int set_pkeys(struct ipath_devdata *dd, u16 *pkeys, u8 port) { struct ipath_portdata *pd; int i; @@ -759,6 +759,7 @@ static int set_pkeys(struct ipath_devdata *dd, u16 *pkeys) } if (changed) { u64 pkey; + struct ib_event event; pkey = (u64) dd->ipath_pkeys[0] | ((u64) dd->ipath_pkeys[1] << 16) | @@ -768,12 +769,17 @@ static int set_pkeys(struct ipath_devdata *dd, u16 *pkeys) (unsigned long long) pkey); ipath_write_kreg(dd, dd->ipath_kregs->kr_partitionkey, pkey); + + event.event = IB_EVENT_PKEY_CHANGE; + event.device = &dd->verbs_dev->ibdev; + event.element.port_num = port; + ib_dispatch_event(&event); } return 0; } static int recv_subn_set_pkeytable(struct ib_smp *smp, - struct ib_device *ibdev) + struct ib_device *ibdev, u8 port) { u32 startpx = 32 * (be32_to_cpu(smp->attr_mod) & 0xffff); __be16 *p = (__be16 *) smp->data; @@ -784,7 +790,7 @@ static int recv_subn_set_pkeytable(struct ib_smp *smp, for (i = 0; i < n; i++) q[i] = be16_to_cpu(p[i]); - if (startpx != 0 || set_pkeys(dev->dd, q) != 0) + if (startpx != 0 || set_pkeys(dev->dd, q, port) != 0) smp->status |= IB_SMP_INVALID_FIELD; return recv_subn_get_pkeytable(smp, ibdev); @@ -1342,7 +1348,7 @@ static int process_subn(struct ib_device *ibdev, int mad_flags, ret = recv_subn_set_portinfo(smp, ibdev, port_num); goto bail; case IB_SMP_ATTR_PKEY_TABLE: - ret = recv_subn_set_pkeytable(smp, ibdev); + ret = recv_subn_set_pkeytable(smp, ibdev, port_num); goto bail; case IB_SMP_ATTR_SM_INFO: if (dev->port_cap_flags & IB_PORT_SM_DISABLED) { diff --git a/drivers/infiniband/hw/mlx4/mad.c b/drivers/infiniband/hw/mlx4/mad.c index 287ad0564acd..82a7dd87089b 100644 --- a/drivers/infiniband/hw/mlx4/mad.c +++ b/drivers/infiniband/hw/mlx4/mad.c @@ -891,7 +891,7 @@ int mlx4_ib_mad_init(struct mlx4_ib_dev *dev) agent = ib_register_mad_agent(&dev->ib_dev, p + 1, q ? IB_QPT_GSI : IB_QPT_SMI, NULL, 0, send_handler, - NULL, NULL); + NULL, NULL, 0); if (IS_ERR(agent)) { ret = PTR_ERR(agent); goto err; diff --git a/drivers/infiniband/hw/mlx4/main.c b/drivers/infiniband/hw/mlx4/main.c index 0f7027e7db13..e1e558a3d692 100644 --- a/drivers/infiniband/hw/mlx4/main.c +++ b/drivers/infiniband/hw/mlx4/main.c @@ -910,8 +910,7 @@ static int __mlx4_ib_default_rules_match(struct ib_qp *qp, const struct default_rules *pdefault_rules = default_table; u8 link_layer = rdma_port_get_link_layer(qp->device, flow_attr->port); - for (i = 0; i < sizeof(default_table)/sizeof(default_table[0]); i++, - pdefault_rules++) { + for (i = 0; i < ARRAY_SIZE(default_table); i++, pdefault_rules++) { __u32 field_types[IB_FLOW_SPEC_SUPPORT_LAYERS]; memset(&field_types, 0, sizeof(field_types)); @@ -965,8 +964,7 @@ static int __mlx4_ib_create_default_rules( int size = 0; int i; - for (i = 0; i < sizeof(pdefault_rules->rules_create_list)/ - sizeof(pdefault_rules->rules_create_list[0]); i++) { + for (i = 0; i < ARRAY_SIZE(pdefault_rules->rules_create_list); i++) { int ret; union ib_flow_spec ib_spec; switch (pdefault_rules->rules_create_list[i]) { @@ -2007,6 +2005,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) (1ull << IB_USER_VERBS_CMD_ALLOC_PD) | (1ull << IB_USER_VERBS_CMD_DEALLOC_PD) | (1ull << IB_USER_VERBS_CMD_REG_MR) | + (1ull << IB_USER_VERBS_CMD_REREG_MR) | (1ull << IB_USER_VERBS_CMD_DEREG_MR) | (1ull << IB_USER_VERBS_CMD_CREATE_COMP_CHANNEL) | (1ull << IB_USER_VERBS_CMD_CREATE_CQ) | @@ -2059,6 +2058,7 @@ static void *mlx4_ib_add(struct mlx4_dev *dev) ibdev->ib_dev.req_notify_cq = mlx4_ib_arm_cq; ibdev->ib_dev.get_dma_mr = mlx4_ib_get_dma_mr; ibdev->ib_dev.reg_user_mr = mlx4_ib_reg_user_mr; + ibdev->ib_dev.rereg_user_mr = mlx4_ib_rereg_user_mr; ibdev->ib_dev.dereg_mr = mlx4_ib_dereg_mr; ibdev->ib_dev.alloc_fast_reg_mr = mlx4_ib_alloc_fast_reg_mr; ibdev->ib_dev.alloc_fast_reg_page_list = mlx4_ib_alloc_fast_reg_page_list; diff --git a/drivers/infiniband/hw/mlx4/mlx4_ib.h b/drivers/infiniband/hw/mlx4/mlx4_ib.h index 369da3ca5d64..e8cad3926bfc 100644 --- a/drivers/infiniband/hw/mlx4/mlx4_ib.h +++ b/drivers/infiniband/hw/mlx4/mlx4_ib.h @@ -788,5 +788,9 @@ int mlx4_ib_steer_qp_alloc(struct mlx4_ib_dev *dev, int count, int *qpn); void mlx4_ib_steer_qp_free(struct mlx4_ib_dev *dev, u32 qpn, int count); int mlx4_ib_steer_qp_reg(struct mlx4_ib_dev *mdev, struct mlx4_ib_qp *mqp, int is_attach); +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); #endif /* MLX4_IB_H */ diff --git a/drivers/infiniband/hw/mlx4/mr.c b/drivers/infiniband/hw/mlx4/mr.c index cb2a8727f3fb..9b0e80e59b08 100644 --- a/drivers/infiniband/hw/mlx4/mr.c +++ b/drivers/infiniband/hw/mlx4/mr.c @@ -144,8 +144,10 @@ struct ib_mr *mlx4_ib_reg_user_mr(struct ib_pd *pd, u64 start, u64 length, if (!mr) return ERR_PTR(-ENOMEM); + /* Force registering the memory as writable. */ + /* Used for memory re-registeration. HCA protects the access */ mr->umem = ib_umem_get(pd->uobject->context, start, length, - access_flags, 0); + access_flags | IB_ACCESS_LOCAL_WRITE, 0); if (IS_ERR(mr->umem)) { err = PTR_ERR(mr->umem); goto err_free; @@ -183,6 +185,90 @@ err_free: return ERR_PTR(err); } +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) +{ + struct mlx4_ib_dev *dev = to_mdev(mr->device); + struct mlx4_ib_mr *mmr = to_mmr(mr); + struct mlx4_mpt_entry *mpt_entry; + struct mlx4_mpt_entry **pmpt_entry = &mpt_entry; + int err; + + /* Since we synchronize this call and mlx4_ib_dereg_mr via uverbs, + * we assume that the calls can't run concurrently. Otherwise, a + * race exists. + */ + err = mlx4_mr_hw_get_mpt(dev->dev, &mmr->mmr, &pmpt_entry); + + if (err) + return err; + + if (flags & IB_MR_REREG_PD) { + err = mlx4_mr_hw_change_pd(dev->dev, *pmpt_entry, + to_mpd(pd)->pdn); + + if (err) + goto release_mpt_entry; + } + + if (flags & IB_MR_REREG_ACCESS) { + err = mlx4_mr_hw_change_access(dev->dev, *pmpt_entry, + convert_access(mr_access_flags)); + + if (err) + goto release_mpt_entry; + } + + if (flags & IB_MR_REREG_TRANS) { + int shift; + int err; + int n; + + mlx4_mr_rereg_mem_cleanup(dev->dev, &mmr->mmr); + ib_umem_release(mmr->umem); + mmr->umem = ib_umem_get(mr->uobject->context, start, length, + mr_access_flags | + IB_ACCESS_LOCAL_WRITE, + 0); + if (IS_ERR(mmr->umem)) { + err = PTR_ERR(mmr->umem); + mmr->umem = NULL; + goto release_mpt_entry; + } + n = ib_umem_page_count(mmr->umem); + shift = ilog2(mmr->umem->page_size); + + mmr->mmr.iova = virt_addr; + mmr->mmr.size = length; + err = mlx4_mr_rereg_mem_write(dev->dev, &mmr->mmr, + virt_addr, length, n, shift, + *pmpt_entry); + if (err) { + ib_umem_release(mmr->umem); + goto release_mpt_entry; + } + + err = mlx4_ib_umem_write_mtt(dev, &mmr->mmr.mtt, mmr->umem); + if (err) { + mlx4_mr_rereg_mem_cleanup(dev->dev, &mmr->mmr); + ib_umem_release(mmr->umem); + goto release_mpt_entry; + } + } + + /* If we couldn't transfer the MR to the HCA, just remember to + * return a failure. But dereg_mr will free the resources. + */ + err = mlx4_mr_hw_write_mpt(dev->dev, &mmr->mmr, pmpt_entry); + +release_mpt_entry: + mlx4_mr_hw_put_mpt(dev->dev, pmpt_entry); + + return err; +} + int mlx4_ib_dereg_mr(struct ib_mr *ibmr) { struct mlx4_ib_mr *mr = to_mmr(ibmr); diff --git a/drivers/infiniband/hw/mlx5/qp.c b/drivers/infiniband/hw/mlx5/qp.c index bbbcf389272c..416cb7244224 100644 --- a/drivers/infiniband/hw/mlx5/qp.c +++ b/drivers/infiniband/hw/mlx5/qp.c @@ -2501,7 +2501,7 @@ int mlx5_ib_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr, spin_lock_irqsave(&qp->sq.lock, flags); for (nreq = 0; wr; nreq++, wr = wr->next) { - if (unlikely(wr->opcode >= sizeof(mlx5_ib_opcode) / sizeof(mlx5_ib_opcode[0]))) { + if (unlikely(wr->opcode >= ARRAY_SIZE(mlx5_ib_opcode))) { mlx5_ib_warn(dev, "\n"); err = -EINVAL; *bad_wr = wr; diff --git a/drivers/infiniband/hw/mthca/mthca_mad.c b/drivers/infiniband/hw/mthca/mthca_mad.c index b6f7f457fc55..8881fa376e06 100644 --- a/drivers/infiniband/hw/mthca/mthca_mad.c +++ b/drivers/infiniband/hw/mthca/mthca_mad.c @@ -294,7 +294,7 @@ int mthca_create_agents(struct mthca_dev *dev) agent = ib_register_mad_agent(&dev->ib_dev, p + 1, q ? IB_QPT_GSI : IB_QPT_SMI, NULL, 0, send_handler, - NULL, NULL); + NULL, NULL, 0); if (IS_ERR(agent)) { ret = PTR_ERR(agent); goto err; diff --git a/drivers/infiniband/hw/qib/qib_mad.c b/drivers/infiniband/hw/qib/qib_mad.c index 22c720e5740d..636be117b578 100644 --- a/drivers/infiniband/hw/qib/qib_mad.c +++ b/drivers/infiniband/hw/qib/qib_mad.c @@ -2476,7 +2476,7 @@ int qib_create_agents(struct qib_ibdev *dev) ibp = &dd->pport[p].ibport_data; agent = ib_register_mad_agent(&dev->ibdev, p + 1, IB_QPT_SMI, NULL, 0, send_handler, - NULL, NULL); + NULL, NULL, 0); if (IS_ERR(agent)) { ret = PTR_ERR(agent); goto err; diff --git a/drivers/infiniband/ulp/ipoib/ipoib.h b/drivers/infiniband/ulp/ipoib/ipoib.h index c639f90cfda4..3edce617c31b 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib.h +++ b/drivers/infiniband/ulp/ipoib/ipoib.h @@ -86,7 +86,6 @@ enum { IPOIB_FLAG_INITIALIZED = 1, IPOIB_FLAG_ADMIN_UP = 2, IPOIB_PKEY_ASSIGNED = 3, - IPOIB_PKEY_STOP = 4, IPOIB_FLAG_SUBINTERFACE = 5, IPOIB_MCAST_RUN = 6, IPOIB_STOP_REAPER = 7, @@ -312,7 +311,6 @@ struct ipoib_dev_priv { struct list_head multicast_list; struct rb_root multicast_tree; - struct delayed_work pkey_poll_task; struct delayed_work mcast_task; struct work_struct carrier_on_task; struct work_struct flush_light; @@ -473,10 +471,11 @@ void ipoib_ib_dev_flush_heavy(struct work_struct *work); void ipoib_pkey_event(struct work_struct *work); void ipoib_ib_dev_cleanup(struct net_device *dev); -int ipoib_ib_dev_open(struct net_device *dev); +int ipoib_ib_dev_open(struct net_device *dev, int flush); int ipoib_ib_dev_up(struct net_device *dev); int ipoib_ib_dev_down(struct net_device *dev, int flush); int ipoib_ib_dev_stop(struct net_device *dev, int flush); +void ipoib_pkey_dev_check_presence(struct net_device *dev); int ipoib_dev_init(struct net_device *dev, struct ib_device *ca, int port); void ipoib_dev_cleanup(struct net_device *dev); @@ -532,8 +531,7 @@ int ipoib_set_mode(struct net_device *dev, const char *buf); void ipoib_setup(struct net_device *dev); -void ipoib_pkey_poll(struct work_struct *work); -int ipoib_pkey_dev_delay_open(struct net_device *dev); +void ipoib_pkey_open(struct ipoib_dev_priv *priv); void ipoib_drain_cq(struct net_device *dev); void ipoib_set_ethtool_ops(struct net_device *dev); diff --git a/drivers/infiniband/ulp/ipoib/ipoib_fs.c b/drivers/infiniband/ulp/ipoib/ipoib_fs.c index 50061854616e..6bd5740e2691 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_fs.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_fs.c @@ -281,10 +281,8 @@ void ipoib_delete_debug_files(struct net_device *dev) { struct ipoib_dev_priv *priv = netdev_priv(dev); - if (priv->mcg_dentry) - debugfs_remove(priv->mcg_dentry); - if (priv->path_dentry) - debugfs_remove(priv->path_dentry); + debugfs_remove(priv->mcg_dentry); + debugfs_remove(priv->path_dentry); } int ipoib_register_debugfs(void) diff --git a/drivers/infiniband/ulp/ipoib/ipoib_ib.c b/drivers/infiniband/ulp/ipoib/ipoib_ib.c index 6a7003ddb0be..72626c348174 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c @@ -664,17 +664,18 @@ static void ipoib_ib_tx_timer_func(unsigned long ctx) drain_tx_cq((struct net_device *)ctx); } -int ipoib_ib_dev_open(struct net_device *dev) +int ipoib_ib_dev_open(struct net_device *dev, int flush) { struct ipoib_dev_priv *priv = netdev_priv(dev); int ret; - if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &priv->pkey_index)) { - ipoib_warn(priv, "P_Key 0x%04x not found\n", priv->pkey); - clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags); + ipoib_pkey_dev_check_presence(dev); + + if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) { + ipoib_warn(priv, "P_Key 0x%04x is %s\n", priv->pkey, + (!(priv->pkey & 0x7fff) ? "Invalid" : "not found")); return -1; } - set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags); ret = ipoib_init_qp(dev); if (ret) { @@ -705,16 +706,17 @@ int ipoib_ib_dev_open(struct net_device *dev) dev_stop: if (!test_and_set_bit(IPOIB_FLAG_INITIALIZED, &priv->flags)) napi_enable(&priv->napi); - ipoib_ib_dev_stop(dev, 1); + ipoib_ib_dev_stop(dev, flush); return -1; } -static void ipoib_pkey_dev_check_presence(struct net_device *dev) +void ipoib_pkey_dev_check_presence(struct net_device *dev) { struct ipoib_dev_priv *priv = netdev_priv(dev); - u16 pkey_index = 0; - if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &pkey_index)) + if (!(priv->pkey & 0x7fff) || + ib_find_pkey(priv->ca, priv->port, priv->pkey, + &priv->pkey_index)) clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags); else set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags); @@ -745,14 +747,6 @@ int ipoib_ib_dev_down(struct net_device *dev, int flush) clear_bit(IPOIB_FLAG_OPER_UP, &priv->flags); netif_carrier_off(dev); - /* Shutdown the P_Key thread if still active */ - if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) { - mutex_lock(&pkey_mutex); - set_bit(IPOIB_PKEY_STOP, &priv->flags); - cancel_delayed_work_sync(&priv->pkey_poll_task); - mutex_unlock(&pkey_mutex); - } - ipoib_mcast_stop_thread(dev, flush); ipoib_mcast_dev_flush(dev); @@ -924,7 +918,7 @@ int ipoib_ib_dev_init(struct net_device *dev, struct ib_device *ca, int port) (unsigned long) dev); if (dev->flags & IFF_UP) { - if (ipoib_ib_dev_open(dev)) { + if (ipoib_ib_dev_open(dev, 1)) { ipoib_transport_dev_cleanup(dev); return -ENODEV; } @@ -966,13 +960,27 @@ static inline int update_parent_pkey(struct ipoib_dev_priv *priv) return 1; } +/* + * returns 0 if pkey value was found in a different slot. + */ +static inline int update_child_pkey(struct ipoib_dev_priv *priv) +{ + u16 old_index = priv->pkey_index; + + priv->pkey_index = 0; + ipoib_pkey_dev_check_presence(priv->dev); + + if (test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags) && + (old_index == priv->pkey_index)) + return 1; + return 0; +} static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv, enum ipoib_flush_level level) { struct ipoib_dev_priv *cpriv; struct net_device *dev = priv->dev; - u16 new_index; int result; down_read(&priv->vlan_rwsem); @@ -986,16 +994,20 @@ static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv, up_read(&priv->vlan_rwsem); - if (!test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags)) { - /* for non-child devices must check/update the pkey value here */ - if (level == IPOIB_FLUSH_HEAVY && - !test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags)) - update_parent_pkey(priv); + if (!test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags) && + level != IPOIB_FLUSH_HEAVY) { ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_INITIALIZED not set.\n"); return; } if (!test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags)) { + /* interface is down. update pkey and leave. */ + if (level == IPOIB_FLUSH_HEAVY) { + if (!test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags)) + update_parent_pkey(priv); + else + update_child_pkey(priv); + } ipoib_dbg(priv, "Not flushing - IPOIB_FLAG_ADMIN_UP not set.\n"); return; } @@ -1005,20 +1017,13 @@ static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv, * (parent) devices should always takes what present in pkey index 0 */ if (test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags)) { - if (ib_find_pkey(priv->ca, priv->port, priv->pkey, &new_index)) { - clear_bit(IPOIB_PKEY_ASSIGNED, &priv->flags); - ipoib_ib_dev_down(dev, 0); - ipoib_ib_dev_stop(dev, 0); - if (ipoib_pkey_dev_delay_open(dev)) - return; - } - /* restart QP only if P_Key index is changed */ - if (test_and_set_bit(IPOIB_PKEY_ASSIGNED, &priv->flags) && - new_index == priv->pkey_index) { + result = update_child_pkey(priv); + if (result) { + /* restart QP only if P_Key index is changed */ ipoib_dbg(priv, "Not flushing - P_Key index not changed.\n"); return; } - priv->pkey_index = new_index; + } else { result = update_parent_pkey(priv); /* restart QP only if P_Key value changed */ @@ -1038,8 +1043,12 @@ static void __ipoib_ib_dev_flush(struct ipoib_dev_priv *priv, ipoib_ib_dev_down(dev, 0); if (level == IPOIB_FLUSH_HEAVY) { - ipoib_ib_dev_stop(dev, 0); - ipoib_ib_dev_open(dev); + if (test_bit(IPOIB_FLAG_INITIALIZED, &priv->flags)) + ipoib_ib_dev_stop(dev, 0); + if (ipoib_ib_dev_open(dev, 0) != 0) + return; + if (netif_queue_stopped(dev)) + netif_start_queue(dev); } /* @@ -1094,54 +1103,4 @@ void ipoib_ib_dev_cleanup(struct net_device *dev) ipoib_transport_dev_cleanup(dev); } -/* - * Delayed P_Key Assigment Interim Support - * - * The following is initial implementation of delayed P_Key assigment - * mechanism. It is using the same approach implemented for the multicast - * group join. The single goal of this implementation is to quickly address - * Bug #2507. This implementation will probably be removed when the P_Key - * change async notification is available. - */ - -void ipoib_pkey_poll(struct work_struct *work) -{ - struct ipoib_dev_priv *priv = - container_of(work, struct ipoib_dev_priv, pkey_poll_task.work); - struct net_device *dev = priv->dev; - - ipoib_pkey_dev_check_presence(dev); - - if (test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) - ipoib_open(dev); - else { - mutex_lock(&pkey_mutex); - if (!test_bit(IPOIB_PKEY_STOP, &priv->flags)) - queue_delayed_work(ipoib_workqueue, - &priv->pkey_poll_task, - HZ); - mutex_unlock(&pkey_mutex); - } -} - -int ipoib_pkey_dev_delay_open(struct net_device *dev) -{ - struct ipoib_dev_priv *priv = netdev_priv(dev); - - /* Look for the interface pkey value in the IB Port P_Key table and */ - /* set the interface pkey assigment flag */ - ipoib_pkey_dev_check_presence(dev); - /* P_Key value not assigned yet - start polling */ - if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) { - mutex_lock(&pkey_mutex); - clear_bit(IPOIB_PKEY_STOP, &priv->flags); - queue_delayed_work(ipoib_workqueue, - &priv->pkey_poll_task, - HZ); - mutex_unlock(&pkey_mutex); - return 1; - } - - return 0; -} diff --git a/drivers/infiniband/ulp/ipoib/ipoib_main.c b/drivers/infiniband/ulp/ipoib/ipoib_main.c index 5786a78ff8bc..217cb77157d8 100644 --- a/drivers/infiniband/ulp/ipoib/ipoib_main.c +++ b/drivers/infiniband/ulp/ipoib/ipoib_main.c @@ -108,11 +108,11 @@ int ipoib_open(struct net_device *dev) set_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags); - if (ipoib_pkey_dev_delay_open(dev)) - return 0; - - if (ipoib_ib_dev_open(dev)) + if (ipoib_ib_dev_open(dev, 1)) { + if (!test_bit(IPOIB_PKEY_ASSIGNED, &priv->flags)) + return 0; goto err_disable; + } if (ipoib_ib_dev_up(dev)) goto err_stop; @@ -1379,7 +1379,6 @@ void ipoib_setup(struct net_device *dev) INIT_LIST_HEAD(&priv->dead_ahs); INIT_LIST_HEAD(&priv->multicast_list); - INIT_DELAYED_WORK(&priv->pkey_poll_task, ipoib_pkey_poll); INIT_DELAYED_WORK(&priv->mcast_task, ipoib_mcast_join_task); INIT_WORK(&priv->carrier_on_task, ipoib_mcast_carrier_on_task); INIT_WORK(&priv->flush_light, ipoib_ib_dev_flush_light); diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.c b/drivers/infiniband/ulp/iser/iscsi_iser.c index eb7973957a6e..61ee91d88380 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.c +++ b/drivers/infiniband/ulp/iser/iscsi_iser.c @@ -596,20 +596,28 @@ iscsi_iser_ep_connect(struct Scsi_Host *shost, struct sockaddr *dst_addr, struct iser_conn *ib_conn; struct iscsi_endpoint *ep; - ep = iscsi_create_endpoint(sizeof(*ib_conn)); + ep = iscsi_create_endpoint(0); if (!ep) return ERR_PTR(-ENOMEM); - ib_conn = ep->dd_data; + ib_conn = kzalloc(sizeof(*ib_conn), GFP_KERNEL); + if (!ib_conn) { + err = -ENOMEM; + goto failure; + } + + ep->dd_data = ib_conn; ib_conn->ep = ep; iser_conn_init(ib_conn); - err = iser_connect(ib_conn, NULL, (struct sockaddr_in *)dst_addr, - non_blocking); + err = iser_connect(ib_conn, NULL, dst_addr, non_blocking); if (err) - return ERR_PTR(err); + goto failure; return ep; +failure: + iscsi_destroy_endpoint(ep); + return ERR_PTR(err); } static int @@ -619,15 +627,16 @@ iscsi_iser_ep_poll(struct iscsi_endpoint *ep, int timeout_ms) int rc; ib_conn = ep->dd_data; - rc = wait_event_interruptible_timeout(ib_conn->wait, - ib_conn->state == ISER_CONN_UP, - msecs_to_jiffies(timeout_ms)); - + rc = wait_for_completion_interruptible_timeout(&ib_conn->up_completion, + msecs_to_jiffies(timeout_ms)); /* if conn establishment failed, return error code to iscsi */ - if (!rc && - (ib_conn->state == ISER_CONN_TERMINATING || - ib_conn->state == ISER_CONN_DOWN)) - rc = -1; + if (rc == 0) { + mutex_lock(&ib_conn->state_mutex); + if (ib_conn->state == ISER_CONN_TERMINATING || + ib_conn->state == ISER_CONN_DOWN) + rc = -1; + mutex_unlock(&ib_conn->state_mutex); + } iser_info("ib conn %p rc = %d\n", ib_conn, rc); @@ -646,19 +655,25 @@ iscsi_iser_ep_disconnect(struct iscsi_endpoint *ep) ib_conn = ep->dd_data; iser_info("ep %p ib conn %p state %d\n", ep, ib_conn, ib_conn->state); + mutex_lock(&ib_conn->state_mutex); iser_conn_terminate(ib_conn); /* - * if iser_conn and iscsi_conn are bound, we must wait iscsi_conn_stop - * call and ISER_CONN_DOWN state before freeing the iser resources. - * otherwise we are safe to free resources immediately. + * if iser_conn and iscsi_conn are bound, we must wait for + * iscsi_conn_stop and flush errors completion before freeing + * the iser resources. Otherwise we are safe to free resources + * immediately. */ if (ib_conn->iscsi_conn) { INIT_WORK(&ib_conn->release_work, iser_release_work); queue_work(release_wq, &ib_conn->release_work); + mutex_unlock(&ib_conn->state_mutex); } else { + ib_conn->state = ISER_CONN_DOWN; + mutex_unlock(&ib_conn->state_mutex); iser_conn_release(ib_conn); } + iscsi_destroy_endpoint(ep); } static umode_t iser_attr_is_visible(int param_type, int param) diff --git a/drivers/infiniband/ulp/iser/iscsi_iser.h b/drivers/infiniband/ulp/iser/iscsi_iser.h index 97cd385bf7f7..c877dad381cb 100644 --- a/drivers/infiniband/ulp/iser/iscsi_iser.h +++ b/drivers/infiniband/ulp/iser/iscsi_iser.h @@ -326,7 +326,6 @@ struct iser_conn { struct iser_device *device; /* device context */ struct rdma_cm_id *cma_id; /* CMA ID */ struct ib_qp *qp; /* QP */ - wait_queue_head_t wait; /* waitq for conn/disconn */ unsigned qp_max_recv_dtos; /* num of rx buffers */ unsigned qp_max_recv_dtos_mask; /* above minus 1 */ unsigned min_posted_rx; /* qp_max_recv_dtos >> 2 */ @@ -335,6 +334,9 @@ struct iser_conn { char name[ISER_OBJECT_NAME_SIZE]; struct work_struct release_work; struct completion stop_completion; + struct mutex state_mutex; + struct completion flush_completion; + struct completion up_completion; struct list_head conn_list; /* entry in ig conn list */ char *login_buf; @@ -448,8 +450,8 @@ int iser_reg_rdma_mem_fastreg(struct iscsi_iser_task *task, enum iser_data_dir cmd_dir); int iser_connect(struct iser_conn *ib_conn, - struct sockaddr_in *src_addr, - struct sockaddr_in *dst_addr, + struct sockaddr *src_addr, + struct sockaddr *dst_addr, int non_blocking); int iser_reg_page_vec(struct iser_conn *ib_conn, diff --git a/drivers/infiniband/ulp/iser/iser_verbs.c b/drivers/infiniband/ulp/iser/iser_verbs.c index ea01075f9f9b..3ef167f97d6f 100644 --- a/drivers/infiniband/ulp/iser/iser_verbs.c +++ b/drivers/infiniband/ulp/iser/iser_verbs.c @@ -491,10 +491,9 @@ out_err: } /** - * releases the QP objects, returns 0 on success, - * -1 on failure + * releases the QP object */ -static int iser_free_ib_conn_res(struct iser_conn *ib_conn) +static void iser_free_ib_conn_res(struct iser_conn *ib_conn) { int cq_index; BUG_ON(ib_conn == NULL); @@ -513,8 +512,6 @@ static int iser_free_ib_conn_res(struct iser_conn *ib_conn) } ib_conn->qp = NULL; - - return 0; } /** @@ -568,31 +565,40 @@ static void iser_device_try_release(struct iser_device *device) mutex_unlock(&ig.device_list_mutex); } +/** + * Called with state mutex held + **/ static int iser_conn_state_comp_exch(struct iser_conn *ib_conn, enum iser_ib_conn_state comp, enum iser_ib_conn_state exch) { int ret; - spin_lock_bh(&ib_conn->lock); if ((ret = (ib_conn->state == comp))) ib_conn->state = exch; - spin_unlock_bh(&ib_conn->lock); return ret; } void iser_release_work(struct work_struct *work) { struct iser_conn *ib_conn; + int rc; ib_conn = container_of(work, struct iser_conn, release_work); /* wait for .conn_stop callback */ - wait_for_completion(&ib_conn->stop_completion); + rc = wait_for_completion_timeout(&ib_conn->stop_completion, 30 * HZ); + WARN_ON(rc == 0); /* wait for the qp`s post send and post receive buffers to empty */ - wait_event_interruptible(ib_conn->wait, - ib_conn->state == ISER_CONN_DOWN); + rc = wait_for_completion_timeout(&ib_conn->flush_completion, 30 * HZ); + WARN_ON(rc == 0); + + ib_conn->state = ISER_CONN_DOWN; + + mutex_lock(&ib_conn->state_mutex); + ib_conn->state = ISER_CONN_DOWN; + mutex_unlock(&ib_conn->state_mutex); iser_conn_release(ib_conn); } @@ -604,23 +610,27 @@ void iser_conn_release(struct iser_conn *ib_conn) { struct iser_device *device = ib_conn->device; - BUG_ON(ib_conn->state == ISER_CONN_UP); - mutex_lock(&ig.connlist_mutex); list_del(&ib_conn->conn_list); mutex_unlock(&ig.connlist_mutex); + + mutex_lock(&ib_conn->state_mutex); + BUG_ON(ib_conn->state != ISER_CONN_DOWN); + iser_free_rx_descriptors(ib_conn); iser_free_ib_conn_res(ib_conn); ib_conn->device = NULL; /* on EVENT_ADDR_ERROR there's no device yet for this conn */ if (device != NULL) iser_device_try_release(device); + mutex_unlock(&ib_conn->state_mutex); + /* if cma handler context, the caller actually destroy the id */ if (ib_conn->cma_id != NULL) { rdma_destroy_id(ib_conn->cma_id); ib_conn->cma_id = NULL; } - iscsi_destroy_endpoint(ib_conn->ep); + kfree(ib_conn); } /** @@ -642,22 +652,31 @@ void iser_conn_terminate(struct iser_conn *ib_conn) ib_conn,err); } +/** + * Called with state mutex held + **/ static void iser_connect_error(struct rdma_cm_id *cma_id) { struct iser_conn *ib_conn; ib_conn = (struct iser_conn *)cma_id->context; - ib_conn->state = ISER_CONN_DOWN; - wake_up_interruptible(&ib_conn->wait); } +/** + * Called with state mutex held + **/ static void iser_addr_handler(struct rdma_cm_id *cma_id) { struct iser_device *device; struct iser_conn *ib_conn; int ret; + ib_conn = (struct iser_conn *)cma_id->context; + if (ib_conn->state != ISER_CONN_PENDING) + /* bailout */ + return; + device = iser_device_find_by_ib_device(cma_id); if (!device) { iser_err("device lookup/creation failed\n"); @@ -665,7 +684,6 @@ static void iser_addr_handler(struct rdma_cm_id *cma_id) return; } - ib_conn = (struct iser_conn *)cma_id->context; ib_conn->device = device; /* connection T10-PI support */ @@ -689,18 +707,27 @@ static void iser_addr_handler(struct rdma_cm_id *cma_id) } } +/** + * Called with state mutex held + **/ static void iser_route_handler(struct rdma_cm_id *cma_id) { struct rdma_conn_param conn_param; int ret; struct iser_cm_hdr req_hdr; + struct iser_conn *ib_conn = (struct iser_conn *)cma_id->context; + struct iser_device *device = ib_conn->device; + + if (ib_conn->state != ISER_CONN_PENDING) + /* bailout */ + return; ret = iser_create_ib_conn_res((struct iser_conn *)cma_id->context); if (ret) goto failure; memset(&conn_param, 0, sizeof conn_param); - conn_param.responder_resources = 4; + conn_param.responder_resources = device->dev_attr.max_qp_rd_atom; conn_param.initiator_depth = 1; conn_param.retry_count = 7; conn_param.rnr_retry_count = 6; @@ -728,12 +755,16 @@ static void iser_connected_handler(struct rdma_cm_id *cma_id) struct ib_qp_attr attr; struct ib_qp_init_attr init_attr; + ib_conn = (struct iser_conn *)cma_id->context; + if (ib_conn->state != ISER_CONN_PENDING) + /* bailout */ + return; + (void)ib_query_qp(cma_id->qp, &attr, ~0, &init_attr); iser_info("remote qpn:%x my qpn:%x\n", attr.dest_qp_num, cma_id->qp->qp_num); - ib_conn = (struct iser_conn *)cma_id->context; - if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_PENDING, ISER_CONN_UP)) - wake_up_interruptible(&ib_conn->wait); + ib_conn->state = ISER_CONN_UP; + complete(&ib_conn->up_completion); } static void iser_disconnected_handler(struct rdma_cm_id *cma_id) @@ -752,19 +783,25 @@ static void iser_disconnected_handler(struct rdma_cm_id *cma_id) iser_err("iscsi_iser connection isn't bound\n"); } - /* Complete the termination process if no posts are pending */ + /* Complete the termination process if no posts are pending. This code + * block also exists in iser_handle_comp_error(), but it is needed here + * for cases of no flushes at all, e.g. discovery over rdma. + */ if (ib_conn->post_recv_buf_count == 0 && (atomic_read(&ib_conn->post_send_buf_count) == 0)) { - ib_conn->state = ISER_CONN_DOWN; - wake_up_interruptible(&ib_conn->wait); + complete(&ib_conn->flush_completion); } } static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *event) { + struct iser_conn *ib_conn; + + ib_conn = (struct iser_conn *)cma_id->context; iser_info("event %d status %d conn %p id %p\n", event->event, event->status, cma_id->context, cma_id); + mutex_lock(&ib_conn->state_mutex); switch (event->event) { case RDMA_CM_EVENT_ADDR_RESOLVED: iser_addr_handler(cma_id); @@ -785,24 +822,28 @@ static int iser_cma_handler(struct rdma_cm_id *cma_id, struct rdma_cm_event *eve case RDMA_CM_EVENT_DISCONNECTED: case RDMA_CM_EVENT_DEVICE_REMOVAL: case RDMA_CM_EVENT_ADDR_CHANGE: + case RDMA_CM_EVENT_TIMEWAIT_EXIT: iser_disconnected_handler(cma_id); break; default: iser_err("Unexpected RDMA CM event (%d)\n", event->event); break; } + mutex_unlock(&ib_conn->state_mutex); return 0; } void iser_conn_init(struct iser_conn *ib_conn) { ib_conn->state = ISER_CONN_INIT; - init_waitqueue_head(&ib_conn->wait); ib_conn->post_recv_buf_count = 0; atomic_set(&ib_conn->post_send_buf_count, 0); init_completion(&ib_conn->stop_completion); + init_completion(&ib_conn->flush_completion); + init_completion(&ib_conn->up_completion); INIT_LIST_HEAD(&ib_conn->conn_list); spin_lock_init(&ib_conn->lock); + mutex_init(&ib_conn->state_mutex); } /** @@ -810,22 +851,21 @@ void iser_conn_init(struct iser_conn *ib_conn) * sleeps until the connection is established or rejected */ int iser_connect(struct iser_conn *ib_conn, - struct sockaddr_in *src_addr, - struct sockaddr_in *dst_addr, + struct sockaddr *src_addr, + struct sockaddr *dst_addr, int non_blocking) { - struct sockaddr *src, *dst; int err = 0; - sprintf(ib_conn->name, "%pI4:%d", - &dst_addr->sin_addr.s_addr, dst_addr->sin_port); + mutex_lock(&ib_conn->state_mutex); + + sprintf(ib_conn->name, "%pISp", dst_addr); + + iser_info("connecting to: %s\n", ib_conn->name); /* the device is known only --after-- address resolution */ ib_conn->device = NULL; - iser_info("connecting to: %pI4, port 0x%x\n", - &dst_addr->sin_addr, dst_addr->sin_port); - ib_conn->state = ISER_CONN_PENDING; ib_conn->cma_id = rdma_create_id(iser_cma_handler, @@ -837,23 +877,21 @@ int iser_connect(struct iser_conn *ib_conn, goto id_failure; } - src = (struct sockaddr *)src_addr; - dst = (struct sockaddr *)dst_addr; - err = rdma_resolve_addr(ib_conn->cma_id, src, dst, 1000); + err = rdma_resolve_addr(ib_conn->cma_id, src_addr, dst_addr, 1000); if (err) { iser_err("rdma_resolve_addr failed: %d\n", err); goto addr_failure; } if (!non_blocking) { - wait_event_interruptible(ib_conn->wait, - (ib_conn->state != ISER_CONN_PENDING)); + wait_for_completion_interruptible(&ib_conn->up_completion); if (ib_conn->state != ISER_CONN_UP) { err = -EIO; goto connect_failure; } } + mutex_unlock(&ib_conn->state_mutex); mutex_lock(&ig.connlist_mutex); list_add(&ib_conn->conn_list, &ig.connlist); @@ -865,6 +903,7 @@ id_failure: addr_failure: ib_conn->state = ISER_CONN_DOWN; connect_failure: + mutex_unlock(&ib_conn->state_mutex); iser_conn_release(ib_conn); return err; } @@ -1049,18 +1088,19 @@ static void iser_handle_comp_error(struct iser_tx_desc *desc, if (ib_conn->post_recv_buf_count == 0 && atomic_read(&ib_conn->post_send_buf_count) == 0) { - /* getting here when the state is UP means that the conn is * - * being terminated asynchronously from the iSCSI layer's * - * perspective. */ - if (iser_conn_state_comp_exch(ib_conn, ISER_CONN_UP, - ISER_CONN_TERMINATING)) + /** + * getting here when the state is UP means that the conn is + * being terminated asynchronously from the iSCSI layer's + * perspective. It is safe to peek at the connection state + * since iscsi_conn_failure is allowed to be called twice. + **/ + if (ib_conn->state == ISER_CONN_UP) iscsi_conn_failure(ib_conn->iscsi_conn, ISCSI_ERR_CONN_FAILED); /* no more non completed posts to the QP, complete the * termination process w.o worrying on disconnect event */ - ib_conn->state = ISER_CONN_DOWN; - wake_up_interruptible(&ib_conn->wait); + complete(&ib_conn->flush_completion); } } diff --git a/drivers/infiniband/ulp/srp/ib_srp.c b/drivers/infiniband/ulp/srp/ib_srp.c index e3c2c5b4297f..62d2a18e1b41 100644 --- a/drivers/infiniband/ulp/srp/ib_srp.c +++ b/drivers/infiniband/ulp/srp/ib_srp.c @@ -130,6 +130,7 @@ static void srp_send_completion(struct ib_cq *cq, void *target_ptr); static int srp_cm_handler(struct ib_cm_id *cm_id, struct ib_cm_event *event); static struct scsi_transport_template *ib_srp_transport_template; +static struct workqueue_struct *srp_remove_wq; static struct ib_client srp_client = { .name = "srp", @@ -731,7 +732,7 @@ static bool srp_queue_remove_work(struct srp_target_port *target) spin_unlock_irq(&target->lock); if (changed) - queue_work(system_long_wq, &target->remove_work); + queue_work(srp_remove_wq, &target->remove_work); return changed; } @@ -1643,10 +1644,14 @@ static void srp_process_rsp(struct srp_target_port *target, struct srp_rsp *rsp) SCSI_SENSE_BUFFERSIZE)); } - if (rsp->flags & (SRP_RSP_FLAG_DOOVER | SRP_RSP_FLAG_DOUNDER)) - scsi_set_resid(scmnd, be32_to_cpu(rsp->data_out_res_cnt)); - else if (rsp->flags & (SRP_RSP_FLAG_DIOVER | SRP_RSP_FLAG_DIUNDER)) + if (unlikely(rsp->flags & SRP_RSP_FLAG_DIUNDER)) scsi_set_resid(scmnd, be32_to_cpu(rsp->data_in_res_cnt)); + else if (unlikely(rsp->flags & SRP_RSP_FLAG_DIOVER)) + scsi_set_resid(scmnd, -be32_to_cpu(rsp->data_in_res_cnt)); + else if (unlikely(rsp->flags & SRP_RSP_FLAG_DOUNDER)) + scsi_set_resid(scmnd, be32_to_cpu(rsp->data_out_res_cnt)); + else if (unlikely(rsp->flags & SRP_RSP_FLAG_DOOVER)) + scsi_set_resid(scmnd, -be32_to_cpu(rsp->data_out_res_cnt)); srp_free_req(target, req, scmnd, be32_to_cpu(rsp->req_lim_delta)); @@ -3261,9 +3266,10 @@ static void srp_remove_one(struct ib_device *device) spin_unlock(&host->target_lock); /* - * Wait for target port removal tasks. + * Wait for tl_err and target port removal tasks. */ flush_workqueue(system_long_wq); + flush_workqueue(srp_remove_wq); kfree(host); } @@ -3313,16 +3319,22 @@ static int __init srp_init_module(void) indirect_sg_entries = cmd_sg_entries; } + srp_remove_wq = create_workqueue("srp_remove"); + if (!srp_remove_wq) { + ret = -ENOMEM; + goto out; + } + + ret = -ENOMEM; ib_srp_transport_template = srp_attach_transport(&ib_srp_transport_functions); if (!ib_srp_transport_template) - return -ENOMEM; + goto destroy_wq; ret = class_register(&srp_class); if (ret) { pr_err("couldn't register class infiniband_srp\n"); - srp_release_transport(ib_srp_transport_template); - return ret; + goto release_tr; } ib_sa_register_client(&srp_sa_client); @@ -3330,13 +3342,22 @@ static int __init srp_init_module(void) ret = ib_register_client(&srp_client); if (ret) { pr_err("couldn't register IB client\n"); - srp_release_transport(ib_srp_transport_template); - ib_sa_unregister_client(&srp_sa_client); - class_unregister(&srp_class); - return ret; + goto unreg_sa; } - return 0; +out: + return ret; + +unreg_sa: + ib_sa_unregister_client(&srp_sa_client); + class_unregister(&srp_class); + +release_tr: + srp_release_transport(ib_srp_transport_template); + +destroy_wq: + destroy_workqueue(srp_remove_wq); + goto out; } static void __exit srp_cleanup_module(void) @@ -3345,6 +3366,7 @@ static void __exit srp_cleanup_module(void) ib_sa_unregister_client(&srp_sa_client); class_unregister(&srp_class); srp_release_transport(ib_srp_transport_template); + destroy_workqueue(srp_remove_wq); } module_init(srp_init_module); diff --git a/drivers/infiniband/ulp/srpt/ib_srpt.c b/drivers/infiniband/ulp/srpt/ib_srpt.c index fe09f2788b15..d28a8c284da9 100644 --- a/drivers/infiniband/ulp/srpt/ib_srpt.c +++ b/drivers/infiniband/ulp/srpt/ib_srpt.c @@ -198,6 +198,7 @@ static void srpt_event_handler(struct ib_event_handler *handler, case IB_EVENT_PKEY_CHANGE: case IB_EVENT_SM_CHANGE: case IB_EVENT_CLIENT_REREGISTER: + case IB_EVENT_GID_CHANGE: /* Refresh port data asynchronously. */ if (event->element.port_num <= sdev->device->phys_port_cnt) { sport = &sdev->port[event->element.port_num - 1]; @@ -563,7 +564,7 @@ static int srpt_refresh_port(struct srpt_port *sport) ®_req, 0, srpt_mad_send_handler, srpt_mad_recv_handler, - sport); + sport, 0); if (IS_ERR(sport->mad_agent)) { ret = PTR_ERR(sport->mad_agent); sport->mad_agent = NULL; diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c index 5d940a26055c..65a4a0f88ea0 100644 --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c @@ -1311,6 +1311,15 @@ static struct mlx4_cmd_info cmd_info[] = { .wrapper = mlx4_MAD_IFC_wrapper }, { + .opcode = MLX4_CMD_MAD_DEMUX, + .has_inbox = false, + .has_outbox = false, + .out_is_imm = false, + .encode_slave_id = false, + .verify = NULL, + .wrapper = mlx4_CMD_EPERM_wrapper + }, + { .opcode = MLX4_CMD_QUERY_IF_STAT, .has_inbox = false, .has_outbox = true, diff --git a/drivers/net/ethernet/mellanox/mlx4/fw.c b/drivers/net/ethernet/mellanox/mlx4/fw.c index 688e1eabab29..494753e44ae3 100644 --- a/drivers/net/ethernet/mellanox/mlx4/fw.c +++ b/drivers/net/ethernet/mellanox/mlx4/fw.c @@ -136,7 +136,8 @@ static void dump_dev_cap_flags2(struct mlx4_dev *dev, u64 flags) [7] = "FSM (MAC anti-spoofing) support", [8] = "Dynamic QP updates support", [9] = "Device managed flow steering IPoIB support", - [10] = "TCP/IP offloads/flow-steering for VXLAN support" + [10] = "TCP/IP offloads/flow-steering for VXLAN support", + [11] = "MAD DEMUX (Secure-Host) support" }; int i; @@ -571,6 +572,7 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) #define QUERY_DEV_CAP_MAX_ICM_SZ_OFFSET 0xa0 #define QUERY_DEV_CAP_FW_REASSIGN_MAC 0x9d #define QUERY_DEV_CAP_VXLAN 0x9e +#define QUERY_DEV_CAP_MAD_DEMUX_OFFSET 0xb0 dev_cap->flags2 = 0; mailbox = mlx4_alloc_cmd_mailbox(dev); @@ -748,6 +750,11 @@ int mlx4_QUERY_DEV_CAP(struct mlx4_dev *dev, struct mlx4_dev_cap *dev_cap) MLX4_GET(dev_cap->max_counters, outbox, QUERY_DEV_CAP_MAX_COUNTERS_OFFSET); + MLX4_GET(field32, outbox, + QUERY_DEV_CAP_MAD_DEMUX_OFFSET); + if (field32 & (1 << 0)) + dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_MAD_DEMUX; + MLX4_GET(field32, outbox, QUERY_DEV_CAP_EXT_2_FLAGS_OFFSET); if (field32 & (1 << 16)) dev_cap->flags2 |= MLX4_DEV_CAP_FLAG2_UPDATE_QP; @@ -2016,3 +2023,85 @@ void mlx4_opreq_action(struct work_struct *work) out: mlx4_free_cmd_mailbox(dev, mailbox); } + +static int mlx4_check_smp_firewall_active(struct mlx4_dev *dev, + struct mlx4_cmd_mailbox *mailbox) +{ +#define MLX4_CMD_MAD_DEMUX_SET_ATTR_OFFSET 0x10 +#define MLX4_CMD_MAD_DEMUX_GETRESP_ATTR_OFFSET 0x20 +#define MLX4_CMD_MAD_DEMUX_TRAP_ATTR_OFFSET 0x40 +#define MLX4_CMD_MAD_DEMUX_TRAP_REPRESS_ATTR_OFFSET 0x70 + + u32 set_attr_mask, getresp_attr_mask; + u32 trap_attr_mask, traprepress_attr_mask; + + MLX4_GET(set_attr_mask, mailbox->buf, + MLX4_CMD_MAD_DEMUX_SET_ATTR_OFFSET); + mlx4_dbg(dev, "SMP firewall set_attribute_mask = 0x%x\n", + set_attr_mask); + + MLX4_GET(getresp_attr_mask, mailbox->buf, + MLX4_CMD_MAD_DEMUX_GETRESP_ATTR_OFFSET); + mlx4_dbg(dev, "SMP firewall getresp_attribute_mask = 0x%x\n", + getresp_attr_mask); + + MLX4_GET(trap_attr_mask, mailbox->buf, + MLX4_CMD_MAD_DEMUX_TRAP_ATTR_OFFSET); + mlx4_dbg(dev, "SMP firewall trap_attribute_mask = 0x%x\n", + trap_attr_mask); + + MLX4_GET(traprepress_attr_mask, mailbox->buf, + MLX4_CMD_MAD_DEMUX_TRAP_REPRESS_ATTR_OFFSET); + mlx4_dbg(dev, "SMP firewall traprepress_attribute_mask = 0x%x\n", + traprepress_attr_mask); + + if (set_attr_mask && getresp_attr_mask && trap_attr_mask && + traprepress_attr_mask) + return 1; + + return 0; +} + +int mlx4_config_mad_demux(struct mlx4_dev *dev) +{ + struct mlx4_cmd_mailbox *mailbox; + int secure_host_active; + int err; + + /* Check if mad_demux is supported */ + if (!(dev->caps.flags2 & MLX4_DEV_CAP_FLAG2_MAD_DEMUX)) + return 0; + + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR(mailbox)) { + mlx4_warn(dev, "Failed to allocate mailbox for cmd MAD_DEMUX"); + return -ENOMEM; + } + + /* Query mad_demux to find out which MADs are handled by internal sma */ + err = mlx4_cmd_box(dev, 0, mailbox->dma, 0x01 /* subn mgmt class */, + MLX4_CMD_MAD_DEMUX_QUERY_RESTR, MLX4_CMD_MAD_DEMUX, + MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); + if (err) { + mlx4_warn(dev, "MLX4_CMD_MAD_DEMUX: query restrictions failed (%d)\n", + err); + goto out; + } + + secure_host_active = mlx4_check_smp_firewall_active(dev, mailbox); + + /* Config mad_demux to handle all MADs returned by the query above */ + err = mlx4_cmd(dev, mailbox->dma, 0x01 /* subn mgmt class */, + MLX4_CMD_MAD_DEMUX_CONFIG, MLX4_CMD_MAD_DEMUX, + MLX4_CMD_TIME_CLASS_B, MLX4_CMD_NATIVE); + if (err) { + mlx4_warn(dev, "MLX4_CMD_MAD_DEMUX: configure failed (%d)\n", err); + goto out; + } + + if (secure_host_active) + mlx4_warn(dev, "HCA operating in secure-host mode. SMP firewall activated.\n"); +out: + mlx4_free_cmd_mailbox(dev, mailbox); + return err; +} diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c index 82ab427290c3..f2c8e8ba23fe 100644 --- a/drivers/net/ethernet/mellanox/mlx4/main.c +++ b/drivers/net/ethernet/mellanox/mlx4/main.c @@ -1831,6 +1831,11 @@ static int mlx4_setup_hca(struct mlx4_dev *dev) mlx4_err(dev, "Failed to initialize multicast group table, aborting\n"); goto err_mr_table_free; } + err = mlx4_config_mad_demux(dev); + if (err) { + mlx4_err(dev, "Failed in config_mad_demux, aborting\n"); + goto err_mcg_table_free; + } } err = mlx4_init_eq_table(dev); diff --git a/drivers/net/ethernet/mellanox/mlx4/mlx4.h b/drivers/net/ethernet/mellanox/mlx4/mlx4.h index 1d8af7336807..3398ff7e67e9 100644 --- a/drivers/net/ethernet/mellanox/mlx4/mlx4.h +++ b/drivers/net/ethernet/mellanox/mlx4/mlx4.h @@ -279,6 +279,8 @@ struct mlx4_icm_table { #define MLX4_MPT_FLAG_PHYSICAL (1 << 9) #define MLX4_MPT_FLAG_REGION (1 << 8) +#define MLX4_MPT_PD_MASK (0x1FFFFUL) +#define MLX4_MPT_PD_VF_MASK (0xFE0000UL) #define MLX4_MPT_PD_FLAG_FAST_REG (1 << 27) #define MLX4_MPT_PD_FLAG_RAE (1 << 28) #define MLX4_MPT_PD_FLAG_EN_INV (3 << 24) @@ -1311,5 +1313,6 @@ void mlx4_init_quotas(struct mlx4_dev *dev); int mlx4_get_slave_num_gids(struct mlx4_dev *dev, int slave, int port); /* Returns the VF index of slave */ int mlx4_get_vf_indx(struct mlx4_dev *dev, int slave); +int mlx4_config_mad_demux(struct mlx4_dev *dev); #endif /* MLX4_H */ diff --git a/drivers/net/ethernet/mellanox/mlx4/mr.c b/drivers/net/ethernet/mellanox/mlx4/mr.c index 2839abb878a6..7d717eccb7b0 100644 --- a/drivers/net/ethernet/mellanox/mlx4/mr.c +++ b/drivers/net/ethernet/mellanox/mlx4/mr.c @@ -298,6 +298,131 @@ static int mlx4_HW2SW_MPT(struct mlx4_dev *dev, struct mlx4_cmd_mailbox *mailbox MLX4_CMD_TIME_CLASS_B, MLX4_CMD_WRAPPED); } +int mlx4_mr_hw_get_mpt(struct mlx4_dev *dev, struct mlx4_mr *mmr, + struct mlx4_mpt_entry ***mpt_entry) +{ + int err; + int key = key_to_hw_index(mmr->key) & (dev->caps.num_mpts - 1); + struct mlx4_cmd_mailbox *mailbox = NULL; + + /* Make sure that at this point we have single-threaded access only */ + + if (mmr->enabled != MLX4_MPT_EN_HW) + return -EINVAL; + + err = mlx4_HW2SW_MPT(dev, NULL, key); + + if (err) { + mlx4_warn(dev, "HW2SW_MPT failed (%d).", err); + mlx4_warn(dev, "Most likely the MR has MWs bound to it.\n"); + return err; + } + + mmr->enabled = MLX4_MPT_EN_SW; + + if (!mlx4_is_mfunc(dev)) { + **mpt_entry = mlx4_table_find( + &mlx4_priv(dev)->mr_table.dmpt_table, + key, NULL); + } else { + mailbox = mlx4_alloc_cmd_mailbox(dev); + if (IS_ERR_OR_NULL(mailbox)) + return PTR_ERR(mailbox); + + err = mlx4_cmd_box(dev, 0, mailbox->dma, key, + 0, MLX4_CMD_QUERY_MPT, + MLX4_CMD_TIME_CLASS_B, + MLX4_CMD_WRAPPED); + + if (err) + goto free_mailbox; + + *mpt_entry = (struct mlx4_mpt_entry **)&mailbox->buf; + } + + if (!(*mpt_entry) || !(**mpt_entry)) { + err = -ENOMEM; + goto free_mailbox; + } + + return 0; + +free_mailbox: + mlx4_free_cmd_mailbox(dev, mailbox); + return err; +} +EXPORT_SYMBOL_GPL(mlx4_mr_hw_get_mpt); + +int mlx4_mr_hw_write_mpt(struct mlx4_dev *dev, struct mlx4_mr *mmr, + struct mlx4_mpt_entry **mpt_entry) +{ + int err; + + if (!mlx4_is_mfunc(dev)) { + /* Make sure any changes to this entry are flushed */ + wmb(); + + *(u8 *)(*mpt_entry) = MLX4_MPT_STATUS_HW; + + /* Make sure the new status is written */ + wmb(); + + err = mlx4_SYNC_TPT(dev); + } else { + int key = key_to_hw_index(mmr->key) & (dev->caps.num_mpts - 1); + + struct mlx4_cmd_mailbox *mailbox = + container_of((void *)mpt_entry, struct mlx4_cmd_mailbox, + buf); + + err = mlx4_SW2HW_MPT(dev, mailbox, key); + } + + mmr->pd = be32_to_cpu((*mpt_entry)->pd_flags) & MLX4_MPT_PD_MASK; + if (!err) + mmr->enabled = MLX4_MPT_EN_HW; + return err; +} +EXPORT_SYMBOL_GPL(mlx4_mr_hw_write_mpt); + +void mlx4_mr_hw_put_mpt(struct mlx4_dev *dev, + struct mlx4_mpt_entry **mpt_entry) +{ + if (mlx4_is_mfunc(dev)) { + struct mlx4_cmd_mailbox *mailbox = + container_of((void *)mpt_entry, struct mlx4_cmd_mailbox, + buf); + mlx4_free_cmd_mailbox(dev, mailbox); + } +} +EXPORT_SYMBOL_GPL(mlx4_mr_hw_put_mpt); + +int mlx4_mr_hw_change_pd(struct mlx4_dev *dev, struct mlx4_mpt_entry *mpt_entry, + u32 pdn) +{ + u32 pd_flags = be32_to_cpu(mpt_entry->pd_flags); + /* The wrapper function will put the slave's id here */ + if (mlx4_is_mfunc(dev)) + pd_flags &= ~MLX4_MPT_PD_VF_MASK; + mpt_entry->pd_flags = cpu_to_be32((pd_flags & ~MLX4_MPT_PD_MASK) | + (pdn & MLX4_MPT_PD_MASK) + | MLX4_MPT_PD_FLAG_EN_INV); + return 0; +} +EXPORT_SYMBOL_GPL(mlx4_mr_hw_change_pd); + +int mlx4_mr_hw_change_access(struct mlx4_dev *dev, + struct mlx4_mpt_entry *mpt_entry, + u32 access) +{ + u32 flags = (be32_to_cpu(mpt_entry->flags) & ~MLX4_PERM_MASK) | + (access & MLX4_PERM_MASK); + + mpt_entry->flags = cpu_to_be32(flags); + return 0; +} +EXPORT_SYMBOL_GPL(mlx4_mr_hw_change_access); + static int mlx4_mr_alloc_reserved(struct mlx4_dev *dev, u32 mridx, u32 pd, u64 iova, u64 size, u32 access, int npages, int page_shift, struct mlx4_mr *mr) @@ -463,6 +588,41 @@ int mlx4_mr_free(struct mlx4_dev *dev, struct mlx4_mr *mr) } EXPORT_SYMBOL_GPL(mlx4_mr_free); +void mlx4_mr_rereg_mem_cleanup(struct mlx4_dev *dev, struct mlx4_mr *mr) +{ + mlx4_mtt_cleanup(dev, &mr->mtt); +} +EXPORT_SYMBOL_GPL(mlx4_mr_rereg_mem_cleanup); + +int mlx4_mr_rereg_mem_write(struct mlx4_dev *dev, struct mlx4_mr *mr, + u64 iova, u64 size, int npages, + int page_shift, struct mlx4_mpt_entry *mpt_entry) +{ + int err; + + mpt_entry->start = cpu_to_be64(mr->iova); + mpt_entry->length = cpu_to_be64(mr->size); + mpt_entry->entity_size = cpu_to_be32(mr->mtt.page_shift); + + err = mlx4_mtt_init(dev, npages, page_shift, &mr->mtt); + if (err) + return err; + + if (mr->mtt.order < 0) { + mpt_entry->flags |= cpu_to_be32(MLX4_MPT_FLAG_PHYSICAL); + mpt_entry->mtt_addr = 0; + } else { + mpt_entry->mtt_addr = cpu_to_be64(mlx4_mtt_addr(dev, + &mr->mtt)); + if (mr->mtt.page_shift == 0) + mpt_entry->mtt_sz = cpu_to_be32(1 << mr->mtt.order); + } + mr->enabled = MLX4_MPT_EN_SW; + + return 0; +} +EXPORT_SYMBOL_GPL(mlx4_mr_rereg_mem_write); + int mlx4_mr_enable(struct mlx4_dev *dev, struct mlx4_mr *mr) { struct mlx4_cmd_mailbox *mailbox; diff --git a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c index 0efc1368e5a8..1089367fed22 100644 --- a/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c +++ b/drivers/net/ethernet/mellanox/mlx4/resource_tracker.c @@ -2613,12 +2613,34 @@ int mlx4_QUERY_MPT_wrapper(struct mlx4_dev *dev, int slave, if (err) return err; - if (mpt->com.from_state != RES_MPT_HW) { + if (mpt->com.from_state == RES_MPT_MAPPED) { + /* In order to allow rereg in SRIOV, we need to alter the MPT entry. To do + * that, the VF must read the MPT. But since the MPT entry memory is not + * in the VF's virtual memory space, it must use QUERY_MPT to obtain the + * entry contents. To guarantee that the MPT cannot be changed, the driver + * must perform HW2SW_MPT before this query and return the MPT entry to HW + * ownership fofollowing the change. The change here allows the VF to + * perform QUERY_MPT also when the entry is in SW ownership. + */ + struct mlx4_mpt_entry *mpt_entry = mlx4_table_find( + &mlx4_priv(dev)->mr_table.dmpt_table, + mpt->key, NULL); + + if (NULL == mpt_entry || NULL == outbox->buf) { + err = -EINVAL; + goto out; + } + + memcpy(outbox->buf, mpt_entry, sizeof(*mpt_entry)); + + err = 0; + } else if (mpt->com.from_state == RES_MPT_HW) { + err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd); + } else { err = -EBUSY; goto out; } - err = mlx4_DMA_wrapper(dev, slave, vhcr, inbox, outbox, cmd); out: put_res(dev, slave, id, RES_MPT); diff --git a/drivers/scsi/scsi_transport_srp.c b/drivers/scsi/scsi_transport_srp.c index 13e898332e45..a0c5bfdc5366 100644 --- a/drivers/scsi/scsi_transport_srp.c +++ b/drivers/scsi/scsi_transport_srp.c @@ -473,7 +473,8 @@ static void __srp_start_tl_fail_timers(struct srp_rport *rport) if (delay > 0) queue_delayed_work(system_long_wq, &rport->reconnect_work, 1UL * delay * HZ); - if (srp_rport_set_state(rport, SRP_RPORT_BLOCKED) == 0) { + if ((fast_io_fail_tmo >= 0 || dev_loss_tmo >= 0) && + srp_rport_set_state(rport, SRP_RPORT_BLOCKED) == 0) { pr_debug("%s new state: %d\n", dev_name(&shost->shost_gendev), rport->state); scsi_target_block(&shost->shost_gendev); diff --git a/include/linux/mlx4/cmd.h b/include/linux/mlx4/cmd.h index c8450366c130..379c02648ab3 100644 --- a/include/linux/mlx4/cmd.h +++ b/include/linux/mlx4/cmd.h @@ -116,6 +116,7 @@ enum { /* special QP and management commands */ MLX4_CMD_CONF_SPECIAL_QP = 0x23, MLX4_CMD_MAD_IFC = 0x24, + MLX4_CMD_MAD_DEMUX = 0x203, /* multicast commands */ MLX4_CMD_READ_MCG = 0x25, @@ -186,6 +187,12 @@ enum { }; enum { + MLX4_CMD_MAD_DEMUX_CONFIG = 0, + MLX4_CMD_MAD_DEMUX_QUERY_STATE = 1, + MLX4_CMD_MAD_DEMUX_QUERY_RESTR = 2, /* Query mad demux restrictions */ +}; + +enum { MLX4_CMD_WRAPPED, MLX4_CMD_NATIVE }; diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h index 35b51e7af886..be4ab553a317 100644 --- a/include/linux/mlx4/device.h +++ b/include/linux/mlx4/device.h @@ -172,6 +172,7 @@ enum { MLX4_DEV_CAP_FLAG2_UPDATE_QP = 1LL << 8, MLX4_DEV_CAP_FLAG2_DMFS_IPOIB = 1LL << 9, MLX4_DEV_CAP_FLAG2_VXLAN_OFFLOADS = 1LL << 10, + MLX4_DEV_CAP_FLAG2_MAD_DEMUX = 1LL << 11, }; enum { @@ -262,6 +263,7 @@ enum { MLX4_PERM_REMOTE_WRITE = 1 << 13, MLX4_PERM_ATOMIC = 1 << 14, MLX4_PERM_BIND_MW = 1 << 15, + MLX4_PERM_MASK = 0xFC00 }; enum { @@ -1243,4 +1245,19 @@ int mlx4_vf_smi_enabled(struct mlx4_dev *dev, int slave, int port); int mlx4_vf_get_enable_smi_admin(struct mlx4_dev *dev, int slave, int port); int mlx4_vf_set_enable_smi_admin(struct mlx4_dev *dev, int slave, int port, int enable); +int mlx4_mr_hw_get_mpt(struct mlx4_dev *dev, struct mlx4_mr *mmr, + struct mlx4_mpt_entry ***mpt_entry); +int mlx4_mr_hw_write_mpt(struct mlx4_dev *dev, struct mlx4_mr *mmr, + struct mlx4_mpt_entry **mpt_entry); +int mlx4_mr_hw_change_pd(struct mlx4_dev *dev, struct mlx4_mpt_entry *mpt_entry, + u32 pdn); +int mlx4_mr_hw_change_access(struct mlx4_dev *dev, + struct mlx4_mpt_entry *mpt_entry, + u32 access); +void mlx4_mr_hw_put_mpt(struct mlx4_dev *dev, + struct mlx4_mpt_entry **mpt_entry); +void mlx4_mr_rereg_mem_cleanup(struct mlx4_dev *dev, struct mlx4_mr *mr); +int mlx4_mr_rereg_mem_write(struct mlx4_dev *dev, struct mlx4_mr *mr, + u64 iova, u64 size, int npages, + int page_shift, struct mlx4_mpt_entry *mpt_entry); #endif /* MLX4_DEVICE_H */ diff --git a/include/rdma/ib_mad.h b/include/rdma/ib_mad.h index 3d81b90cc315..9bb99e983f58 100644 --- a/include/rdma/ib_mad.h +++ b/include/rdma/ib_mad.h @@ -40,6 +40,7 @@ #include <linux/list.h> #include <rdma/ib_verbs.h> +#include <uapi/rdma/ib_user_mad.h> /* Management base version */ #define IB_MGMT_BASE_VERSION 1 @@ -355,9 +356,13 @@ typedef void (*ib_mad_recv_handler)(struct ib_mad_agent *mad_agent, * @hi_tid: Access layer assigned transaction ID for this client. * Unsolicited MADs sent by this client will have the upper 32-bits * of their TID set to this value. + * @flags: registration flags * @port_num: Port number on which QP is registered * @rmpp_version: If set, indicates the RMPP version used by this agent. */ +enum { + IB_MAD_USER_RMPP = IB_USER_MAD_USER_RMPP, +}; struct ib_mad_agent { struct ib_device *device; struct ib_qp *qp; @@ -367,6 +372,7 @@ struct ib_mad_agent { ib_mad_snoop_handler snoop_handler; void *context; u32 hi_tid; + u32 flags; u8 port_num; u8 rmpp_version; }; @@ -426,6 +432,7 @@ struct ib_mad_recv_wc { * in the range from 0x30 to 0x4f. Otherwise not used. * @method_mask: The caller will receive unsolicited MADs for any method * where @method_mask = 1. + * */ struct ib_mad_reg_req { u8 mgmt_class; @@ -451,6 +458,7 @@ struct ib_mad_reg_req { * @recv_handler: The completion callback routine invoked for a received * MAD. * @context: User specified context associated with the registration. + * @registration_flags: Registration flags to set for this agent */ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, u8 port_num, @@ -459,7 +467,8 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, u8 rmpp_version, ib_mad_send_handler send_handler, ib_mad_recv_handler recv_handler, - void *context); + void *context, + u32 registration_flags); enum ib_mad_snoop_flags { /*IB_MAD_SNOOP_POSTED_SENDS = 1,*/ @@ -661,4 +670,11 @@ void *ib_get_rmpp_segment(struct ib_mad_send_buf *send_buf, int seg_num); */ void ib_free_send_mad(struct ib_mad_send_buf *send_buf); +/** + * ib_mad_kernel_rmpp_agent - Returns if the agent is performing RMPP. + * @agent: the agent in question + * @return: true if agent is performing rmpp, false otherwise. + */ +int ib_mad_kernel_rmpp_agent(struct ib_mad_agent *agent); + #endif /* IB_MAD_H */ diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h index 7ccef342f724..ed44cc07a7b3 100644 --- a/include/rdma/ib_verbs.h +++ b/include/rdma/ib_verbs.h @@ -1097,7 +1097,8 @@ struct ib_mr_attr { enum ib_mr_rereg_flags { IB_MR_REREG_TRANS = 1, IB_MR_REREG_PD = (1<<1), - IB_MR_REREG_ACCESS = (1<<2) + IB_MR_REREG_ACCESS = (1<<2), + IB_MR_REREG_SUPPORTED = ((IB_MR_REREG_ACCESS << 1) - 1) }; /** @@ -1547,6 +1548,13 @@ struct ib_device { u64 virt_addr, int mr_access_flags, struct ib_udata *udata); + int (*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 (*query_mr)(struct ib_mr *mr, struct ib_mr_attr *mr_attr); int (*dereg_mr)(struct ib_mr *mr); diff --git a/include/uapi/rdma/ib_user_mad.h b/include/uapi/rdma/ib_user_mad.h index d6fce1cbdb90..09f809f323ea 100644 --- a/include/uapi/rdma/ib_user_mad.h +++ b/include/uapi/rdma/ib_user_mad.h @@ -191,6 +191,45 @@ struct ib_user_mad_reg_req { __u8 rmpp_version; }; +/** + * ib_user_mad_reg_req2 - MAD registration request + * + * @id - Set by the _kernel_; used by userspace to identify the + * registered agent in future requests. + * @qpn - Queue pair number; must be 0 or 1. + * @mgmt_class - Indicates which management class of MADs should be + * receive by the caller. This field is only required if + * the user wishes to receive unsolicited MADs, otherwise + * it should be 0. + * @mgmt_class_version - Indicates which version of MADs for the given + * management class to receive. + * @res - Ignored. + * @flags - additional registration flags; Must be in the set of + * flags defined in IB_USER_MAD_REG_FLAGS_CAP + * @method_mask - The caller wishes to receive unsolicited MADs for the + * methods whose bit(s) is(are) set. + * @oui - Indicates IEEE OUI to use when mgmt_class is a vendor + * class in the range from 0x30 to 0x4f. Otherwise not + * used. + * @rmpp_version - If set, indicates the RMPP version to use. + */ +enum { + IB_USER_MAD_USER_RMPP = (1 << 0), +}; +#define IB_USER_MAD_REG_FLAGS_CAP (IB_USER_MAD_USER_RMPP) +struct ib_user_mad_reg_req2 { + __u32 id; + __u32 qpn; + __u8 mgmt_class; + __u8 mgmt_class_version; + __u16 res; + __u32 flags; + __u64 method_mask[2]; + __u32 oui; + __u8 rmpp_version; + __u8 reserved[3]; +}; + #define IB_IOCTL_MAGIC 0x1b #define IB_USER_MAD_REGISTER_AGENT _IOWR(IB_IOCTL_MAGIC, 1, \ @@ -200,4 +239,7 @@ struct ib_user_mad_reg_req { #define IB_USER_MAD_ENABLE_PKEY _IO(IB_IOCTL_MAGIC, 3) +#define IB_USER_MAD_REGISTER_AGENT2 _IOWR(IB_IOCTL_MAGIC, 4, \ + struct ib_user_mad_reg_req2) + #endif /* IB_USER_MAD_H */ diff --git a/include/uapi/rdma/ib_user_verbs.h b/include/uapi/rdma/ib_user_verbs.h index cbfdd4ca9510..26daf55ff76e 100644 --- a/include/uapi/rdma/ib_user_verbs.h +++ b/include/uapi/rdma/ib_user_verbs.h @@ -276,6 +276,22 @@ struct ib_uverbs_reg_mr_resp { __u32 rkey; }; +struct ib_uverbs_rereg_mr { + __u64 response; + __u32 mr_handle; + __u32 flags; + __u64 start; + __u64 length; + __u64 hca_va; + __u32 pd_handle; + __u32 access_flags; +}; + +struct ib_uverbs_rereg_mr_resp { + __u32 lkey; + __u32 rkey; +}; + struct ib_uverbs_dereg_mr { __u32 mr_handle; }; diff --git a/include/uapi/rdma/rdma_user_cm.h b/include/uapi/rdma/rdma_user_cm.h index 99b80abf360a..3066718eb120 100644 --- a/include/uapi/rdma/rdma_user_cm.h +++ b/include/uapi/rdma/rdma_user_cm.h @@ -34,6 +34,7 @@ #define RDMA_USER_CM_H #include <linux/types.h> +#include <linux/socket.h> #include <linux/in6.h> #include <rdma/ib_user_verbs.h> #include <rdma/ib_user_sa.h> |