[deliverable/linux.git] / drivers / infiniband / core / addr.c

/*
 * 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 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/mutex.h>
#include <linux/inetdevice.h>
#include <linux/slab.h>
#include <linux/workqueue.h>
#include <linux/module.h>
#include <net/arp.h>
#include <net/neighbour.h>
#include <net/route.h>
#include <net/netevent.h>
#include <net/addrconf.h>
#include <net/ip6_route.h>
#include <rdma/ib_addr.h>

MODULE_AUTHOR("Sean Hefty");
MODULE_DESCRIPTION("IB Address Translation");
MODULE_LICENSE("Dual BSD/GPL");

struct addr_req {
	struct list_head list;
	struct sockaddr_storage src_addr;
	struct sockaddr_storage dst_addr;
	struct rdma_dev_addr *addr;
	struct rdma_addr_client *client;
	void *context;
	void (*callback)(int status, struct sockaddr *src_addr,
			 struct rdma_dev_addr *addr, void *context);
	unsigned long timeout;
	int status;
};

static void process_req(struct work_struct *work);

static DEFINE_MUTEX(lock);
static LIST_HEAD(req_list);
static DECLARE_DELAYED_WORK(work, process_req);
static struct workqueue_struct *addr_wq;

void rdma_addr_register_client(struct rdma_addr_client *client)
{
	atomic_set(&client->refcount, 1);
	init_completion(&client->comp);
}
EXPORT_SYMBOL(rdma_addr_register_client);

static inline void put_client(struct rdma_addr_client *client)
{
	if (atomic_dec_and_test(&client->refcount))
		complete(&client->comp);
}

void rdma_addr_unregister_client(struct rdma_addr_client *client)
{
	put_client(client);
	wait_for_completion(&client->comp);
}
EXPORT_SYMBOL(rdma_addr_unregister_client);

int rdma_copy_addr(struct rdma_dev_addr *dev_addr, struct net_device *dev,
		     const unsigned char *dst_dev_addr)
{
	dev_addr->dev_type = dev->type;
	memcpy(dev_addr->src_dev_addr, dev->dev_addr, MAX_ADDR_LEN);
	memcpy(dev_addr->broadcast, dev->broadcast, MAX_ADDR_LEN);
	if (dst_dev_addr)
		memcpy(dev_addr->dst_dev_addr, dst_dev_addr, MAX_ADDR_LEN);
	dev_addr->bound_dev_if = dev->ifindex;
	return 0;
}
EXPORT_SYMBOL(rdma_copy_addr);

int rdma_translate_ip(struct sockaddr *addr, struct rdma_dev_addr *dev_addr)
{
	struct net_device *dev;
	int ret = -EADDRNOTAVAIL;

	if (dev_addr->bound_dev_if) {
		dev = dev_get_by_index(&init_net, dev_addr->bound_dev_if);
		if (!dev)
			return -ENODEV;
		ret = rdma_copy_addr(dev_addr, dev, NULL);
		dev_put(dev);
		return ret;
	}

	switch (addr->sa_family) {
	case AF_INET:
		dev = ip_dev_find(&init_net,
			((struct sockaddr_in *) addr)->sin_addr.s_addr);

		if (!dev)
			return ret;

		ret = rdma_copy_addr(dev_addr, dev, NULL);
		dev_put(dev);
		break;

#if IS_ENABLED(CONFIG_IPV6)
	case AF_INET6:
		rcu_read_lock();
		for_each_netdev_rcu(&init_net, dev) {
			if (ipv6_chk_addr(&init_net,
					  &((struct sockaddr_in6 *) addr)->sin6_addr,
					  dev, 1)) {
				ret = rdma_copy_addr(dev_addr, dev, NULL);
				break;
			}
		}
		rcu_read_unlock();
		break;
#endif
	}
	return ret;
}
EXPORT_SYMBOL(rdma_translate_ip);

static void set_timeout(unsigned long time)
{
	unsigned long delay;

	cancel_delayed_work(&work);

	delay = time - jiffies;
	if ((long)delay <= 0)
		delay = 1;

	queue_delayed_work(addr_wq, &work, delay);
}

static void queue_req(struct addr_req *req)
{
	struct addr_req *temp_req;

	mutex_lock(&lock);
	list_for_each_entry_reverse(temp_req, &req_list, list) {
		if (time_after_eq(req->timeout, temp_req->timeout))
			break;
	}

	list_add(&req->list, &temp_req->list);

	if (req_list.next == &req->list)
		set_timeout(req->timeout);
	mutex_unlock(&lock);
}

static int dst_fetch_ha(struct dst_entry *dst, struct rdma_dev_addr *dev_addr, void *daddr)
{
	struct neighbour *n;
	int ret;

	n = dst_neigh_lookup(dst, daddr);

	rcu_read_lock();
	if (!n || !(n->nud_state & NUD_VALID)) {
		if (n)
			neigh_event_send(n, NULL);
		ret = -ENODATA;
	} else {
		ret = rdma_copy_addr(dev_addr, dst->dev, n->ha);
	}
	rcu_read_unlock();

	if (n)
		neigh_release(n);

	return ret;
}

static int addr4_resolve(struct sockaddr_in *src_in,
			 struct sockaddr_in *dst_in,
			 struct rdma_dev_addr *addr)
{
	__be32 src_ip = src_in->sin_addr.s_addr;
	__be32 dst_ip = dst_in->sin_addr.s_addr;
	struct rtable *rt;
	struct flowi4 fl4;
	int ret;

	memset(&fl4, 0, sizeof(fl4));
	fl4.daddr = dst_ip;
	fl4.saddr = src_ip;
	fl4.flowi4_oif = addr->bound_dev_if;
	rt = ip_route_output_key(&init_net, &fl4);
	if (IS_ERR(rt)) {
		ret = PTR_ERR(rt);
		goto out;
	}
	src_in->sin_family = AF_INET;
	src_in->sin_addr.s_addr = fl4.saddr;

	if (rt->dst.dev->flags & IFF_LOOPBACK) {
		ret = rdma_translate_ip((struct sockaddr *) dst_in, addr);
		if (!ret)
			memcpy(addr->dst_dev_addr, addr->src_dev_addr, MAX_ADDR_LEN);
		goto put;
	}

	/* If the device does ARP internally, return 'done' */
	if (rt->dst.dev->flags & IFF_NOARP) {
		ret = rdma_copy_addr(addr, rt->dst.dev, NULL);
		goto put;
	}

	ret = dst_fetch_ha(&rt->dst, addr, &fl4.daddr);
put:
	ip_rt_put(rt);
out:
	return ret;
}

#if IS_ENABLED(CONFIG_IPV6)
static int addr6_resolve(struct sockaddr_in6 *src_in,
			 struct sockaddr_in6 *dst_in,
			 struct rdma_dev_addr *addr)
{
	struct flowi6 fl6;
	struct dst_entry *dst;
	int ret;

	memset(&fl6, 0, sizeof fl6);
	fl6.daddr = dst_in->sin6_addr;
	fl6.saddr = src_in->sin6_addr;
	fl6.flowi6_oif = addr->bound_dev_if;

	dst = ip6_route_output(&init_net, NULL, &fl6);
	if ((ret = dst->error))
		goto put;

	if (ipv6_addr_any(&fl6.saddr)) {
		ret = ipv6_dev_get_saddr(&init_net, ip6_dst_idev(dst)->dev,
					 &fl6.daddr, 0, &fl6.saddr);
		if (ret)
			goto put;

		src_in->sin6_family = AF_INET6;
		src_in->sin6_addr = fl6.saddr;
	}

	if (dst->dev->flags & IFF_LOOPBACK) {
		ret = rdma_translate_ip((struct sockaddr *) dst_in, addr);
		if (!ret)
			memcpy(addr->dst_dev_addr, addr->src_dev_addr, MAX_ADDR_LEN);
		goto put;
	}

	/* If the device does ARP internally, return 'done' */
	if (dst->dev->flags & IFF_NOARP) {
		ret = rdma_copy_addr(addr, dst->dev, NULL);
		goto put;
	}

	ret = dst_fetch_ha(dst, addr, &fl6.daddr);
put:
	dst_release(dst);
	return ret;
}
#else
static int addr6_resolve(struct sockaddr_in6 *src_in,
			 struct sockaddr_in6 *dst_in,
			 struct rdma_dev_addr *addr)
{
	return -EADDRNOTAVAIL;
}
#endif

static int addr_resolve(struct sockaddr *src_in,
			struct sockaddr *dst_in,
			struct rdma_dev_addr *addr)
{
	if (src_in->sa_family == AF_INET) {
		return addr4_resolve((struct sockaddr_in *) src_in,
			(struct sockaddr_in *) dst_in, addr);
	} else
		return addr6_resolve((struct sockaddr_in6 *) src_in,
			(struct sockaddr_in6 *) dst_in, addr);
}

static void process_req(struct work_struct *work)
{
	struct addr_req *req, *temp_req;
	struct sockaddr *src_in, *dst_in;
	struct list_head done_list;

	INIT_LIST_HEAD(&done_list);

	mutex_lock(&lock);
	list_for_each_entry_safe(req, temp_req, &req_list, list) {
		if (req->status == -ENODATA) {
			src_in = (struct sockaddr *) &req->src_addr;
			dst_in = (struct sockaddr *) &req->dst_addr;
			req->status = addr_resolve(src_in, dst_in, req->addr);
			if (req->status && time_after_eq(jiffies, req->timeout))
				req->status = -ETIMEDOUT;
			else if (req->status == -ENODATA)
				continue;
		}
		list_move_tail(&req->list, &done_list);
	}

	if (!list_empty(&req_list)) {
		req = list_entry(req_list.next, struct addr_req, list);
		set_timeout(req->timeout);
	}
	mutex_unlock(&lock);

	list_for_each_entry_safe(req, temp_req, &done_list, list) {
		list_del(&req->list);
		req->callback(req->status, (struct sockaddr *) &req->src_addr,
			req->addr, req->context);
		put_client(req->client);
		kfree(req);
	}
}

int rdma_resolve_ip(struct rdma_addr_client *client,
		    struct sockaddr *src_addr, struct sockaddr *dst_addr,
		    struct rdma_dev_addr *addr, int timeout_ms,
		    void (*callback)(int status, struct sockaddr *src_addr,
				     struct rdma_dev_addr *addr, void *context),
		    void *context)
{
	struct sockaddr *src_in, *dst_in;
	struct addr_req *req;
	int ret = 0;

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

	src_in = (struct sockaddr *) &req->src_addr;
	dst_in = (struct sockaddr *) &req->dst_addr;

	if (src_addr) {
		if (src_addr->sa_family != dst_addr->sa_family) {
			ret = -EINVAL;
			goto err;
		}

		memcpy(src_in, src_addr, ip_addr_size(src_addr));
	} else {
		src_in->sa_family = dst_addr->sa_family;
	}

	memcpy(dst_in, dst_addr, ip_addr_size(dst_addr));
	req->addr = addr;
	req->callback = callback;
	req->context = context;
	req->client = client;
	atomic_inc(&client->refcount);

	req->status = addr_resolve(src_in, dst_in, addr);
	switch (req->status) {
	case 0:
		req->timeout = jiffies;
		queue_req(req);
		break;
	case -ENODATA:
		req->timeout = msecs_to_jiffies(timeout_ms) + jiffies;
		queue_req(req);
		break;
	default:
		ret = req->status;
		atomic_dec(&client->refcount);
		goto err;
	}
	return ret;
err:
	kfree(req);
	return ret;
}
EXPORT_SYMBOL(rdma_resolve_ip);

void rdma_addr_cancel(struct rdma_dev_addr *addr)
{
	struct addr_req *req, *temp_req;

	mutex_lock(&lock);
	list_for_each_entry_safe(req, temp_req, &req_list, list) {
		if (req->addr == addr) {
			req->status = -ECANCELED;
			req->timeout = jiffies;
			list_move(&req->list, &req_list);
			set_timeout(req->timeout);
			break;
		}
	}
	mutex_unlock(&lock);
}
EXPORT_SYMBOL(rdma_addr_cancel);

static int netevent_callback(struct notifier_block *self, unsigned long event,
	void *ctx)
{
	if (event == NETEVENT_NEIGH_UPDATE) {
		struct neighbour *neigh = ctx;

		if (neigh->nud_state & NUD_VALID) {
			set_timeout(jiffies);
		}
	}
	return 0;
}

static struct notifier_block nb = {
	.notifier_call = netevent_callback
};

static int __init addr_init(void)
{
	addr_wq = create_singlethread_workqueue("ib_addr");
	if (!addr_wq)
		return -ENOMEM;

	register_netevent_notifier(&nb);
	return 0;
}

static void __exit addr_cleanup(void)
{
	unregister_netevent_notifier(&nb);
	destroy_workqueue(addr_wq);
}

module_init(addr_init);
module_exit(addr_cleanup);
Commit	Line	Data
7025fcd3 SH	1	/*
	2	* Copyright (c) 2005 Voltaire Inc. All rights reserved.
	3	* Copyright (c) 2002-2005, Network Appliance, Inc. All rights reserved.
	4	* Copyright (c) 1999-2005, Mellanox Technologies, Inc. All rights reserved.
	5	* Copyright (c) 2005 Intel Corporation. All rights reserved.
	6	*
a9474917 SH	7	* This software is available to you under a choice of one of two
	8	* licenses. You may choose to be licensed under the terms of the GNU
	9	* General Public License (GPL) Version 2, available from the file
	10	* COPYING in the main directory of this source tree, or the
	11	* OpenIB.org BSD license below:
7025fcd3	12	*
a9474917 SH	13	* Redistribution and use in source and binary forms, with or
	14	* without modification, are permitted provided that the following
	15	* conditions are met:
7025fcd3	16	*
a9474917 SH	17	* - Redistributions of source code must retain the above
	18	* copyright notice, this list of conditions and the following
	19	* disclaimer.
7025fcd3	20	*
a9474917 SH	21	* - Redistributions in binary form must reproduce the above
	22	* copyright notice, this list of conditions and the following
	23	* disclaimer in the documentation and/or other materials
	24	* provided with the distribution.
7025fcd3	25	*
a9474917 SH	26	* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
	27	* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
	28	* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
	29	* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
	30	* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
	31	* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
	32	* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
	33	* SOFTWARE.
7025fcd3 SH	34	*/
	35
	36	#include <linux/mutex.h>
	37	#include <linux/inetdevice.h>
5a0e3ad6	38	#include <linux/slab.h>
7025fcd3	39	#include <linux/workqueue.h>
e4dd23d7	40	#include <linux/module.h>
7025fcd3 SH	41	#include <net/arp.h>
	42	#include <net/neighbour.h>
	43	#include <net/route.h>
e795d092	44	#include <net/netevent.h>
38617c64 AS	45	#include <net/addrconf.h>
38617c64 AS	46	#include <net/ip6_route.h>
7025fcd3 SH	47	#include <rdma/ib_addr.h>
	48
	49	MODULE_AUTHOR("Sean Hefty");
	50	MODULE_DESCRIPTION("IB Address Translation");
	51	MODULE_LICENSE("Dual BSD/GPL");
	52
	53	struct addr_req {
	54	struct list_head list;
38617c64 AS	55	struct sockaddr_storage src_addr;
38617c64 AS	56	struct sockaddr_storage dst_addr;
7025fcd3	57	struct rdma_dev_addr *addr;
7a118df3	58	struct rdma_addr_client *client;
7025fcd3 SH	59	void *context;
	60	void (callback)(int status, struct sockaddr src_addr,
	61	struct rdma_dev_addr addr, void context);
	62	unsigned long timeout;
	63	int status;
	64	};
	65
c4028958	66	static void process_req(struct work_struct *work);
7025fcd3 SH	67
	68	static DEFINE_MUTEX(lock);
	69	static LIST_HEAD(req_list);
c4028958	70	static DECLARE_DELAYED_WORK(work, process_req);
7025fcd3 SH	71	static struct workqueue_struct *addr_wq;
7025fcd3 SH	72
7a118df3 SH	73	void rdma_addr_register_client(struct rdma_addr_client *client)
	74	{
	75	atomic_set(&client->refcount, 1);
	76	init_completion(&client->comp);
	77	}
	78	EXPORT_SYMBOL(rdma_addr_register_client);
	79
	80	static inline void put_client(struct rdma_addr_client *client)
	81	{
	82	if (atomic_dec_and_test(&client->refcount))
	83	complete(&client->comp);
	84	}
	85
	86	void rdma_addr_unregister_client(struct rdma_addr_client *client)
	87	{
	88	put_client(client);
	89	wait_for_completion(&client->comp);
	90	}
	91	EXPORT_SYMBOL(rdma_addr_unregister_client);
	92
07ebafba TT	93	int rdma_copy_addr(struct rdma_dev_addr dev_addr, struct net_device dev,
07ebafba TT	94	const unsigned char *dst_dev_addr)
7025fcd3	95	{
c4315d85	96	dev_addr->dev_type = dev->type;
7025fcd3 SH	97	memcpy(dev_addr->src_dev_addr, dev->dev_addr, MAX_ADDR_LEN);
	98	memcpy(dev_addr->broadcast, dev->broadcast, MAX_ADDR_LEN);
	99	if (dst_dev_addr)
	100	memcpy(dev_addr->dst_dev_addr, dst_dev_addr, MAX_ADDR_LEN);
6266ed6e	101	dev_addr->bound_dev_if = dev->ifindex;
7025fcd3 SH	102	return 0;
7025fcd3 SH	103	}
07ebafba	104	EXPORT_SYMBOL(rdma_copy_addr);
7025fcd3 SH	105
	106	int rdma_translate_ip(struct sockaddr addr, struct rdma_dev_addr dev_addr)
	107	{
	108	struct net_device *dev;
38617c64	109	int ret = -EADDRNOTAVAIL;
7025fcd3	110
6266ed6e SH	111	if (dev_addr->bound_dev_if) {
	112	dev = dev_get_by_index(&init_net, dev_addr->bound_dev_if);
	113	if (!dev)
	114	return -ENODEV;
	115	ret = rdma_copy_addr(dev_addr, dev, NULL);
	116	dev_put(dev);
	117	return ret;
	118	}
	119
38617c64 AS	120	switch (addr->sa_family) {
	121	case AF_INET:
	122	dev = ip_dev_find(&init_net,
	123	((struct sockaddr_in *) addr)->sin_addr.s_addr);
	124
	125	if (!dev)
	126	return ret;
7025fcd3	127
38617c64 AS	128	ret = rdma_copy_addr(dev_addr, dev, NULL);
	129	dev_put(dev);
	130	break;
2c4ab624	131
d90f9b35	132	#if IS_ENABLED(CONFIG_IPV6)
38617c64	133	case AF_INET6:
22f4fbd9 ED	134	rcu_read_lock();
22f4fbd9 ED	135	for_each_netdev_rcu(&init_net, dev) {
38617c64 AS	136	if (ipv6_chk_addr(&init_net,
	137	&((struct sockaddr_in6 *) addr)->sin6_addr,
	138	dev, 1)) {
	139	ret = rdma_copy_addr(dev_addr, dev, NULL);
	140	break;
	141	}
	142	}
22f4fbd9	143	rcu_read_unlock();
38617c64	144	break;
2c4ab624	145	#endif
38617c64	146	}
7025fcd3 SH	147	return ret;
	148	}
	149	EXPORT_SYMBOL(rdma_translate_ip);
	150
	151	static void set_timeout(unsigned long time)
	152	{
	153	unsigned long delay;
	154
	155	cancel_delayed_work(&work);
	156
	157	delay = time - jiffies;
	158	if ((long)delay <= 0)
	159	delay = 1;
	160
	161	queue_delayed_work(addr_wq, &work, delay);
	162	}
	163
	164	static void queue_req(struct addr_req *req)
	165	{
	166	struct addr_req *temp_req;
	167
	168	mutex_lock(&lock);
	169	list_for_each_entry_reverse(temp_req, &req_list, list) {
f115db48	170	if (time_after_eq(req->timeout, temp_req->timeout))
7025fcd3 SH	171	break;
	172	}
	173
	174	list_add(&req->list, &temp_req->list);
	175
	176	if (req_list.next == &req->list)
	177	set_timeout(req->timeout);
	178	mutex_unlock(&lock);
	179	}
	180
02b61955	181	static int dst_fetch_ha(struct dst_entry dst, struct rdma_dev_addr dev_addr, void *daddr)
51d45974 DM	182	{
	183	struct neighbour *n;
	184	int ret;
	185
02b61955 DM	186	n = dst_neigh_lookup(dst, daddr);
02b61955 DM	187
51d45974	188	rcu_read_lock();
51d45974 DM	189	if (!n \|\| !(n->nud_state & NUD_VALID)) {
	190	if (n)
	191	neigh_event_send(n, NULL);
	192	ret = -ENODATA;
	193	} else {
02b61955	194	ret = rdma_copy_addr(dev_addr, dst->dev, n->ha);
51d45974 DM	195	}
	196	rcu_read_unlock();
	197
02b61955 DM	198	if (n)
	199	neigh_release(n);
	200
51d45974 DM	201	return ret;
	202	}
	203
923c100e SH	204	static int addr4_resolve(struct sockaddr_in *src_in,
	205	struct sockaddr_in *dst_in,
	206	struct rdma_dev_addr *addr)
7025fcd3	207	{
1b90c137 AV	208	__be32 src_ip = src_in->sin_addr.s_addr;
1b90c137 AV	209	__be32 dst_ip = dst_in->sin_addr.s_addr;
7025fcd3	210	struct rtable *rt;
5fc3590c	211	struct flowi4 fl4;
7025fcd3 SH	212	int ret;
7025fcd3 SH	213
5fc3590c DM	214	memset(&fl4, 0, sizeof(fl4));
	215	fl4.daddr = dst_ip;
	216	fl4.saddr = src_ip;
	217	fl4.flowi4_oif = addr->bound_dev_if;
	218	rt = ip_route_output_key(&init_net, &fl4);
b23dd4fe DM	219	if (IS_ERR(rt)) {
b23dd4fe DM	220	ret = PTR_ERR(rt);
7025fcd3	221	goto out;
b23dd4fe	222	}
923c100e	223	src_in->sin_family = AF_INET;
5fc3590c	224	src_in->sin_addr.s_addr = fl4.saddr;
923c100e	225
72cdd1d9	226	if (rt->dst.dev->flags & IFF_LOOPBACK) {
923c100e SH	227	ret = rdma_translate_ip((struct sockaddr *) dst_in, addr);
	228	if (!ret)
	229	memcpy(addr->dst_dev_addr, addr->src_dev_addr, MAX_ADDR_LEN);
	230	goto put;
	231	}
	232
7025fcd3	233	/* If the device does ARP internally, return 'done' */
72cdd1d9	234	if (rt->dst.dev->flags & IFF_NOARP) {
1bdd6384	235	ret = rdma_copy_addr(addr, rt->dst.dev, NULL);
7025fcd3 SH	236	goto put;
	237	}
	238
02b61955	239	ret = dst_fetch_ha(&rt->dst, addr, &fl4.daddr);
7025fcd3 SH	240	put:
	241	ip_rt_put(rt);
	242	out:
	243	return ret;
	244	}
	245
d90f9b35	246	#if IS_ENABLED(CONFIG_IPV6)
d14714df SH	247	static int addr6_resolve(struct sockaddr_in6 *src_in,
	248	struct sockaddr_in6 *dst_in,
	249	struct rdma_dev_addr *addr)
38617c64	250	{
4c9483b2	251	struct flowi6 fl6;
38617c64	252	struct dst_entry *dst;
d14714df	253	int ret;
38617c64	254
4c9483b2	255	memset(&fl6, 0, sizeof fl6);
4e3fd7a0 AD	256	fl6.daddr = dst_in->sin6_addr;
4e3fd7a0 AD	257	fl6.saddr = src_in->sin6_addr;
4c9483b2	258	fl6.flowi6_oif = addr->bound_dev_if;
38617c64	259
4c9483b2	260	dst = ip6_route_output(&init_net, NULL, &fl6);
d14714df SH	261	if ((ret = dst->error))
	262	goto put;
	263
4c9483b2	264	if (ipv6_addr_any(&fl6.saddr)) {
d14714df	265	ret = ipv6_dev_get_saddr(&init_net, ip6_dst_idev(dst)->dev,
4c9483b2	266	&fl6.daddr, 0, &fl6.saddr);
d14714df SH	267	if (ret)
d14714df SH	268	goto put;
38617c64	269
d14714df	270	src_in->sin6_family = AF_INET6;
4e3fd7a0	271	src_in->sin6_addr = fl6.saddr;
d14714df SH	272	}
	273
	274	if (dst->dev->flags & IFF_LOOPBACK) {
	275	ret = rdma_translate_ip((struct sockaddr *) dst_in, addr);
	276	if (!ret)
	277	memcpy(addr->dst_dev_addr, addr->src_dev_addr, MAX_ADDR_LEN);
	278	goto put;
	279	}
	280
	281	/* If the device does ARP internally, return 'done' */
38617c64 AS	282	if (dst->dev->flags & IFF_NOARP) {
38617c64 AS	283	ret = rdma_copy_addr(addr, dst->dev, NULL);
d14714df	284	goto put;
38617c64 AS	285	}
38617c64 AS	286
02b61955	287	ret = dst_fetch_ha(dst, addr, &fl6.daddr);
d14714df	288	put:
38617c64 AS	289	dst_release(dst);
	290	return ret;
	291	}
2c4ab624	292	#else
d14714df SH	293	static int addr6_resolve(struct sockaddr_in6 *src_in,
	294	struct sockaddr_in6 *dst_in,
	295	struct rdma_dev_addr *addr)
2c4ab624 RD	296	{
	297	return -EADDRNOTAVAIL;
	298	}
	299	#endif
38617c64	300
923c100e SH	301	static int addr_resolve(struct sockaddr *src_in,
	302	struct sockaddr *dst_in,
	303	struct rdma_dev_addr *addr)
38617c64 AS	304	{
38617c64 AS	305	if (src_in->sa_family == AF_INET) {
923c100e	306	return addr4_resolve((struct sockaddr_in *) src_in,
38617c64 AS	307	(struct sockaddr_in *) dst_in, addr);
38617c64 AS	308	} else
d14714df	309	return addr6_resolve((struct sockaddr_in6 *) src_in,
38617c64 AS	310	(struct sockaddr_in6 *) dst_in, addr);
	311	}
	312
c4028958	313	static void process_req(struct work_struct *work)
7025fcd3 SH	314	{
7025fcd3 SH	315	struct addr_req req, temp_req;
38617c64	316	struct sockaddr src_in, dst_in;
7025fcd3 SH	317	struct list_head done_list;
	318
	319	INIT_LIST_HEAD(&done_list);
	320
	321	mutex_lock(&lock);
	322	list_for_each_entry_safe(req, temp_req, &req_list, list) {
c78bb844	323	if (req->status == -ENODATA) {
38617c64 AS	324	src_in = (struct sockaddr *) &req->src_addr;
38617c64 AS	325	dst_in = (struct sockaddr *) &req->dst_addr;
923c100e	326	req->status = addr_resolve(src_in, dst_in, req->addr);
c78bb844 KK	327	if (req->status && time_after_eq(jiffies, req->timeout))
	328	req->status = -ETIMEDOUT;
	329	else if (req->status == -ENODATA)
	330	continue;
7025fcd3	331	}
04699a1f	332	list_move_tail(&req->list, &done_list);
7025fcd3 SH	333	}
	334
	335	if (!list_empty(&req_list)) {
	336	req = list_entry(req_list.next, struct addr_req, list);
	337	set_timeout(req->timeout);
	338	}
	339	mutex_unlock(&lock);
	340
	341	list_for_each_entry_safe(req, temp_req, &done_list, list) {
	342	list_del(&req->list);
38617c64 AS	343	req->callback(req->status, (struct sockaddr *) &req->src_addr,
38617c64 AS	344	req->addr, req->context);
7a118df3	345	put_client(req->client);
7025fcd3 SH	346	kfree(req);
	347	}
	348	}
	349
7a118df3 SH	350	int rdma_resolve_ip(struct rdma_addr_client *client,
7a118df3 SH	351	struct sockaddr src_addr, struct sockaddr dst_addr,
7025fcd3 SH	352	struct rdma_dev_addr *addr, int timeout_ms,
	353	void (callback)(int status, struct sockaddr src_addr,
	354	struct rdma_dev_addr addr, void context),
	355	void *context)
	356	{
38617c64	357	struct sockaddr src_in, dst_in;
7025fcd3 SH	358	struct addr_req *req;
	359	int ret = 0;
	360
dd00cc48	361	req = kzalloc(sizeof *req, GFP_KERNEL);
7025fcd3 SH	362	if (!req)
7025fcd3 SH	363	return -ENOMEM;
7025fcd3	364
d2e08862 SH	365	src_in = (struct sockaddr *) &req->src_addr;
	366	dst_in = (struct sockaddr *) &req->dst_addr;
	367
	368	if (src_addr) {
	369	if (src_addr->sa_family != dst_addr->sa_family) {
	370	ret = -EINVAL;
	371	goto err;
	372	}
	373
	374	memcpy(src_in, src_addr, ip_addr_size(src_addr));
	375	} else {
	376	src_in->sa_family = dst_addr->sa_family;
	377	}
	378
	379	memcpy(dst_in, dst_addr, ip_addr_size(dst_addr));
7025fcd3 SH	380	req->addr = addr;
	381	req->callback = callback;
	382	req->context = context;
7a118df3 SH	383	req->client = client;
7a118df3 SH	384	atomic_inc(&client->refcount);
7025fcd3	385
d14714df	386	req->status = addr_resolve(src_in, dst_in, addr);
7025fcd3 SH	387	switch (req->status) {
	388	case 0:
	389	req->timeout = jiffies;
	390	queue_req(req);
	391	break;
	392	case -ENODATA:
	393	req->timeout = msecs_to_jiffies(timeout_ms) + jiffies;
	394	queue_req(req);
7025fcd3 SH	395	break;
	396	default:
	397	ret = req->status;
7a118df3	398	atomic_dec(&client->refcount);
d2e08862	399	goto err;
7025fcd3 SH	400	}
7025fcd3 SH	401	return ret;
d2e08862 SH	402	err:
	403	kfree(req);
	404	return ret;
7025fcd3 SH	405	}
	406	EXPORT_SYMBOL(rdma_resolve_ip);
	407
	408	void rdma_addr_cancel(struct rdma_dev_addr *addr)
	409	{
	410	struct addr_req req, temp_req;
	411
	412	mutex_lock(&lock);
	413	list_for_each_entry_safe(req, temp_req, &req_list, list) {
	414	if (req->addr == addr) {
	415	req->status = -ECANCELED;
	416	req->timeout = jiffies;
04699a1f	417	list_move(&req->list, &req_list);
7025fcd3 SH	418	set_timeout(req->timeout);
	419	break;
	420	}
	421	}
	422	mutex_unlock(&lock);
	423	}
	424	EXPORT_SYMBOL(rdma_addr_cancel);
	425
3cd96564	426	static int netevent_callback(struct notifier_block *self, unsigned long event,
e795d092	427	void *ctx)
7025fcd3	428	{
3cd96564	429	if (event == NETEVENT_NEIGH_UPDATE) {
e795d092	430	struct neighbour *neigh = ctx;
7025fcd3	431
1f126670	432	if (neigh->nud_state & NUD_VALID) {
e795d092 TT	433	set_timeout(jiffies);
	434	}
	435	}
7025fcd3 SH	436	return 0;
	437	}
	438
e795d092 TT	439	static struct notifier_block nb = {
e795d092 TT	440	.notifier_call = netevent_callback
7025fcd3 SH	441	};
7025fcd3 SH	442
716abb1f	443	static int __init addr_init(void)
7025fcd3	444	{
c7f743a6	445	addr_wq = create_singlethread_workqueue("ib_addr");
7025fcd3 SH	446	if (!addr_wq)
	447	return -ENOMEM;
	448
e795d092	449	register_netevent_notifier(&nb);
7025fcd3 SH	450	return 0;
	451	}
	452
716abb1f	453	static void __exit addr_cleanup(void)
7025fcd3	454	{
e795d092	455	unregister_netevent_notifier(&nb);
7025fcd3 SH	456	destroy_workqueue(addr_wq);
	457	}
	458
	459	module_init(addr_init);
	460	module_exit(addr_cleanup);