xref: /illumos-gate/usr/src/uts/common/io/ib/clients/rdsv3/send.c (revision 27954b0d964ffcb749cf19296906e7fecdf3da1b)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 /*
22  * Copyright (c) 2010, Oracle and/or its affiliates. All rights reserved.
23  */
24 
25 /*
26  * Copyright (c) 2006 Oracle.  All rights reserved.
27  *
28  * This software is available to you under a choice of one of two
29  * licenses.  You may choose to be licensed under the terms of the GNU
30  * General Public License (GPL) Version 2, available from the file
31  * COPYING in the main directory of this source tree, or the
32  * OpenIB.org BSD license below:
33  *
34  *     Redistribution and use in source and binary forms, with or
35  *     without modification, are permitted provided that the following
36  *     conditions are met:
37  *
38  *      - Redistributions of source code must retain the above
39  *        copyright notice, this list of conditions and the following
40  *        disclaimer.
41  *
42  *      - Redistributions in binary form must reproduce the above
43  *        copyright notice, this list of conditions and the following
44  *        disclaimer in the documentation and/or other materials
45  *        provided with the distribution.
46  *
47  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
48  * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
49  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
50  * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
51  * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
52  * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
53  * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
54  * SOFTWARE.
55  *
56  */
57 #include <sys/stropts.h>
58 #include <sys/systm.h>
59 
60 #include <sys/rds.h>
61 #include <sys/socket.h>
62 #include <sys/socketvar.h>
63 
64 #include <sys/ib/clients/rdsv3/rdsv3.h>
65 #include <sys/ib/clients/rdsv3/rdma.h>
66 #include <sys/ib/clients/rdsv3/rdsv3_debug.h>
67 
68 /*
69  * When transmitting messages in rdsv3_send_xmit, we need to emerge from
70  * time to time and briefly release the CPU. Otherwise the softlock watchdog
71  * will kick our shin.
72  * Also, it seems fairer to not let one busy connection stall all the
73  * others.
74  *
75  * send_batch_count is the number of times we'll loop in send_xmit. Setting
76  * it to 0 will restore the old behavior (where we looped until we had
77  * drained the queue).
78  */
79 static int send_batch_count = 64;
80 
81 extern void rdsv3_ib_send_unmap_rdma(void *ic, struct rdsv3_rdma_op *op);
82 /*
83  * Reset the send state. Caller must hold c_send_lock when calling here.
84  */
85 void
86 rdsv3_send_reset(struct rdsv3_connection *conn)
87 {
88 	struct rdsv3_message *rm, *tmp;
89 	struct rdsv3_rdma_op *ro;
90 
91 	RDSV3_DPRINTF4("rdsv3_send_reset", "Enter(conn: %p)", conn);
92 
93 	if (conn->c_xmit_rm) {
94 		rm = conn->c_xmit_rm;
95 		ro = rm->m_rdma_op;
96 		if (ro && ro->r_mapped) {
97 			RDSV3_DPRINTF2("rdsv3_send_reset",
98 			    "rm %p mflg 0x%x map %d mihdl %p sgl %p",
99 			    rm, rm->m_flags, ro->r_mapped,
100 			    ro->r_rdma_sg[0].mihdl,
101 			    ro->r_rdma_sg[0].swr.wr_sgl);
102 			rdsv3_ib_send_unmap_rdma(conn->c_transport_data, ro);
103 		}
104 		/*
105 		 * Tell the user the RDMA op is no longer mapped by the
106 		 * transport. This isn't entirely true (it's flushed out
107 		 * independently) but as the connection is down, there's
108 		 * no ongoing RDMA to/from that memory
109 		 */
110 		rdsv3_message_unmapped(conn->c_xmit_rm);
111 		rdsv3_message_put(conn->c_xmit_rm);
112 		conn->c_xmit_rm = NULL;
113 	}
114 	conn->c_xmit_sg = 0;
115 	conn->c_xmit_hdr_off = 0;
116 	conn->c_xmit_data_off = 0;
117 	conn->c_xmit_rdma_sent = 0;
118 
119 	conn->c_map_queued = 0;
120 
121 	conn->c_unacked_packets = rdsv3_sysctl_max_unacked_packets;
122 	conn->c_unacked_bytes = rdsv3_sysctl_max_unacked_bytes;
123 
124 	/* Mark messages as retransmissions, and move them to the send q */
125 	mutex_enter(&conn->c_lock);
126 	RDSV3_FOR_EACH_LIST_NODE_SAFE(rm, tmp, &conn->c_retrans, m_conn_item) {
127 		set_bit(RDSV3_MSG_ACK_REQUIRED, &rm->m_flags);
128 		set_bit(RDSV3_MSG_RETRANSMITTED, &rm->m_flags);
129 		if (rm->m_rdma_op && rm->m_rdma_op->r_mapped) {
130 			RDSV3_DPRINTF4("_send_reset",
131 			    "RT rm %p mflg 0x%x sgl %p",
132 			    rm, rm->m_flags,
133 			    rm->m_rdma_op->r_rdma_sg[0].swr.wr_sgl);
134 		}
135 	}
136 	list_move_tail(&conn->c_send_queue, &conn->c_retrans);
137 	mutex_exit(&conn->c_lock);
138 
139 	RDSV3_DPRINTF4("rdsv3_send_reset", "Return(conn: %p)", conn);
140 }
141 
142 /*
143  * We're making the concious trade-off here to only send one message
144  * down the connection at a time.
145  *   Pro:
146  *      - tx queueing is a simple fifo list
147  *   	- reassembly is optional and easily done by transports per conn
148  *      - no per flow rx lookup at all, straight to the socket
149  *   	- less per-frag memory and wire overhead
150  *   Con:
151  *      - queued acks can be delayed behind large messages
152  *   Depends:
153  *      - small message latency is higher behind queued large messages
154  *      - large message latency isn't starved by intervening small sends
155  */
156 int
157 rdsv3_send_xmit(struct rdsv3_connection *conn)
158 {
159 	struct rdsv3_message *rm;
160 	unsigned int tmp;
161 	unsigned int send_quota = send_batch_count;
162 	struct rdsv3_scatterlist *sg;
163 	int ret = 0;
164 	int was_empty = 0;
165 	list_t to_be_dropped;
166 
167 	RDSV3_DPRINTF4("rdsv3_send_xmit", "Enter(conn: %p)", conn);
168 
169 	list_create(&to_be_dropped, sizeof (struct rdsv3_message),
170 	    offsetof(struct rdsv3_message, m_conn_item));
171 
172 	/*
173 	 * sendmsg calls here after having queued its message on the send
174 	 * queue.  We only have one task feeding the connection at a time.  If
175 	 * another thread is already feeding the queue then we back off.  This
176 	 * avoids blocking the caller and trading per-connection data between
177 	 * caches per message.
178 	 *
179 	 * The sem holder will issue a retry if they notice that someone queued
180 	 * a message after they stopped walking the send queue but before they
181 	 * dropped the sem.
182 	 */
183 	if (!mutex_tryenter(&conn->c_send_lock)) {
184 		RDSV3_DPRINTF4("rdsv3_send_xmit",
185 		    "Another thread running(conn: %p)", conn);
186 		rdsv3_stats_inc(s_send_sem_contention);
187 		ret = -ENOMEM;
188 		goto out;
189 	}
190 
191 	if (conn->c_trans->xmit_prepare)
192 		conn->c_trans->xmit_prepare(conn);
193 
194 	/*
195 	 * spin trying to push headers and data down the connection until
196 	 * the connection doens't make forward progress.
197 	 */
198 	while (--send_quota) {
199 		/*
200 		 * See if need to send a congestion map update if we're
201 		 * between sending messages.  The send_sem protects our sole
202 		 * use of c_map_offset and _bytes.
203 		 * Note this is used only by transports that define a special
204 		 * xmit_cong_map function. For all others, we create allocate
205 		 * a cong_map message and treat it just like any other send.
206 		 */
207 		if (conn->c_map_bytes) {
208 			ret = conn->c_trans->xmit_cong_map(conn, conn->c_lcong,
209 			    conn->c_map_offset);
210 			if (ret <= 0)
211 				break;
212 
213 			conn->c_map_offset += ret;
214 			conn->c_map_bytes -= ret;
215 			if (conn->c_map_bytes)
216 				continue;
217 		}
218 
219 		/*
220 		 * If we're done sending the current message, clear the
221 		 * offset and S/G temporaries.
222 		 */
223 		rm = conn->c_xmit_rm;
224 		if (rm != NULL &&
225 		    conn->c_xmit_hdr_off == sizeof (struct rdsv3_header) &&
226 		    conn->c_xmit_sg == rm->m_nents) {
227 			conn->c_xmit_rm = NULL;
228 			conn->c_xmit_sg = 0;
229 			conn->c_xmit_hdr_off = 0;
230 			conn->c_xmit_data_off = 0;
231 			conn->c_xmit_rdma_sent = 0;
232 
233 			/* Release the reference to the previous message. */
234 			rdsv3_message_put(rm);
235 			rm = NULL;
236 		}
237 
238 		/* If we're asked to send a cong map update, do so. */
239 		if (rm == NULL && test_and_clear_bit(0, &conn->c_map_queued)) {
240 			if (conn->c_trans->xmit_cong_map != NULL) {
241 				conn->c_map_offset = 0;
242 				conn->c_map_bytes =
243 				    sizeof (struct rdsv3_header) +
244 				    RDSV3_CONG_MAP_BYTES;
245 				continue;
246 			}
247 
248 			rm = rdsv3_cong_update_alloc(conn);
249 			if (IS_ERR(rm)) {
250 				ret = PTR_ERR(rm);
251 				break;
252 			}
253 
254 			conn->c_xmit_rm = rm;
255 		}
256 
257 		/*
258 		 * Grab the next message from the send queue, if there is one.
259 		 *
260 		 * c_xmit_rm holds a ref while we're sending this message down
261 		 * the connction.  We can use this ref while holding the
262 		 * send_sem.. rdsv3_send_reset() is serialized with it.
263 		 */
264 		if (rm == NULL) {
265 			unsigned int len;
266 
267 			mutex_enter(&conn->c_lock);
268 
269 			if (!list_is_empty(&conn->c_send_queue)) {
270 				rm = list_remove_head(&conn->c_send_queue);
271 				rdsv3_message_addref(rm);
272 
273 				/*
274 				 * Move the message from the send queue to
275 				 * the retransmit
276 				 * list right away.
277 				 */
278 				list_insert_tail(&conn->c_retrans, rm);
279 			}
280 
281 			mutex_exit(&conn->c_lock);
282 
283 			if (rm == NULL) {
284 				was_empty = 1;
285 				break;
286 			}
287 
288 			/*
289 			 * Unfortunately, the way Infiniband deals with
290 			 * RDMA to a bad MR key is by moving the entire
291 			 * queue pair to error state. We cold possibly
292 			 * recover from that, but right now we drop the
293 			 * connection.
294 			 * Therefore, we never retransmit messages with
295 			 * RDMA ops.
296 			 */
297 			if (rm->m_rdma_op &&
298 			    test_bit(RDSV3_MSG_RETRANSMITTED, &rm->m_flags)) {
299 				mutex_enter(&conn->c_lock);
300 				if (test_and_clear_bit(RDSV3_MSG_ON_CONN,
301 				    &rm->m_flags))
302 					list_remove_node(&rm->m_conn_item);
303 					list_insert_tail(&to_be_dropped, rm);
304 				mutex_exit(&conn->c_lock);
305 				rdsv3_message_put(rm);
306 				continue;
307 			}
308 
309 			/* Require an ACK every once in a while */
310 			len = ntohl(rm->m_inc.i_hdr.h_len);
311 			if (conn->c_unacked_packets == 0 ||
312 			    conn->c_unacked_bytes < len) {
313 				set_bit(RDSV3_MSG_ACK_REQUIRED, &rm->m_flags);
314 
315 				conn->c_unacked_packets =
316 				    rdsv3_sysctl_max_unacked_packets;
317 				conn->c_unacked_bytes =
318 				    rdsv3_sysctl_max_unacked_bytes;
319 				rdsv3_stats_inc(s_send_ack_required);
320 			} else {
321 				conn->c_unacked_bytes -= len;
322 				conn->c_unacked_packets--;
323 			}
324 
325 			conn->c_xmit_rm = rm;
326 		}
327 
328 		/*
329 		 * Try and send an rdma message.  Let's see if we can
330 		 * keep this simple and require that the transport either
331 		 * send the whole rdma or none of it.
332 		 */
333 		if (rm->m_rdma_op && !conn->c_xmit_rdma_sent) {
334 			ret = conn->c_trans->xmit_rdma(conn, rm->m_rdma_op);
335 			if (ret)
336 				break;
337 			conn->c_xmit_rdma_sent = 1;
338 			/*
339 			 * The transport owns the mapped memory for now.
340 			 * You can't unmap it while it's on the send queue
341 			 */
342 			set_bit(RDSV3_MSG_MAPPED, &rm->m_flags);
343 		}
344 
345 		if (conn->c_xmit_hdr_off < sizeof (struct rdsv3_header) ||
346 		    conn->c_xmit_sg < rm->m_nents) {
347 			ret = conn->c_trans->xmit(conn, rm,
348 			    conn->c_xmit_hdr_off,
349 			    conn->c_xmit_sg,
350 			    conn->c_xmit_data_off);
351 			if (ret <= 0)
352 				break;
353 
354 			if (conn->c_xmit_hdr_off <
355 			    sizeof (struct rdsv3_header)) {
356 				tmp = min(ret,
357 				    sizeof (struct rdsv3_header) -
358 				    conn->c_xmit_hdr_off);
359 				conn->c_xmit_hdr_off += tmp;
360 				ret -= tmp;
361 			}
362 
363 			sg = &rm->m_sg[conn->c_xmit_sg];
364 			while (ret) {
365 				tmp = min(ret, rdsv3_sg_len(sg) -
366 				    conn->c_xmit_data_off);
367 				conn->c_xmit_data_off += tmp;
368 				ret -= tmp;
369 				if (conn->c_xmit_data_off == rdsv3_sg_len(sg)) {
370 					conn->c_xmit_data_off = 0;
371 					sg++;
372 					conn->c_xmit_sg++;
373 					ASSERT(!(ret != 0 &&
374 					    conn->c_xmit_sg == rm->m_nents));
375 				}
376 			}
377 		}
378 	}
379 
380 	/* Nuke any messages we decided not to retransmit. */
381 	if (!list_is_empty(&to_be_dropped))
382 		rdsv3_send_remove_from_sock(&to_be_dropped, RDSV3_RDMA_DROPPED);
383 
384 	if (conn->c_trans->xmit_complete)
385 		conn->c_trans->xmit_complete(conn);
386 
387 	/*
388 	 * We might be racing with another sender who queued a message but
389 	 * backed off on noticing that we held the c_send_lock.  If we check
390 	 * for queued messages after dropping the sem then either we'll
391 	 * see the queued message or the queuer will get the sem.  If we
392 	 * notice the queued message then we trigger an immediate retry.
393 	 *
394 	 * We need to be careful only to do this when we stopped processing
395 	 * the send queue because it was empty.  It's the only way we
396 	 * stop processing the loop when the transport hasn't taken
397 	 * responsibility for forward progress.
398 	 */
399 	mutex_exit(&conn->c_send_lock);
400 
401 	if (conn->c_map_bytes || (send_quota == 0 && !was_empty)) {
402 		/*
403 		 * We exhausted the send quota, but there's work left to
404 		 * do. Return and (re-)schedule the send worker.
405 		 */
406 		ret = -EAGAIN;
407 	}
408 
409 	if (ret == 0 && was_empty) {
410 		/*
411 		 * A simple bit test would be way faster than taking the
412 		 * spin lock
413 		 */
414 		mutex_enter(&conn->c_lock);
415 		if (!list_is_empty(&conn->c_send_queue)) {
416 			rdsv3_stats_inc(s_send_sem_queue_raced);
417 			ret = -EAGAIN;
418 		}
419 		mutex_exit(&conn->c_lock);
420 	}
421 
422 out:
423 	RDSV3_DPRINTF4("rdsv3_send_xmit", "Return(conn: %p, ret: %d)",
424 	    conn, ret);
425 	return (ret);
426 }
427 
428 static void
429 rdsv3_send_sndbuf_remove(struct rdsv3_sock *rs, struct rdsv3_message *rm)
430 {
431 	uint32_t len = ntohl(rm->m_inc.i_hdr.h_len);
432 
433 	ASSERT(mutex_owned(&rs->rs_lock));
434 
435 	ASSERT(rs->rs_snd_bytes >= len);
436 	rs->rs_snd_bytes -= len;
437 
438 	if (rs->rs_snd_bytes == 0)
439 		rdsv3_stats_inc(s_send_queue_empty);
440 }
441 
442 static inline int
443 rdsv3_send_is_acked(struct rdsv3_message *rm, uint64_t ack,
444     is_acked_func is_acked)
445 {
446 	if (is_acked)
447 		return (is_acked(rm, ack));
448 	return (ntohll(rm->m_inc.i_hdr.h_sequence) <= ack);
449 }
450 
451 /*
452  * Returns true if there are no messages on the send and retransmit queues
453  * which have a sequence number greater than or equal to the given sequence
454  * number.
455  */
456 int
457 rdsv3_send_acked_before(struct rdsv3_connection *conn, uint64_t seq)
458 {
459 	struct rdsv3_message *rm;
460 	int ret = 1;
461 
462 	RDSV3_DPRINTF4("rdsv3_send_acked_before", "Enter(conn: %p)", conn);
463 
464 	mutex_enter(&conn->c_lock);
465 
466 	/* XXX - original code spits out warning */
467 	rm = list_head(&conn->c_retrans);
468 	if (ntohll(rm->m_inc.i_hdr.h_sequence) < seq)
469 		ret = 0;
470 
471 	/* XXX - original code spits out warning */
472 	rm = list_head(&conn->c_send_queue);
473 	if (ntohll(rm->m_inc.i_hdr.h_sequence) < seq)
474 		ret = 0;
475 
476 	mutex_exit(&conn->c_lock);
477 
478 	RDSV3_DPRINTF4("rdsv3_send_acked_before", "Return(conn: %p)", conn);
479 
480 	return (ret);
481 }
482 
483 /*
484  * This is pretty similar to what happens below in the ACK
485  * handling code - except that we call here as soon as we get
486  * the IB send completion on the RDMA op and the accompanying
487  * message.
488  */
489 void
490 rdsv3_rdma_send_complete(struct rdsv3_message *rm, int status)
491 {
492 	struct rdsv3_sock *rs = NULL;
493 	struct rdsv3_rdma_op *ro;
494 	struct rdsv3_notifier *notifier;
495 
496 	RDSV3_DPRINTF4("rdsv3_rdma_send_complete", "Enter(rm: %p)", rm);
497 
498 	mutex_enter(&rm->m_rs_lock);
499 
500 	ro = rm->m_rdma_op;
501 	if (test_bit(RDSV3_MSG_ON_SOCK, &rm->m_flags) &&
502 	    ro && ro->r_notify && ro->r_notifier) {
503 		notifier = ro->r_notifier;
504 		rs = rm->m_rs;
505 		rdsv3_sk_sock_hold(rdsv3_rs_to_sk(rs));
506 
507 		notifier->n_status = status;
508 		mutex_enter(&rs->rs_lock);
509 		list_insert_tail(&rs->rs_notify_queue, notifier);
510 		mutex_exit(&rs->rs_lock);
511 
512 		ro->r_notifier = NULL;
513 	}
514 
515 	mutex_exit(&rm->m_rs_lock);
516 
517 	if (rs) {
518 		rdsv3_wake_sk_sleep(rs);
519 		rdsv3_sk_sock_put(rdsv3_rs_to_sk(rs));
520 	}
521 
522 	RDSV3_DPRINTF4("rdsv3_rdma_send_complete", "Return(rm: %p)", rm);
523 }
524 
525 /*
526  * This is the same as rdsv3_rdma_send_complete except we
527  * don't do any locking - we have all the ingredients (message,
528  * socket, socket lock) and can just move the notifier.
529  */
530 static inline void
531 __rdsv3_rdma_send_complete(struct rdsv3_sock *rs, struct rdsv3_message *rm,
532     int status)
533 {
534 	struct rdsv3_rdma_op *ro;
535 	void *ic;
536 
537 	RDSV3_DPRINTF4("__rdsv3_rdma_send_complete",
538 	    "Enter(rs: %p, rm: %p)", rs, rm);
539 
540 	ro = rm->m_rdma_op;
541 	if (ro && ro->r_notify && ro->r_notifier) {
542 		ro->r_notifier->n_status = status;
543 		list_insert_tail(&rs->rs_notify_queue, ro->r_notifier);
544 		ro->r_notifier = NULL;
545 	}
546 
547 	/* No need to wake the app - caller does this */
548 }
549 
550 /*
551  * This is called from the IB send completion when we detect
552  * a RDMA operation that failed with remote access error.
553  * So speed is not an issue here.
554  */
555 struct rdsv3_message *
556 rdsv3_send_get_message(struct rdsv3_connection *conn,
557     struct rdsv3_rdma_op *op)
558 {
559 	struct rdsv3_message *rm, *tmp, *found = NULL;
560 
561 	RDSV3_DPRINTF4("rdsv3_send_get_message", "Enter(conn: %p)", conn);
562 
563 	mutex_enter(&conn->c_lock);
564 
565 	RDSV3_FOR_EACH_LIST_NODE_SAFE(rm, tmp, &conn->c_retrans, m_conn_item) {
566 		if (rm->m_rdma_op == op) {
567 			atomic_add_32(&rm->m_refcount, 1);
568 			found = rm;
569 			goto out;
570 		}
571 	}
572 
573 	RDSV3_FOR_EACH_LIST_NODE_SAFE(rm, tmp, &conn->c_send_queue,
574 	    m_conn_item) {
575 		if (rm->m_rdma_op == op) {
576 			atomic_add_32(&rm->m_refcount, 1);
577 			found = rm;
578 			break;
579 		}
580 	}
581 
582 out:
583 	mutex_exit(&conn->c_lock);
584 
585 	return (found);
586 }
587 
588 /*
589  * This removes messages from the socket's list if they're on it.  The list
590  * argument must be private to the caller, we must be able to modify it
591  * without locks.  The messages must have a reference held for their
592  * position on the list.  This function will drop that reference after
593  * removing the messages from the 'messages' list regardless of if it found
594  * the messages on the socket list or not.
595  */
596 void
597 rdsv3_send_remove_from_sock(struct list *messages, int status)
598 {
599 	struct rdsv3_sock *rs = NULL;
600 	struct rdsv3_message *rm;
601 
602 	RDSV3_DPRINTF4("rdsv3_send_remove_from_sock", "Enter");
603 
604 	while (!list_is_empty(messages)) {
605 		int was_on_sock = 0;
606 		rm = list_remove_head(messages);
607 
608 		/*
609 		 * If we see this flag cleared then we're *sure* that someone
610 		 * else beat us to removing it from the sock.  If we race
611 		 * with their flag update we'll get the lock and then really
612 		 * see that the flag has been cleared.
613 		 *
614 		 * The message spinlock makes sure nobody clears rm->m_rs
615 		 * while we're messing with it. It does not prevent the
616 		 * message from being removed from the socket, though.
617 		 */
618 		mutex_enter(&rm->m_rs_lock);
619 		if (!test_bit(RDSV3_MSG_ON_SOCK, &rm->m_flags))
620 			goto unlock_and_drop;
621 
622 		if (rs != rm->m_rs) {
623 			if (rs) {
624 				rdsv3_wake_sk_sleep(rs);
625 				rdsv3_sk_sock_put(rdsv3_rs_to_sk(rs));
626 			}
627 			rs = rm->m_rs;
628 			rdsv3_sk_sock_hold(rdsv3_rs_to_sk(rs));
629 		}
630 
631 		mutex_enter(&rs->rs_lock);
632 		if (test_and_clear_bit(RDSV3_MSG_ON_SOCK, &rm->m_flags)) {
633 			struct rdsv3_rdma_op *ro = rm->m_rdma_op;
634 			struct rdsv3_notifier *notifier;
635 
636 			list_remove_node(&rm->m_sock_item);
637 			rdsv3_send_sndbuf_remove(rs, rm);
638 			if (ro && ro->r_notifier &&
639 			    (status || ro->r_notify)) {
640 				notifier = ro->r_notifier;
641 				list_insert_tail(&rs->rs_notify_queue,
642 				    notifier);
643 				if (!notifier->n_status)
644 					notifier->n_status = status;
645 				rm->m_rdma_op->r_notifier = NULL;
646 			}
647 			was_on_sock = 1;
648 			rm->m_rs = NULL;
649 		}
650 		mutex_exit(&rs->rs_lock);
651 
652 unlock_and_drop:
653 		mutex_exit(&rm->m_rs_lock);
654 		rdsv3_message_put(rm);
655 		if (was_on_sock)
656 			rdsv3_message_put(rm);
657 	}
658 
659 	if (rs) {
660 		rdsv3_wake_sk_sleep(rs);
661 		rdsv3_sk_sock_put(rdsv3_rs_to_sk(rs));
662 	}
663 
664 	RDSV3_DPRINTF4("rdsv3_send_remove_from_sock", "Return");
665 }
666 
667 /*
668  * Transports call here when they've determined that the receiver queued
669  * messages up to, and including, the given sequence number.  Messages are
670  * moved to the retrans queue when rdsv3_send_xmit picks them off the send
671  * queue. This means that in the TCP case, the message may not have been
672  * assigned the m_ack_seq yet - but that's fine as long as tcp_is_acked
673  * checks the RDSV3_MSG_HAS_ACK_SEQ bit.
674  *
675  * XXX It's not clear to me how this is safely serialized with socket
676  * destruction.  Maybe it should bail if it sees SOCK_DEAD.
677  */
678 void
679 rdsv3_send_drop_acked(struct rdsv3_connection *conn, uint64_t ack,
680     is_acked_func is_acked)
681 {
682 	struct rdsv3_message *rm, *tmp;
683 	list_t list;
684 
685 	RDSV3_DPRINTF4("rdsv3_send_drop_acked", "Enter(conn: %p)", conn);
686 
687 	list_create(&list, sizeof (struct rdsv3_message),
688 	    offsetof(struct rdsv3_message, m_conn_item));
689 
690 	mutex_enter(&conn->c_lock);
691 
692 	RDSV3_FOR_EACH_LIST_NODE_SAFE(rm, tmp, &conn->c_retrans, m_conn_item) {
693 		if (!rdsv3_send_is_acked(rm, ack, is_acked))
694 			break;
695 
696 		list_remove_node(&rm->m_conn_item);
697 		list_insert_tail(&list, rm);
698 		clear_bit(RDSV3_MSG_ON_CONN, &rm->m_flags);
699 	}
700 
701 #if 0
702 XXX
703 	/* order flag updates with spin locks */
704 	if (!list_is_empty(&list))
705 		smp_mb__after_clear_bit();
706 #endif
707 
708 	mutex_exit(&conn->c_lock);
709 
710 	/* now remove the messages from the sock list as needed */
711 	rdsv3_send_remove_from_sock(&list, RDSV3_RDMA_SUCCESS);
712 
713 	RDSV3_DPRINTF4("rdsv3_send_drop_acked", "Return(conn: %p)", conn);
714 }
715 
716 void
717 rdsv3_send_drop_to(struct rdsv3_sock *rs, struct sockaddr_in *dest)
718 {
719 	struct rdsv3_message *rm, *tmp;
720 	struct rdsv3_connection *conn;
721 	list_t list;
722 	int wake = 0;
723 
724 	RDSV3_DPRINTF4("rdsv3_send_drop_to", "Enter(rs: %p)", rs);
725 
726 	list_create(&list, sizeof (struct rdsv3_message),
727 	    offsetof(struct rdsv3_message, m_sock_item));
728 
729 	/* get all the messages we're dropping under the rs lock */
730 	mutex_enter(&rs->rs_lock);
731 
732 	RDSV3_FOR_EACH_LIST_NODE_SAFE(rm, tmp, &rs->rs_send_queue,
733 	    m_sock_item) {
734 		if (dest && (dest->sin_addr.s_addr != rm->m_daddr ||
735 		    dest->sin_port != rm->m_inc.i_hdr.h_dport))
736 			continue;
737 
738 		wake = 1;
739 		list_remove(&rs->rs_send_queue, rm);
740 		list_insert_tail(&list, rm);
741 		rdsv3_send_sndbuf_remove(rs, rm);
742 		clear_bit(RDSV3_MSG_ON_SOCK, &rm->m_flags);
743 	}
744 
745 	mutex_exit(&rs->rs_lock);
746 
747 	conn = NULL;
748 
749 	/* now remove the messages from the conn list as needed */
750 	RDSV3_FOR_EACH_LIST_NODE(rm, &list, m_sock_item) {
751 		/*
752 		 * We do this here rather than in the loop above, so that
753 		 * we don't have to nest m_rs_lock under rs->rs_lock
754 		 */
755 		mutex_enter(&rm->m_rs_lock);
756 		/* If this is a RDMA operation, notify the app. */
757 		__rdsv3_rdma_send_complete(rs, rm, RDSV3_RDMA_CANCELED);
758 		rm->m_rs = NULL;
759 		mutex_exit(&rm->m_rs_lock);
760 
761 		/*
762 		 * If we see this flag cleared then we're *sure* that someone
763 		 * else beat us to removing it from the conn.  If we race
764 		 * with their flag update we'll get the lock and then really
765 		 * see that the flag has been cleared.
766 		 */
767 		if (!test_bit(RDSV3_MSG_ON_CONN, &rm->m_flags))
768 			continue;
769 
770 		if (conn != rm->m_inc.i_conn) {
771 			if (conn)
772 				mutex_exit(&conn->c_lock);
773 			conn = rm->m_inc.i_conn;
774 			mutex_enter(&conn->c_lock);
775 		}
776 
777 		if (test_and_clear_bit(RDSV3_MSG_ON_CONN, &rm->m_flags)) {
778 			list_remove_node(&rm->m_conn_item);
779 			rdsv3_message_put(rm);
780 		}
781 	}
782 
783 	if (conn)
784 		mutex_exit(&conn->c_lock);
785 
786 	if (wake)
787 		rdsv3_wake_sk_sleep(rs);
788 
789 	while (!list_is_empty(&list)) {
790 		rm = list_remove_head(&list);
791 
792 		rdsv3_message_wait(rm);
793 		rdsv3_message_put(rm);
794 	}
795 
796 	RDSV3_DPRINTF4("rdsv3_send_drop_to", "Return(rs: %p)", rs);
797 }
798 
799 /*
800  * we only want this to fire once so we use the callers 'queued'.  It's
801  * possible that another thread can race with us and remove the
802  * message from the flow with RDSV3_CANCEL_SENT_TO.
803  */
804 static int
805 rdsv3_send_queue_rm(struct rdsv3_sock *rs, struct rdsv3_connection *conn,
806     struct rdsv3_message *rm, uint16_be_t sport,
807     uint16_be_t dport, int *queued)
808 {
809 	uint32_t len;
810 
811 	RDSV3_DPRINTF4("rdsv3_send_queue_rm", "Enter(rs: %p, rm: %p)", rs, rm);
812 
813 	if (*queued)
814 		goto out;
815 
816 	len = ntohl(rm->m_inc.i_hdr.h_len);
817 
818 	/*
819 	 * this is the only place which holds both the socket's rs_lock
820 	 * and the connection's c_lock
821 	 */
822 	mutex_enter(&rs->rs_lock);
823 
824 	/*
825 	 * If there is a little space in sndbuf, we don't queue anything,
826 	 * and userspace gets -EAGAIN. But poll() indicates there's send
827 	 * room. This can lead to bad behavior (spinning) if snd_bytes isn't
828 	 * freed up by incoming acks. So we check the *old* value of
829 	 * rs_snd_bytes here to allow the last msg to exceed the buffer,
830 	 * and poll() now knows no more data can be sent.
831 	 */
832 	if (rs->rs_snd_bytes < rdsv3_sk_sndbuf(rs)) {
833 		rs->rs_snd_bytes += len;
834 
835 		/*
836 		 * let recv side know we are close to send space exhaustion.
837 		 * This is probably not the optimal way to do it, as this
838 		 * means we set the flag on *all* messages as soon as our
839 		 * throughput hits a certain threshold.
840 		 */
841 		if (rs->rs_snd_bytes >= rdsv3_sk_sndbuf(rs) / 2)
842 			set_bit(RDSV3_MSG_ACK_REQUIRED, &rm->m_flags);
843 
844 		list_insert_tail(&rs->rs_send_queue, rm);
845 		set_bit(RDSV3_MSG_ON_SOCK, &rm->m_flags);
846 
847 		rdsv3_message_addref(rm);
848 		rm->m_rs = rs;
849 
850 		/*
851 		 * The code ordering is a little weird, but we're
852 		 * trying to minimize the time we hold c_lock
853 		 */
854 		rdsv3_message_populate_header(&rm->m_inc.i_hdr, sport,
855 		    dport, 0);
856 		rm->m_inc.i_conn = conn;
857 		rdsv3_message_addref(rm);	/* XXX - called twice */
858 
859 		mutex_enter(&conn->c_lock);
860 		rm->m_inc.i_hdr.h_sequence = htonll(conn->c_next_tx_seq++);
861 		list_insert_tail(&conn->c_send_queue, rm);
862 		set_bit(RDSV3_MSG_ON_CONN, &rm->m_flags);
863 		mutex_exit(&conn->c_lock);
864 
865 		RDSV3_DPRINTF5("rdsv3_send_queue_rm",
866 		    "queued msg %p len %d, rs %p bytes %d seq %llu",
867 		    rm, len, rs, rs->rs_snd_bytes,
868 		    (unsigned long long)ntohll(
869 		    rm->m_inc.i_hdr.h_sequence));
870 
871 		*queued = 1;
872 	}
873 
874 	mutex_exit(&rs->rs_lock);
875 
876 	RDSV3_DPRINTF4("rdsv3_send_queue_rm", "Return(rs: %p)", rs);
877 out:
878 	return (*queued);
879 }
880 
881 static int
882 rdsv3_cmsg_send(struct rdsv3_sock *rs, struct rdsv3_message *rm,
883     struct msghdr *msg, int *allocated_mr)
884 {
885 	struct cmsghdr *cmsg;
886 	int ret = 0;
887 
888 	RDSV3_DPRINTF4("rdsv3_cmsg_send", "Enter(rs: %p)", rs);
889 
890 	for (cmsg = CMSG_FIRSTHDR(msg); cmsg; cmsg = CMSG_NXTHDR(msg, cmsg)) {
891 
892 		if (cmsg->cmsg_level != SOL_RDS)
893 			continue;
894 
895 		RDSV3_DPRINTF4("rdsv3_cmsg_send", "cmsg(%p, %p) type %d",
896 		    cmsg, rm, cmsg->cmsg_type);
897 		/*
898 		 * As a side effect, RDMA_DEST and RDMA_MAP will set
899 		 * rm->m_rdma_cookie and rm->m_rdma_mr.
900 		 */
901 		switch (cmsg->cmsg_type) {
902 		case RDSV3_CMSG_RDMA_ARGS:
903 			ret = rdsv3_cmsg_rdma_args(rs, rm, cmsg);
904 			break;
905 
906 		case RDSV3_CMSG_RDMA_DEST:
907 			ret = rdsv3_cmsg_rdma_dest(rs, rm, cmsg);
908 			break;
909 
910 		case RDSV3_CMSG_RDMA_MAP:
911 			ret = rdsv3_cmsg_rdma_map(rs, rm, cmsg);
912 			if (ret)
913 				*allocated_mr = 1;
914 			break;
915 
916 		default:
917 			return (-EINVAL);
918 		}
919 
920 		if (ret)
921 			break;
922 	}
923 
924 	RDSV3_DPRINTF4("rdsv3_cmsg_send", "Return(rs: %p)", rs);
925 
926 	return (ret);
927 }
928 
929 int
930 rdsv3_sendmsg(struct rdsv3_sock *rs, uio_t *uio, struct nmsghdr *msg,
931     size_t payload_len)
932 {
933 	struct rsock *sk = rdsv3_rs_to_sk(rs);
934 	struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
935 	uint32_be_t daddr;
936 	uint16_be_t dport;
937 	struct rdsv3_message *rm = NULL;
938 	struct rdsv3_connection *conn;
939 	int ret = 0;
940 	int queued = 0, allocated_mr = 0;
941 	int nonblock = msg->msg_flags & MSG_DONTWAIT;
942 	long timeo = rdsv3_sndtimeo(sk, nonblock);
943 
944 	RDSV3_DPRINTF4("rdsv3_sendmsg", "Enter(rs: %p)", rs);
945 
946 	if (msg->msg_namelen) {
947 		/* XXX fail non-unicast destination IPs? */
948 		if (msg->msg_namelen < sizeof (*usin) ||
949 		    usin->sin_family != AF_INET_OFFLOAD) {
950 			ret = -EINVAL;
951 			RDSV3_DPRINTF2("rdsv3_sendmsg", "returning: %d", -ret);
952 			goto out;
953 		}
954 		daddr = usin->sin_addr.s_addr;
955 		dport = usin->sin_port;
956 	} else {
957 		/* We only care about consistency with ->connect() */
958 		mutex_enter(&sk->sk_lock);
959 		daddr = rs->rs_conn_addr;
960 		dport = rs->rs_conn_port;
961 		mutex_exit(&sk->sk_lock);
962 	}
963 
964 	/* racing with another thread binding seems ok here */
965 	if (daddr == 0 || rs->rs_bound_addr == 0) {
966 		ret = -ENOTCONN; /* XXX not a great errno */
967 		RDSV3_DPRINTF2("rdsv3_sendmsg", "returning: %d", -ret);
968 		goto out;
969 	}
970 
971 	rm = rdsv3_message_copy_from_user(uio, payload_len);
972 	if (IS_ERR(rm)) {
973 		ret = PTR_ERR(rm);
974 		RDSV3_DPRINTF2("rdsv3_sendmsg",
975 		    "rdsv3_message_copy_from_user failed %d", -ret);
976 		rm = NULL;
977 		goto out;
978 	}
979 
980 	rm->m_daddr = daddr;
981 
982 	/* Parse any control messages the user may have included. */
983 	ret = rdsv3_cmsg_send(rs, rm, msg, &allocated_mr);
984 	if (ret) {
985 		RDSV3_DPRINTF2("rdsv3_sendmsg",
986 		    "rdsv3_cmsg_send(rs: %p rm: %p msg: %p) returned: %d",
987 		    rs, rm, msg, ret);
988 		goto out;
989 	}
990 
991 	/*
992 	 * rdsv3_conn_create has a spinlock that runs with IRQ off.
993 	 * Caching the conn in the socket helps a lot.
994 	 */
995 	mutex_enter(&rs->rs_conn_lock);
996 	if (rs->rs_conn && rs->rs_conn->c_faddr == daddr) {
997 		conn = rs->rs_conn;
998 	} else {
999 		conn = rdsv3_conn_create_outgoing(rs->rs_bound_addr,
1000 		    daddr, rs->rs_transport, KM_NOSLEEP);
1001 		if (IS_ERR(conn)) {
1002 			mutex_exit(&rs->rs_conn_lock);
1003 			ret = PTR_ERR(conn);
1004 			RDSV3_DPRINTF2("rdsv3_sendmsg",
1005 			    "rdsv3_conn_create_outgoing failed %d",
1006 			    -ret);
1007 			goto out;
1008 		}
1009 		rs->rs_conn = conn;
1010 	}
1011 	mutex_exit(&rs->rs_conn_lock);
1012 
1013 	if ((rm->m_rdma_cookie || rm->m_rdma_op) &&
1014 	    conn->c_trans->xmit_rdma == NULL) {
1015 		RDSV3_DPRINTF2("rdsv3_sendmsg", "rdma_op %p conn xmit_rdma %p",
1016 		    rm->m_rdma_op, conn->c_trans->xmit_rdma);
1017 		ret = -EOPNOTSUPP;
1018 		goto out;
1019 	}
1020 
1021 	/*
1022 	 * If the connection is down, trigger a connect. We may
1023 	 * have scheduled a delayed reconnect however - in this case
1024 	 * we should not interfere.
1025 	 */
1026 	if (rdsv3_conn_state(conn) == RDSV3_CONN_DOWN &&
1027 	    !test_and_set_bit(RDSV3_RECONNECT_PENDING, &conn->c_flags))
1028 		rdsv3_queue_delayed_work(rdsv3_wq, &conn->c_conn_w, 0);
1029 
1030 	ret = rdsv3_cong_wait(conn->c_fcong, dport, nonblock, rs);
1031 	if (ret) {
1032 		mutex_enter(&rdsv3_poll_waitq.waitq_mutex);
1033 		rs->rs_seen_congestion = 1;
1034 		cv_signal(&rdsv3_poll_waitq.waitq_cv);
1035 		mutex_exit(&rdsv3_poll_waitq.waitq_mutex);
1036 
1037 		RDSV3_DPRINTF2("rdsv3_sendmsg",
1038 		    "rdsv3_cong_wait (dport: %d) returned: %d", dport, ret);
1039 		goto out;
1040 	}
1041 
1042 	(void) rdsv3_send_queue_rm(rs, conn, rm, rs->rs_bound_port, dport,
1043 	    &queued);
1044 	if (!queued) {
1045 		/* rdsv3_stats_inc(s_send_queue_full); */
1046 		/* XXX make sure this is reasonable */
1047 		if (payload_len > rdsv3_sk_sndbuf(rs)) {
1048 			ret = -EMSGSIZE;
1049 			RDSV3_DPRINTF2("rdsv3_sendmsg",
1050 			    "msgsize(%d) too big, returning: %d",
1051 			    payload_len, -ret);
1052 			goto out;
1053 		}
1054 		if (nonblock) {
1055 			ret = -EAGAIN;
1056 			RDSV3_DPRINTF3("rdsv3_sendmsg",
1057 			    "send queue full (%d), returning: %d",
1058 			    payload_len, -ret);
1059 			goto out;
1060 		}
1061 
1062 #if 0
1063 		ret = rdsv3_wait_sig(sk->sk_sleep,
1064 		    (rdsv3_send_queue_rm(rs, conn, rm, rs->rs_bound_port,
1065 		    dport, &queued)));
1066 		if (ret == 0) {
1067 			/* signal/timeout pending */
1068 			RDSV3_DPRINTF2("rdsv3_sendmsg",
1069 			    "woke due to signal: %d", ret);
1070 			ret = -ERESTART;
1071 			goto out;
1072 		}
1073 #else
1074 		mutex_enter(&sk->sk_sleep->waitq_mutex);
1075 		sk->sk_sleep->waitq_waiters++;
1076 		while (!rdsv3_send_queue_rm(rs, conn, rm, rs->rs_bound_port,
1077 		    dport, &queued)) {
1078 			ret = cv_wait_sig(&sk->sk_sleep->waitq_cv,
1079 			    &sk->sk_sleep->waitq_mutex);
1080 			if (ret == 0) {
1081 				/* signal/timeout pending */
1082 				RDSV3_DPRINTF2("rdsv3_sendmsg",
1083 				    "woke due to signal: %d", ret);
1084 				ret = -ERESTART;
1085 				sk->sk_sleep->waitq_waiters--;
1086 				mutex_exit(&sk->sk_sleep->waitq_mutex);
1087 				goto out;
1088 			}
1089 		}
1090 		sk->sk_sleep->waitq_waiters--;
1091 		mutex_exit(&sk->sk_sleep->waitq_mutex);
1092 #endif
1093 
1094 		RDSV3_DPRINTF5("rdsv3_sendmsg", "sendmsg woke queued %d",
1095 		    queued);
1096 
1097 		ASSERT(queued);
1098 		ret = 0;
1099 	}
1100 
1101 	/*
1102 	 * By now we've committed to the send.  We reuse rdsv3_send_worker()
1103 	 * to retry sends in the rds thread if the transport asks us to.
1104 	 */
1105 	rdsv3_stats_inc(s_send_queued);
1106 
1107 	if (!test_bit(RDSV3_LL_SEND_FULL, &conn->c_flags))
1108 		rdsv3_send_worker(&conn->c_send_w.work);
1109 
1110 	rdsv3_message_put(rm);
1111 	RDSV3_DPRINTF4("rdsv3_sendmsg", "Return(rs: %p, len: %d)",
1112 	    rs, payload_len);
1113 	return (payload_len);
1114 
1115 out:
1116 	/*
1117 	 * If the user included a RDMA_MAP cmsg, we allocated a MR on the fly.
1118 	 * If the sendmsg goes through, we keep the MR. If it fails with EAGAIN
1119 	 * or in any other way, we need to destroy the MR again
1120 	 */
1121 	if (allocated_mr)
1122 		rdsv3_rdma_unuse(rs, rdsv3_rdma_cookie_key(rm->m_rdma_cookie),
1123 		    1);
1124 
1125 	if (rm)
1126 		rdsv3_message_put(rm);
1127 	return (ret);
1128 }
1129 
1130 /*
1131  * Reply to a ping packet.
1132  */
1133 int
1134 rdsv3_send_pong(struct rdsv3_connection *conn, uint16_be_t dport)
1135 {
1136 	struct rdsv3_message *rm;
1137 	int ret = 0;
1138 
1139 	RDSV3_DPRINTF4("rdsv3_send_pong", "Enter(conn: %p)", conn);
1140 
1141 	rm = rdsv3_message_alloc(0, KM_NOSLEEP);
1142 	if (rm == NULL) {
1143 		ret = -ENOMEM;
1144 		goto out;
1145 	}
1146 
1147 	rm->m_daddr = conn->c_faddr;
1148 
1149 	/*
1150 	 * If the connection is down, trigger a connect. We may
1151 	 * have scheduled a delayed reconnect however - in this case
1152 	 * we should not interfere.
1153 	 */
1154 	if (rdsv3_conn_state(conn) == RDSV3_CONN_DOWN &&
1155 	    !test_and_set_bit(RDSV3_RECONNECT_PENDING, &conn->c_flags))
1156 		rdsv3_queue_delayed_work(rdsv3_wq, &conn->c_conn_w, 0);
1157 
1158 	ret = rdsv3_cong_wait(conn->c_fcong, dport, 1, NULL);
1159 	if (ret)
1160 		goto out;
1161 
1162 	mutex_enter(&conn->c_lock);
1163 	list_insert_tail(&conn->c_send_queue, rm);
1164 	set_bit(RDSV3_MSG_ON_CONN, &rm->m_flags);
1165 	rdsv3_message_addref(rm);
1166 	rm->m_inc.i_conn = conn;
1167 
1168 	rdsv3_message_populate_header(&rm->m_inc.i_hdr, 0, dport,
1169 	    conn->c_next_tx_seq);
1170 	conn->c_next_tx_seq++;
1171 	mutex_exit(&conn->c_lock);
1172 
1173 	rdsv3_stats_inc(s_send_queued);
1174 	rdsv3_stats_inc(s_send_pong);
1175 
1176 	rdsv3_queue_delayed_work(rdsv3_wq, &conn->c_send_w, 0);
1177 	rdsv3_message_put(rm);
1178 
1179 	RDSV3_DPRINTF4("rdsv3_send_pong", "Return(conn: %p)", conn);
1180 	return (0);
1181 
1182 out:
1183 	if (rm)
1184 		rdsv3_message_put(rm);
1185 	return (ret);
1186 }
1187