multicast.c

来自「linux 内核源代码」· C语言 代码 · 共 837 行 · 第 1/2 页

/*
 * Copyright (c) 2006 Intel Corporation.  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
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer in the documentation and/or other materials
 *        provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#include <linux/completion.h>
#include <linux/dma-mapping.h>
#include <linux/err.h>
#include <linux/interrupt.h>
#include <linux/bitops.h>
#include <linux/random.h>

#include <rdma/ib_cache.h>
#include "sa.h"

static void mcast_add_one(struct ib_device *device);
static void mcast_remove_one(struct ib_device *device);

static struct ib_client mcast_client = {
	.name   = "ib_multicast",
	.add    = mcast_add_one,
	.remove = mcast_remove_one
};

static struct ib_sa_client	sa_client;
static struct workqueue_struct	*mcast_wq;
static union ib_gid mgid0;

struct mcast_device;

struct mcast_port {
	struct mcast_device	*dev;
	spinlock_t		lock;
	struct rb_root		table;
	atomic_t		refcount;
	struct completion	comp;
	u8			port_num;
};

struct mcast_device {
	struct ib_device	*device;
	struct ib_event_handler	event_handler;
	int			start_port;
	int			end_port;
	struct mcast_port	port[0];
};

enum mcast_state {
	MCAST_IDLE,
	MCAST_JOINING,
	MCAST_MEMBER,
	MCAST_BUSY,
	MCAST_ERROR
};

struct mcast_member;

struct mcast_group {
	struct ib_sa_mcmember_rec rec;
	struct rb_node		node;
	struct mcast_port	*port;
	spinlock_t		lock;
	struct work_struct	work;
	struct list_head	pending_list;
	struct list_head	active_list;
	struct mcast_member	*last_join;
	int			members[3];
	atomic_t		refcount;
	enum mcast_state	state;
	struct ib_sa_query	*query;
	int			query_id;
};

struct mcast_member {
	struct ib_sa_multicast	multicast;
	struct ib_sa_client	*client;
	struct mcast_group	*group;
	struct list_head	list;
	enum mcast_state	state;
	atomic_t		refcount;
	struct completion	comp;
};

static void join_handler(int status, struct ib_sa_mcmember_rec *rec,
			 void *context);
static void leave_handler(int status, struct ib_sa_mcmember_rec *rec,
			  void *context);

static struct mcast_group *mcast_find(struct mcast_port *port,
				      union ib_gid *mgid)
{
	struct rb_node *node = port->table.rb_node;
	struct mcast_group *group;
	int ret;

	while (node) {
		group = rb_entry(node, struct mcast_group, node);
		ret = memcmp(mgid->raw, group->rec.mgid.raw, sizeof *mgid);
		if (!ret)
			return group;

		if (ret < 0)
			node = node->rb_left;
		else
			node = node->rb_right;
	}
	return NULL;
}

static struct mcast_group *mcast_insert(struct mcast_port *port,
					struct mcast_group *group,
					int allow_duplicates)
{
	struct rb_node **link = &port->table.rb_node;
	struct rb_node *parent = NULL;
	struct mcast_group *cur_group;
	int ret;

	while (*link) {
		parent = *link;
		cur_group = rb_entry(parent, struct mcast_group, node);

		ret = memcmp(group->rec.mgid.raw, cur_group->rec.mgid.raw,
			     sizeof group->rec.mgid);
		if (ret < 0)
			link = &(*link)->rb_left;
		else if (ret > 0)
			link = &(*link)->rb_right;
		else if (allow_duplicates)
			link = &(*link)->rb_left;
		else
			return cur_group;
	}
	rb_link_node(&group->node, parent, link);
	rb_insert_color(&group->node, &port->table);
	return NULL;
}

static void deref_port(struct mcast_port *port)
{
	if (atomic_dec_and_test(&port->refcount))
		complete(&port->comp);
}

static void release_group(struct mcast_group *group)
{
	struct mcast_port *port = group->port;
	unsigned long flags;

	spin_lock_irqsave(&port->lock, flags);
	if (atomic_dec_and_test(&group->refcount)) {
		rb_erase(&group->node, &port->table);
		spin_unlock_irqrestore(&port->lock, flags);
		kfree(group);
		deref_port(port);
	} else
		spin_unlock_irqrestore(&port->lock, flags);
}

static void deref_member(struct mcast_member *member)
{
	if (atomic_dec_and_test(&member->refcount))
		complete(&member->comp);
}

static void queue_join(struct mcast_member *member)
{
	struct mcast_group *group = member->group;
	unsigned long flags;

	spin_lock_irqsave(&group->lock, flags);
	list_add_tail(&member->list, &group->pending_list);
	if (group->state == MCAST_IDLE) {
		group->state = MCAST_BUSY;
		atomic_inc(&group->refcount);
		queue_work(mcast_wq, &group->work);
	}
	spin_unlock_irqrestore(&group->lock, flags);
}

/*
 * A multicast group has three types of members: full member, non member, and
 * send only member.  We need to keep track of the number of members of each
 * type based on their join state.  Adjust the number of members the belong to
 * the specified join states.
 */
static void adjust_membership(struct mcast_group *group, u8 join_state, int inc)
{
	int i;

	for (i = 0; i < 3; i++, join_state >>= 1)
		if (join_state & 0x1)
			group->members[i] += inc;
}

/*
 * If a multicast group has zero members left for a particular join state, but
 * the group is still a member with the SA, we need to leave that join state.
 * Determine which join states we still belong to, but that do not have any
 * active members.
 */
static u8 get_leave_state(struct mcast_group *group)
{
	u8 leave_state = 0;
	int i;

	for (i = 0; i < 3; i++)
		if (!group->members[i])
			leave_state |= (0x1 << i);

	return leave_state & group->rec.join_state;
}

static int check_selector(ib_sa_comp_mask comp_mask,
			  ib_sa_comp_mask selector_mask,
			  ib_sa_comp_mask value_mask,
			  u8 selector, u8 src_value, u8 dst_value)
{
	int err;

	if (!(comp_mask & selector_mask) || !(comp_mask & value_mask))
		return 0;

	switch (selector) {
	case IB_SA_GT:
		err = (src_value <= dst_value);
		break;
	case IB_SA_LT:
		err = (src_value >= dst_value);
		break;
	case IB_SA_EQ:
		err = (src_value != dst_value);
		break;
	default:
		err = 0;
		break;
	}

	return err;
}

static int cmp_rec(struct ib_sa_mcmember_rec *src,
		   struct ib_sa_mcmember_rec *dst, ib_sa_comp_mask comp_mask)
{
	/* MGID must already match */

	if (comp_mask & IB_SA_MCMEMBER_REC_PORT_GID &&
	    memcmp(&src->port_gid, &dst->port_gid, sizeof src->port_gid))
		return -EINVAL;
	if (comp_mask & IB_SA_MCMEMBER_REC_QKEY && src->qkey != dst->qkey)
		return -EINVAL;
	if (comp_mask & IB_SA_MCMEMBER_REC_MLID && src->mlid != dst->mlid)
		return -EINVAL;
	if (check_selector(comp_mask, IB_SA_MCMEMBER_REC_MTU_SELECTOR,
			   IB_SA_MCMEMBER_REC_MTU, dst->mtu_selector,
			   src->mtu, dst->mtu))
		return -EINVAL;
	if (comp_mask & IB_SA_MCMEMBER_REC_TRAFFIC_CLASS &&
	    src->traffic_class != dst->traffic_class)
		return -EINVAL;
	if (comp_mask & IB_SA_MCMEMBER_REC_PKEY && src->pkey != dst->pkey)
		return -EINVAL;
	if (check_selector(comp_mask, IB_SA_MCMEMBER_REC_RATE_SELECTOR,
			   IB_SA_MCMEMBER_REC_RATE, dst->rate_selector,
			   src->rate, dst->rate))
		return -EINVAL;
	if (check_selector(comp_mask,
			   IB_SA_MCMEMBER_REC_PACKET_LIFE_TIME_SELECTOR,
			   IB_SA_MCMEMBER_REC_PACKET_LIFE_TIME,
			   dst->packet_life_time_selector,
			   src->packet_life_time, dst->packet_life_time))
		return -EINVAL;
	if (comp_mask & IB_SA_MCMEMBER_REC_SL && src->sl != dst->sl)
		return -EINVAL;
	if (comp_mask & IB_SA_MCMEMBER_REC_FLOW_LABEL &&
	    src->flow_label != dst->flow_label)
		return -EINVAL;
	if (comp_mask & IB_SA_MCMEMBER_REC_HOP_LIMIT &&
	    src->hop_limit != dst->hop_limit)
		return -EINVAL;
	if (comp_mask & IB_SA_MCMEMBER_REC_SCOPE && src->scope != dst->scope)
		return -EINVAL;

	/* join_state checked separately, proxy_join ignored */

	return 0;
}

static int send_join(struct mcast_group *group, struct mcast_member *member)
{
	struct mcast_port *port = group->port;
	int ret;

	group->last_join = member;
	ret = ib_sa_mcmember_rec_query(&sa_client, port->dev->device,
				       port->port_num, IB_MGMT_METHOD_SET,
				       &member->multicast.rec,
				       member->multicast.comp_mask,
				       3000, GFP_KERNEL, join_handler, group,
				       &group->query);
	if (ret >= 0) {
		group->query_id = ret;
		ret = 0;
	}
	return ret;
}

static int send_leave(struct mcast_group *group, u8 leave_state)
{
	struct mcast_port *port = group->port;
	struct ib_sa_mcmember_rec rec;
	int ret;

	rec = group->rec;
	rec.join_state = leave_state;

	ret = ib_sa_mcmember_rec_query(&sa_client, port->dev->device,
				       port->port_num, IB_SA_METHOD_DELETE, &rec,
				       IB_SA_MCMEMBER_REC_MGID     |
				       IB_SA_MCMEMBER_REC_PORT_GID |
				       IB_SA_MCMEMBER_REC_JOIN_STATE,
				       3000, GFP_KERNEL, leave_handler,
				       group, &group->query);
	if (ret >= 0) {
		group->query_id = ret;
		ret = 0;
	}
	return ret;
}

static void join_group(struct mcast_group *group, struct mcast_member *member,
		       u8 join_state)
{
	member->state = MCAST_MEMBER;
	adjust_membership(group, join_state, 1);
	group->rec.join_state |= join_state;
	member->multicast.rec = group->rec;
	member->multicast.rec.join_state = join_state;
	list_move(&member->list, &group->active_list);
}

static int fail_join(struct mcast_group *group, struct mcast_member *member,
		     int status)
{
	spin_lock_irq(&group->lock);
	list_del_init(&member->list);
	spin_unlock_irq(&group->lock);
	return member->multicast.callback(status, &member->multicast);
}

static void process_group_error(struct mcast_group *group)
{
	struct mcast_member *member;
	int ret;

	spin_lock_irq(&group->lock);
	while (!list_empty(&group->active_list)) {
		member = list_entry(group->active_list.next,
				    struct mcast_member, list);
		atomic_inc(&member->refcount);
		list_del_init(&member->list);
		adjust_membership(group, member->multicast.rec.join_state, -1);
		member->state = MCAST_ERROR;
		spin_unlock_irq(&group->lock);

		ret = member->multicast.callback(-ENETRESET,
						 &member->multicast);
		deref_member(member);
		if (ret)
			ib_sa_free_multicast(&member->multicast);
		spin_lock_irq(&group->lock);
	}

	group->rec.join_state = 0;
	group->state = MCAST_BUSY;
	spin_unlock_irq(&group->lock);
}

static void mcast_work_handler(struct work_struct *work)
{
	struct mcast_group *group;
	struct mcast_member *member;
	struct ib_sa_multicast *multicast;
	int status, ret;
	u8 join_state;

	group = container_of(work, typeof(*group), work);
retest:
	spin_lock_irq(&group->lock);
	while (!list_empty(&group->pending_list) ||
	       (group->state == MCAST_ERROR)) {