cma.c

linux 内核源代码

/*
 * Copyright (c) 2005 Voltaire Inc.  All rights reserved.
 * Copyright (c) 2002-2005, Network Appliance, Inc. All rights reserved.
 * Copyright (c) 1999-2005, Mellanox Technologies, Inc. All rights reserved.
 * Copyright (c) 2005-2006 Intel Corporation.  All rights reserved.
 *
 * This Software is licensed under one of the following licenses:
 *
 * 1) under the terms of the "Common Public License 1.0" a copy of which is
 *    available from the Open Source Initiative, see
 *    http://www.opensource.org/licenses/cpl.php.
 *
 * 2) under the terms of the "The BSD License" a copy of which is
 *    available from the Open Source Initiative, see
 *    http://www.opensource.org/licenses/bsd-license.php.
 *
 * 3) under the terms of the "GNU General Public License (GPL) Version 2" a
 *    copy of which is available from the Open Source Initiative, see
 *    http://www.opensource.org/licenses/gpl-license.php.
 *
 * Licensee has the right to choose one of the above licenses.
 *
 * Redistributions of source code must retain the above copyright
 * notice and one of the license notices.
 *
 * Redistributions in binary form must reproduce both the above copyright
 * notice, one of the license notices in the documentation
 * and/or other materials provided with the distribution.
 *
 */

#include <linux/completion.h>
#include <linux/in.h>
#include <linux/in6.h>
#include <linux/mutex.h>
#include <linux/random.h>
#include <linux/idr.h>
#include <linux/inetdevice.h>

#include <net/tcp.h>

#include <rdma/rdma_cm.h>
#include <rdma/rdma_cm_ib.h>
#include <rdma/ib_cache.h>
#include <rdma/ib_cm.h>
#include <rdma/ib_sa.h>
#include <rdma/iw_cm.h>

MODULE_AUTHOR("Sean Hefty");
MODULE_DESCRIPTION("Generic RDMA CM Agent");
MODULE_LICENSE("Dual BSD/GPL");

#define CMA_CM_RESPONSE_TIMEOUT 20
#define CMA_MAX_CM_RETRIES 15
#define CMA_CM_MRA_SETTING (IB_CM_MRA_FLAG_DELAY | 24)

static void cma_add_one(struct ib_device *device);
static void cma_remove_one(struct ib_device *device);

static struct ib_client cma_client = {
	.name   = "cma",
	.add    = cma_add_one,
	.remove = cma_remove_one
};

static struct ib_sa_client sa_client;
static struct rdma_addr_client addr_client;
static LIST_HEAD(dev_list);
static LIST_HEAD(listen_any_list);
static DEFINE_MUTEX(lock);
static struct workqueue_struct *cma_wq;
static DEFINE_IDR(sdp_ps);
static DEFINE_IDR(tcp_ps);
static DEFINE_IDR(udp_ps);
static DEFINE_IDR(ipoib_ps);
static int next_port;

struct cma_device {
	struct list_head	list;
	struct ib_device	*device;
	struct completion	comp;
	atomic_t		refcount;
	struct list_head	id_list;
};

enum cma_state {
	CMA_IDLE,
	CMA_ADDR_QUERY,
	CMA_ADDR_RESOLVED,
	CMA_ROUTE_QUERY,
	CMA_ROUTE_RESOLVED,
	CMA_CONNECT,
	CMA_DISCONNECT,
	CMA_ADDR_BOUND,
	CMA_LISTEN,
	CMA_DEVICE_REMOVAL,
	CMA_DESTROYING
};

struct rdma_bind_list {
	struct idr		*ps;
	struct hlist_head	owners;
	unsigned short		port;
};

/*
 * Device removal can occur at anytime, so we need extra handling to
 * serialize notifying the user of device removal with other callbacks.
 * We do this by disabling removal notification while a callback is in process,
 * and reporting it after the callback completes.
 */
struct rdma_id_private {
	struct rdma_cm_id	id;

	struct rdma_bind_list	*bind_list;
	struct hlist_node	node;
	struct list_head	list; /* listen_any_list or cma_device.list */
	struct list_head	listen_list; /* per device listens */
	struct cma_device	*cma_dev;
	struct list_head	mc_list;

	int			internal_id;
	enum cma_state		state;
	spinlock_t		lock;
	struct mutex		qp_mutex;

	struct completion	comp;
	atomic_t		refcount;
	wait_queue_head_t	wait_remove;
	atomic_t		dev_remove;

	int			backlog;
	int			timeout_ms;
	struct ib_sa_query	*query;
	int			query_id;
	union {
		struct ib_cm_id	*ib;
		struct iw_cm_id	*iw;
	} cm_id;

	u32			seq_num;
	u32			qkey;
	u32			qp_num;
	u8			srq;
	u8			tos;
};

struct cma_multicast {
	struct rdma_id_private *id_priv;
	union {
		struct ib_sa_multicast *ib;
	} multicast;
	struct list_head	list;
	void			*context;
	struct sockaddr		addr;
	u8			pad[sizeof(struct sockaddr_in6) -
				    sizeof(struct sockaddr)];
};

struct cma_work {
	struct work_struct	work;
	struct rdma_id_private	*id;
	enum cma_state		old_state;
	enum cma_state		new_state;
	struct rdma_cm_event	event;
};

union cma_ip_addr {
	struct in6_addr ip6;
	struct {
		__u32 pad[3];
		__u32 addr;
	} ip4;
};

struct cma_hdr {
	u8 cma_version;
	u8 ip_version;	/* IP version: 7:4 */
	__u16 port;
	union cma_ip_addr src_addr;
	union cma_ip_addr dst_addr;
};

struct sdp_hh {
	u8 bsdh[16];
	u8 sdp_version; /* Major version: 7:4 */
	u8 ip_version;	/* IP version: 7:4 */
	u8 sdp_specific1[10];
	__u16 port;
	__u16 sdp_specific2;
	union cma_ip_addr src_addr;
	union cma_ip_addr dst_addr;
};

struct sdp_hah {
	u8 bsdh[16];
	u8 sdp_version;
};

#define CMA_VERSION 0x00
#define SDP_MAJ_VERSION 0x2

static int cma_comp(struct rdma_id_private *id_priv, enum cma_state comp)
{
	unsigned long flags;
	int ret;

	spin_lock_irqsave(&id_priv->lock, flags);
	ret = (id_priv->state == comp);
	spin_unlock_irqrestore(&id_priv->lock, flags);
	return ret;
}

static int cma_comp_exch(struct rdma_id_private *id_priv,
			 enum cma_state comp, enum cma_state exch)
{
	unsigned long flags;
	int ret;

	spin_lock_irqsave(&id_priv->lock, flags);
	if ((ret = (id_priv->state == comp)))
		id_priv->state = exch;
	spin_unlock_irqrestore(&id_priv->lock, flags);
	return ret;
}

static enum cma_state cma_exch(struct rdma_id_private *id_priv,
			       enum cma_state exch)
{
	unsigned long flags;
	enum cma_state old;

	spin_lock_irqsave(&id_priv->lock, flags);
	old = id_priv->state;
	id_priv->state = exch;
	spin_unlock_irqrestore(&id_priv->lock, flags);
	return old;
}

static inline u8 cma_get_ip_ver(struct cma_hdr *hdr)
{
	return hdr->ip_version >> 4;
}

static inline void cma_set_ip_ver(struct cma_hdr *hdr, u8 ip_ver)
{
	hdr->ip_version = (ip_ver << 4) | (hdr->ip_version & 0xF);
}

static inline u8 sdp_get_majv(u8 sdp_version)
{
	return sdp_version >> 4;
}

static inline u8 sdp_get_ip_ver(struct sdp_hh *hh)
{
	return hh->ip_version >> 4;
}

static inline void sdp_set_ip_ver(struct sdp_hh *hh, u8 ip_ver)
{
	hh->ip_version = (ip_ver << 4) | (hh->ip_version & 0xF);
}

static inline int cma_is_ud_ps(enum rdma_port_space ps)
{
	return (ps == RDMA_PS_UDP || ps == RDMA_PS_IPOIB);
}

static void cma_attach_to_dev(struct rdma_id_private *id_priv,
			      struct cma_device *cma_dev)
{
	atomic_inc(&cma_dev->refcount);
	id_priv->cma_dev = cma_dev;
	id_priv->id.device = cma_dev->device;
	list_add_tail(&id_priv->list, &cma_dev->id_list);
}

static inline void cma_deref_dev(struct cma_device *cma_dev)
{
	if (atomic_dec_and_test(&cma_dev->refcount))
		complete(&cma_dev->comp);
}

static void cma_detach_from_dev(struct rdma_id_private *id_priv)
{
	list_del(&id_priv->list);
	cma_deref_dev(id_priv->cma_dev);
	id_priv->cma_dev = NULL;
}

static int cma_set_qkey(struct ib_device *device, u8 port_num,
			enum rdma_port_space ps,
			struct rdma_dev_addr *dev_addr, u32 *qkey)
{
	struct ib_sa_mcmember_rec rec;
	int ret = 0;

	switch (ps) {
	case RDMA_PS_UDP:
		*qkey = RDMA_UDP_QKEY;
		break;
	case RDMA_PS_IPOIB:
		ib_addr_get_mgid(dev_addr, &rec.mgid);
		ret = ib_sa_get_mcmember_rec(device, port_num, &rec.mgid, &rec);
		*qkey = be32_to_cpu(rec.qkey);
		break;
	default:
		break;
	}
	return ret;
}

static int cma_acquire_dev(struct rdma_id_private *id_priv)
{
	struct rdma_dev_addr *dev_addr = &id_priv->id.route.addr.dev_addr;
	struct cma_device *cma_dev;
	union ib_gid gid;
	int ret = -ENODEV;

	switch (rdma_node_get_transport(dev_addr->dev_type)) {
	case RDMA_TRANSPORT_IB:
		ib_addr_get_sgid(dev_addr, &gid);
		break;
	case RDMA_TRANSPORT_IWARP:
		iw_addr_get_sgid(dev_addr, &gid);
		break;
	default:
		return -ENODEV;
	}

	list_for_each_entry(cma_dev, &dev_list, list) {
		ret = ib_find_cached_gid(cma_dev->device, &gid,
					 &id_priv->id.port_num, NULL);
		if (!ret) {
			ret = cma_set_qkey(cma_dev->device,
					   id_priv->id.port_num,
					   id_priv->id.ps, dev_addr,
					   &id_priv->qkey);
			if (!ret)
				cma_attach_to_dev(id_priv, cma_dev);
			break;
		}
	}
	return ret;
}

static void cma_deref_id(struct rdma_id_private *id_priv)
{
	if (atomic_dec_and_test(&id_priv->refcount))
		complete(&id_priv->comp);
}

static int cma_disable_remove(struct rdma_id_private *id_priv,
			      enum cma_state state)
{
	unsigned long flags;
	int ret;

	spin_lock_irqsave(&id_priv->lock, flags);
	if (id_priv->state == state) {
		atomic_inc(&id_priv->dev_remove);
		ret = 0;
	} else
		ret = -EINVAL;
	spin_unlock_irqrestore(&id_priv->lock, flags);
	return ret;
}

static void cma_enable_remove(struct rdma_id_private *id_priv)
{
	if (atomic_dec_and_test(&id_priv->dev_remove))
		wake_up(&id_priv->wait_remove);
}

static int cma_has_cm_dev(struct rdma_id_private *id_priv)
{
	return (id_priv->id.device && id_priv->cm_id.ib);
}

struct rdma_cm_id *rdma_create_id(rdma_cm_event_handler event_handler,
				  void *context, enum rdma_port_space ps)
{
	struct rdma_id_private *id_priv;

	id_priv = kzalloc(sizeof *id_priv, GFP_KERNEL);
	if (!id_priv)
		return ERR_PTR(-ENOMEM);

	id_priv->state = CMA_IDLE;
	id_priv->id.context = context;
	id_priv->id.event_handler = event_handler;
	id_priv->id.ps = ps;
	spin_lock_init(&id_priv->lock);
	mutex_init(&id_priv->qp_mutex);
	init_completion(&id_priv->comp);
	atomic_set(&id_priv->refcount, 1);
	init_waitqueue_head(&id_priv->wait_remove);
	atomic_set(&id_priv->dev_remove, 0);
	INIT_LIST_HEAD(&id_priv->listen_list);
	INIT_LIST_HEAD(&id_priv->mc_list);
	get_random_bytes(&id_priv->seq_num, sizeof id_priv->seq_num);

	return &id_priv->id;
}
EXPORT_SYMBOL(rdma_create_id);

static int cma_init_ud_qp(struct rdma_id_private *id_priv, struct ib_qp *qp)
{
	struct ib_qp_attr qp_attr;
	int qp_attr_mask, ret;

	qp_attr.qp_state = IB_QPS_INIT;
	ret = rdma_init_qp_attr(&id_priv->id, &qp_attr, &qp_attr_mask);
	if (ret)
		return ret;

	ret = ib_modify_qp(qp, &qp_attr, qp_attr_mask);
	if (ret)
		return ret;

	qp_attr.qp_state = IB_QPS_RTR;
	ret = ib_modify_qp(qp, &qp_attr, IB_QP_STATE);
	if (ret)
		return ret;

	qp_attr.qp_state = IB_QPS_RTS;
	qp_attr.sq_psn = 0;
	ret = ib_modify_qp(qp, &qp_attr, IB_QP_STATE | IB_QP_SQ_PSN);

	return ret;
}

static int cma_init_conn_qp(struct rdma_id_private *id_priv, struct ib_qp *qp)
{
	struct ib_qp_attr qp_attr;
	int qp_attr_mask, ret;

	qp_attr.qp_state = IB_QPS_INIT;
	ret = rdma_init_qp_attr(&id_priv->id, &qp_attr, &qp_attr_mask);
	if (ret)
		return ret;

	return ib_modify_qp(qp, &qp_attr, qp_attr_mask);
}

int rdma_create_qp(struct rdma_cm_id *id, struct ib_pd *pd,
		   struct ib_qp_init_attr *qp_init_attr)
{
	struct rdma_id_private *id_priv;
	struct ib_qp *qp;
	int ret;

	id_priv = container_of(id, struct rdma_id_private, id);
	if (id->device != pd->device)
		return -EINVAL;

	qp = ib_create_qp(pd, qp_init_attr);
	if (IS_ERR(qp))
		return PTR_ERR(qp);

	if (cma_is_ud_ps(id_priv->id.ps))
		ret = cma_init_ud_qp(id_priv, qp);
	else
		ret = cma_init_conn_qp(id_priv, qp);
	if (ret)
		goto err;

	id->qp = qp;
	id_priv->qp_num = qp->qp_num;
	id_priv->srq = (qp->srq != NULL);
	return 0;
err:
	ib_destroy_qp(qp);
	return ret;
}
EXPORT_SYMBOL(rdma_create_qp);

void rdma_destroy_qp(struct rdma_cm_id *id)
{
	struct rdma_id_private *id_priv;

	id_priv = container_of(id, struct rdma_id_private, id);
	mutex_lock(&id_priv->qp_mutex);
	ib_destroy_qp(id_priv->id.qp);
	id_priv->id.qp = NULL;
	mutex_unlock(&id_priv->qp_mutex);
}
EXPORT_SYMBOL(rdma_destroy_qp);

static int cma_modify_qp_rtr(struct rdma_id_private *id_priv)
{
	struct ib_qp_attr qp_attr;
	int qp_attr_mask, ret;

	mutex_lock(&id_priv->qp_mutex);
	if (!id_priv->id.qp) {
		ret = 0;
		goto out;
	}

	/* Need to update QP attributes from default values. */
	qp_attr.qp_state = IB_QPS_INIT;
	ret = rdma_init_qp_attr(&id_priv->id, &qp_attr, &qp_attr_mask);
	if (ret)
		goto out;

	ret = ib_modify_qp(id_priv->id.qp, &qp_attr, qp_attr_mask);
	if (ret)
		goto out;

	qp_attr.qp_state = IB_QPS_RTR;
	ret = rdma_init_qp_attr(&id_priv->id, &qp_attr, &qp_attr_mask);
	if (ret)
		goto out;

	ret = ib_modify_qp(id_priv->id.qp, &qp_attr, qp_attr_mask);
out:
	mutex_unlock(&id_priv->qp_mutex);
	return ret;
}

static int cma_modify_qp_rts(struct rdma_id_private *id_priv)