diff options
Diffstat (limited to 'drivers/infiniband/core/mad.c')
| -rw-r--r-- | drivers/infiniband/core/mad.c | 1546 |
1 files changed, 965 insertions, 581 deletions
diff --git a/drivers/infiniband/core/mad.c b/drivers/infiniband/core/mad.c index 23628c622a5..ab31f136d04 100644 --- a/drivers/infiniband/core/mad.c +++ b/drivers/infiniband/core/mad.c @@ -1,5 +1,8 @@ /* - * Copyright (c) 2004, 2005 Voltaire, Inc. All rights reserved. + * Copyright (c) 2004-2007 Voltaire, Inc. All rights reserved. + * Copyright (c) 2005 Intel Corporation. All rights reserved. + * Copyright (c) 2005 Mellanox Technologies Ltd. All rights reserved. + * Copyright (c) 2009 HNR Consulting. All rights reserved. * * This software is available to you under a choice of one of two * licenses. You may choose to be licensed under the terms of the GNU @@ -29,12 +32,14 @@ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. * - * $Id: mad.c 1389 2004-12-27 22:56:47Z roland $ */ - #include <linux/dma-mapping.h> +#include <linux/slab.h> +#include <linux/module.h> +#include <rdma/ib_cache.h> #include "mad_priv.h" +#include "mad_rmpp.h" #include "smi.h" #include "agent.h" @@ -43,14 +48,21 @@ MODULE_DESCRIPTION("kernel IB MAD API"); MODULE_AUTHOR("Hal Rosenstock"); MODULE_AUTHOR("Sean Hefty"); +static int mad_sendq_size = IB_MAD_QP_SEND_SIZE; +static int mad_recvq_size = IB_MAD_QP_RECV_SIZE; + +module_param_named(send_queue_size, mad_sendq_size, int, 0444); +MODULE_PARM_DESC(send_queue_size, "Size of send queue in number of work requests"); +module_param_named(recv_queue_size, mad_recvq_size, int, 0444); +MODULE_PARM_DESC(recv_queue_size, "Size of receive queue in number of work requests"); + +static struct kmem_cache *ib_mad_cache; -kmem_cache_t *ib_mad_cache; static struct list_head ib_mad_port_list; static u32 ib_mad_client_id = 0; /* Port list lock */ -static spinlock_t ib_mad_port_list_lock; - +static DEFINE_SPINLOCK(ib_mad_port_list_lock); /* Forward declarations */ static int method_in_use(struct ib_mad_mgmt_method_table **method, @@ -58,16 +70,12 @@ static int method_in_use(struct ib_mad_mgmt_method_table **method, static void remove_mad_reg_req(struct ib_mad_agent_private *priv); static struct ib_mad_agent_private *find_mad_agent( struct ib_mad_port_private *port_priv, - struct ib_mad *mad, int solicited); + struct ib_mad *mad); static int ib_mad_post_receive_mads(struct ib_mad_qp_info *qp_info, struct ib_mad_private *mad); static void cancel_mads(struct ib_mad_agent_private *mad_agent_priv); -static void ib_mad_complete_send_wr(struct ib_mad_send_wr_private *mad_send_wr, - struct ib_mad_send_wc *mad_send_wc); -static void timeout_sends(void *data); -static void cancel_sends(void *data); -static void local_completions(void *data); -static int solicited_mad(struct ib_mad *mad); +static void timeout_sends(struct work_struct *work); +static void local_completions(struct work_struct *work); static int add_nonoui_reg_req(struct ib_mad_reg_req *mad_reg_req, struct ib_mad_agent_private *agent_priv, u8 mgmt_class); @@ -168,6 +176,15 @@ static int is_vendor_method_in_use( return 0; } +int ib_response_mad(struct ib_mad *mad) +{ + return ((mad->mad_hdr.method & IB_MGMT_METHOD_RESP) || + (mad->mad_hdr.method == IB_MGMT_METHOD_TRAP_REPRESS) || + ((mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_BM) && + (mad->mad_hdr.attr_mod & IB_BM_ATTR_MOD_RESP))); +} +EXPORT_SYMBOL(ib_response_mad); + /* * ib_register_mad_agent - Register to send/receive MADs */ @@ -197,8 +214,8 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, if (qpn == -1) goto error1; - if (rmpp_version) - goto error1; /* XXX: until RMPP implemented */ + if (rmpp_version && rmpp_version != IB_MGMT_RMPP_VERSION) + goto error1; /* Validate MAD registration request if supplied */ if (mad_reg_req) { @@ -228,6 +245,11 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, if (!is_vendor_oui(mad_reg_req->oui)) goto error1; } + /* Make sure class supplied is consistent with RMPP */ + if (!ib_is_mad_class_rmpp(mad_reg_req->mgmt_class)) { + if (rmpp_version) + goto error1; + } /* Make sure class supplied is consistent with QP type */ if (qp_type == IB_QPT_SMI) { if ((mad_reg_req->mgmt_class != @@ -255,34 +277,55 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, goto error1; } + /* Verify the QP requested is supported. For example, Ethernet devices + * will not have QP0 */ + if (!port_priv->qp_info[qpn].qp) { + ret = ERR_PTR(-EPROTONOSUPPORT); + goto error1; + } + /* Allocate structures */ - mad_agent_priv = kmalloc(sizeof *mad_agent_priv, GFP_KERNEL); + mad_agent_priv = kzalloc(sizeof *mad_agent_priv, GFP_KERNEL); if (!mad_agent_priv) { ret = ERR_PTR(-ENOMEM); goto error1; } + mad_agent_priv->agent.mr = ib_get_dma_mr(port_priv->qp_info[qpn].qp->pd, + IB_ACCESS_LOCAL_WRITE); + if (IS_ERR(mad_agent_priv->agent.mr)) { + ret = ERR_PTR(-ENOMEM); + goto error2; + } + if (mad_reg_req) { - reg_req = kmalloc(sizeof *reg_req, GFP_KERNEL); + reg_req = kmemdup(mad_reg_req, sizeof *reg_req, GFP_KERNEL); if (!reg_req) { ret = ERR_PTR(-ENOMEM); - goto error2; + goto error3; } - /* Make a copy of the MAD registration request */ - memcpy(reg_req, mad_reg_req, sizeof *reg_req); } /* Now, fill in the various structures */ - memset(mad_agent_priv, 0, sizeof *mad_agent_priv); mad_agent_priv->qp_info = &port_priv->qp_info[qpn]; mad_agent_priv->reg_req = reg_req; - mad_agent_priv->rmpp_version = rmpp_version; + mad_agent_priv->agent.rmpp_version = rmpp_version; mad_agent_priv->agent.device = device; mad_agent_priv->agent.recv_handler = recv_handler; mad_agent_priv->agent.send_handler = send_handler; 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; + spin_lock_init(&mad_agent_priv->lock); + INIT_LIST_HEAD(&mad_agent_priv->send_list); + INIT_LIST_HEAD(&mad_agent_priv->wait_list); + INIT_LIST_HEAD(&mad_agent_priv->done_list); + INIT_LIST_HEAD(&mad_agent_priv->rmpp_list); + INIT_DELAYED_WORK(&mad_agent_priv->timed_work, timeout_sends); + INIT_LIST_HEAD(&mad_agent_priv->local_list); + INIT_WORK(&mad_agent_priv->local_work, local_completions); + atomic_set(&mad_agent_priv->refcount, 1); + init_completion(&mad_agent_priv->comp); spin_lock_irqsave(&port_priv->reg_lock, flags); mad_agent_priv->agent.hi_tid = ++ib_mad_client_id; @@ -301,7 +344,7 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, if (method) { if (method_in_use(&method, mad_reg_req)) - goto error3; + goto error4; } } ret2 = add_nonoui_reg_req(mad_reg_req, mad_agent_priv, @@ -317,14 +360,14 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, if (is_vendor_method_in_use( vendor_class, mad_reg_req)) - goto error3; + goto error4; } } ret2 = add_oui_reg_req(mad_reg_req, mad_agent_priv); } if (ret2) { ret = ERR_PTR(ret2); - goto error3; + goto error4; } } @@ -332,23 +375,13 @@ struct ib_mad_agent *ib_register_mad_agent(struct ib_device *device, list_add_tail(&mad_agent_priv->agent_list, &port_priv->agent_list); spin_unlock_irqrestore(&port_priv->reg_lock, flags); - spin_lock_init(&mad_agent_priv->lock); - INIT_LIST_HEAD(&mad_agent_priv->send_list); - INIT_LIST_HEAD(&mad_agent_priv->wait_list); - INIT_WORK(&mad_agent_priv->timed_work, timeout_sends, mad_agent_priv); - INIT_LIST_HEAD(&mad_agent_priv->local_list); - INIT_WORK(&mad_agent_priv->local_work, local_completions, - mad_agent_priv); - INIT_LIST_HEAD(&mad_agent_priv->canceled_list); - INIT_WORK(&mad_agent_priv->canceled_work, cancel_sends, mad_agent_priv); - atomic_set(&mad_agent_priv->refcount, 1); - init_waitqueue_head(&mad_agent_priv->wait); - return &mad_agent_priv->agent; -error3: +error4: spin_unlock_irqrestore(&port_priv->reg_lock, flags); kfree(reg_req); +error3: + ib_dereg_mr(mad_agent_priv->agent.mr); error2: kfree(mad_agent_priv); error1: @@ -387,19 +420,15 @@ static int register_snoop_agent(struct ib_mad_qp_info *qp_info, if (i == qp_info->snoop_table_size) { /* Grow table. */ - new_snoop_table = kmalloc(sizeof mad_snoop_priv * - qp_info->snoop_table_size + 1, - GFP_ATOMIC); + new_snoop_table = krealloc(qp_info->snoop_table, + sizeof mad_snoop_priv * + (qp_info->snoop_table_size + 1), + GFP_ATOMIC); if (!new_snoop_table) { i = -ENOMEM; goto out; } - if (qp_info->snoop_table) { - memcpy(new_snoop_table, qp_info->snoop_table, - sizeof mad_snoop_priv * - qp_info->snoop_table_size); - kfree(qp_info->snoop_table); - } + qp_info->snoop_table = new_snoop_table; qp_info->snoop_table_size++; } @@ -440,14 +469,13 @@ struct ib_mad_agent *ib_register_mad_snoop(struct ib_device *device, goto error1; } /* Allocate structures */ - mad_snoop_priv = kmalloc(sizeof *mad_snoop_priv, GFP_KERNEL); + mad_snoop_priv = kzalloc(sizeof *mad_snoop_priv, GFP_KERNEL); if (!mad_snoop_priv) { ret = ERR_PTR(-ENOMEM); goto error1; } /* Now, fill in the various structures */ - memset(mad_snoop_priv, 0, sizeof *mad_snoop_priv); mad_snoop_priv->qp_info = &port_priv->qp_info[qpn]; mad_snoop_priv->agent.device = device; mad_snoop_priv->agent.recv_handler = recv_handler; @@ -456,7 +484,7 @@ struct ib_mad_agent *ib_register_mad_snoop(struct ib_device *device, mad_snoop_priv->agent.qp = port_priv->qp_info[qpn].qp; mad_snoop_priv->agent.port_num = port_num; mad_snoop_priv->mad_snoop_flags = mad_snoop_flags; - init_waitqueue_head(&mad_snoop_priv->wait); + init_completion(&mad_snoop_priv->comp); mad_snoop_priv->snoop_index = register_snoop_agent( &port_priv->qp_info[qpn], mad_snoop_priv); @@ -475,6 +503,18 @@ error1: } EXPORT_SYMBOL(ib_register_mad_snoop); +static inline void deref_mad_agent(struct ib_mad_agent_private *mad_agent_priv) +{ + if (atomic_dec_and_test(&mad_agent_priv->refcount)) + complete(&mad_agent_priv->comp); +} + +static inline void deref_snoop_agent(struct ib_mad_snoop_private *mad_snoop_priv) +{ + if (atomic_dec_and_test(&mad_snoop_priv->refcount)) + complete(&mad_snoop_priv->comp); +} + static void unregister_mad_agent(struct ib_mad_agent_private *mad_agent_priv) { struct ib_mad_port_private *port_priv; @@ -487,25 +527,22 @@ static void unregister_mad_agent(struct ib_mad_agent_private *mad_agent_priv) * MADs, preventing us from queuing additional work */ cancel_mads(mad_agent_priv); - port_priv = mad_agent_priv->qp_info->port_priv; - cancel_delayed_work(&mad_agent_priv->timed_work); - flush_workqueue(port_priv->wq); spin_lock_irqsave(&port_priv->reg_lock, flags); remove_mad_reg_req(mad_agent_priv); list_del(&mad_agent_priv->agent_list); spin_unlock_irqrestore(&port_priv->reg_lock, flags); - /* XXX: Cleanup pending RMPP receives for this agent */ + flush_workqueue(port_priv->wq); + ib_cancel_rmpp_recvs(mad_agent_priv); - atomic_dec(&mad_agent_priv->refcount); - wait_event(mad_agent_priv->wait, - !atomic_read(&mad_agent_priv->refcount)); + deref_mad_agent(mad_agent_priv); + wait_for_completion(&mad_agent_priv->comp); - if (mad_agent_priv->reg_req) - kfree(mad_agent_priv->reg_req); + kfree(mad_agent_priv->reg_req); + ib_dereg_mr(mad_agent_priv->agent.mr); kfree(mad_agent_priv); } @@ -520,9 +557,8 @@ static void unregister_mad_snoop(struct ib_mad_snoop_private *mad_snoop_priv) atomic_dec(&qp_info->snoop_count); spin_unlock_irqrestore(&qp_info->snoop_lock, flags); - atomic_dec(&mad_snoop_priv->refcount); - wait_event(mad_snoop_priv->wait, - !atomic_read(&mad_snoop_priv->refcount)); + deref_snoop_agent(mad_snoop_priv); + wait_for_completion(&mad_snoop_priv->comp); kfree(mad_snoop_priv); } @@ -565,7 +601,7 @@ static void dequeue_mad(struct ib_mad_list_head *mad_list) } static void snoop_send(struct ib_mad_qp_info *qp_info, - struct ib_send_wr *send_wr, + struct ib_mad_send_buf *send_buf, struct ib_mad_send_wc *mad_send_wc, int mad_snoop_flags) { @@ -583,9 +619,8 @@ static void snoop_send(struct ib_mad_qp_info *qp_info, atomic_inc(&mad_snoop_priv->refcount); spin_unlock_irqrestore(&qp_info->snoop_lock, flags); mad_snoop_priv->agent.snoop_handler(&mad_snoop_priv->agent, - send_wr, mad_send_wc); - if (atomic_dec_and_test(&mad_snoop_priv->refcount)) - wake_up(&mad_snoop_priv->wait); + send_buf, mad_send_wc); + deref_snoop_agent(mad_snoop_priv); spin_lock_irqsave(&qp_info->snoop_lock, flags); } spin_unlock_irqrestore(&qp_info->snoop_lock, flags); @@ -610,14 +645,14 @@ static void snoop_recv(struct ib_mad_qp_info *qp_info, spin_unlock_irqrestore(&qp_info->snoop_lock, flags); mad_snoop_priv->agent.recv_handler(&mad_snoop_priv->agent, mad_recv_wc); - if (atomic_dec_and_test(&mad_snoop_priv->refcount)) - wake_up(&mad_snoop_priv->wait); + deref_snoop_agent(mad_snoop_priv); spin_lock_irqsave(&qp_info->snoop_lock, flags); } spin_unlock_irqrestore(&qp_info->snoop_lock, flags); } -static void build_smp_wc(u64 wr_id, u16 slid, u16 pkey_index, u8 port_num, +static void build_smp_wc(struct ib_qp *qp, + u64 wr_id, u16 slid, u16 pkey_index, u8 port_num, struct ib_wc *wc) { memset(wc, 0, sizeof *wc); @@ -627,7 +662,7 @@ static void build_smp_wc(u64 wr_id, u16 slid, u16 pkey_index, u8 port_num, wc->pkey_index = pkey_index; wc->byte_len = sizeof(struct ib_mad) + sizeof(struct ib_grh); wc->src_qp = IB_QP0; - wc->qp_num = IB_QP0; + wc->qp = qp; wc->slid = slid; wc->sl = 0; wc->dlid_path_bits = 0; @@ -640,27 +675,44 @@ static void build_smp_wc(u64 wr_id, u16 slid, u16 pkey_index, u8 port_num, * Return < 0 if error */ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, - struct ib_smp *smp, - struct ib_send_wr *send_wr) + struct ib_mad_send_wr_private *mad_send_wr) { - int ret, solicited; + int ret = 0; + struct ib_smp *smp = mad_send_wr->send_buf.mad; unsigned long flags; struct ib_mad_local_private *local; struct ib_mad_private *mad_priv; struct ib_mad_port_private *port_priv; struct ib_mad_agent_private *recv_mad_agent = NULL; struct ib_device *device = mad_agent_priv->agent.device; - u8 port_num = mad_agent_priv->agent.port_num; + u8 port_num; struct ib_wc mad_wc; + struct ib_send_wr *send_wr = &mad_send_wr->send_wr; + + if (device->node_type == RDMA_NODE_IB_SWITCH && + smp->mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) + port_num = send_wr->wr.ud.port_num; + else + port_num = mad_agent_priv->agent.port_num; - if (!smi_handle_dr_smp_send(smp, device->node_type, port_num)) { + /* + * Directed route handling starts if the initial LID routed part of + * a request or the ending LID routed part of a response is empty. + * If we are at the start of the LID routed part, don't update the + * hop_ptr or hop_cnt. See section 14.2.2, Vol 1 IB spec. + */ + if ((ib_get_smp_direction(smp) ? smp->dr_dlid : smp->dr_slid) == + IB_LID_PERMISSIVE && + smi_handle_dr_smp_send(smp, device->node_type, port_num) == + IB_SMI_DISCARD) { ret = -EINVAL; printk(KERN_ERR PFX "Invalid directed route\n"); goto out; } + /* Check to post send on QP or process locally */ - ret = smi_check_local_dr_smp(smp, device, port_num); - if (!ret || !device->process_mad) + if (smi_check_local_smp(smp, device) == IB_SMI_DISCARD && + smi_check_local_returning_smp(smp, device) == IB_SMI_DISCARD) goto out; local = kmalloc(sizeof *local, GFP_ATOMIC); @@ -679,7 +731,9 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, goto out; } - build_smp_wc(send_wr->wr_id, smp->dr_slid, send_wr->wr.ud.pkey_index, + build_smp_wc(mad_agent_priv->agent.qp, + send_wr->wr_id, be16_to_cpu(smp->dr_slid), + send_wr->wr.ud.pkey_index, send_wr->wr.ud.port_num, &mad_wc); /* No GRH for DR SMP */ @@ -689,11 +743,7 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, switch (ret) { case IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY: - /* - * See if response is solicited and - * there is a recv handler - */ - if (solicited_mad(&mad_priv->mad.mad) && + if (ib_response_mad(&mad_priv->mad.mad) && mad_agent_priv->agent.recv_handler) { local->mad_priv = mad_priv; local->recv_mad_agent = mad_agent_priv; @@ -710,21 +760,20 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, break; case IB_MAD_RESULT_SUCCESS: /* Treat like an incoming receive MAD */ - solicited = solicited_mad(&mad_priv->mad.mad); port_priv = ib_get_mad_port(mad_agent_priv->agent.device, mad_agent_priv->agent.port_num); if (port_priv) { - mad_priv->mad.mad.mad_hdr.tid = - ((struct ib_mad *)smp)->mad_hdr.tid; + memcpy(&mad_priv->mad.mad, smp, sizeof(struct ib_mad)); recv_mad_agent = find_mad_agent(port_priv, - &mad_priv->mad.mad, - solicited); + &mad_priv->mad.mad); } if (!port_priv || !recv_mad_agent) { + /* + * No receiving agent so drop packet and + * generate send completion. + */ kmem_cache_free(ib_mad_cache, mad_priv); - kfree(local); - ret = 0; - goto out; + break; } local->mad_priv = mad_priv; local->recv_mad_agent = recv_mad_agent; @@ -736,13 +785,7 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, goto out; } - local->send_wr = *send_wr; - local->send_wr.sg_list = local->sg_list; - memcpy(local->sg_list, send_wr->sg_list, - sizeof *send_wr->sg_list * send_wr->num_sge); - local->send_wr.next = NULL; - local->tid = send_wr->wr.ud.mad_hdr->tid; - local->wr_id = send_wr->wr_id; + local->mad_send_wr = mad_send_wr; /* Reference MAD agent until send side of local completion handled */ atomic_inc(&mad_agent_priv->refcount); /* Queue local completion to local list */ @@ -750,42 +793,274 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv, list_add_tail(&local->completion_list, &mad_agent_priv->local_list); spin_unlock_irqrestore(&mad_agent_priv->lock, flags); queue_work(mad_agent_priv->qp_info->port_priv->wq, - &mad_agent_priv->local_work); + &mad_agent_priv->local_work); ret = 1; out: return ret; } -static int ib_send_mad(struct ib_mad_agent_private *mad_agent_priv, - struct ib_mad_send_wr_private *mad_send_wr) +static int get_pad_size(int hdr_len, int data_len) +{ + int seg_size, pad; + + seg_size = sizeof(struct ib_mad) - hdr_len; + if (data_len && seg_size) { + pad = seg_size - data_len % seg_size; + return pad == seg_size ? 0 : pad; + } else + return seg_size; +} + +static void free_send_rmpp_list(struct ib_mad_send_wr_private *mad_send_wr) +{ + struct ib_rmpp_segment *s, *t; + + list_for_each_entry_safe(s, t, &mad_send_wr->rmpp_list, list) { + list_del(&s->list); + kfree(s); + } +} + +static int alloc_send_rmpp_list(struct ib_mad_send_wr_private *send_wr, + gfp_t gfp_mask) +{ + struct ib_mad_send_buf *send_buf = &send_wr->send_buf; + struct ib_rmpp_mad *rmpp_mad = send_buf->mad; + struct ib_rmpp_segment *seg = NULL; + int left, seg_size, pad; + + send_buf->seg_size = sizeof (struct ib_mad) - send_buf->hdr_len; + seg_size = send_buf->seg_size; + pad = send_wr->pad; + + /* Allocate data segments. */ + 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); + free_send_rmpp_list(send_wr); + return -ENOMEM; + } + seg->num = ++send_buf->seg_count; + list_add_tail(&seg->list, &send_wr->rmpp_list); + } + + /* Zero any padding */ + if (pad) + memset(seg->data + seg_size - pad, 0, pad); + + rmpp_mad->rmpp_hdr.rmpp_version = send_wr->mad_agent_priv-> + agent.rmpp_version; + rmpp_mad->rmpp_hdr.rmpp_type = IB_MGMT_RMPP_TYPE_DATA; + ib_set_rmpp_flags(&rmpp_mad->rmpp_hdr, IB_MGMT_RMPP_FLAG_ACTIVE); + + send_wr->cur_seg = container_of(send_wr->rmpp_list.next, + struct ib_rmpp_segment, list); + send_wr->last_ack_seg = send_wr->cur_seg; + return 0; +} + +struct ib_mad_send_buf * ib_create_send_mad(struct ib_mad_agent *mad_agent, + u32 remote_qpn, u16 pkey_index, + int rmpp_active, + int hdr_len, int data_len, + gfp_t gfp_mask) +{ + struct ib_mad_agent_private *mad_agent_priv; + struct ib_mad_send_wr_private *mad_send_wr; + int pad, message_size, ret, size; + void *buf; + + mad_agent_priv = container_of(mad_agent, struct ib_mad_agent_private, + 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); + + size = rmpp_active ? hdr_len : sizeof(struct ib_mad); + buf = kzalloc(sizeof *mad_send_wr + size, gfp_mask); + if (!buf) + return ERR_PTR(-ENOMEM); + + mad_send_wr = buf + size; + INIT_LIST_HEAD(&mad_send_wr->rmpp_list); + mad_send_wr->send_buf.mad = buf; + mad_send_wr->send_buf.hdr_len = hdr_len; + mad_send_wr->send_buf.data_len = data_len; + mad_send_wr->pad = pad; + + mad_send_wr->mad_agent_priv = mad_agent_priv; + mad_send_wr->sg_list[0].length = hdr_len; + mad_send_wr->sg_list[0].lkey = mad_agent->mr->lkey; + mad_send_wr->sg_list[1].length = sizeof(struct ib_mad) - hdr_len; + mad_send_wr->sg_list[1].lkey = mad_agent->mr->lkey; + + mad_send_wr->send_wr.wr_id = (unsigned long) mad_send_wr; + mad_send_wr->send_wr.sg_list = mad_send_wr->sg_list; + mad_send_wr->send_wr.num_sge = 2; + mad_send_wr->send_wr.opcode = IB_WR_SEND; + mad_send_wr->send_wr.send_flags = IB_SEND_SIGNALED; + mad_send_wr->send_wr.wr.ud.remote_qpn = remote_qpn; + mad_send_wr->send_wr.wr.ud.remote_qkey = IB_QP_SET_QKEY; + mad_send_wr->send_wr.wr.ud.pkey_index = pkey_index; + + if (rmpp_active) { + ret = alloc_send_rmpp_list(mad_send_wr, gfp_mask); + if (ret) { + kfree(buf); + return ERR_PTR(ret); + } + } + + mad_send_wr->send_buf.mad_agent = mad_agent; + atomic_inc(&mad_agent_priv->refcount); + return &mad_send_wr->send_buf; +} +EXPORT_SYMBOL(ib_create_send_mad); + +int ib_get_mad_data_offset(u8 mgmt_class) +{ + if (mgmt_class == IB_MGMT_CLASS_SUBN_ADM) + return IB_MGMT_SA_HDR; + else if ((mgmt_class == IB_MGMT_CLASS_DEVICE_MGMT) || + (mgmt_class == IB_MGMT_CLASS_DEVICE_ADM) || + (mgmt_class == IB_MGMT_CLASS_BIS)) + return IB_MGMT_DEVICE_HDR; + else if ((mgmt_class >= IB_MGMT_CLASS_VENDOR_RANGE2_START) && + (mgmt_class <= IB_MGMT_CLASS_VENDOR_RANGE2_END)) + return IB_MGMT_VENDOR_HDR; + else + return IB_MGMT_MAD_HDR; +} +EXPORT_SYMBOL(ib_get_mad_data_offset); + +int ib_is_mad_class_rmpp(u8 mgmt_class) +{ + if ((mgmt_class == IB_MGMT_CLASS_SUBN_ADM) || + (mgmt_class == IB_MGMT_CLASS_DEVICE_MGMT) || + (mgmt_class == IB_MGMT_CLASS_DEVICE_ADM) || + (mgmt_class == IB_MGMT_CLASS_BIS) || + ((mgmt_class >= IB_MGMT_CLASS_VENDOR_RANGE2_START) && + (mgmt_class <= IB_MGMT_CLASS_VENDOR_RANGE2_END))) + return 1; + return 0; +} +EXPORT_SYMBOL(ib_is_mad_class_rmpp); + +void *ib_get_rmpp_segment(struct ib_mad_send_buf *send_buf, int seg_num) +{ + struct ib_mad_send_wr_private *mad_send_wr; + struct list_head *list; + + mad_send_wr = container_of(send_buf, struct ib_mad_send_wr_private, + send_buf); + list = &mad_send_wr->cur_seg->list; + + if (mad_send_wr->cur_seg->num < seg_num) { + list_for_each_entry(mad_send_wr->cur_seg, list, list) + if (mad_send_wr->cur_seg->num == seg_num) + break; + } else if (mad_send_wr->cur_seg->num > seg_num) { + list_for_each_entry_reverse(mad_send_wr->cur_seg, list, list) + if (mad_send_wr->cur_seg->num == seg_num) + break; + } + return mad_send_wr->cur_seg->data; +} +EXPORT_SYMBOL(ib_get_rmpp_segment); + +static inline void *ib_get_payload(struct ib_mad_send_wr_private *mad_send_wr) +{ + if (mad_send_wr->send_buf.seg_count) + return ib_get_rmpp_segment(&mad_send_wr->send_buf, + mad_send_wr->seg_num); + else + return mad_send_wr->send_buf.mad + + mad_send_wr->send_buf.hdr_len; +} + +void ib_free_send_mad(struct ib_mad_send_buf *send_buf) +{ + struct ib_mad_agent_private *mad_agent_priv; + struct ib_mad_send_wr_private *mad_send_wr; + + mad_agent_priv = container_of(send_buf->mad_agent, + struct ib_mad_agent_private, agent); + mad_send_wr = container_of(send_buf, struct ib_mad_send_wr_private, + send_buf); + + free_send_rmpp_list(mad_send_wr); + kfree(send_buf->mad); + deref_mad_agent(mad_agent_priv); +} +EXPORT_SYMBOL(ib_free_send_mad); + +int ib_send_mad(struct ib_mad_send_wr_private *mad_send_wr) { struct ib_mad_qp_info *qp_info; + struct list_head *list; struct ib_send_wr *bad_send_wr; + struct ib_mad_agent *mad_agent; + struct ib_sge *sge; unsigned long flags; int ret; - /* Replace user's WR ID with our own to find WR upon completion */ - qp_info = mad_agent_priv->qp_info; - mad_send_wr->wr_id = mad_send_wr->send_wr.wr_id; + /* Set WR ID to find mad_send_wr upon completion */ + qp_info = mad_send_wr->mad_agent_priv->qp_info; mad_send_wr->send_wr.wr_id = (unsigned long)&mad_send_wr->mad_list; mad_send_wr->mad_list.mad_queue = &qp_info->send_queue; + mad_agent = mad_send_wr->send_buf.mad_agent; + sge = mad_send_wr->sg_list; + sge[0].addr = ib_dma_map_single(mad_agent->device, + mad_send_wr->send_buf.mad, + sge[0].length, + DMA_TO_DEVICE); + if (unlikely(ib_dma_mapping_error(mad_agent->device, sge[0].addr))) + return -ENOMEM; + + mad_send_wr->header_mapping = sge[0].addr; + + sge[1].addr = ib_dma_map_single(mad_agent->device, + ib_get_payload(mad_send_wr), + sge[1].length, + DMA_TO_DEVICE); + if (unlikely(ib_dma_mapping_error(mad_agent->device, sge[1].addr))) { + ib_dma_unmap_single(mad_agent->device, + mad_send_wr->header_mapping, + sge[0].length, DMA_TO_DEVICE); + return -ENOMEM; + } + mad_send_wr->payload_mapping = sge[1].addr; + spin_lock_irqsave(&qp_info->send_queue.lock, flags); - if (qp_info->send_queue.count++ < qp_info->send_queue.max_active) { - list_add_tail(&mad_send_wr->mad_list.list, - &qp_info->send_queue.list); - spin_unlock_irqrestore(&qp_info->send_queue.lock, flags); - ret = ib_post_send(mad_agent_priv->agent.qp, - &mad_send_wr->send_wr, &bad_send_wr); - if (ret) { - printk(KERN_ERR PFX "ib_post_send failed: %d\n", ret); - dequeue_mad(&mad_send_wr->mad_list); - } + if (qp_info->send_queue.count < qp_info->send_queue.max_active) { + ret = ib_post_send(mad_agent->qp, &mad_send_wr->send_wr, + &bad_send_wr); + list = &qp_info->send_queue.list; } else { - list_add_tail(&mad_send_wr->mad_list.list, - &qp_info->overflow_list); - spin_unlock_irqrestore(&qp_info->send_queue.lock, flags); ret = 0; + list = &qp_info->overflow_list; + } + + if (!ret) { + qp_info->send_queue.count++; + list_add_tail(&mad_send_wr->mad_list.list, list); + } + spin_unlock_irqrestore(&qp_info->send_queue.lock, flags); + if (ret) { + ib_dma_unmap_single(mad_agent->device, + mad_send_wr->header_mapping, + sge[0].length, DMA_TO_DEVICE); + ib_dma_unmap_single(mad_agent->device, + mad_send_wr->payload_mapping, + sge[1].length, DMA_TO_DEVICE); } return ret; } @@ -794,45 +1069,35 @@ static int ib_send_mad(struct ib_mad_agent_private *mad_agent_priv, * ib_post_send_mad - Posts MAD(s) to the send queue of the QP associated * with the registered client */ -int ib_post_send_mad(struct ib_mad_agent *mad_agent, - struct ib_send_wr *send_wr, - struct ib_send_wr **bad_send_wr) +int ib_post_send_mad(struct ib_mad_send_buf *send_buf, + struct ib_mad_send_buf **bad_send_buf) { - int ret = -EINVAL; struct ib_mad_agent_private *mad_agent_priv; - - /* Validate supplied parameters */ - if (!bad_send_wr) - goto error1; - - if (!mad_agent || !send_wr) - goto error2; - - if (!mad_agent->send_handler) - goto error2; - - mad_agent_priv = container_of(mad_agent, - struct ib_mad_agent_private, - agent); + struct ib_mad_send_buf *next_send_buf; + struct ib_mad_send_wr_private *mad_send_wr; + unsigned long flags; + int ret = -EINVAL; /* Walk list of send WRs and post each on send list */ - while (send_wr) { - unsigned long flags; - struct ib_send_wr *next_send_wr; - struct ib_mad_send_wr_private *mad_send_wr; - struct ib_smp *smp; - - /* Validate more parameters */ - if (send_wr->num_sge > IB_MAD_SEND_REQ_MAX_SG) - goto error2; + for (; send_buf; send_buf = next_send_buf) { - if (send_wr->wr.ud.timeout_ms && !mad_agent->recv_handler) - goto error2; + mad_send_wr = container_of(send_buf, + struct ib_mad_send_wr_private, + send_buf); + mad_agent_priv = mad_send_wr->mad_agent_priv; + + if (!send_buf->mad_agent->send_handler || + (send_buf->timeout_ms && + !send_buf->mad_agent->recv_handler)) { + ret = -EINVAL; + goto error; + } - if (!send_wr->wr.ud.mad_hdr) { - printk(KERN_ERR PFX "MAD header must be supplied " - "in WR %p\n", send_wr); - goto error2; + if (!ib_is_mad_class_rmpp(((struct ib_mad_hdr *) send_buf->mad)->mgmt_class)) { + if (mad_agent_priv->agent.rmpp_version) { + ret = -EINVAL; + goto error; + } } /* @@ -840,39 +1105,26 @@ int ib_post_send_mad(struct ib_mad_agent *mad_agent, * current one completes, and the user modifies the work * request associated with the completion */ - next_send_wr = (struct ib_send_wr *)send_wr->next; + next_send_buf = send_buf->next; + mad_send_wr->send_wr.wr.ud.ah = send_buf->ah; - smp = (struct ib_smp *)send_wr->wr.ud.mad_hdr; - if (smp->mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) { - ret = handle_outgoing_dr_smp(mad_agent_priv, smp, - send_wr); + if (((struct ib_mad_hdr *) send_buf->mad)->mgmt_class == + IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) { + ret = handle_outgoing_dr_smp(mad_agent_priv, + mad_send_wr); if (ret < 0) /* error */ - goto error2; + goto error; else if (ret == 1) /* locally consumed */ - goto next; - } - - /* Allocate MAD send WR tracking structure */ - mad_send_wr = kmalloc(sizeof *mad_send_wr, GFP_ATOMIC); - if (!mad_send_wr) { - printk(KERN_ERR PFX "No memory for " - "ib_mad_send_wr_private\n"); - ret = -ENOMEM; - goto error2; + continue; } - mad_send_wr->send_wr = *send_wr; - mad_send_wr->send_wr.sg_list = mad_send_wr->sg_list; - memcpy(mad_send_wr->sg_list, send_wr->sg_list, - sizeof *send_wr->sg_list * send_wr->num_sge); - mad_send_wr->send_wr.next = NULL; - mad_send_wr->tid = send_wr->wr.ud.mad_hdr->tid; - mad_send_wr->agent = mad_agent; + mad_send_wr->tid = ((struct ib_mad_hdr *) send_buf->mad)->tid; /* Timeout will be updated after send completes */ - mad_send_wr->timeout = msecs_to_jiffies(send_wr->wr. - ud.timeout_ms); - mad_send_wr->retry = 0; - /* One reference for each work request to QP + response */ + mad_send_wr->timeout = msecs_to_jiffies(send_buf->timeout_ms); + mad_send_wr->max_retries = send_buf->retries; + mad_send_wr->retries_left = send_buf->retries; + send_buf->retries = 0; + /* Reference for work request to QP + response */ mad_send_wr->refcount = 1 + (mad_send_wr->timeout > 0); mad_send_wr->status = IB_WC_SUCCESS; @@ -883,23 +1135,25 @@ int ib_post_send_mad(struct ib_mad_agent *mad_agent, &mad_agent_priv->send_list); spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - ret = ib_send_mad(mad_agent_priv, mad_send_wr); - if (ret) { + if (mad_agent_priv->agent.rmpp_version) { + ret = ib_send_rmpp_mad(mad_send_wr); + if (ret >= 0 && ret != IB_RMPP_RESULT_CONSUMED) + ret = ib_send_mad(mad_send_wr); + } else + ret = ib_send_mad(mad_send_wr); + if (ret < 0) { /* Fail send request */ spin_lock_irqsave(&mad_agent_priv->lock, flags); list_del(&mad_send_wr->agent_list); spin_unlock_irqrestore(&mad_agent_priv->lock, flags); atomic_dec(&mad_agent_priv->refcount); - goto error2; + goto error; } -next: - send_wr = next_send_wr; } return 0; - -error2: - *bad_send_wr = send_wr; -error1: +error: + if (bad_send_buf) + *bad_send_buf = send_buf; return ret; } EXPORT_SYMBOL(ib_post_send_mad); @@ -910,41 +1164,28 @@ EXPORT_SYMBOL(ib_post_send_mad); */ void ib_free_recv_mad(struct ib_mad_recv_wc *mad_recv_wc) { - struct ib_mad_recv_buf *entry; + struct ib_mad_recv_buf *mad_recv_buf, *temp_recv_buf; struct ib_mad_private_header *mad_priv_hdr; struct ib_mad_private *priv; + struct list_head free_list; - mad_priv_hdr = container_of(mad_recv_wc, - struct ib_mad_private_header, - recv_wc); - priv = container_of(mad_priv_hdr, struct ib_mad_private, header); + INIT_LIST_HEAD(&free_list); + list_splice_init(&mad_recv_wc->rmpp_list, &free_list); - /* - * Walk receive buffer list associated with this WC - * No need to remove them from list of receive buffers - */ - list_for_each_entry(entry, &mad_recv_wc->recv_buf.list, list) { - /* Free previous receive buffer */ - kmem_cache_free(ib_mad_cache, priv); + list_for_each_entry_safe(mad_recv_buf, temp_recv_buf, + &free_list, list) { + mad_recv_wc = container_of(mad_recv_buf, struct ib_mad_recv_wc, + recv_buf); mad_priv_hdr = container_of(mad_recv_wc, struct ib_mad_private_header, recv_wc); priv = container_of(mad_priv_hdr, struct ib_mad_private, header); + kmem_cache_free(ib_mad_cache, priv); } - - /* Free last buffer */ - kmem_cache_free(ib_mad_cache, priv); } EXPORT_SYMBOL(ib_free_recv_mad); -void ib_coalesce_recv_mad(struct ib_mad_recv_wc *mad_recv_wc, - void *buf) -{ - printk(KERN_ERR PFX "ib_coalesce_recv_mad() not implemented yet\n"); -} -EXPORT_SYMBOL(ib_coalesce_recv_mad); - struct ib_mad_agent *ib_redirect_mad_qp(struct ib_qp *qp, u8 rmpp_version, ib_mad_send_handler send_handler, @@ -968,10 +1209,7 @@ static int method_in_use(struct ib_mad_mgmt_method_table **method, { int i; - for (i = find_first_bit(mad_reg_req->method_mask, IB_MGMT_MAX_METHODS); - i < IB_MGMT_MAX_METHODS; - i = find_next_bit(mad_reg_req->method_mask, IB_MGMT_MAX_METHODS, - 1+i)) { + 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); return -EINVAL; @@ -983,14 +1221,12 @@ static int method_in_use(struct ib_mad_mgmt_method_table **method, static int allocate_method_table(struct ib_mad_mgmt_method_table **method) { /* Allocate management method table */ - *method = kmalloc(sizeof **method, GFP_ATOMIC); + *method = kzalloc(sizeof **method, GFP_ATOMIC); if (!*method) { printk(KERN_ERR PFX "No memory for " "ib_mad_mgmt_method_table\n"); return -ENOMEM; } - /* Clear management method table */ - memset(*method, 0, sizeof **method); return 0; } @@ -1037,8 +1273,8 @@ static int find_vendor_oui(struct ib_mad_mgmt_vendor_class *vendor_class, int i; for (i = 0; i < MAX_MGMT_OUI; i++) - /* Is there matching OUI for this vendor class ? */ - if (!memcmp(vendor_class->oui[i], oui, 3)) + /* Is there matching OUI for this vendor class ? */ + if (!memcmp(vendor_class->oui[i], oui, 3)) return i; return -1; @@ -1081,15 +1317,14 @@ static int add_nonoui_reg_req(struct ib_mad_reg_req *mad_reg_req, class = &port_priv->version[mad_reg_req->mgmt_class_version].class; if (!*class) { /* Allocate management class table for "new" class version */ - *class = kmalloc(sizeof **class, GFP_ATOMIC); + *class = kzalloc(sizeof **class, GFP_ATOMIC); if (!*class) { printk(KERN_ERR PFX "No memory for " "ib_mad_mgmt_class_table\n"); ret = -ENOMEM; goto error1; } - /* Clear management class table */ - memset(*class, 0, sizeof(**class)); + /* Allocate method table for this management class */ method = &(*class)->method_table[mgmt_class]; if ((ret = allocate_method_table(method))) @@ -1108,13 +1343,9 @@ static int add_nonoui_reg_req(struct ib_mad_reg_req *mad_reg_req, goto error3; /* Finally, add in methods being registered */ - for (i = find_first_bit(mad_reg_req->method_mask, - IB_MGMT_MAX_METHODS); - i < IB_MGMT_MAX_METHODS; - i = find_next_bit(mad_reg_req->method_mask, IB_MGMT_MAX_METHODS, - 1+i)) { + for_each_set_bit(i, mad_reg_req->method_mask, IB_MGMT_MAX_METHODS) (*method)->agent[i] = agent_priv; - } + return 0; error3: @@ -1153,25 +1384,24 @@ static int add_oui_reg_req(struct ib_mad_reg_req *mad_reg_req, mad_reg_req->mgmt_class_version].vendor; if (!*vendor_table) { /* Allocate mgmt vendor class table for "new" class version */ - vendor = kmalloc(sizeof *vendor, GFP_ATOMIC); + vendor = kzalloc(sizeof *vendor, GFP_ATOMIC); if (!vendor) { printk(KERN_ERR PFX "No memory for " "ib_mad_mgmt_vendor_class_table\n"); goto error1; } - /* Clear management vendor class table */ - memset(vendor, 0, sizeof(*vendor)); + *vendor_table = vendor; } if (!(*vendor_table)->vendor_class[vclass]) { /* Allocate table for this management vendor class */ - vendor_class = kmalloc(sizeof *vendor_class, GFP_ATOMIC); + vendor_class = kzalloc(sizeof *vendor_class, GFP_ATOMIC); if (!vendor_class) { printk(KERN_ERR PFX "No memory for " "ib_mad_mgmt_vendor_class\n"); goto error2; } - memset(vendor_class, 0, sizeof(*vendor_class)); + (*vendor_table)->vendor_class[vclass] = vendor_class; } for (i = 0; i < MAX_MGMT_OUI; i++) { @@ -1208,13 +1438,9 @@ check_in_use: goto error4; /* Finally, add in methods being registered */ - for (i = find_first_bit(mad_reg_req->method_mask, - IB_MGMT_MAX_METHODS); - i < IB_MGMT_MAX_METHODS; - i = find_next_bit(mad_reg_req->method_mask, IB_MGMT_MAX_METHODS, - 1+i)) { + for_each_set_bit(i, mad_reg_req->method_mask, IB_MGMT_MAX_METHODS) (*method)->agent[i] = agent_priv; - } + return 0; error4: @@ -1338,42 +1564,15 @@ out: return; } -static int response_mad(struct ib_mad *mad) -{ - /* Trap represses are responses although response bit is reset */ - return ((mad->mad_hdr.method == IB_MGMT_METHOD_TRAP_REPRESS) || - (mad->mad_hdr.method & IB_MGMT_METHOD_RESP)); -} - -static int solicited_mad(struct ib_mad *mad) -{ - /* CM MADs are never solicited */ - if (mad->mad_hdr.mgmt_class == IB_MGMT_CLASS_CM) { - return 0; - } - - /* XXX: Determine whether MAD is using RMPP */ - - /* Not using RMPP */ - /* Is this MAD a response to a previous MAD ? */ - return response_mad(mad); -} - static struct ib_mad_agent_private * find_mad_agent(struct ib_mad_port_private *port_priv, - struct ib_mad *mad, - int solicited) + struct ib_mad *mad) { struct ib_mad_agent_private *mad_agent = NULL; unsigned long flags; spin_lock_irqsave(&port_priv->reg_lock, flags); - - /* - * Whether MAD was solicited determines type of routing to - * MAD client. - */ - if (solicited) { + if (ib_response_mad(mad)) { u32 hi_tid; struct ib_mad_agent_private *entry; @@ -1382,8 +1581,7 @@ find_mad_agent(struct ib_mad_port_private *port_priv, * of MAD. */ hi_tid = be64_to_cpu(mad->mad_hdr.tid) >> 32; - list_for_each_entry(entry, &port_priv->agent_list, - agent_list) { + list_for_each_entry(entry, &port_priv->agent_list, agent_list) { if (entry->agent.hi_tid == hi_tid) { mad_agent = entry; break; @@ -1408,6 +1606,9 @@ find_mad_agent(struct ib_mad_port_private *port_priv, mad->mad_hdr.class_version].class; if (!class) goto out; + if (convert_mgmt_class(mad->mad_hdr.mgmt_class) >= + IB_MGMT_MAX_METHODS) + goto out; method = class->method_table[convert_mgmt_class( mad->mad_hdr.mgmt_class)]; if (method) @@ -1477,111 +1678,209 @@ out: return valid; } -/* - * Return start of fully reassembled MAD, or NULL, if MAD isn't assembled yet - */ -static struct ib_mad_private * -reassemble_recv(struct ib_mad_agent_private *mad_agent_priv, - struct ib_mad_private *recv) +static int is_data_mad(struct ib_mad_agent_private *mad_agent_priv, + struct ib_mad_hdr *mad_hdr) { - /* Until we have RMPP, all receives are reassembled!... */ - INIT_LIST_HEAD(&recv->header.recv_wc.recv_buf.list); - return recv; + struct ib_rmpp_mad *rmpp_mad; + + rmpp_mad = (struct ib_rmpp_mad *)mad_hdr; + return !mad_agent_priv->agent.rmpp_version || + !(ib_get_rmpp_flags(&rmpp_mad->rmpp_hdr) & + IB_MGMT_RMPP_FLAG_ACTIVE) || + (rmpp_mad->rmpp_hdr.rmpp_type == IB_MGMT_RMPP_TYPE_DATA); } -static struct ib_mad_send_wr_private* -find_send_req(struct ib_mad_agent_private *mad_agent_priv, - u64 tid) +static inline int rcv_has_same_class(struct ib_mad_send_wr_private *wr, + struct ib_mad_recv_wc *rwc) { - struct ib_mad_send_wr_private *mad_send_wr; + return ((struct ib_mad *)(wr->send_buf.mad))->mad_hdr.mgmt_class == + rwc->recv_buf.mad->mad_hdr.mgmt_class; +} - list_for_each_entry(mad_send_wr, &mad_agent_priv->wait_list, - agent_list) { - if (mad_send_wr->tid == tid) - return mad_send_wr; +static inline int rcv_has_same_gid(struct ib_mad_agent_private *mad_agent_priv, + struct ib_mad_send_wr_private *wr, + struct ib_mad_recv_wc *rwc ) +{ + struct ib_ah_attr attr; + u8 send_resp, rcv_resp; + union ib_gid sgid; + struct ib_device *device = mad_agent_priv->agent.device; + u8 port_num = mad_agent_priv->agent.port_num; + u8 lmc; + + send_resp = ib_response_mad((struct ib_mad *)wr->send_buf.mad); + rcv_resp = ib_response_mad(rwc->recv_buf.mad); + + if (send_resp == rcv_resp) + /* both requests, or both responses. GIDs different */ + return 0; + + if (ib_query_ah(wr->send_buf.ah, &attr)) + /* Assume not equal, to avoid false positives. */ + return 0; + + if (!!(attr.ah_flags & IB_AH_GRH) != + !!(rwc->wc->wc_flags & IB_WC_GRH)) + /* one has GID, other does not. Assume different */ + return 0; + + if (!send_resp && rcv_resp) { + /* is request/response. */ + if (!(attr.ah_flags & IB_AH_GRH)) { + if (ib_get_cached_lmc(device, port_num, &lmc)) + return 0; + return (!lmc || !((attr.src_path_bits ^ + rwc->wc->dlid_path_bits) & + ((1 << lmc) - 1))); + } else { + if (ib_get_cached_gid(device, port_num, + attr.grh.sgid_index, &sgid)) + return 0; + return !memcmp(sgid.raw, rwc->recv_buf.grh->dgid.raw, + 16); + } + } + + if (!(attr.ah_flags & IB_AH_GRH)) + return attr.dlid == rwc->wc->slid; + else + return !memcmp(attr.grh.dgid.raw, rwc->recv_buf.grh->sgid.raw, + 16); +} + +static inline int is_direct(u8 class) +{ + return (class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE); +} + +struct ib_mad_send_wr_private* +ib_find_send_mad(struct ib_mad_agent_private *mad_agent_priv, + struct ib_mad_recv_wc *wc) +{ + struct ib_mad_send_wr_private *wr; + struct ib_mad *mad; + + mad = (struct ib_mad *)wc->recv_buf.mad; + + list_for_each_entry(wr, &mad_agent_priv->wait_list, agent_list) { + if ((wr->tid == mad->mad_hdr.tid) && + rcv_has_same_class(wr, wc) && + /* + * Don't check GID for direct routed MADs. + * These might have permissive LIDs. + */ + (is_direct(wc->recv_buf.mad->mad_hdr.mgmt_class) || + rcv_has_same_gid(mad_agent_priv, wr, wc))) + return (wr->status == IB_WC_SUCCESS) ? wr : NULL; } /* * It's possible to receive the response before we've * been notified that the send has completed */ - list_for_each_entry(mad_send_wr, &mad_agent_priv->send_list, - agent_list) { - if (mad_send_wr->tid == tid && mad_send_wr->timeout) { + list_for_each_entry(wr, &mad_agent_priv->send_list, agent_list) { + if (is_data_mad(mad_agent_priv, wr->send_buf.mad) && + wr->tid == mad->mad_hdr.tid && + wr->timeout && + rcv_has_same_class(wr, wc) && + /* + * Don't check GID for direct routed MADs. + * These might have permissive LIDs. + */ + (is_direct(wc->recv_buf.mad->mad_hdr.mgmt_class) || + rcv_has_same_gid(mad_agent_priv, wr, wc))) /* Verify request has not been canceled */ - return (mad_send_wr->status == IB_WC_SUCCESS) ? - mad_send_wr : NULL; - } + return (wr->status == IB_WC_SUCCESS) ? wr : NULL; } return NULL; } +void ib_mark_mad_done(struct ib_mad_send_wr_private *mad_send_wr) +{ + mad_send_wr->timeout = 0; + if (mad_send_wr->refcount == 1) + list_move_tail(&mad_send_wr->agent_list, + &mad_send_wr->mad_agent_priv->done_list); +} + static void ib_mad_complete_recv(struct ib_mad_agent_private *mad_agent_priv, - struct ib_mad_private *recv, - int solicited) + struct ib_mad_recv_wc *mad_recv_wc) { struct ib_mad_send_wr_private *mad_send_wr; struct ib_mad_send_wc mad_send_wc; unsigned long flags; - /* Fully reassemble receive before processing */ - recv = reassemble_recv(mad_agent_priv, recv); - if (!recv) { - if (atomic_dec_and_test(&mad_agent_priv->refcount)) - wake_up(&mad_agent_priv->wait); - return; + 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) { + mad_recv_wc = ib_process_rmpp_recv_wc(mad_agent_priv, + mad_recv_wc); + if (!mad_recv_wc) { + deref_mad_agent(mad_agent_priv); + return; + } } /* Complete corresponding request */ - if (solicited) { + if (ib_response_mad(mad_recv_wc->recv_buf.mad)) { spin_lock_irqsave(&mad_agent_priv->lock, flags); - mad_send_wr = find_send_req(mad_agent_priv, - recv->mad.mad.mad_hdr.tid); + 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(&recv->header.recv_wc); - if (atomic_dec_and_test(&mad_agent_priv->refcount)) - wake_up(&mad_agent_priv->wait); + ib_free_recv_mad(mad_recv_wc); + deref_mad_agent(mad_agent_priv); return; } - /* Timeout = 0 means that we won't wait for a response */ - mad_send_wr->timeout = 0; + ib_mark_mad_done(mad_send_wr); spin_unlock_irqrestore(&mad_agent_priv->lock, flags); /* Defined behavior is to complete response before request */ - recv->header.recv_wc.wc->wr_id = mad_send_wr->wr_id; - mad_agent_priv->agent.recv_handler( - &mad_agent_priv->agent, - &recv->header.recv_wc); + 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.wr_id = mad_send_wr->wr_id; + 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, - &recv->header.recv_wc); - if (atomic_dec_and_test(&mad_agent_priv->refcount)) - wake_up(&mad_agent_priv->wait); + mad_agent_priv->agent.recv_handler(&mad_agent_priv->agent, + mad_recv_wc); + deref_mad_agent(mad_agent_priv); } } +static bool generate_unmatched_resp(struct ib_mad_private *recv, + struct ib_mad_private *response) +{ + if (recv->mad.mad.mad_hdr.method == IB_MGMT_METHOD_GET || + recv->mad.mad.mad_hdr.method == IB_MGMT_METHOD_SET) { + memcpy(response, recv, sizeof *response); + response->header.recv_wc.wc = &response->header.wc; + response->header.recv_wc.recv_buf.mad = &response->mad.mad; + response->header.recv_wc.recv_buf.grh = &response->grh; + response->mad.mad.mad_hdr.method = IB_MGMT_METHOD_GET_RESP; + response->mad.mad.mad_hdr.status = + cpu_to_be16(IB_MGMT_MAD_STATUS_UNSUPPORTED_METHOD_ATTRIB); + if (recv->mad.mad.mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) + response->mad.mad.mad_hdr.status |= IB_SMP_DIRECTION; + + return true; + } else { + return false; + } +} static void ib_mad_recv_done_handler(struct ib_mad_port_private *port_priv, struct ib_wc *wc) { struct ib_mad_qp_info *qp_info; struct ib_mad_private_header *mad_priv_hdr; - struct ib_mad_private *recv, *response; + struct ib_mad_private *recv, *response = NULL; struct ib_mad_list_head *mad_list; struct ib_mad_agent_private *mad_agent; - int solicited; - - 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"); + int port_num; + int ret = IB_MAD_RESULT_SUCCESS; mad_list = (struct ib_mad_list_head *)(unsigned long)wc->wr_id; qp_info = mad_list->mad_queue->qp_info; @@ -1590,11 +1889,11 @@ static void ib_mad_recv_done_handler(struct ib_mad_port_private *port_priv, mad_priv_hdr = container_of(mad_list, struct ib_mad_private_header, mad_list); recv = container_of(mad_priv_hdr, struct ib_mad_private, header); - dma_unmap_single(port_priv->device->dma_device, - pci_unmap_addr(&recv->header, mapping), - sizeof(struct ib_mad_private) - - sizeof(struct ib_mad_private_header), - DMA_FROM_DEVICE); + ib_dma_unmap_single(port_priv->device, + recv->header.mapping, + sizeof(struct ib_mad_private) - + sizeof(struct ib_mad_private_header), + DMA_FROM_DEVICE); /* Setup MAD receive work completion from "normal" work completion */ recv->header.wc = *wc; @@ -1610,39 +1909,61 @@ static void ib_mad_recv_done_handler(struct ib_mad_port_private *port_priv, if (!validate_mad(&recv->mad.mad, qp_info->qp->qp_num)) goto out; + 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"); + goto out; + } + + if (port_priv->device->node_type == RDMA_NODE_IB_SWITCH) + port_num = wc->port_num; + else + port_num = port_priv->port_num; + if (recv->mad.mad.mad_hdr.mgmt_class == IB_MGMT_CLASS_SUBN_DIRECTED_ROUTE) { - if (!smi_handle_dr_smp_recv(&recv->mad.smp, - port_priv->device->node_type, - port_priv->port_num, - port_priv->device->phys_port_cnt)) + enum smi_forward_action retsmi; + + if (smi_handle_dr_smp_recv(&recv->mad.smp, + port_priv->device->node_type, + port_num, + port_priv->device->phys_port_cnt) == + IB_SMI_DISCARD) goto out; - if (!smi_check_forward_dr_smp(&recv->mad.smp)) + + retsmi = smi_check_forward_dr_smp(&recv->mad.smp); + if (retsmi == IB_SMI_LOCAL) goto local; - if (!smi_handle_dr_smp_send(&recv->mad.smp, - port_priv->device->node_type, - port_priv->port_num)) - goto out; - if (!smi_check_local_dr_smp(&recv->mad.smp, + + if (retsmi == IB_SMI_SEND) { /* don't forward */ + if (smi_handle_dr_smp_send(&recv->mad.smp, + port_priv->device->node_type, + port_num) == IB_SMI_DISCARD) + goto out; + + if (smi_check_local_smp(&recv->mad.smp, port_priv->device) == IB_SMI_DISCARD) + goto out; + } else if (port_priv->device->node_type == RDMA_NODE_IB_SWITCH) { + /* forward case for switches */ + memcpy(response, recv, sizeof(*response)); + response->header.recv_wc.wc = &response->header.wc; + response->header.recv_wc.recv_buf.mad = &response->mad.mad; + response->header.recv_wc.recv_buf.grh = &response->grh; + + agent_send_response(&response->mad.mad, + &response->grh, wc, port_priv->device, - port_priv->port_num)) + smi_get_fwd_port(&recv->mad.smp), + qp_info->qp->qp_num); + goto out; + } } local: /* Give driver "right of first refusal" on incoming MAD */ if (port_priv->device->process_mad) { - int ret; - - if (!response) { - printk(KERN_ERR PFX "No memory for response MAD\n"); - /* - * Is it better to assume that - * it wouldn't be processed ? - */ - goto out; - } - ret = port_priv->device->process_mad(port_priv->device, 0, port_priv->port_num, wc, &recv->grh, @@ -1652,26 +1973,28 @@ local: if (ret & IB_MAD_RESULT_CONSUMED) goto out; if (ret & IB_MAD_RESULT_REPLY) { - /* Send response */ - if (!agent_send(response, &recv->grh, wc, - port_priv->device, - port_priv->port_num)) - response = NULL; + agent_send_response(&response->mad.mad, + &recv->grh, wc, + port_priv->device, + port_num, + qp_info->qp->qp_num); goto out; } } } - /* Determine corresponding MAD agent for incoming receive MAD */ - solicited = solicited_mad(&recv->mad.mad); - mad_agent = find_mad_agent(port_priv, &recv->mad.mad, solicited); + mad_agent = find_mad_agent(port_priv, &recv->mad.mad); if (mad_agent) { - ib_mad_complete_recv(mad_agent, recv, solicited); + ib_mad_complete_recv(mad_agent, &recv->header.recv_wc); /* * recv is freed up in error cases in ib_mad_complete_recv * or via recv_handler in ib_mad_complete_recv() */ recv = NULL; + } else if ((ret & IB_MAD_RESULT_SUCCESS) && + generate_unmatched_resp(recv, response)) { + agent_send_response(&response->mad.mad, &recv->grh, wc, + port_priv->device, port_num, qp_info->qp->qp_num); } out: @@ -1699,60 +2022,74 @@ static void adjust_timeout(struct ib_mad_agent_private *mad_agent_priv) if (time_after(mad_agent_priv->timeout, mad_send_wr->timeout)) { mad_agent_priv->timeout = mad_send_wr->timeout; - cancel_delayed_work(&mad_agent_priv->timed_work); delay = mad_send_wr->timeout - jiffies; if ((long)delay <= 0) delay = 1; - queue_delayed_work(mad_agent_priv->qp_info-> - port_priv->wq, - &mad_agent_priv->timed_work, delay); + mod_delayed_work(mad_agent_priv->qp_info->port_priv->wq, + &mad_agent_priv->timed_work, delay); } } } -static void wait_for_response(struct ib_mad_agent_private *mad_agent_priv, - struct ib_mad_send_wr_private *mad_send_wr ) +static void wait_for_response(struct ib_mad_send_wr_private *mad_send_wr) { + struct ib_mad_agent_private *mad_agent_priv; struct ib_mad_send_wr_private *temp_mad_send_wr; struct list_head *list_item; unsigned long delay; + mad_agent_priv = mad_send_wr->mad_agent_priv; list_del(&mad_send_wr->agent_list); delay = mad_send_wr->timeout; mad_send_wr->timeout += jiffies; - list_for_each_prev(list_item, &mad_agent_priv->wait_list) { - temp_mad_send_wr = list_entry(list_item, - struct ib_mad_send_wr_private, - agent_list); - if (time_after(mad_send_wr->timeout, - temp_mad_send_wr->timeout)) - break; + if (delay) { + list_for_each_prev(list_item, &mad_agent_priv->wait_list) { + temp_mad_send_wr = list_entry(list_item, + struct ib_mad_send_wr_private, + agent_list); + if (time_after(mad_send_wr->timeout, + temp_mad_send_wr->timeout)) + break; + } } + else + list_item = &mad_agent_priv->wait_list; list_add(&mad_send_wr->agent_list, list_item); /* Reschedule a work item if we have a shorter timeout */ - if (mad_agent_priv->wait_list.next == &mad_send_wr->agent_list) { - cancel_delayed_work(&mad_agent_priv->timed_work); - queue_delayed_work(mad_agent_priv->qp_info->port_priv->wq, - &mad_agent_priv->timed_work, delay); - } + if (mad_agent_priv->wait_list.next == &mad_send_wr->agent_list) + mod_delayed_work(mad_agent_priv->qp_info->port_priv->wq, + &mad_agent_priv->timed_work, delay); +} + +void ib_reset_mad_timeout(struct ib_mad_send_wr_private *mad_send_wr, + int timeout_ms) +{ + mad_send_wr->timeout = msecs_to_jiffies(timeout_ms); + wait_for_response(mad_send_wr); } /* * Process a send work completion */ -static void ib_mad_complete_send_wr(struct ib_mad_send_wr_private *mad_send_wr, - struct ib_mad_send_wc *mad_send_wc) +void ib_mad_complete_send_wr(struct ib_mad_send_wr_private *mad_send_wr, + struct ib_mad_send_wc *mad_send_wc) { struct ib_mad_agent_private *mad_agent_priv; unsigned long flags; + int ret; - mad_agent_priv = container_of(mad_send_wr->agent, - struct ib_mad_agent_private, agent); - + mad_agent_priv = mad_send_wr->mad_agent_priv; spin_lock_irqsave(&mad_agent_priv->lock, flags); + if (mad_agent_priv->agent.rmpp_version) { + ret = ib_process_rmpp_send_wc(mad_send_wr, mad_send_wc); + if (ret == IB_RMPP_RESULT_CONSUMED) + goto done; + } else + ret = IB_RMPP_RESULT_UNHANDLED; + if (mad_send_wc->status != IB_WC_SUCCESS && mad_send_wr->status == IB_WC_SUCCESS) { mad_send_wr->status = mad_send_wc->status; @@ -1762,10 +2099,9 @@ static void ib_mad_complete_send_wr(struct ib_mad_send_wr_private *mad_send_wr, if (--mad_send_wr->refcount > 0) { if (mad_send_wr->refcount == 1 && mad_send_wr->timeout && mad_send_wr->status == IB_WC_SUCCESS) { - wait_for_response(mad_agent_priv, mad_send_wr); + wait_for_response(mad_send_wr); } - spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - return; + goto done; } /* Remove send from MAD agent and notify client of completion */ @@ -1775,14 +2111,17 @@ static void ib_mad_complete_send_wr(struct ib_mad_send_wr_private *mad_send_wr, if (mad_send_wr->status != IB_WC_SUCCESS ) mad_send_wc->status = mad_send_wr->status; - mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, - mad_send_wc); + if (ret == IB_RMPP_RESULT_INTERNAL) + ib_rmpp_send_handler(mad_send_wc); + else + mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, + mad_send_wc); /* Release reference on agent taken when sending */ - if (atomic_dec_and_test(&mad_agent_priv->refcount)) - wake_up(&mad_agent_priv->wait); - - kfree(mad_send_wr); + deref_mad_agent(mad_agent_priv); + return; +done: + spin_unlock_irqrestore(&mad_agent_priv->lock, flags); } static void ib_mad_send_done_handler(struct ib_mad_port_private *port_priv, @@ -1793,6 +2132,7 @@ static void ib_mad_send_done_handler(struct ib_mad_port_private *port_priv, struct ib_mad_qp_info *qp_info; struct ib_mad_queue *send_queue; struct ib_send_wr *bad_send_wr; + struct ib_mad_send_wc mad_send_wc; unsigned long flags; int ret; @@ -1803,6 +2143,12 @@ static void ib_mad_send_done_handler(struct ib_mad_port_private *port_priv, qp_info = send_queue->qp_info; retry: + ib_dma_unmap_single(mad_send_wr->send_buf.mad_agent->device, + mad_send_wr->header_mapping, + mad_send_wr->sg_list[0].length, DMA_TO_DEVICE); + ib_dma_unmap_single(mad_send_wr->send_buf.mad_agent->device, + mad_send_wr->payload_mapping, + mad_send_wr->sg_list[1].length, DMA_TO_DEVICE); queued_send_wr = NULL; spin_lock_irqsave(&send_queue->lock, flags); list_del(&mad_list->list); @@ -1814,22 +2160,21 @@ retry: queued_send_wr = container_of(mad_list, struct ib_mad_send_wr_private, mad_list); - list_del(&mad_list->list); - list_add_tail(&mad_list->list, &send_queue->list); + list_move_tail(&mad_list->list, &send_queue->list); } spin_unlock_irqrestore(&send_queue->lock, flags); - /* Restore client wr_id in WC and complete send */ - wc->wr_id = mad_send_wr->wr_id; + mad_send_wc.send_buf = &mad_send_wr->send_buf; + mad_send_wc.status = wc->status; + mad_send_wc.vendor_err = wc->vendor_err; if (atomic_read(&qp_info->snoop_count)) - snoop_send(qp_info, &mad_send_wr->send_wr, - (struct ib_mad_send_wc *)wc, + snoop_send(qp_info, &mad_send_wr->send_buf, &mad_send_wc, IB_MAD_SNOOP_SEND_COMPLETIONS); - ib_mad_complete_send_wr(mad_send_wr, (struct ib_mad_send_wc *)wc); + ib_mad_complete_send_wr(mad_send_wr, &mad_send_wc); if (queued_send_wr) { ret = ib_post_send(qp_info->qp, &queued_send_wr->send_wr, - &bad_send_wr); + &bad_send_wr); if (ret) { printk(KERN_ERR PFX "ib_post_send failed: %d\n", ret); mad_send_wr = queued_send_wr; @@ -1915,12 +2260,12 @@ static void mad_error_handler(struct ib_mad_port_private *port_priv, /* * IB MAD completion callback */ -static void ib_mad_completion_handler(void *data) +static void ib_mad_completion_handler(struct work_struct *work) { struct ib_mad_port_private *port_priv; struct ib_wc wc; - port_priv = (struct ib_mad_port_private *)data; + port_priv = container_of(work, struct ib_mad_port_private, work); ib_req_notify_cq(port_priv->cq, IB_CQ_NEXT_COMP); while (ib_poll_cq(port_priv->cq, 1, &wc) == 1) { @@ -1954,7 +2299,7 @@ static void cancel_mads(struct ib_mad_agent_private *mad_agent_priv) list_for_each_entry_safe(mad_send_wr, temp_mad_send_wr, &mad_agent_priv->send_list, agent_list) { if (mad_send_wr->status == IB_WC_SUCCESS) { - mad_send_wr->status = IB_WC_WR_FLUSH_ERR; + mad_send_wr->status = IB_WC_WR_FLUSH_ERR; mad_send_wr->refcount -= (mad_send_wr->timeout > 0); } } @@ -1969,128 +2314,102 @@ static void cancel_mads(struct ib_mad_agent_private *mad_agent_priv) list_for_each_entry_safe(mad_send_wr, temp_mad_send_wr, &cancel_list, agent_list) { - mad_send_wc.wr_id = mad_send_wr->wr_id; + mad_send_wc.send_buf = &mad_send_wr->send_buf; + list_del(&mad_send_wr->agent_list); mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, &mad_send_wc); - - list_del(&mad_send_wr->agent_list); - kfree(mad_send_wr); atomic_dec(&mad_agent_priv->refcount); } } static struct ib_mad_send_wr_private* -find_send_by_wr_id(struct ib_mad_agent_private *mad_agent_priv, - u64 wr_id) +find_send_wr(struct ib_mad_agent_private *mad_agent_priv, + struct ib_mad_send_buf *send_buf) { struct ib_mad_send_wr_private *mad_send_wr; list_for_each_entry(mad_send_wr, &mad_agent_priv->wait_list, agent_list) { - if (mad_send_wr->wr_id == wr_id) + if (&mad_send_wr->send_buf == send_buf) return mad_send_wr; } list_for_each_entry(mad_send_wr, &mad_agent_priv->send_list, agent_list) { - if (mad_send_wr->wr_id == wr_id) + if (is_data_mad(mad_agent_priv, mad_send_wr->send_buf.mad) && + &mad_send_wr->send_buf == send_buf) return mad_send_wr; } return NULL; } -void cancel_sends(void *data) -{ - struct ib_mad_agent_private *mad_agent_priv; - struct ib_mad_send_wr_private *mad_send_wr; - struct ib_mad_send_wc mad_send_wc; - unsigned long flags; - - mad_agent_priv = data; - - mad_send_wc.status = IB_WC_WR_FLUSH_ERR; - mad_send_wc.vendor_err = 0; - - spin_lock_irqsave(&mad_agent_priv->lock, flags); - while (!list_empty(&mad_agent_priv->canceled_list)) { - mad_send_wr = list_entry(mad_agent_priv->canceled_list.next, - struct ib_mad_send_wr_private, - agent_list); - - list_del(&mad_send_wr->agent_list); - spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - - mad_send_wc.wr_id = mad_send_wr->wr_id; - mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, - &mad_send_wc); - - kfree(mad_send_wr); - if (atomic_dec_and_test(&mad_agent_priv->refcount)) - wake_up(&mad_agent_priv->wait); - spin_lock_irqsave(&mad_agent_priv->lock, flags); - } - spin_unlock_irqrestore(&mad_agent_priv->lock, flags); -} - -void ib_cancel_mad(struct ib_mad_agent *mad_agent, - u64 wr_id) +int ib_modify_mad(struct ib_mad_agent *mad_agent, + struct ib_mad_send_buf *send_buf, u32 timeout_ms) { struct ib_mad_agent_private *mad_agent_priv; struct ib_mad_send_wr_private *mad_send_wr; unsigned long flags; + int active; mad_agent_priv = container_of(mad_agent, struct ib_mad_agent_private, agent); spin_lock_irqsave(&mad_agent_priv->lock, flags); - mad_send_wr = find_send_by_wr_id(mad_agent_priv, wr_id); - if (!mad_send_wr) { + mad_send_wr = find_send_wr(mad_agent_priv, send_buf); + if (!mad_send_wr || mad_send_wr->status != IB_WC_SUCCESS) { spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - goto out; + return -EINVAL; } - if (mad_send_wr->status == IB_WC_SUCCESS) - mad_send_wr->refcount -= (mad_send_wr->timeout > 0); - - if (mad_send_wr->refcount != 0) { + active = (!mad_send_wr->timeout || mad_send_wr->refcount > 1); + if (!timeout_ms) { mad_send_wr->status = IB_WC_WR_FLUSH_ERR; - spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - goto out; + mad_send_wr->refcount -= (mad_send_wr->timeout > 0); } - list_del(&mad_send_wr->agent_list); - list_add_tail(&mad_send_wr->agent_list, &mad_agent_priv->canceled_list); - adjust_timeout(mad_agent_priv); + mad_send_wr->send_buf.timeout_ms = timeout_ms; + if (active) + mad_send_wr->timeout = msecs_to_jiffies(timeout_ms); + else + ib_reset_mad_timeout(mad_send_wr, timeout_ms); + spin_unlock_irqrestore(&mad_agent_priv->lock, flags); + return 0; +} +EXPORT_SYMBOL(ib_modify_mad); - queue_work(mad_agent_priv->qp_info->port_priv->wq, - &mad_agent_priv->canceled_work); -out: - return; +void ib_cancel_mad(struct ib_mad_agent *mad_agent, + struct ib_mad_send_buf *send_buf) +{ + ib_modify_mad(mad_agent, send_buf, 0); } EXPORT_SYMBOL(ib_cancel_mad); -static void local_completions(void *data) +static void local_completions(struct work_struct *work) { struct ib_mad_agent_private *mad_agent_priv; struct ib_mad_local_private *local; struct ib_mad_agent_private *recv_mad_agent; unsigned long flags; + int free_mad; struct ib_wc wc; struct ib_mad_send_wc mad_send_wc; - mad_agent_priv = (struct ib_mad_agent_private *)data; + mad_agent_priv = + container_of(work, struct ib_mad_agent_private, local_work); spin_lock_irqsave(&mad_agent_priv->lock, flags); while (!list_empty(&mad_agent_priv->local_list)) { local = list_entry(mad_agent_priv->local_list.next, struct ib_mad_local_private, completion_list); + list_del(&local->completion_list); spin_unlock_irqrestore(&mad_agent_priv->lock, flags); + free_mad = 0; 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"); - kmem_cache_free(ib_mad_cache, local->mad_priv); + free_mad = 1; goto local_send_completion; } @@ -2098,14 +2417,17 @@ static void local_completions(void *data) * Defined behavior is to complete response * before request */ - build_smp_wc(local->wr_id, IB_LID_PERMISSIVE, - 0 /* pkey index */, - recv_mad_agent->agent.port_num, &wc); + build_smp_wc(recv_mad_agent->agent.qp, + (unsigned long) local->mad_send_wr, + be16_to_cpu(IB_LID_PERMISSIVE), + 0, recv_mad_agent->agent.port_num, &wc); local->mad_priv->header.recv_wc.wc = &wc; local->mad_priv->header.recv_wc.mad_len = sizeof(struct ib_mad); - INIT_LIST_HEAD(&local->mad_priv->header.recv_wc.recv_buf.list); + INIT_LIST_HEAD(&local->mad_priv->header.recv_wc.rmpp_list); + list_add(&local->mad_priv->header.recv_wc.recv_buf.list, + &local->mad_priv->header.recv_wc.rmpp_list); local->mad_priv->header.recv_wc.recv_buf.grh = NULL; local->mad_priv->header.recv_wc.recv_buf.mad = &local->mad_priv->mad.mad; @@ -2125,32 +2447,68 @@ local_send_completion: /* Complete send */ mad_send_wc.status = IB_WC_SUCCESS; mad_send_wc.vendor_err = 0; - mad_send_wc.wr_id = local->wr_id; + mad_send_wc.send_buf = &local->mad_send_wr->send_buf; if (atomic_read(&mad_agent_priv->qp_info->snoop_count)) - snoop_send(mad_agent_priv->qp_info, &local->send_wr, - &mad_send_wc, - IB_MAD_SNOOP_SEND_COMPLETIONS); + snoop_send(mad_agent_priv->qp_info, + &local->mad_send_wr->send_buf, + &mad_send_wc, IB_MAD_SNOOP_SEND_COMPLETIONS); mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, &mad_send_wc); spin_lock_irqsave(&mad_agent_priv->lock, flags); - list_del(&local->completion_list); atomic_dec(&mad_agent_priv->refcount); + if (free_mad) + kmem_cache_free(ib_mad_cache, local->mad_priv); kfree(local); } spin_unlock_irqrestore(&mad_agent_priv->lock, flags); } -static void timeout_sends(void *data) +static int retry_send(struct ib_mad_send_wr_private *mad_send_wr) +{ + int ret; + + if (!mad_send_wr->retries_left) + return -ETIMEDOUT; + + mad_send_wr->retries_left--; + mad_send_wr->send_buf.retries++; + + mad_send_wr->timeout = msecs_to_jiffies(mad_send_wr->send_buf.timeout_ms); + + if (mad_send_wr->mad_agent_priv->agent.rmpp_version) { + ret = ib_retry_rmpp(mad_send_wr); + switch (ret) { + case IB_RMPP_RESULT_UNHANDLED: + ret = ib_send_mad(mad_send_wr); + break; + case IB_RMPP_RESULT_CONSUMED: + ret = 0; + break; + default: + ret = -ECOMM; + break; + } + } else + ret = ib_send_mad(mad_send_wr); + + if (!ret) { + mad_send_wr->refcount++; + list_add_tail(&mad_send_wr->agent_list, + &mad_send_wr->mad_agent_priv->send_list); + } + return ret; +} + +static void timeout_sends(struct work_struct *work) { struct ib_mad_agent_private *mad_agent_priv; struct ib_mad_send_wr_private *mad_send_wr; struct ib_mad_send_wc mad_send_wc; unsigned long flags, delay; - mad_agent_priv = (struct ib_mad_agent_private *)data; - - mad_send_wc.status = IB_WC_RESP_TIMEOUT_ERR; + mad_agent_priv = container_of(work, struct ib_mad_agent_private, + timed_work.work); mad_send_wc.vendor_err = 0; spin_lock_irqsave(&mad_agent_priv->lock, flags); @@ -2170,24 +2528,35 @@ static void timeout_sends(void *data) } list_del(&mad_send_wr->agent_list); + if (mad_send_wr->status == IB_WC_SUCCESS && + !retry_send(mad_send_wr)) + continue; + spin_unlock_irqrestore(&mad_agent_priv->lock, flags); - mad_send_wc.wr_id = mad_send_wr->wr_id; + if (mad_send_wr->status == IB_WC_SUCCESS) + mad_send_wc.status = IB_WC_RESP_TIMEOUT_ERR; + else + mad_send_wc.status = mad_send_wr->status; + mad_send_wc.send_buf = &mad_send_wr->send_buf; mad_agent_priv->agent.send_handler(&mad_agent_priv->agent, &mad_send_wc); - kfree(mad_send_wr); atomic_dec(&mad_agent_priv->refcount); spin_lock_irqsave(&mad_agent_priv->lock, flags); } spin_unlock_irqrestore(&mad_agent_priv->lock, flags); } -static void ib_mad_thread_completion_handler(struct ib_cq *cq) +static void ib_mad_thread_completion_handler(struct ib_cq *cq, void *arg) { struct ib_mad_port_private *port_priv = cq->cq_context; + unsigned long flags; - queue_work(port_priv->wq, &port_priv->work); + spin_lock_irqsave(&ib_mad_port_list_lock, flags); + if (!list_empty(&port_priv->port_list)) + queue_work(port_priv->wq, &port_priv->work); + spin_unlock_irqrestore(&ib_mad_port_list_lock, flags); } /* @@ -2225,13 +2594,17 @@ static int ib_mad_post_receive_mads(struct ib_mad_qp_info *qp_info, break; } } - sg_list.addr = dma_map_single(qp_info->port_priv-> - device->dma_device, - &mad_priv->grh, - sizeof *mad_priv - - sizeof mad_priv->header, - DMA_FROM_DEVICE); - pci_unmap_addr_set(&mad_priv->header, mapping, sg_list.addr); + sg_list.addr = ib_dma_map_single(qp_info->port_priv->device, + &mad_priv->grh, + sizeof *mad_priv - + sizeof mad_priv->header, + DMA_FROM_DEVICE); + if (unlikely(ib_dma_mapping_error(qp_info->port_priv->device, + sg_list.addr))) { + ret = -ENOMEM; + break; + } + mad_priv->header.mapping = sg_list.addr; recv_wr.wr_id = (unsigned long)&mad_priv->header.mad_list; mad_priv->header.mad_list.mad_queue = recv_queue; @@ -2246,12 +2619,11 @@ static int ib_mad_post_receive_mads(struct ib_mad_qp_info *qp_info, list_del(&mad_priv->header.mad_list.list); recv_queue->count--; spin_unlock_irqrestore(&recv_queue->lock, flags); - dma_unmap_single(qp_info->port_priv->device->dma_device, - pci_unmap_addr(&mad_priv->header, - mapping), - sizeof *mad_priv - - sizeof mad_priv->header, - DMA_FROM_DEVICE); + ib_dma_unmap_single(qp_info->port_priv->device, + mad_priv->header.mapping, + sizeof *mad_priv - + 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); break; @@ -2270,6 +2642,9 @@ static void cleanup_recv_queue(struct ib_mad_qp_info *qp_info) struct ib_mad_private *recv; struct ib_mad_list_head *mad_list; + if (!qp_info->qp) + return; + while (!list_empty(&qp_info->recv_queue.list)) { mad_list = list_entry(qp_info->recv_queue.list.next, @@ -2283,11 +2658,11 @@ static void cleanup_recv_queue(struct ib_mad_qp_info *qp_info) /* Remove from posted receive MAD list */ list_del(&mad_list->list); - dma_unmap_single(qp_info->port_priv->device->dma_device, - pci_unmap_addr(&recv->header, mapping), - sizeof(struct ib_mad_private) - - sizeof(struct ib_mad_private_header), - DMA_FROM_DEVICE); + ib_dma_unmap_single(qp_info->port_priv->device, + recv->header.mapping, + sizeof(struct ib_mad_private) - + sizeof(struct ib_mad_private_header), + DMA_FROM_DEVICE); kmem_cache_free(ib_mad_cache, recv); } @@ -2302,21 +2677,30 @@ static int ib_mad_port_start(struct ib_mad_port_private *port_priv) int ret, i; struct ib_qp_attr *attr; struct ib_qp *qp; + u16 pkey_index; attr = kmalloc(sizeof *attr, GFP_KERNEL); - if (!attr) { + if (!attr) { printk(KERN_ERR PFX "Couldn't kmalloc ib_qp_attr\n"); return -ENOMEM; } + ret = ib_find_pkey(port_priv->device, port_priv->port_num, + IB_DEFAULT_PKEY_FULL, &pkey_index); + if (ret) + pkey_index = 0; + for (i = 0; i < IB_MAD_QPS_CORE; i++) { qp = port_priv->qp_info[i].qp; + if (!qp) + continue; + /* * PKey index for QP1 is irrelevant but * one is needed for the Reset to Init transition */ attr->qp_state = IB_QPS_INIT; - attr->pkey_index = 0; + attr->pkey_index = pkey_index; attr->qkey = (qp->qp_num == 0) ? 0 : IB_QP1_QKEY; ret = ib_modify_qp(qp, attr, IB_QP_STATE | IB_QP_PKEY_INDEX | IB_QP_QKEY); @@ -2352,6 +2736,9 @@ static int ib_mad_port_start(struct ib_mad_port_private *port_priv) } for (i = 0; i < IB_MAD_QPS_CORE; i++) { + if (!port_priv->qp_info[i].qp) + continue; + ret = ib_mad_post_receive_mads(&port_priv->qp_info[i], NULL); if (ret) { printk(KERN_ERR PFX "Couldn't post receive WRs\n"); @@ -2404,8 +2791,8 @@ static int create_mad_qp(struct ib_mad_qp_info *qp_info, qp_init_attr.send_cq = qp_info->port_priv->cq; qp_init_attr.recv_cq = qp_info->port_priv->cq; qp_init_attr.sq_sig_type = IB_SIGNAL_ALL_WR; - qp_init_attr.cap.max_send_wr = IB_MAD_QP_SEND_SIZE; - qp_init_attr.cap.max_recv_wr = IB_MAD_QP_RECV_SIZE; + qp_init_attr.cap.max_send_wr = mad_sendq_size; + qp_init_attr.cap.max_recv_wr = mad_recvq_size; qp_init_attr.cap.max_send_sge = IB_MAD_SEND_REQ_MAX_SG; qp_init_attr.cap.max_recv_sge = IB_MAD_RECV_REQ_MAX_SG; qp_init_attr.qp_type = qp_type; @@ -2420,8 +2807,8 @@ static int create_mad_qp(struct ib_mad_qp_info *qp_info, goto error; } /* Use minimum queue sizes unless the CQ is resized */ - qp_info->send_queue.max_active = IB_MAD_QP_SEND_SIZE; - qp_info->recv_queue.max_active = IB_MAD_QP_RECV_SIZE; + qp_info->send_queue.max_active = mad_sendq_size; + qp_info->recv_queue.max_active = mad_recvq_size; return 0; error: @@ -2430,9 +2817,11 @@ error: static void destroy_mad_qp(struct ib_mad_qp_info *qp_info) { + if (!qp_info->qp) + return; + ib_destroy_qp(qp_info->qp); - if (qp_info->snoop_table) - kfree(qp_info->snoop_table); + kfree(qp_info->snoop_table); } /* @@ -2446,22 +2835,15 @@ static int ib_mad_port_open(struct ib_device *device, struct ib_mad_port_private *port_priv; unsigned long flags; char name[sizeof "ib_mad123"]; - - /* First, check if port already open at MAD layer */ - port_priv = ib_get_mad_port(device, port_num); - if (port_priv) { - printk(KERN_DEBUG PFX "%s port %d already open\n", - device->name, port_num); - return 0; - } + int has_smi; /* Create new device info */ - port_priv = kmalloc(sizeof *port_priv, GFP_KERNEL); + port_priv = kzalloc(sizeof *port_priv, GFP_KERNEL); if (!port_priv) { printk(KERN_ERR PFX "No memory for ib_mad_port_private\n"); return -ENOMEM; } - memset(port_priv, 0, sizeof *port_priv); + port_priv->device = device; port_priv->port_num = port_num; spin_lock_init(&port_priv->reg_lock); @@ -2469,11 +2851,14 @@ static int ib_mad_port_open(struct ib_device *device, init_mad_qp(port_priv, &port_priv->qp_info[0]); init_mad_qp(port_priv, &port_priv->qp_info[1]); - cq_size = (IB_MAD_QP_SEND_SIZE + IB_MAD_QP_RECV_SIZE) * 2; + cq_size = mad_sendq_size + mad_recvq_size; + has_smi = rdma_port_get_link_layer(device, port_num) == IB_LINK_LAYER_INFINIBAND; + if (has_smi) + cq_size *= 2; + port_priv->cq = ib_create_cq(port_priv->device, - (ib_comp_handler) - ib_mad_thread_completion_handler, - NULL, port_priv, cq_size); + 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"); ret = PTR_ERR(port_priv->cq); @@ -2494,9 +2879,11 @@ static int ib_mad_port_open(struct ib_device *device, goto error5; } - ret = create_mad_qp(&port_priv->qp_info[0], IB_QPT_SMI); - if (ret) - goto error6; + if (has_smi) { + ret = create_mad_qp(&port_priv->qp_info[0], IB_QPT_SMI); + if (ret) + goto error6; + } ret = create_mad_qp(&port_priv->qp_info[1], IB_QPT_GSI); if (ret) goto error7; @@ -2507,7 +2894,11 @@ static int ib_mad_port_open(struct ib_device *device, ret = -ENOMEM; goto error8; } - INIT_WORK(&port_priv->work, ib_mad_completion_handler, port_priv); + INIT_WORK(&port_priv->work, ib_mad_completion_handler); + + spin_lock_irqsave(&ib_mad_port_list_lock, flags); + list_add_tail(&port_priv->port_list, &ib_mad_port_list); + spin_unlock_irqrestore(&ib_mad_port_list_lock, flags); ret = ib_mad_port_start(port_priv); if (ret) { @@ -2515,12 +2906,13 @@ static int ib_mad_port_open(struct ib_device *device, goto error9; } - spin_lock_irqsave(&ib_mad_port_list_lock, flags); - list_add_tail(&port_priv->port_list, &ib_mad_port_list); - spin_unlock_irqrestore(&ib_mad_port_list_lock, flags); return 0; error9: + spin_lock_irqsave(&ib_mad_port_list_lock, flags); + list_del_init(&port_priv->port_list); + spin_unlock_irqrestore(&ib_mad_port_list_lock, flags); + destroy_workqueue(port_priv->wq); error8: destroy_mad_qp(&port_priv->qp_info[1]); @@ -2557,11 +2949,9 @@ static int ib_mad_port_close(struct ib_device *device, int port_num) printk(KERN_ERR PFX "Port %d not found\n", port_num); return -ENODEV; } - list_del(&port_priv->port_list); + list_del_init(&port_priv->port_list); spin_unlock_irqrestore(&ib_mad_port_list_lock, flags); - /* Stop processing completions. */ - flush_workqueue(port_priv->wq); destroy_workqueue(port_priv->wq); destroy_mad_qp(&port_priv->qp_info[1]); destroy_mad_qp(&port_priv->qp_info[0]); @@ -2579,59 +2969,62 @@ static int ib_mad_port_close(struct ib_device *device, int port_num) static void ib_mad_init_device(struct ib_device *device) { - int ret, num_ports, cur_port, i, ret2; + int start, end, i; - if (device->node_type == IB_NODE_SWITCH) { - num_ports = 1; - cur_port = 0; + if (rdma_node_get_transport(device->node_type) != RDMA_TRANSPORT_IB) + return; + + if (device->node_type == RDMA_NODE_IB_SWITCH) { + start = 0; + end = 0; } else { - num_ports = device->phys_port_cnt; - cur_port = 1; + start = 1; + end = device->phys_port_cnt; } - for (i = 0; i < num_ports; i++, cur_port++) { - ret = ib_mad_port_open(device, cur_port); - if (ret) { + + 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, cur_port); - goto error_device_open; + device->name, i); + goto error; } - ret = ib_agent_port_open(device, cur_port); - if (ret) { + if (ib_agent_port_open(device, i)) { printk(KERN_ERR PFX "Couldn't open %s port %d " "for agents\n", - device->name, cur_port); - goto error_device_open; + device->name, i); + goto error_agent; } } + return; - goto error_device_query; +error_agent: + if (ib_mad_port_close(device, i)) + printk(KERN_ERR PFX "Couldn't close %s port %d\n", + device->name, i); -error_device_open: - while (i > 0) { - cur_port--; - ret2 = ib_agent_port_close(device, cur_port); - if (ret2) { +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, cur_port); - } - ret2 = ib_mad_port_close(device, cur_port); - if (ret2) { + device->name, i); + if (ib_mad_port_close(device, i)) printk(KERN_ERR PFX "Couldn't close %s port %d\n", - device->name, cur_port); - } + device->name, i); i--; } - -error_device_query: - return; } static void ib_mad_remove_device(struct ib_device *device) { - int ret = 0, i, num_ports, cur_port, ret2; + int i, num_ports, cur_port; - if (device->node_type == IB_NODE_SWITCH) { + if (rdma_node_get_transport(device->node_type) != RDMA_TRANSPORT_IB) + return; + + if (device->node_type == RDMA_NODE_IB_SWITCH) { num_ports = 1; cur_port = 0; } else { @@ -2639,21 +3032,13 @@ static void ib_mad_remove_device(struct ib_device *device) cur_port = 1; } for (i = 0; i < num_ports; i++, cur_port++) { - ret2 = ib_agent_port_close(device, cur_port); - if (ret2) { + 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); - if (!ret) - ret = ret2; - } - ret2 = ib_mad_port_close(device, cur_port); - if (ret2) { + if (ib_mad_port_close(device, cur_port)) printk(KERN_ERR PFX "Couldn't close %s port %d\n", device->name, cur_port); - if (!ret) - ret = ret2; - } } } @@ -2667,14 +3052,16 @@ static int __init ib_mad_init_module(void) { int ret; - spin_lock_init(&ib_mad_port_list_lock); - spin_lock_init(&ib_agent_port_list_lock); + mad_recvq_size = min(mad_recvq_size, IB_MAD_QP_MAX_SIZE); + mad_recvq_size = max(mad_recvq_size, IB_MAD_QP_MIN_SIZE); + + mad_sendq_size = min(mad_sendq_size, IB_MAD_QP_MAX_SIZE); + mad_sendq_size = max(mad_sendq_size, IB_MAD_QP_MIN_SIZE); ib_mad_cache = kmem_cache_create("ib_mad", sizeof(struct ib_mad_private), 0, SLAB_HWCACHE_ALIGN, - NULL, NULL); if (!ib_mad_cache) { printk(KERN_ERR PFX "Couldn't create ib_mad cache\n"); @@ -2701,10 +3088,7 @@ error1: static void __exit ib_mad_cleanup_module(void) { ib_unregister_client(&mad_client); - - if (kmem_cache_destroy(ib_mad_cache)) { - printk(KERN_DEBUG PFX "Failed to destroy ib_mad cache\n"); - } + kmem_cache_destroy(ib_mad_cache); } module_init(ib_mad_init_module); |