mad.c

linux-2.6.15.6

	mad_snoop_priv->agent.context = context;
	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);
	mad_snoop_priv->snoop_index = register_snoop_agent(
						&port_priv->qp_info[qpn],
						mad_snoop_priv);
	if (mad_snoop_priv->snoop_index < 0) {
		ret = ERR_PTR(mad_snoop_priv->snoop_index);
		goto error2;
	}

	atomic_set(&mad_snoop_priv->refcount, 1);
	return &mad_snoop_priv->agent;

error2:
	kfree(mad_snoop_priv);
error1:
	return ret;
}
EXPORT_SYMBOL(ib_register_mad_snoop);

static void unregister_mad_agent(struct ib_mad_agent_private *mad_agent_priv)
{
	struct ib_mad_port_private *port_priv;
	unsigned long flags;

	/* Note that we could still be handling received MADs */

	/*
	 * Canceling all sends results in dropping received response
	 * 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);

	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);

	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));

	kfree(mad_agent_priv->reg_req);
	ib_dereg_mr(mad_agent_priv->agent.mr);
	kfree(mad_agent_priv);
}

static void unregister_mad_snoop(struct ib_mad_snoop_private *mad_snoop_priv)
{
	struct ib_mad_qp_info *qp_info;
	unsigned long flags;

	qp_info = mad_snoop_priv->qp_info;
	spin_lock_irqsave(&qp_info->snoop_lock, flags);
	qp_info->snoop_table[mad_snoop_priv->snoop_index] = NULL;
	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));

	kfree(mad_snoop_priv);
}

/*
 * ib_unregister_mad_agent - Unregisters a client from using MAD services
 */
int ib_unregister_mad_agent(struct ib_mad_agent *mad_agent)
{
	struct ib_mad_agent_private *mad_agent_priv;
	struct ib_mad_snoop_private *mad_snoop_priv;

	/* If the TID is zero, the agent can only snoop. */
	if (mad_agent->hi_tid) {
		mad_agent_priv = container_of(mad_agent,
					      struct ib_mad_agent_private,
					      agent);
		unregister_mad_agent(mad_agent_priv);
	} else {
		mad_snoop_priv = container_of(mad_agent,
					      struct ib_mad_snoop_private,
					      agent);
		unregister_mad_snoop(mad_snoop_priv);
	}
	return 0;
}
EXPORT_SYMBOL(ib_unregister_mad_agent);

static inline 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 void dequeue_mad(struct ib_mad_list_head *mad_list)
{
	struct ib_mad_queue *mad_queue;
	unsigned long flags;

	BUG_ON(!mad_list->mad_queue);
	mad_queue = mad_list->mad_queue;
	spin_lock_irqsave(&mad_queue->lock, flags);
	list_del(&mad_list->list);
	mad_queue->count--;
	spin_unlock_irqrestore(&mad_queue->lock, flags);
}

static void snoop_send(struct ib_mad_qp_info *qp_info,
		       struct ib_mad_send_buf *send_buf,
		       struct ib_mad_send_wc *mad_send_wc,
		       int mad_snoop_flags)
{
	struct ib_mad_snoop_private *mad_snoop_priv;
	unsigned long flags;
	int i;

	spin_lock_irqsave(&qp_info->snoop_lock, flags);
	for (i = 0; i < qp_info->snoop_table_size; i++) {
		mad_snoop_priv = qp_info->snoop_table[i];
		if (!mad_snoop_priv ||
		    !(mad_snoop_priv->mad_snoop_flags & mad_snoop_flags))
			continue;

		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_buf, mad_send_wc);
		if (atomic_dec_and_test(&mad_snoop_priv->refcount))
			wake_up(&mad_snoop_priv->wait);
		spin_lock_irqsave(&qp_info->snoop_lock, flags);
	}
	spin_unlock_irqrestore(&qp_info->snoop_lock, flags);
}

static void snoop_recv(struct ib_mad_qp_info *qp_info,
		       struct ib_mad_recv_wc *mad_recv_wc,
		       int mad_snoop_flags)
{
	struct ib_mad_snoop_private *mad_snoop_priv;
	unsigned long flags;
	int i;

	spin_lock_irqsave(&qp_info->snoop_lock, flags);
	for (i = 0; i < qp_info->snoop_table_size; i++) {
		mad_snoop_priv = qp_info->snoop_table[i];
		if (!mad_snoop_priv ||
		    !(mad_snoop_priv->mad_snoop_flags & mad_snoop_flags))
			continue;

		atomic_inc(&mad_snoop_priv->refcount);
		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);
		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,
			 struct ib_wc *wc)
{
	memset(wc, 0, sizeof *wc);
	wc->wr_id = wr_id;
	wc->status = IB_WC_SUCCESS;
	wc->opcode = IB_WC_RECV;
	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->slid = slid;
	wc->sl = 0;
	wc->dlid_path_bits = 0;
	wc->port_num = port_num;
}

/*
 * Return 0 if SMP is to be sent
 * Return 1 if SMP was consumed locally (whether or not solicited)
 * Return < 0 if error
 */
static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv,
				  struct ib_mad_send_wr_private *mad_send_wr)
{
	int ret;
	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;
	struct ib_wc mad_wc;
	struct ib_send_wr *send_wr = &mad_send_wr->send_wr;

	if (!smi_handle_dr_smp_send(smp, device->node_type, port_num)) {
		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)
		goto out;

	local = kmalloc(sizeof *local, GFP_ATOMIC);
	if (!local) {
		ret = -ENOMEM;
		printk(KERN_ERR PFX "No memory for ib_mad_local_private\n");
		goto out;
	}
	local->mad_priv = NULL;
	local->recv_mad_agent = NULL;
	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");
		kfree(local);
		goto out;
	}

	build_smp_wc(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 */
	ret = device->process_mad(device, 0, port_num, &mad_wc, NULL,
				  (struct ib_mad *)smp,
				  (struct ib_mad *)&mad_priv->mad);
	switch (ret)
	{
	case IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_REPLY:
		if (response_mad(&mad_priv->mad.mad) &&
		    mad_agent_priv->agent.recv_handler) {
			local->mad_priv = mad_priv;
			local->recv_mad_agent = mad_agent_priv;
			/*
			 * Reference MAD agent until receive
			 * side of local completion handled
			 */
			atomic_inc(&mad_agent_priv->refcount);
		} else
			kmem_cache_free(ib_mad_cache, mad_priv);
		break;
	case IB_MAD_RESULT_SUCCESS | IB_MAD_RESULT_CONSUMED:
		kmem_cache_free(ib_mad_cache, mad_priv);
		break;
	case IB_MAD_RESULT_SUCCESS:
		/* Treat like an incoming receive 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;
			recv_mad_agent = find_mad_agent(port_priv,
						        &mad_priv->mad.mad);
		}
		if (!port_priv || !recv_mad_agent) {
			kmem_cache_free(ib_mad_cache, mad_priv);
			kfree(local);
			ret = 0;
			goto out;
		}
		local->mad_priv = mad_priv;
		local->recv_mad_agent = recv_mad_agent;
		break;
	default:
		kmem_cache_free(ib_mad_cache, mad_priv);
		kfree(local);
		ret = -EINVAL;
		goto out;
	}

	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 */
	spin_lock_irqsave(&mad_agent_priv->lock, flags);
	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);
	ret = 1;
out:
	return ret;
}

static int get_buf_length(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;
		if (pad == seg_size)
			pad = 0;
	} else
		pad = seg_size;
	return hdr_len + data_len + pad;
}

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 buf_size;
	void *buf;

	mad_agent_priv = container_of(mad_agent, struct ib_mad_agent_private,
				      agent);
	buf_size = get_buf_length(hdr_len, data_len);

	if ((!mad_agent->rmpp_version &&
	     (rmpp_active || buf_size > sizeof(struct ib_mad))) ||
	    (!rmpp_active && buf_size > sizeof(struct ib_mad)))
		return ERR_PTR(-EINVAL);

	buf = kzalloc(sizeof *mad_send_wr + buf_size, gfp_mask);
	if (!buf)
		return ERR_PTR(-ENOMEM);

	mad_send_wr = buf + buf_size;
	mad_send_wr->send_buf.mad = buf;

	mad_send_wr->mad_agent_priv = mad_agent_priv;
	mad_send_wr->sg_list[0].length = buf_size;
	mad_send_wr->sg_list[0].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 = 1;
	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) {
		struct ib_rmpp_mad *rmpp_mad = mad_send_wr->send_buf.mad;
		rmpp_mad->rmpp_hdr.paylen_newwin = cpu_to_be32(hdr_len -
						   IB_MGMT_RMPP_HDR + data_len);
		rmpp_mad->rmpp_hdr.rmpp_version = mad_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);
	}

	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);

void ib_free_send_mad(struct ib_mad_send_buf *send_buf)
{
	struct ib_mad_agent_private *mad_agent_priv;

	mad_agent_priv = container_of(send_buf->mad_agent,
				      struct ib_mad_agent_private, agent);
	kfree(send_buf->mad);

	if (atomic_dec_and_test(&mad_agent_priv->refcount))
		wake_up(&mad_agent_priv->wait);
}
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;

	/* 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->addr = dma_map_single(mad_agent->device->dma_device,
				   mad_send_wr->send_buf.mad, sge->length,
				   DMA_TO_DEVICE);
	pci_unmap_addr_set(mad_send_wr, mapping, sge->addr);

	spin_lock_irqsave(&qp_info->send_queue.lock, flags);
	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 {
		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)
		dma_unmap_single(mad_agent->device->dma_device,
				 pci_unmap_addr(mad_send_wr, mapping),
				 sge->length, DMA_TO_DEVICE);

	return ret;
}

/*
 * 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_send_buf *send_buf,
		     struct ib_mad_send_buf **bad_send_buf)
{
	struct ib_mad_agent_private *mad_agent_priv;
	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 */
	for (; send_buf; send_buf = next_send_buf) {

		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;
		}

		/*
		 * Save pointer to next work request to post in case the
		 * current one completes, and the user modifies the work
		 * request associated with the completion
		 */