Linux Audio

Check our new training course

Real-Time Linux with PREEMPT_RT training

Feb 18-20, 2025
Register
Loading...
v5.9
  1/*
  2 * Copyright (c) 2006, 2018 Oracle and/or its affiliates. All rights reserved.
  3 *
  4 * This software is available to you under a choice of one of two
  5 * licenses.  You may choose to be licensed under the terms of the GNU
  6 * General Public License (GPL) Version 2, available from the file
  7 * COPYING in the main directory of this source tree, or the
  8 * OpenIB.org BSD license below:
  9 *
 10 *     Redistribution and use in source and binary forms, with or
 11 *     without modification, are permitted provided that the following
 12 *     conditions are met:
 13 *
 14 *      - Redistributions of source code must retain the above
 15 *        copyright notice, this list of conditions and the following
 16 *        disclaimer.
 17 *
 18 *      - Redistributions in binary form must reproduce the above
 19 *        copyright notice, this list of conditions and the following
 20 *        disclaimer in the documentation and/or other materials
 21 *        provided with the distribution.
 22 *
 23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 30 * SOFTWARE.
 31 *
 32 */
 33#include <linux/kernel.h>
 34#include <linux/list.h>
 35#include <linux/slab.h>
 36#include <linux/export.h>
 37#include <net/ipv6.h>
 38#include <net/inet6_hashtables.h>
 39#include <net/addrconf.h>
 40
 41#include "rds.h"
 42#include "loop.h"
 43
 44#define RDS_CONNECTION_HASH_BITS 12
 45#define RDS_CONNECTION_HASH_ENTRIES (1 << RDS_CONNECTION_HASH_BITS)
 46#define RDS_CONNECTION_HASH_MASK (RDS_CONNECTION_HASH_ENTRIES - 1)
 47
 48/* converting this to RCU is a chore for another day.. */
 49static DEFINE_SPINLOCK(rds_conn_lock);
 50static unsigned long rds_conn_count;
 51static struct hlist_head rds_conn_hash[RDS_CONNECTION_HASH_ENTRIES];
 52static struct kmem_cache *rds_conn_slab;
 53
 54static struct hlist_head *rds_conn_bucket(const struct in6_addr *laddr,
 55					  const struct in6_addr *faddr)
 56{
 57	static u32 rds6_hash_secret __read_mostly;
 58	static u32 rds_hash_secret __read_mostly;
 59
 60	u32 lhash, fhash, hash;
 61
 62	net_get_random_once(&rds_hash_secret, sizeof(rds_hash_secret));
 63	net_get_random_once(&rds6_hash_secret, sizeof(rds6_hash_secret));
 64
 65	lhash = (__force u32)laddr->s6_addr32[3];
 66#if IS_ENABLED(CONFIG_IPV6)
 67	fhash = __ipv6_addr_jhash(faddr, rds6_hash_secret);
 68#else
 69	fhash = (__force u32)faddr->s6_addr32[3];
 70#endif
 71	hash = __inet_ehashfn(lhash, 0, fhash, 0, rds_hash_secret);
 72
 
 
 
 
 73	return &rds_conn_hash[hash & RDS_CONNECTION_HASH_MASK];
 74}
 75
 76#define rds_conn_info_set(var, test, suffix) do {		\
 77	if (test)						\
 78		var |= RDS_INFO_CONNECTION_FLAG_##suffix;	\
 79} while (0)
 80
 81/* rcu read lock must be held or the connection spinlock */
 82static struct rds_connection *rds_conn_lookup(struct net *net,
 83					      struct hlist_head *head,
 84					      const struct in6_addr *laddr,
 85					      const struct in6_addr *faddr,
 86					      struct rds_transport *trans,
 87					      u8 tos, int dev_if)
 88{
 89	struct rds_connection *conn, *ret = NULL;
 90
 91	hlist_for_each_entry_rcu(conn, head, c_hash_node) {
 92		if (ipv6_addr_equal(&conn->c_faddr, faddr) &&
 93		    ipv6_addr_equal(&conn->c_laddr, laddr) &&
 94		    conn->c_trans == trans &&
 95		    conn->c_tos == tos &&
 96		    net == rds_conn_net(conn) &&
 97		    conn->c_dev_if == dev_if) {
 98			ret = conn;
 99			break;
100		}
101	}
102	rdsdebug("returning conn %p for %pI6c -> %pI6c\n", ret,
103		 laddr, faddr);
104	return ret;
105}
106
107/*
108 * This is called by transports as they're bringing down a connection.
109 * It clears partial message state so that the transport can start sending
110 * and receiving over this connection again in the future.  It is up to
111 * the transport to have serialized this call with its send and recv.
112 */
113static void rds_conn_path_reset(struct rds_conn_path *cp)
114{
115	struct rds_connection *conn = cp->cp_conn;
116
117	rdsdebug("connection %pI6c to %pI6c reset\n",
118		 &conn->c_laddr, &conn->c_faddr);
119
120	rds_stats_inc(s_conn_reset);
121	rds_send_path_reset(cp);
122	cp->cp_flags = 0;
123
124	/* Do not clear next_rx_seq here, else we cannot distinguish
125	 * retransmitted packets from new packets, and will hand all
126	 * of them to the application. That is not consistent with the
127	 * reliability guarantees of RDS. */
128}
129
130static void __rds_conn_path_init(struct rds_connection *conn,
131				 struct rds_conn_path *cp, bool is_outgoing)
132{
133	spin_lock_init(&cp->cp_lock);
134	cp->cp_next_tx_seq = 1;
135	init_waitqueue_head(&cp->cp_waitq);
136	INIT_LIST_HEAD(&cp->cp_send_queue);
137	INIT_LIST_HEAD(&cp->cp_retrans);
138
139	cp->cp_conn = conn;
140	atomic_set(&cp->cp_state, RDS_CONN_DOWN);
141	cp->cp_send_gen = 0;
 
 
 
 
 
142	cp->cp_reconnect_jiffies = 0;
143	cp->cp_conn->c_proposed_version = RDS_PROTOCOL_VERSION;
144	INIT_DELAYED_WORK(&cp->cp_send_w, rds_send_worker);
145	INIT_DELAYED_WORK(&cp->cp_recv_w, rds_recv_worker);
146	INIT_DELAYED_WORK(&cp->cp_conn_w, rds_connect_worker);
147	INIT_WORK(&cp->cp_down_w, rds_shutdown_worker);
148	mutex_init(&cp->cp_cm_lock);
149	cp->cp_flags = 0;
150}
151
152/*
153 * There is only every one 'conn' for a given pair of addresses in the
154 * system at a time.  They contain messages to be retransmitted and so
155 * span the lifetime of the actual underlying transport connections.
156 *
157 * For now they are not garbage collected once they're created.  They
158 * are torn down as the module is removed, if ever.
159 */
160static struct rds_connection *__rds_conn_create(struct net *net,
161						const struct in6_addr *laddr,
162						const struct in6_addr *faddr,
163						struct rds_transport *trans,
164						gfp_t gfp, u8 tos,
165						int is_outgoing,
166						int dev_if)
167{
168	struct rds_connection *conn, *parent = NULL;
169	struct hlist_head *head = rds_conn_bucket(laddr, faddr);
170	struct rds_transport *loop_trans;
171	unsigned long flags;
172	int ret, i;
173	int npaths = (trans->t_mp_capable ? RDS_MPATH_WORKERS : 1);
174
175	rcu_read_lock();
176	conn = rds_conn_lookup(net, head, laddr, faddr, trans, tos, dev_if);
177	if (conn &&
178	    conn->c_loopback &&
179	    conn->c_trans != &rds_loop_transport &&
180	    ipv6_addr_equal(laddr, faddr) &&
181	    !is_outgoing) {
182		/* This is a looped back IB connection, and we're
183		 * called by the code handling the incoming connect.
184		 * We need a second connection object into which we
185		 * can stick the other QP. */
186		parent = conn;
187		conn = parent->c_passive;
188	}
189	rcu_read_unlock();
190	if (conn)
191		goto out;
192
193	conn = kmem_cache_zalloc(rds_conn_slab, gfp);
194	if (!conn) {
195		conn = ERR_PTR(-ENOMEM);
196		goto out;
197	}
198	conn->c_path = kcalloc(npaths, sizeof(struct rds_conn_path), gfp);
199	if (!conn->c_path) {
200		kmem_cache_free(rds_conn_slab, conn);
201		conn = ERR_PTR(-ENOMEM);
202		goto out;
203	}
204
205	INIT_HLIST_NODE(&conn->c_hash_node);
206	conn->c_laddr = *laddr;
207	conn->c_isv6 = !ipv6_addr_v4mapped(laddr);
208	conn->c_faddr = *faddr;
209	conn->c_dev_if = dev_if;
210	conn->c_tos = tos;
211
212#if IS_ENABLED(CONFIG_IPV6)
213	/* If the local address is link local, set c_bound_if to be the
214	 * index used for this connection.  Otherwise, set it to 0 as
215	 * the socket is not bound to an interface.  c_bound_if is used
216	 * to look up a socket when a packet is received
217	 */
218	if (ipv6_addr_type(laddr) & IPV6_ADDR_LINKLOCAL)
219		conn->c_bound_if = dev_if;
220	else
221#endif
222		conn->c_bound_if = 0;
223
224	rds_conn_net_set(conn, net);
225
226	ret = rds_cong_get_maps(conn);
227	if (ret) {
228		kfree(conn->c_path);
229		kmem_cache_free(rds_conn_slab, conn);
230		conn = ERR_PTR(ret);
231		goto out;
232	}
233
234	/*
235	 * This is where a connection becomes loopback.  If *any* RDS sockets
236	 * can bind to the destination address then we'd rather the messages
237	 * flow through loopback rather than either transport.
238	 */
239	loop_trans = rds_trans_get_preferred(net, faddr, conn->c_dev_if);
240	if (loop_trans) {
241		rds_trans_put(loop_trans);
242		conn->c_loopback = 1;
243		if (is_outgoing && trans->t_prefer_loopback) {
244			/* "outgoing" connection - and the transport
245			 * says it wants the connection handled by the
246			 * loopback transport. This is what TCP does.
247			 */
248			trans = &rds_loop_transport;
249		}
250	}
251
252	conn->c_trans = trans;
253
254	init_waitqueue_head(&conn->c_hs_waitq);
255	for (i = 0; i < npaths; i++) {
256		__rds_conn_path_init(conn, &conn->c_path[i],
257				     is_outgoing);
258		conn->c_path[i].cp_index = i;
259	}
260	rcu_read_lock();
261	if (rds_destroy_pending(conn))
262		ret = -ENETDOWN;
263	else
264		ret = trans->conn_alloc(conn, GFP_ATOMIC);
265	if (ret) {
266		rcu_read_unlock();
267		kfree(conn->c_path);
268		kmem_cache_free(rds_conn_slab, conn);
269		conn = ERR_PTR(ret);
270		goto out;
271	}
272
273	rdsdebug("allocated conn %p for %pI6c -> %pI6c over %s %s\n",
274		 conn, laddr, faddr,
275		 strnlen(trans->t_name, sizeof(trans->t_name)) ?
276		 trans->t_name : "[unknown]", is_outgoing ? "(outgoing)" : "");
277
278	/*
279	 * Since we ran without holding the conn lock, someone could
280	 * have created the same conn (either normal or passive) in the
281	 * interim. We check while holding the lock. If we won, we complete
282	 * init and return our conn. If we lost, we rollback and return the
283	 * other one.
284	 */
285	spin_lock_irqsave(&rds_conn_lock, flags);
286	if (parent) {
287		/* Creating passive conn */
288		if (parent->c_passive) {
289			trans->conn_free(conn->c_path[0].cp_transport_data);
290			kfree(conn->c_path);
291			kmem_cache_free(rds_conn_slab, conn);
292			conn = parent->c_passive;
293		} else {
294			parent->c_passive = conn;
295			rds_cong_add_conn(conn);
296			rds_conn_count++;
297		}
298	} else {
299		/* Creating normal conn */
300		struct rds_connection *found;
301
302		found = rds_conn_lookup(net, head, laddr, faddr, trans,
303					tos, dev_if);
304		if (found) {
305			struct rds_conn_path *cp;
306			int i;
307
308			for (i = 0; i < npaths; i++) {
309				cp = &conn->c_path[i];
310				/* The ->conn_alloc invocation may have
311				 * allocated resource for all paths, so all
312				 * of them may have to be freed here.
313				 */
314				if (cp->cp_transport_data)
315					trans->conn_free(cp->cp_transport_data);
316			}
317			kfree(conn->c_path);
318			kmem_cache_free(rds_conn_slab, conn);
319			conn = found;
320		} else {
321			conn->c_my_gen_num = rds_gen_num;
322			conn->c_peer_gen_num = 0;
323			hlist_add_head_rcu(&conn->c_hash_node, head);
324			rds_cong_add_conn(conn);
325			rds_conn_count++;
326		}
327	}
328	spin_unlock_irqrestore(&rds_conn_lock, flags);
329	rcu_read_unlock();
330
331out:
332	return conn;
333}
334
335struct rds_connection *rds_conn_create(struct net *net,
336				       const struct in6_addr *laddr,
337				       const struct in6_addr *faddr,
338				       struct rds_transport *trans, u8 tos,
339				       gfp_t gfp, int dev_if)
340{
341	return __rds_conn_create(net, laddr, faddr, trans, gfp, tos, 0, dev_if);
342}
343EXPORT_SYMBOL_GPL(rds_conn_create);
344
345struct rds_connection *rds_conn_create_outgoing(struct net *net,
346						const struct in6_addr *laddr,
347						const struct in6_addr *faddr,
348						struct rds_transport *trans,
349						u8 tos, gfp_t gfp, int dev_if)
350{
351	return __rds_conn_create(net, laddr, faddr, trans, gfp, tos, 1, dev_if);
352}
353EXPORT_SYMBOL_GPL(rds_conn_create_outgoing);
354
355void rds_conn_shutdown(struct rds_conn_path *cp)
356{
357	struct rds_connection *conn = cp->cp_conn;
358
359	/* shut it down unless it's down already */
360	if (!rds_conn_path_transition(cp, RDS_CONN_DOWN, RDS_CONN_DOWN)) {
361		/*
362		 * Quiesce the connection mgmt handlers before we start tearing
363		 * things down. We don't hold the mutex for the entire
364		 * duration of the shutdown operation, else we may be
365		 * deadlocking with the CM handler. Instead, the CM event
366		 * handler is supposed to check for state DISCONNECTING
367		 */
368		mutex_lock(&cp->cp_cm_lock);
369		if (!rds_conn_path_transition(cp, RDS_CONN_UP,
370					      RDS_CONN_DISCONNECTING) &&
371		    !rds_conn_path_transition(cp, RDS_CONN_ERROR,
372					      RDS_CONN_DISCONNECTING)) {
373			rds_conn_path_error(cp,
374					    "shutdown called in state %d\n",
375					    atomic_read(&cp->cp_state));
376			mutex_unlock(&cp->cp_cm_lock);
377			return;
378		}
379		mutex_unlock(&cp->cp_cm_lock);
380
381		wait_event(cp->cp_waitq,
382			   !test_bit(RDS_IN_XMIT, &cp->cp_flags));
383		wait_event(cp->cp_waitq,
384			   !test_bit(RDS_RECV_REFILL, &cp->cp_flags));
385
386		conn->c_trans->conn_path_shutdown(cp);
387		rds_conn_path_reset(cp);
388
389		if (!rds_conn_path_transition(cp, RDS_CONN_DISCONNECTING,
390					      RDS_CONN_DOWN) &&
391		    !rds_conn_path_transition(cp, RDS_CONN_ERROR,
392					      RDS_CONN_DOWN)) {
393			/* This can happen - eg when we're in the middle of tearing
394			 * down the connection, and someone unloads the rds module.
395			 * Quite reproducible with loopback connections.
396			 * Mostly harmless.
397			 *
398			 * Note that this also happens with rds-tcp because
399			 * we could have triggered rds_conn_path_drop in irq
400			 * mode from rds_tcp_state change on the receipt of
401			 * a FIN, thus we need to recheck for RDS_CONN_ERROR
402			 * here.
403			 */
404			rds_conn_path_error(cp, "%s: failed to transition "
405					    "to state DOWN, current state "
406					    "is %d\n", __func__,
407					    atomic_read(&cp->cp_state));
408			return;
409		}
410	}
411
412	/* Then reconnect if it's still live.
413	 * The passive side of an IB loopback connection is never added
414	 * to the conn hash, so we never trigger a reconnect on this
415	 * conn - the reconnect is always triggered by the active peer. */
416	cancel_delayed_work_sync(&cp->cp_conn_w);
417	rcu_read_lock();
418	if (!hlist_unhashed(&conn->c_hash_node)) {
419		rcu_read_unlock();
420		rds_queue_reconnect(cp);
421	} else {
422		rcu_read_unlock();
423	}
424}
425
426/* destroy a single rds_conn_path. rds_conn_destroy() iterates over
427 * all paths using rds_conn_path_destroy()
428 */
429static void rds_conn_path_destroy(struct rds_conn_path *cp)
430{
431	struct rds_message *rm, *rtmp;
432
433	if (!cp->cp_transport_data)
434		return;
435
 
 
 
436	/* make sure lingering queued work won't try to ref the conn */
437	cancel_delayed_work_sync(&cp->cp_send_w);
438	cancel_delayed_work_sync(&cp->cp_recv_w);
439
440	rds_conn_path_drop(cp, true);
441	flush_work(&cp->cp_down_w);
442
443	/* tear down queued messages */
444	list_for_each_entry_safe(rm, rtmp,
445				 &cp->cp_send_queue,
446				 m_conn_item) {
447		list_del_init(&rm->m_conn_item);
448		BUG_ON(!list_empty(&rm->m_sock_item));
449		rds_message_put(rm);
450	}
451	if (cp->cp_xmit_rm)
452		rds_message_put(cp->cp_xmit_rm);
453
454	WARN_ON(delayed_work_pending(&cp->cp_send_w));
455	WARN_ON(delayed_work_pending(&cp->cp_recv_w));
456	WARN_ON(delayed_work_pending(&cp->cp_conn_w));
457	WARN_ON(work_pending(&cp->cp_down_w));
458
459	cp->cp_conn->c_trans->conn_free(cp->cp_transport_data);
460}
461
462/*
463 * Stop and free a connection.
464 *
465 * This can only be used in very limited circumstances.  It assumes that once
466 * the conn has been shutdown that no one else is referencing the connection.
467 * We can only ensure this in the rmmod path in the current code.
468 */
469void rds_conn_destroy(struct rds_connection *conn)
470{
471	unsigned long flags;
472	int i;
473	struct rds_conn_path *cp;
474	int npaths = (conn->c_trans->t_mp_capable ? RDS_MPATH_WORKERS : 1);
475
476	rdsdebug("freeing conn %p for %pI4 -> "
477		 "%pI4\n", conn, &conn->c_laddr,
478		 &conn->c_faddr);
479
480	/* Ensure conn will not be scheduled for reconnect */
481	spin_lock_irq(&rds_conn_lock);
482	hlist_del_init_rcu(&conn->c_hash_node);
483	spin_unlock_irq(&rds_conn_lock);
484	synchronize_rcu();
485
486	/* shut the connection down */
487	for (i = 0; i < npaths; i++) {
488		cp = &conn->c_path[i];
489		rds_conn_path_destroy(cp);
490		BUG_ON(!list_empty(&cp->cp_retrans));
491	}
492
493	/*
494	 * The congestion maps aren't freed up here.  They're
495	 * freed by rds_cong_exit() after all the connections
496	 * have been freed.
497	 */
498	rds_cong_remove_conn(conn);
499
500	kfree(conn->c_path);
501	kmem_cache_free(rds_conn_slab, conn);
502
503	spin_lock_irqsave(&rds_conn_lock, flags);
504	rds_conn_count--;
505	spin_unlock_irqrestore(&rds_conn_lock, flags);
506}
507EXPORT_SYMBOL_GPL(rds_conn_destroy);
508
509static void __rds_inc_msg_cp(struct rds_incoming *inc,
510			     struct rds_info_iterator *iter,
511			     void *saddr, void *daddr, int flip, bool isv6)
512{
513#if IS_ENABLED(CONFIG_IPV6)
514	if (isv6)
515		rds6_inc_info_copy(inc, iter, saddr, daddr, flip);
516	else
517#endif
518		rds_inc_info_copy(inc, iter, *(__be32 *)saddr,
519				  *(__be32 *)daddr, flip);
520}
521
522static void rds_conn_message_info_cmn(struct socket *sock, unsigned int len,
523				      struct rds_info_iterator *iter,
524				      struct rds_info_lengths *lens,
525				      int want_send, bool isv6)
526{
527	struct hlist_head *head;
528	struct list_head *list;
529	struct rds_connection *conn;
530	struct rds_message *rm;
531	unsigned int total = 0;
532	unsigned long flags;
533	size_t i;
534	int j;
535
536	if (isv6)
537		len /= sizeof(struct rds6_info_message);
538	else
539		len /= sizeof(struct rds_info_message);
540
541	rcu_read_lock();
542
543	for (i = 0, head = rds_conn_hash; i < ARRAY_SIZE(rds_conn_hash);
544	     i++, head++) {
545		hlist_for_each_entry_rcu(conn, head, c_hash_node) {
546			struct rds_conn_path *cp;
547			int npaths;
548
549			if (!isv6 && conn->c_isv6)
550				continue;
551
552			npaths = (conn->c_trans->t_mp_capable ?
553				 RDS_MPATH_WORKERS : 1);
554
555			for (j = 0; j < npaths; j++) {
556				cp = &conn->c_path[j];
557				if (want_send)
558					list = &cp->cp_send_queue;
559				else
560					list = &cp->cp_retrans;
561
562				spin_lock_irqsave(&cp->cp_lock, flags);
563
564				/* XXX too lazy to maintain counts.. */
565				list_for_each_entry(rm, list, m_conn_item) {
566					total++;
567					if (total <= len)
568						__rds_inc_msg_cp(&rm->m_inc,
569								 iter,
570								 &conn->c_laddr,
571								 &conn->c_faddr,
572								 0, isv6);
573				}
574
575				spin_unlock_irqrestore(&cp->cp_lock, flags);
 
 
576			}
577		}
578	}
579	rcu_read_unlock();
580
581	lens->nr = total;
582	if (isv6)
583		lens->each = sizeof(struct rds6_info_message);
584	else
585		lens->each = sizeof(struct rds_info_message);
586}
587
588static void rds_conn_message_info(struct socket *sock, unsigned int len,
589				  struct rds_info_iterator *iter,
590				  struct rds_info_lengths *lens,
591				  int want_send)
592{
593	rds_conn_message_info_cmn(sock, len, iter, lens, want_send, false);
594}
595
596#if IS_ENABLED(CONFIG_IPV6)
597static void rds6_conn_message_info(struct socket *sock, unsigned int len,
598				   struct rds_info_iterator *iter,
599				   struct rds_info_lengths *lens,
600				   int want_send)
601{
602	rds_conn_message_info_cmn(sock, len, iter, lens, want_send, true);
603}
604#endif
605
606static void rds_conn_message_info_send(struct socket *sock, unsigned int len,
607				       struct rds_info_iterator *iter,
608				       struct rds_info_lengths *lens)
609{
610	rds_conn_message_info(sock, len, iter, lens, 1);
611}
612
613#if IS_ENABLED(CONFIG_IPV6)
614static void rds6_conn_message_info_send(struct socket *sock, unsigned int len,
615					struct rds_info_iterator *iter,
616					struct rds_info_lengths *lens)
617{
618	rds6_conn_message_info(sock, len, iter, lens, 1);
619}
620#endif
621
622static void rds_conn_message_info_retrans(struct socket *sock,
623					  unsigned int len,
624					  struct rds_info_iterator *iter,
625					  struct rds_info_lengths *lens)
626{
627	rds_conn_message_info(sock, len, iter, lens, 0);
628}
629
630#if IS_ENABLED(CONFIG_IPV6)
631static void rds6_conn_message_info_retrans(struct socket *sock,
632					   unsigned int len,
633					   struct rds_info_iterator *iter,
634					   struct rds_info_lengths *lens)
635{
636	rds6_conn_message_info(sock, len, iter, lens, 0);
637}
638#endif
639
640void rds_for_each_conn_info(struct socket *sock, unsigned int len,
641			  struct rds_info_iterator *iter,
642			  struct rds_info_lengths *lens,
643			  int (*visitor)(struct rds_connection *, void *),
644			  u64 *buffer,
645			  size_t item_len)
646{
 
647	struct hlist_head *head;
648	struct rds_connection *conn;
649	size_t i;
650
651	rcu_read_lock();
652
653	lens->nr = 0;
654	lens->each = item_len;
655
656	for (i = 0, head = rds_conn_hash; i < ARRAY_SIZE(rds_conn_hash);
657	     i++, head++) {
658		hlist_for_each_entry_rcu(conn, head, c_hash_node) {
659
660			/* XXX no c_lock usage.. */
661			if (!visitor(conn, buffer))
662				continue;
663
664			/* We copy as much as we can fit in the buffer,
665			 * but we count all items so that the caller
666			 * can resize the buffer. */
667			if (len >= item_len) {
668				rds_info_copy(iter, buffer, item_len);
669				len -= item_len;
670			}
671			lens->nr++;
672		}
673	}
674	rcu_read_unlock();
675}
676EXPORT_SYMBOL_GPL(rds_for_each_conn_info);
677
678static void rds_walk_conn_path_info(struct socket *sock, unsigned int len,
679				    struct rds_info_iterator *iter,
680				    struct rds_info_lengths *lens,
681				    int (*visitor)(struct rds_conn_path *, void *),
682				    u64 *buffer,
683				    size_t item_len)
684{
 
685	struct hlist_head *head;
686	struct rds_connection *conn;
687	size_t i;
 
688
689	rcu_read_lock();
690
691	lens->nr = 0;
692	lens->each = item_len;
693
694	for (i = 0, head = rds_conn_hash; i < ARRAY_SIZE(rds_conn_hash);
695	     i++, head++) {
696		hlist_for_each_entry_rcu(conn, head, c_hash_node) {
697			struct rds_conn_path *cp;
698
699			/* XXX We only copy the information from the first
700			 * path for now.  The problem is that if there are
701			 * more than one underlying paths, we cannot report
702			 * information of all of them using the existing
703			 * API.  For example, there is only one next_tx_seq,
704			 * which path's next_tx_seq should we report?  It is
705			 * a bug in the design of MPRDS.
706			 */
707			cp = conn->c_path;
708
709			/* XXX no cp_lock usage.. */
710			if (!visitor(cp, buffer))
711				continue;
 
 
 
712
713			/* We copy as much as we can fit in the buffer,
714			 * but we count all items so that the caller
715			 * can resize the buffer.
716			 */
717			if (len >= item_len) {
718				rds_info_copy(iter, buffer, item_len);
719				len -= item_len;
720			}
721			lens->nr++;
722		}
723	}
724	rcu_read_unlock();
725}
726
727static int rds_conn_info_visitor(struct rds_conn_path *cp, void *buffer)
728{
729	struct rds_info_connection *cinfo = buffer;
730	struct rds_connection *conn = cp->cp_conn;
731
732	if (conn->c_isv6)
733		return 0;
734
735	cinfo->next_tx_seq = cp->cp_next_tx_seq;
736	cinfo->next_rx_seq = cp->cp_next_rx_seq;
737	cinfo->laddr = conn->c_laddr.s6_addr32[3];
738	cinfo->faddr = conn->c_faddr.s6_addr32[3];
739	cinfo->tos = conn->c_tos;
740	strncpy(cinfo->transport, conn->c_trans->t_name,
741		sizeof(cinfo->transport));
742	cinfo->flags = 0;
743
744	rds_conn_info_set(cinfo->flags, test_bit(RDS_IN_XMIT, &cp->cp_flags),
745			  SENDING);
746	/* XXX Future: return the state rather than these funky bits */
747	rds_conn_info_set(cinfo->flags,
748			  atomic_read(&cp->cp_state) == RDS_CONN_CONNECTING,
749			  CONNECTING);
750	rds_conn_info_set(cinfo->flags,
751			  atomic_read(&cp->cp_state) == RDS_CONN_UP,
752			  CONNECTED);
753	return 1;
754}
755
756#if IS_ENABLED(CONFIG_IPV6)
757static int rds6_conn_info_visitor(struct rds_conn_path *cp, void *buffer)
758{
759	struct rds6_info_connection *cinfo6 = buffer;
760	struct rds_connection *conn = cp->cp_conn;
761
762	cinfo6->next_tx_seq = cp->cp_next_tx_seq;
763	cinfo6->next_rx_seq = cp->cp_next_rx_seq;
764	cinfo6->laddr = conn->c_laddr;
765	cinfo6->faddr = conn->c_faddr;
766	strncpy(cinfo6->transport, conn->c_trans->t_name,
767		sizeof(cinfo6->transport));
768	cinfo6->flags = 0;
769
770	rds_conn_info_set(cinfo6->flags, test_bit(RDS_IN_XMIT, &cp->cp_flags),
771			  SENDING);
772	/* XXX Future: return the state rather than these funky bits */
773	rds_conn_info_set(cinfo6->flags,
774			  atomic_read(&cp->cp_state) == RDS_CONN_CONNECTING,
775			  CONNECTING);
776	rds_conn_info_set(cinfo6->flags,
777			  atomic_read(&cp->cp_state) == RDS_CONN_UP,
778			  CONNECTED);
779	/* Just return 1 as there is no error case. This is a helper function
780	 * for rds_walk_conn_path_info() and it wants a return value.
781	 */
782	return 1;
783}
784#endif
785
786static void rds_conn_info(struct socket *sock, unsigned int len,
787			  struct rds_info_iterator *iter,
788			  struct rds_info_lengths *lens)
789{
790	u64 buffer[(sizeof(struct rds_info_connection) + 7) / 8];
791
792	rds_walk_conn_path_info(sock, len, iter, lens,
793				rds_conn_info_visitor,
794				buffer,
795				sizeof(struct rds_info_connection));
796}
797
798#if IS_ENABLED(CONFIG_IPV6)
799static void rds6_conn_info(struct socket *sock, unsigned int len,
800			   struct rds_info_iterator *iter,
801			   struct rds_info_lengths *lens)
802{
803	u64 buffer[(sizeof(struct rds6_info_connection) + 7) / 8];
804
805	rds_walk_conn_path_info(sock, len, iter, lens,
806				rds6_conn_info_visitor,
807				buffer,
808				sizeof(struct rds6_info_connection));
809}
810#endif
811
812int rds_conn_init(void)
813{
814	int ret;
815
816	ret = rds_loop_net_init(); /* register pernet callback */
817	if (ret)
818		return ret;
819
820	rds_conn_slab = kmem_cache_create("rds_connection",
821					  sizeof(struct rds_connection),
822					  0, 0, NULL);
823	if (!rds_conn_slab) {
824		rds_loop_net_exit();
825		return -ENOMEM;
826	}
827
828	rds_info_register_func(RDS_INFO_CONNECTIONS, rds_conn_info);
829	rds_info_register_func(RDS_INFO_SEND_MESSAGES,
830			       rds_conn_message_info_send);
831	rds_info_register_func(RDS_INFO_RETRANS_MESSAGES,
832			       rds_conn_message_info_retrans);
833#if IS_ENABLED(CONFIG_IPV6)
834	rds_info_register_func(RDS6_INFO_CONNECTIONS, rds6_conn_info);
835	rds_info_register_func(RDS6_INFO_SEND_MESSAGES,
836			       rds6_conn_message_info_send);
837	rds_info_register_func(RDS6_INFO_RETRANS_MESSAGES,
838			       rds6_conn_message_info_retrans);
839#endif
840	return 0;
841}
842
843void rds_conn_exit(void)
844{
845	rds_loop_net_exit(); /* unregister pernet callback */
846	rds_loop_exit();
847
848	WARN_ON(!hlist_empty(rds_conn_hash));
849
850	kmem_cache_destroy(rds_conn_slab);
851
852	rds_info_deregister_func(RDS_INFO_CONNECTIONS, rds_conn_info);
853	rds_info_deregister_func(RDS_INFO_SEND_MESSAGES,
854				 rds_conn_message_info_send);
855	rds_info_deregister_func(RDS_INFO_RETRANS_MESSAGES,
856				 rds_conn_message_info_retrans);
857#if IS_ENABLED(CONFIG_IPV6)
858	rds_info_deregister_func(RDS6_INFO_CONNECTIONS, rds6_conn_info);
859	rds_info_deregister_func(RDS6_INFO_SEND_MESSAGES,
860				 rds6_conn_message_info_send);
861	rds_info_deregister_func(RDS6_INFO_RETRANS_MESSAGES,
862				 rds6_conn_message_info_retrans);
863#endif
864}
865
866/*
867 * Force a disconnect
868 */
869void rds_conn_path_drop(struct rds_conn_path *cp, bool destroy)
870{
871	atomic_set(&cp->cp_state, RDS_CONN_ERROR);
872
873	rcu_read_lock();
874	if (!destroy && rds_destroy_pending(cp->cp_conn)) {
875		rcu_read_unlock();
876		return;
877	}
878	queue_work(rds_wq, &cp->cp_down_w);
879	rcu_read_unlock();
880}
881EXPORT_SYMBOL_GPL(rds_conn_path_drop);
882
883void rds_conn_drop(struct rds_connection *conn)
884{
885	WARN_ON(conn->c_trans->t_mp_capable);
886	rds_conn_path_drop(&conn->c_path[0], false);
887}
888EXPORT_SYMBOL_GPL(rds_conn_drop);
889
890/*
891 * If the connection is down, trigger a connect. We may have scheduled a
892 * delayed reconnect however - in this case we should not interfere.
893 */
894void rds_conn_path_connect_if_down(struct rds_conn_path *cp)
895{
896	rcu_read_lock();
897	if (rds_destroy_pending(cp->cp_conn)) {
898		rcu_read_unlock();
899		return;
900	}
901	if (rds_conn_path_state(cp) == RDS_CONN_DOWN &&
902	    !test_and_set_bit(RDS_RECONNECT_PENDING, &cp->cp_flags))
903		queue_delayed_work(rds_wq, &cp->cp_conn_w, 0);
904	rcu_read_unlock();
905}
906EXPORT_SYMBOL_GPL(rds_conn_path_connect_if_down);
907
908/* Check connectivity of all paths
909 */
910void rds_check_all_paths(struct rds_connection *conn)
911{
912	int i = 0;
913
914	do {
915		rds_conn_path_connect_if_down(&conn->c_path[i]);
916	} while (++i < conn->c_npaths);
917}
918
919void rds_conn_connect_if_down(struct rds_connection *conn)
920{
921	WARN_ON(conn->c_trans->t_mp_capable);
922	rds_conn_path_connect_if_down(&conn->c_path[0]);
923}
924EXPORT_SYMBOL_GPL(rds_conn_connect_if_down);
925
926void
927__rds_conn_path_error(struct rds_conn_path *cp, const char *fmt, ...)
928{
929	va_list ap;
930
931	va_start(ap, fmt);
932	vprintk(fmt, ap);
933	va_end(ap);
934
935	rds_conn_path_drop(cp, false);
936}
v4.10.11
  1/*
  2 * Copyright (c) 2006 Oracle.  All rights reserved.
  3 *
  4 * This software is available to you under a choice of one of two
  5 * licenses.  You may choose to be licensed under the terms of the GNU
  6 * General Public License (GPL) Version 2, available from the file
  7 * COPYING in the main directory of this source tree, or the
  8 * OpenIB.org BSD license below:
  9 *
 10 *     Redistribution and use in source and binary forms, with or
 11 *     without modification, are permitted provided that the following
 12 *     conditions are met:
 13 *
 14 *      - Redistributions of source code must retain the above
 15 *        copyright notice, this list of conditions and the following
 16 *        disclaimer.
 17 *
 18 *      - Redistributions in binary form must reproduce the above
 19 *        copyright notice, this list of conditions and the following
 20 *        disclaimer in the documentation and/or other materials
 21 *        provided with the distribution.
 22 *
 23 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 24 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 25 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 26 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 27 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 28 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 29 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 30 * SOFTWARE.
 31 *
 32 */
 33#include <linux/kernel.h>
 34#include <linux/list.h>
 35#include <linux/slab.h>
 36#include <linux/export.h>
 37#include <net/inet_hashtables.h>
 
 
 38
 39#include "rds.h"
 40#include "loop.h"
 41
 42#define RDS_CONNECTION_HASH_BITS 12
 43#define RDS_CONNECTION_HASH_ENTRIES (1 << RDS_CONNECTION_HASH_BITS)
 44#define RDS_CONNECTION_HASH_MASK (RDS_CONNECTION_HASH_ENTRIES - 1)
 45
 46/* converting this to RCU is a chore for another day.. */
 47static DEFINE_SPINLOCK(rds_conn_lock);
 48static unsigned long rds_conn_count;
 49static struct hlist_head rds_conn_hash[RDS_CONNECTION_HASH_ENTRIES];
 50static struct kmem_cache *rds_conn_slab;
 51
 52static struct hlist_head *rds_conn_bucket(__be32 laddr, __be32 faddr)
 
 53{
 
 54	static u32 rds_hash_secret __read_mostly;
 55
 56	unsigned long hash;
 57
 58	net_get_random_once(&rds_hash_secret, sizeof(rds_hash_secret));
 
 
 
 
 
 
 
 
 
 59
 60	/* Pass NULL, don't need struct net for hash */
 61	hash = __inet_ehashfn(be32_to_cpu(laddr), 0,
 62			      be32_to_cpu(faddr), 0,
 63			      rds_hash_secret);
 64	return &rds_conn_hash[hash & RDS_CONNECTION_HASH_MASK];
 65}
 66
 67#define rds_conn_info_set(var, test, suffix) do {		\
 68	if (test)						\
 69		var |= RDS_INFO_CONNECTION_FLAG_##suffix;	\
 70} while (0)
 71
 72/* rcu read lock must be held or the connection spinlock */
 73static struct rds_connection *rds_conn_lookup(struct net *net,
 74					      struct hlist_head *head,
 75					      __be32 laddr, __be32 faddr,
 76					      struct rds_transport *trans)
 
 
 77{
 78	struct rds_connection *conn, *ret = NULL;
 79
 80	hlist_for_each_entry_rcu(conn, head, c_hash_node) {
 81		if (conn->c_faddr == faddr && conn->c_laddr == laddr &&
 82		    conn->c_trans == trans && net == rds_conn_net(conn)) {
 
 
 
 
 83			ret = conn;
 84			break;
 85		}
 86	}
 87	rdsdebug("returning conn %p for %pI4 -> %pI4\n", ret,
 88		 &laddr, &faddr);
 89	return ret;
 90}
 91
 92/*
 93 * This is called by transports as they're bringing down a connection.
 94 * It clears partial message state so that the transport can start sending
 95 * and receiving over this connection again in the future.  It is up to
 96 * the transport to have serialized this call with its send and recv.
 97 */
 98static void rds_conn_path_reset(struct rds_conn_path *cp)
 99{
100	struct rds_connection *conn = cp->cp_conn;
101
102	rdsdebug("connection %pI4 to %pI4 reset\n",
103	  &conn->c_laddr, &conn->c_faddr);
104
105	rds_stats_inc(s_conn_reset);
106	rds_send_path_reset(cp);
107	cp->cp_flags = 0;
108
109	/* Do not clear next_rx_seq here, else we cannot distinguish
110	 * retransmitted packets from new packets, and will hand all
111	 * of them to the application. That is not consistent with the
112	 * reliability guarantees of RDS. */
113}
114
115static void __rds_conn_path_init(struct rds_connection *conn,
116				 struct rds_conn_path *cp, bool is_outgoing)
117{
118	spin_lock_init(&cp->cp_lock);
119	cp->cp_next_tx_seq = 1;
120	init_waitqueue_head(&cp->cp_waitq);
121	INIT_LIST_HEAD(&cp->cp_send_queue);
122	INIT_LIST_HEAD(&cp->cp_retrans);
123
124	cp->cp_conn = conn;
125	atomic_set(&cp->cp_state, RDS_CONN_DOWN);
126	cp->cp_send_gen = 0;
127	/* cp_outgoing is per-path. So we can only set it here
128	 * for the single-path transports.
129	 */
130	if (!conn->c_trans->t_mp_capable)
131		cp->cp_outgoing = (is_outgoing ? 1 : 0);
132	cp->cp_reconnect_jiffies = 0;
 
133	INIT_DELAYED_WORK(&cp->cp_send_w, rds_send_worker);
134	INIT_DELAYED_WORK(&cp->cp_recv_w, rds_recv_worker);
135	INIT_DELAYED_WORK(&cp->cp_conn_w, rds_connect_worker);
136	INIT_WORK(&cp->cp_down_w, rds_shutdown_worker);
137	mutex_init(&cp->cp_cm_lock);
138	cp->cp_flags = 0;
139}
140
141/*
142 * There is only every one 'conn' for a given pair of addresses in the
143 * system at a time.  They contain messages to be retransmitted and so
144 * span the lifetime of the actual underlying transport connections.
145 *
146 * For now they are not garbage collected once they're created.  They
147 * are torn down as the module is removed, if ever.
148 */
149static struct rds_connection *__rds_conn_create(struct net *net,
150						__be32 laddr, __be32 faddr,
151				       struct rds_transport *trans, gfp_t gfp,
152				       int is_outgoing)
 
 
 
153{
154	struct rds_connection *conn, *parent = NULL;
155	struct hlist_head *head = rds_conn_bucket(laddr, faddr);
156	struct rds_transport *loop_trans;
157	unsigned long flags;
158	int ret, i;
 
159
160	rcu_read_lock();
161	conn = rds_conn_lookup(net, head, laddr, faddr, trans);
162	if (conn && conn->c_loopback && conn->c_trans != &rds_loop_transport &&
163	    laddr == faddr && !is_outgoing) {
 
 
 
164		/* This is a looped back IB connection, and we're
165		 * called by the code handling the incoming connect.
166		 * We need a second connection object into which we
167		 * can stick the other QP. */
168		parent = conn;
169		conn = parent->c_passive;
170	}
171	rcu_read_unlock();
172	if (conn)
173		goto out;
174
175	conn = kmem_cache_zalloc(rds_conn_slab, gfp);
176	if (!conn) {
177		conn = ERR_PTR(-ENOMEM);
178		goto out;
179	}
 
 
 
 
 
 
180
181	INIT_HLIST_NODE(&conn->c_hash_node);
182	conn->c_laddr = laddr;
183	conn->c_faddr = faddr;
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
184
185	rds_conn_net_set(conn, net);
186
187	ret = rds_cong_get_maps(conn);
188	if (ret) {
 
189		kmem_cache_free(rds_conn_slab, conn);
190		conn = ERR_PTR(ret);
191		goto out;
192	}
193
194	/*
195	 * This is where a connection becomes loopback.  If *any* RDS sockets
196	 * can bind to the destination address then we'd rather the messages
197	 * flow through loopback rather than either transport.
198	 */
199	loop_trans = rds_trans_get_preferred(net, faddr);
200	if (loop_trans) {
201		rds_trans_put(loop_trans);
202		conn->c_loopback = 1;
203		if (is_outgoing && trans->t_prefer_loopback) {
204			/* "outgoing" connection - and the transport
205			 * says it wants the connection handled by the
206			 * loopback transport. This is what TCP does.
207			 */
208			trans = &rds_loop_transport;
209		}
210	}
211
212	conn->c_trans = trans;
213
214	init_waitqueue_head(&conn->c_hs_waitq);
215	for (i = 0; i < RDS_MPATH_WORKERS; i++) {
216		__rds_conn_path_init(conn, &conn->c_path[i],
217				     is_outgoing);
218		conn->c_path[i].cp_index = i;
219	}
220	ret = trans->conn_alloc(conn, gfp);
 
 
 
 
221	if (ret) {
 
 
222		kmem_cache_free(rds_conn_slab, conn);
223		conn = ERR_PTR(ret);
224		goto out;
225	}
226
227	rdsdebug("allocated conn %p for %pI4 -> %pI4 over %s %s\n",
228	  conn, &laddr, &faddr,
229	  trans->t_name ? trans->t_name : "[unknown]",
230	  is_outgoing ? "(outgoing)" : "");
231
232	/*
233	 * Since we ran without holding the conn lock, someone could
234	 * have created the same conn (either normal or passive) in the
235	 * interim. We check while holding the lock. If we won, we complete
236	 * init and return our conn. If we lost, we rollback and return the
237	 * other one.
238	 */
239	spin_lock_irqsave(&rds_conn_lock, flags);
240	if (parent) {
241		/* Creating passive conn */
242		if (parent->c_passive) {
243			trans->conn_free(conn->c_path[0].cp_transport_data);
 
244			kmem_cache_free(rds_conn_slab, conn);
245			conn = parent->c_passive;
246		} else {
247			parent->c_passive = conn;
248			rds_cong_add_conn(conn);
249			rds_conn_count++;
250		}
251	} else {
252		/* Creating normal conn */
253		struct rds_connection *found;
254
255		found = rds_conn_lookup(net, head, laddr, faddr, trans);
 
256		if (found) {
257			struct rds_conn_path *cp;
258			int i;
259
260			for (i = 0; i < RDS_MPATH_WORKERS; i++) {
261				cp = &conn->c_path[i];
262				/* The ->conn_alloc invocation may have
263				 * allocated resource for all paths, so all
264				 * of them may have to be freed here.
265				 */
266				if (cp->cp_transport_data)
267					trans->conn_free(cp->cp_transport_data);
268			}
 
269			kmem_cache_free(rds_conn_slab, conn);
270			conn = found;
271		} else {
272			conn->c_my_gen_num = rds_gen_num;
273			conn->c_peer_gen_num = 0;
274			hlist_add_head_rcu(&conn->c_hash_node, head);
275			rds_cong_add_conn(conn);
276			rds_conn_count++;
277		}
278	}
279	spin_unlock_irqrestore(&rds_conn_lock, flags);
 
280
281out:
282	return conn;
283}
284
285struct rds_connection *rds_conn_create(struct net *net,
286				       __be32 laddr, __be32 faddr,
287				       struct rds_transport *trans, gfp_t gfp)
 
 
288{
289	return __rds_conn_create(net, laddr, faddr, trans, gfp, 0);
290}
291EXPORT_SYMBOL_GPL(rds_conn_create);
292
293struct rds_connection *rds_conn_create_outgoing(struct net *net,
294						__be32 laddr, __be32 faddr,
295				       struct rds_transport *trans, gfp_t gfp)
 
 
296{
297	return __rds_conn_create(net, laddr, faddr, trans, gfp, 1);
298}
299EXPORT_SYMBOL_GPL(rds_conn_create_outgoing);
300
301void rds_conn_shutdown(struct rds_conn_path *cp)
302{
303	struct rds_connection *conn = cp->cp_conn;
304
305	/* shut it down unless it's down already */
306	if (!rds_conn_path_transition(cp, RDS_CONN_DOWN, RDS_CONN_DOWN)) {
307		/*
308		 * Quiesce the connection mgmt handlers before we start tearing
309		 * things down. We don't hold the mutex for the entire
310		 * duration of the shutdown operation, else we may be
311		 * deadlocking with the CM handler. Instead, the CM event
312		 * handler is supposed to check for state DISCONNECTING
313		 */
314		mutex_lock(&cp->cp_cm_lock);
315		if (!rds_conn_path_transition(cp, RDS_CONN_UP,
316					      RDS_CONN_DISCONNECTING) &&
317		    !rds_conn_path_transition(cp, RDS_CONN_ERROR,
318					      RDS_CONN_DISCONNECTING)) {
319			rds_conn_path_error(cp,
320					    "shutdown called in state %d\n",
321					    atomic_read(&cp->cp_state));
322			mutex_unlock(&cp->cp_cm_lock);
323			return;
324		}
325		mutex_unlock(&cp->cp_cm_lock);
326
327		wait_event(cp->cp_waitq,
328			   !test_bit(RDS_IN_XMIT, &cp->cp_flags));
329		wait_event(cp->cp_waitq,
330			   !test_bit(RDS_RECV_REFILL, &cp->cp_flags));
331
332		conn->c_trans->conn_path_shutdown(cp);
333		rds_conn_path_reset(cp);
334
335		if (!rds_conn_path_transition(cp, RDS_CONN_DISCONNECTING,
 
 
336					      RDS_CONN_DOWN)) {
337			/* This can happen - eg when we're in the middle of tearing
338			 * down the connection, and someone unloads the rds module.
339			 * Quite reproduceable with loopback connections.
340			 * Mostly harmless.
 
 
 
 
 
 
341			 */
342			rds_conn_path_error(cp, "%s: failed to transition "
343					    "to state DOWN, current state "
344					    "is %d\n", __func__,
345					    atomic_read(&cp->cp_state));
346			return;
347		}
348	}
349
350	/* Then reconnect if it's still live.
351	 * The passive side of an IB loopback connection is never added
352	 * to the conn hash, so we never trigger a reconnect on this
353	 * conn - the reconnect is always triggered by the active peer. */
354	cancel_delayed_work_sync(&cp->cp_conn_w);
355	rcu_read_lock();
356	if (!hlist_unhashed(&conn->c_hash_node)) {
357		rcu_read_unlock();
358		rds_queue_reconnect(cp);
359	} else {
360		rcu_read_unlock();
361	}
362}
363
364/* destroy a single rds_conn_path. rds_conn_destroy() iterates over
365 * all paths using rds_conn_path_destroy()
366 */
367static void rds_conn_path_destroy(struct rds_conn_path *cp)
368{
369	struct rds_message *rm, *rtmp;
370
371	if (!cp->cp_transport_data)
372		return;
373
374	rds_conn_path_drop(cp);
375	flush_work(&cp->cp_down_w);
376
377	/* make sure lingering queued work won't try to ref the conn */
378	cancel_delayed_work_sync(&cp->cp_send_w);
379	cancel_delayed_work_sync(&cp->cp_recv_w);
380
 
 
 
381	/* tear down queued messages */
382	list_for_each_entry_safe(rm, rtmp,
383				 &cp->cp_send_queue,
384				 m_conn_item) {
385		list_del_init(&rm->m_conn_item);
386		BUG_ON(!list_empty(&rm->m_sock_item));
387		rds_message_put(rm);
388	}
389	if (cp->cp_xmit_rm)
390		rds_message_put(cp->cp_xmit_rm);
391
 
 
 
 
 
392	cp->cp_conn->c_trans->conn_free(cp->cp_transport_data);
393}
394
395/*
396 * Stop and free a connection.
397 *
398 * This can only be used in very limited circumstances.  It assumes that once
399 * the conn has been shutdown that no one else is referencing the connection.
400 * We can only ensure this in the rmmod path in the current code.
401 */
402void rds_conn_destroy(struct rds_connection *conn)
403{
404	unsigned long flags;
405	int i;
406	struct rds_conn_path *cp;
 
407
408	rdsdebug("freeing conn %p for %pI4 -> "
409		 "%pI4\n", conn, &conn->c_laddr,
410		 &conn->c_faddr);
411
412	/* Ensure conn will not be scheduled for reconnect */
413	spin_lock_irq(&rds_conn_lock);
414	hlist_del_init_rcu(&conn->c_hash_node);
415	spin_unlock_irq(&rds_conn_lock);
416	synchronize_rcu();
417
418	/* shut the connection down */
419	for (i = 0; i < RDS_MPATH_WORKERS; i++) {
420		cp = &conn->c_path[i];
421		rds_conn_path_destroy(cp);
422		BUG_ON(!list_empty(&cp->cp_retrans));
423	}
424
425	/*
426	 * The congestion maps aren't freed up here.  They're
427	 * freed by rds_cong_exit() after all the connections
428	 * have been freed.
429	 */
430	rds_cong_remove_conn(conn);
431
 
432	kmem_cache_free(rds_conn_slab, conn);
433
434	spin_lock_irqsave(&rds_conn_lock, flags);
435	rds_conn_count--;
436	spin_unlock_irqrestore(&rds_conn_lock, flags);
437}
438EXPORT_SYMBOL_GPL(rds_conn_destroy);
439
440static void rds_conn_message_info(struct socket *sock, unsigned int len,
441				  struct rds_info_iterator *iter,
442				  struct rds_info_lengths *lens,
443				  int want_send)
 
 
 
 
 
 
 
 
 
 
 
 
 
444{
445	struct hlist_head *head;
446	struct list_head *list;
447	struct rds_connection *conn;
448	struct rds_message *rm;
449	unsigned int total = 0;
450	unsigned long flags;
451	size_t i;
452	int j;
453
454	len /= sizeof(struct rds_info_message);
 
 
 
455
456	rcu_read_lock();
457
458	for (i = 0, head = rds_conn_hash; i < ARRAY_SIZE(rds_conn_hash);
459	     i++, head++) {
460		hlist_for_each_entry_rcu(conn, head, c_hash_node) {
461			struct rds_conn_path *cp;
 
 
 
 
462
463			for (j = 0; j < RDS_MPATH_WORKERS; j++) {
 
 
 
464				cp = &conn->c_path[j];
465				if (want_send)
466					list = &cp->cp_send_queue;
467				else
468					list = &cp->cp_retrans;
469
470				spin_lock_irqsave(&cp->cp_lock, flags);
471
472				/* XXX too lazy to maintain counts.. */
473				list_for_each_entry(rm, list, m_conn_item) {
474					total++;
475					if (total <= len)
476						rds_inc_info_copy(&rm->m_inc,
477								  iter,
478								  conn->c_laddr,
479								  conn->c_faddr,
480								  0);
481				}
482
483				spin_unlock_irqrestore(&cp->cp_lock, flags);
484				if (!conn->c_trans->t_mp_capable)
485					break;
486			}
487		}
488	}
489	rcu_read_unlock();
490
491	lens->nr = total;
492	lens->each = sizeof(struct rds_info_message);
 
 
 
 
 
 
 
 
 
 
 
493}
494
 
 
 
 
 
 
 
 
 
 
495static void rds_conn_message_info_send(struct socket *sock, unsigned int len,
496				       struct rds_info_iterator *iter,
497				       struct rds_info_lengths *lens)
498{
499	rds_conn_message_info(sock, len, iter, lens, 1);
500}
501
 
 
 
 
 
 
 
 
 
502static void rds_conn_message_info_retrans(struct socket *sock,
503					  unsigned int len,
504					  struct rds_info_iterator *iter,
505					  struct rds_info_lengths *lens)
506{
507	rds_conn_message_info(sock, len, iter, lens, 0);
508}
509
 
 
 
 
 
 
 
 
 
 
510void rds_for_each_conn_info(struct socket *sock, unsigned int len,
511			  struct rds_info_iterator *iter,
512			  struct rds_info_lengths *lens,
513			  int (*visitor)(struct rds_connection *, void *),
 
514			  size_t item_len)
515{
516	uint64_t buffer[(item_len + 7) / 8];
517	struct hlist_head *head;
518	struct rds_connection *conn;
519	size_t i;
520
521	rcu_read_lock();
522
523	lens->nr = 0;
524	lens->each = item_len;
525
526	for (i = 0, head = rds_conn_hash; i < ARRAY_SIZE(rds_conn_hash);
527	     i++, head++) {
528		hlist_for_each_entry_rcu(conn, head, c_hash_node) {
529
530			/* XXX no c_lock usage.. */
531			if (!visitor(conn, buffer))
532				continue;
533
534			/* We copy as much as we can fit in the buffer,
535			 * but we count all items so that the caller
536			 * can resize the buffer. */
537			if (len >= item_len) {
538				rds_info_copy(iter, buffer, item_len);
539				len -= item_len;
540			}
541			lens->nr++;
542		}
543	}
544	rcu_read_unlock();
545}
546EXPORT_SYMBOL_GPL(rds_for_each_conn_info);
547
548void rds_walk_conn_path_info(struct socket *sock, unsigned int len,
549			     struct rds_info_iterator *iter,
550			     struct rds_info_lengths *lens,
551			     int (*visitor)(struct rds_conn_path *, void *),
552			     size_t item_len)
 
553{
554	u64  buffer[(item_len + 7) / 8];
555	struct hlist_head *head;
556	struct rds_connection *conn;
557	size_t i;
558	int j;
559
560	rcu_read_lock();
561
562	lens->nr = 0;
563	lens->each = item_len;
564
565	for (i = 0, head = rds_conn_hash; i < ARRAY_SIZE(rds_conn_hash);
566	     i++, head++) {
567		hlist_for_each_entry_rcu(conn, head, c_hash_node) {
568			struct rds_conn_path *cp;
569
570			for (j = 0; j < RDS_MPATH_WORKERS; j++) {
571				cp = &conn->c_path[j];
 
 
 
 
 
 
 
572
573				/* XXX no cp_lock usage.. */
574				if (!visitor(cp, buffer))
575					continue;
576				if (!conn->c_trans->t_mp_capable)
577					break;
578			}
579
580			/* We copy as much as we can fit in the buffer,
581			 * but we count all items so that the caller
582			 * can resize the buffer.
583			 */
584			if (len >= item_len) {
585				rds_info_copy(iter, buffer, item_len);
586				len -= item_len;
587			}
588			lens->nr++;
589		}
590	}
591	rcu_read_unlock();
592}
593
594static int rds_conn_info_visitor(struct rds_conn_path *cp, void *buffer)
595{
596	struct rds_info_connection *cinfo = buffer;
 
 
 
 
597
598	cinfo->next_tx_seq = cp->cp_next_tx_seq;
599	cinfo->next_rx_seq = cp->cp_next_rx_seq;
600	cinfo->laddr = cp->cp_conn->c_laddr;
601	cinfo->faddr = cp->cp_conn->c_faddr;
602	strncpy(cinfo->transport, cp->cp_conn->c_trans->t_name,
 
603		sizeof(cinfo->transport));
604	cinfo->flags = 0;
605
606	rds_conn_info_set(cinfo->flags, test_bit(RDS_IN_XMIT, &cp->cp_flags),
607			  SENDING);
608	/* XXX Future: return the state rather than these funky bits */
609	rds_conn_info_set(cinfo->flags,
610			  atomic_read(&cp->cp_state) == RDS_CONN_CONNECTING,
611			  CONNECTING);
612	rds_conn_info_set(cinfo->flags,
613			  atomic_read(&cp->cp_state) == RDS_CONN_UP,
614			  CONNECTED);
615	return 1;
616}
617
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
618static void rds_conn_info(struct socket *sock, unsigned int len,
619			  struct rds_info_iterator *iter,
620			  struct rds_info_lengths *lens)
621{
 
 
622	rds_walk_conn_path_info(sock, len, iter, lens,
623				rds_conn_info_visitor,
 
624				sizeof(struct rds_info_connection));
625}
626
 
 
 
 
 
 
 
 
 
 
 
 
 
 
627int rds_conn_init(void)
628{
 
 
 
 
 
 
629	rds_conn_slab = kmem_cache_create("rds_connection",
630					  sizeof(struct rds_connection),
631					  0, 0, NULL);
632	if (!rds_conn_slab)
 
633		return -ENOMEM;
 
634
635	rds_info_register_func(RDS_INFO_CONNECTIONS, rds_conn_info);
636	rds_info_register_func(RDS_INFO_SEND_MESSAGES,
637			       rds_conn_message_info_send);
638	rds_info_register_func(RDS_INFO_RETRANS_MESSAGES,
639			       rds_conn_message_info_retrans);
640
 
 
 
 
 
 
641	return 0;
642}
643
644void rds_conn_exit(void)
645{
 
646	rds_loop_exit();
647
648	WARN_ON(!hlist_empty(rds_conn_hash));
649
650	kmem_cache_destroy(rds_conn_slab);
651
652	rds_info_deregister_func(RDS_INFO_CONNECTIONS, rds_conn_info);
653	rds_info_deregister_func(RDS_INFO_SEND_MESSAGES,
654				 rds_conn_message_info_send);
655	rds_info_deregister_func(RDS_INFO_RETRANS_MESSAGES,
656				 rds_conn_message_info_retrans);
 
 
 
 
 
 
 
657}
658
659/*
660 * Force a disconnect
661 */
662void rds_conn_path_drop(struct rds_conn_path *cp)
663{
664	atomic_set(&cp->cp_state, RDS_CONN_ERROR);
 
 
 
 
 
 
665	queue_work(rds_wq, &cp->cp_down_w);
 
666}
667EXPORT_SYMBOL_GPL(rds_conn_path_drop);
668
669void rds_conn_drop(struct rds_connection *conn)
670{
671	WARN_ON(conn->c_trans->t_mp_capable);
672	rds_conn_path_drop(&conn->c_path[0]);
673}
674EXPORT_SYMBOL_GPL(rds_conn_drop);
675
676/*
677 * If the connection is down, trigger a connect. We may have scheduled a
678 * delayed reconnect however - in this case we should not interfere.
679 */
680void rds_conn_path_connect_if_down(struct rds_conn_path *cp)
681{
 
 
 
 
 
682	if (rds_conn_path_state(cp) == RDS_CONN_DOWN &&
683	    !test_and_set_bit(RDS_RECONNECT_PENDING, &cp->cp_flags))
684		queue_delayed_work(rds_wq, &cp->cp_conn_w, 0);
 
685}
686EXPORT_SYMBOL_GPL(rds_conn_path_connect_if_down);
687
 
 
 
 
 
 
 
 
 
 
 
688void rds_conn_connect_if_down(struct rds_connection *conn)
689{
690	WARN_ON(conn->c_trans->t_mp_capable);
691	rds_conn_path_connect_if_down(&conn->c_path[0]);
692}
693EXPORT_SYMBOL_GPL(rds_conn_connect_if_down);
694
695void
696__rds_conn_path_error(struct rds_conn_path *cp, const char *fmt, ...)
697{
698	va_list ap;
699
700	va_start(ap, fmt);
701	vprintk(fmt, ap);
702	va_end(ap);
703
704	rds_conn_path_drop(cp);
705}