projects / deliverable / linux.git / blame
? search:
summary | shortlog | log | commit | commitdiff | tree
blob | blame (incremental) | history | HEAD
stating/lustre: only build if configured as module
[deliverable/linux.git] / drivers / staging / lustre / lnet / klnds / socklnd / socklnd_cb.c
CommitLineData
d7e09d03
PT		 1/*
2 * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
3 *
4 * Copyright (c) 2011, 2012, Intel Corporation.
5 *
6 * Author: Zach Brown <zab@zabbo.net>
7 * Author: Peter J. Braam <braam@clusterfs.com>
8 * Author: Phil Schwan <phil@clusterfs.com>
9 * Author: Eric Barton <eric@bartonsoftware.com>
10 *
11 * This file is part of Portals, http://www.sf.net/projects/sandiaportals/
12 *
13 * Portals is free software; you can redistribute it and/or
14 * modify it under the terms of version 2 of the GNU General Public
15 * License as published by the Free Software Foundation.
16 *
17 * Portals is distributed in the hope that it will be useful,
18 * but WITHOUT ANY WARRANTY; without even the implied warranty of
19 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
20 * GNU General Public License for more details.
21 *
22 * You should have received a copy of the GNU General Public License
23 * along with Portals; if not, write to the Free Software
24 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
25 */
26
27#include "socklnd.h"
28
29ksock_tx_t *
30ksocknal_alloc_tx(int type, int size)
31{
32 ksock_tx_t *tx = NULL;
33
34 if (type == KSOCK_MSG_NOOP) {
35 LASSERT(size == KSOCK_NOOP_TX_SIZE);
36
37 /* searching for a noop tx in free list */
38 spin_lock(&ksocknal_data.ksnd_tx_lock);
39
40 if (!list_empty(&ksocknal_data.ksnd_idle_noop_txs)) {
41 tx = list_entry(ksocknal_data.ksnd_idle_noop_txs. \
42 next, ksock_tx_t, tx_list);
43 LASSERT(tx->tx_desc_size == size);
44 list_del(&tx->tx_list);
45 }
46
47 spin_unlock(&ksocknal_data.ksnd_tx_lock);
48 }
49
50 if (tx == NULL)
51 LIBCFS_ALLOC(tx, size);
52
53 if (tx == NULL)
54 return NULL;
55
56 atomic_set(&tx->tx_refcount, 1);
57 tx->tx_zc_aborted = 0;
58 tx->tx_zc_capable = 0;
59 tx->tx_zc_checked = 0;
60 tx->tx_desc_size = size;
61
62 atomic_inc(&ksocknal_data.ksnd_nactive_txs);
63
64 return tx;
65}
66
67ksock_tx_t *
68ksocknal_alloc_tx_noop(__u64 cookie, int nonblk)
69{
70 ksock_tx_t *tx;
71
72 tx = ksocknal_alloc_tx(KSOCK_MSG_NOOP, KSOCK_NOOP_TX_SIZE);
73 if (tx == NULL) {
74 CERROR("Can't allocate noop tx desc\n");
75 return NULL;
76 }
77
78 tx->tx_conn = NULL;
79 tx->tx_lnetmsg = NULL;
80 tx->tx_kiov = NULL;
81 tx->tx_nkiov = 0;
82 tx->tx_iov = tx->tx_frags.virt.iov;
83 tx->tx_niov = 1;
84 tx->tx_nonblk = nonblk;
85
86 socklnd_init_msg(&tx->tx_msg, KSOCK_MSG_NOOP);
87 tx->tx_msg.ksm_zc_cookies[1] = cookie;
88
89 return tx;
90}
91
92
93void
94ksocknal_free_tx (ksock_tx_t *tx)
95{
96 atomic_dec(&ksocknal_data.ksnd_nactive_txs);
97
98 if (tx->tx_lnetmsg == NULL && tx->tx_desc_size == KSOCK_NOOP_TX_SIZE) {
99 /* it's a noop tx */
100 spin_lock(&ksocknal_data.ksnd_tx_lock);
101
102 list_add(&tx->tx_list, &ksocknal_data.ksnd_idle_noop_txs);
103
104 spin_unlock(&ksocknal_data.ksnd_tx_lock);
105 } else {
106 LIBCFS_FREE(tx, tx->tx_desc_size);
107 }
108}
109
110int
111ksocknal_send_iov (ksock_conn_t *conn, ksock_tx_t *tx)
112{
113 struct iovec *iov = tx->tx_iov;
114 int nob;
115 int rc;
116
117 LASSERT (tx->tx_niov > 0);
118
119 /* Never touch tx->tx_iov inside ksocknal_lib_send_iov() */
120 rc = ksocknal_lib_send_iov(conn, tx);
121
122 if (rc <= 0) /* sent nothing? */
123 return (rc);
124
125 nob = rc;
126 LASSERT (nob <= tx->tx_resid);
127 tx->tx_resid -= nob;
128
129 /* "consume" iov */
130 do {
131 LASSERT (tx->tx_niov > 0);
132
133 if (nob < (int) iov->iov_len) {
134 iov->iov_base = (void *)((char *)iov->iov_base + nob);
135 iov->iov_len -= nob;
136 return (rc);
137 }
138
139 nob -= iov->iov_len;
140 tx->tx_iov = ++iov;
141 tx->tx_niov--;
142 } while (nob != 0);
143
144 return (rc);
145}
146
147int
148ksocknal_send_kiov (ksock_conn_t *conn, ksock_tx_t *tx)
149{
150 lnet_kiov_t *kiov = tx->tx_kiov;
151 int nob;
152 int rc;
153
154 LASSERT (tx->tx_niov == 0);
155 LASSERT (tx->tx_nkiov > 0);
156
157 /* Never touch tx->tx_kiov inside ksocknal_lib_send_kiov() */
158 rc = ksocknal_lib_send_kiov(conn, tx);
159
160 if (rc <= 0) /* sent nothing? */
161 return (rc);
162
163 nob = rc;
164 LASSERT (nob <= tx->tx_resid);
165 tx->tx_resid -= nob;
166
167 /* "consume" kiov */
168 do {
169 LASSERT(tx->tx_nkiov > 0);
170
171 if (nob < (int)kiov->kiov_len) {
172 kiov->kiov_offset += nob;
173 kiov->kiov_len -= nob;
174 return rc;
175 }
176
177 nob -= (int)kiov->kiov_len;
178 tx->tx_kiov = ++kiov;
179 tx->tx_nkiov--;
180 } while (nob != 0);
181
182 return (rc);
183}
184
185int
186ksocknal_transmit (ksock_conn_t *conn, ksock_tx_t *tx)
187{
188 int rc;
189 int bufnob;
190
191 if (ksocknal_data.ksnd_stall_tx != 0) {
192 cfs_pause(cfs_time_seconds(ksocknal_data.ksnd_stall_tx));
193 }
194
195 LASSERT (tx->tx_resid != 0);
196
197 rc = ksocknal_connsock_addref(conn);
198 if (rc != 0) {
199 LASSERT (conn->ksnc_closing);
200 return (-ESHUTDOWN);
201 }
202
203 do {
204 if (ksocknal_data.ksnd_enomem_tx > 0) {
205 /* testing... */
206 ksocknal_data.ksnd_enomem_tx--;
207 rc = -EAGAIN;
208 } else if (tx->tx_niov != 0) {
209 rc = ksocknal_send_iov (conn, tx);
210 } else {
211 rc = ksocknal_send_kiov (conn, tx);
212 }
213
214 bufnob = cfs_sock_wmem_queued(conn->ksnc_sock);
215 if (rc > 0) /* sent something? */
216 conn->ksnc_tx_bufnob += rc; /* account it */
217
218 if (bufnob < conn->ksnc_tx_bufnob) {
219 /* allocated send buffer bytes < computed; infer
220 * something got ACKed */
221 conn->ksnc_tx_deadline =
222 cfs_time_shift(*ksocknal_tunables.ksnd_timeout);
223 conn->ksnc_peer->ksnp_last_alive = cfs_time_current();
224 conn->ksnc_tx_bufnob = bufnob;
225 mb();
226 }
227
228 if (rc <= 0) { /* Didn't write anything? */
229
230 if (rc == 0) /* some stacks return 0 instead of -EAGAIN */
231 rc = -EAGAIN;
232
233 /* Check if EAGAIN is due to memory pressure */
234 if(rc == -EAGAIN && ksocknal_lib_memory_pressure(conn))
235 rc = -ENOMEM;
236
237 break;
238 }
239
240 /* socket's wmem_queued now includes 'rc' bytes */
241 atomic_sub (rc, &conn->ksnc_tx_nob);
242 rc = 0;
243
244 } while (tx->tx_resid != 0);
245
246 ksocknal_connsock_decref(conn);
247 return (rc);
248}
249
250int
251ksocknal_recv_iov (ksock_conn_t *conn)
252{
253 struct iovec *iov = conn->ksnc_rx_iov;
254 int nob;
255 int rc;
256
257 LASSERT (conn->ksnc_rx_niov > 0);
258
259 /* Never touch conn->ksnc_rx_iov or change connection
260 * status inside ksocknal_lib_recv_iov */
261 rc = ksocknal_lib_recv_iov(conn);
262
263 if (rc <= 0)
264 return (rc);
265
266 /* received something... */
267 nob = rc;
268
269 conn->ksnc_peer->ksnp_last_alive = cfs_time_current();
270 conn->ksnc_rx_deadline =
271 cfs_time_shift(*ksocknal_tunables.ksnd_timeout);
272 mb(); /* order with setting rx_started */
273 conn->ksnc_rx_started = 1;
274
275 conn->ksnc_rx_nob_wanted -= nob;
276 conn->ksnc_rx_nob_left -= nob;
277
278 do {
279 LASSERT (conn->ksnc_rx_niov > 0);
280
281 if (nob < (int)iov->iov_len) {
282 iov->iov_len -= nob;
283 iov->iov_base = (void *)((char *)iov->iov_base + nob);
284 return (-EAGAIN);
285 }
286
287 nob -= iov->iov_len;
288 conn->ksnc_rx_iov = ++iov;
289 conn->ksnc_rx_niov--;
290 } while (nob != 0);
291
292 return (rc);
293}
294
295int
296ksocknal_recv_kiov (ksock_conn_t *conn)
297{
298 lnet_kiov_t *kiov = conn->ksnc_rx_kiov;
299 int nob;
300 int rc;
301 LASSERT (conn->ksnc_rx_nkiov > 0);
302
303 /* Never touch conn->ksnc_rx_kiov or change connection
304 * status inside ksocknal_lib_recv_iov */
305 rc = ksocknal_lib_recv_kiov(conn);
306
307 if (rc <= 0)
308 return (rc);
309
310 /* received something... */
311 nob = rc;
312
313 conn->ksnc_peer->ksnp_last_alive = cfs_time_current();
314 conn->ksnc_rx_deadline =
315 cfs_time_shift(*ksocknal_tunables.ksnd_timeout);
316 mb(); /* order with setting rx_started */
317 conn->ksnc_rx_started = 1;
318
319 conn->ksnc_rx_nob_wanted -= nob;
320 conn->ksnc_rx_nob_left -= nob;
321
322 do {
323 LASSERT (conn->ksnc_rx_nkiov > 0);
324
325 if (nob < (int) kiov->kiov_len) {
326 kiov->kiov_offset += nob;
327 kiov->kiov_len -= nob;
328 return -EAGAIN;
329 }
330
331 nob -= kiov->kiov_len;
332 conn->ksnc_rx_kiov = ++kiov;
333 conn->ksnc_rx_nkiov--;
334 } while (nob != 0);
335
336 return 1;
337}
338
339int
340ksocknal_receive (ksock_conn_t *conn)
341{
342 /* Return 1 on success, 0 on EOF, < 0 on error.
343 * Caller checks ksnc_rx_nob_wanted to determine
344 * progress/completion. */
345 int rc;
346 ENTRY;
347
348 if (ksocknal_data.ksnd_stall_rx != 0) {
349 cfs_pause(cfs_time_seconds (ksocknal_data.ksnd_stall_rx));
350 }
351
352 rc = ksocknal_connsock_addref(conn);
353 if (rc != 0) {
354 LASSERT (conn->ksnc_closing);
355 return (-ESHUTDOWN);
356 }
357
358 for (;;) {
359 if (conn->ksnc_rx_niov != 0)
360 rc = ksocknal_recv_iov (conn);
361 else
362 rc = ksocknal_recv_kiov (conn);
363
364 if (rc <= 0) {
365 /* error/EOF or partial receive */
366 if (rc == -EAGAIN) {
367 rc = 1;
368 } else if (rc == 0 && conn->ksnc_rx_started) {
369 /* EOF in the middle of a message */
370 rc = -EPROTO;
371 }
372 break;
373 }
374
375 /* Completed a fragment */
376
377 if (conn->ksnc_rx_nob_wanted == 0) {
378 rc = 1;
379 break;
380 }
381 }
382
383 ksocknal_connsock_decref(conn);
384 RETURN (rc);
385}
386
387void
388ksocknal_tx_done (lnet_ni_t *ni, ksock_tx_t *tx)
389{
390 lnet_msg_t *lnetmsg = tx->tx_lnetmsg;
391 int rc = (tx->tx_resid == 0 && !tx->tx_zc_aborted) ? 0 : -EIO;
392 ENTRY;
393
394 LASSERT(ni != NULL || tx->tx_conn != NULL);
395
396 if (tx->tx_conn != NULL)
397 ksocknal_conn_decref(tx->tx_conn);
398
399 if (ni == NULL && tx->tx_conn != NULL)
400 ni = tx->tx_conn->ksnc_peer->ksnp_ni;
401
402 ksocknal_free_tx (tx);
403 if (lnetmsg != NULL) /* KSOCK_MSG_NOOP go without lnetmsg */
404 lnet_finalize (ni, lnetmsg, rc);
405
406 EXIT;
407}
408
409void
410ksocknal_txlist_done (lnet_ni_t *ni, struct list_head *txlist, int error)
411{
412 ksock_tx_t *tx;
413
414 while (!list_empty (txlist)) {
415 tx = list_entry (txlist->next, ksock_tx_t, tx_list);
416
417 if (error && tx->tx_lnetmsg != NULL) {
418 CNETERR("Deleting packet type %d len %d %s->%s\n",
419 le32_to_cpu (tx->tx_lnetmsg->msg_hdr.type),
420 le32_to_cpu (tx->tx_lnetmsg->msg_hdr.payload_length),
421 libcfs_nid2str(le64_to_cpu(tx->tx_lnetmsg->msg_hdr.src_nid)),
422 libcfs_nid2str(le64_to_cpu(tx->tx_lnetmsg->msg_hdr.dest_nid)));
423 } else if (error) {
424 CNETERR("Deleting noop packet\n");
425 }
426
427 list_del (&tx->tx_list);
428
429 LASSERT (atomic_read(&tx->tx_refcount) == 1);
430 ksocknal_tx_done (ni, tx);
431 }
432}
433
434static void
435ksocknal_check_zc_req(ksock_tx_t *tx)
436{
437 ksock_conn_t *conn = tx->tx_conn;
438 ksock_peer_t *peer = conn->ksnc_peer;
439
440 /* Set tx_msg.ksm_zc_cookies[0] to a unique non-zero cookie and add tx
441 * to ksnp_zc_req_list if some fragment of this message should be sent
442 * zero-copy. Our peer will send an ACK containing this cookie when
443 * she has received this message to tell us we can signal completion.
444 * tx_msg.ksm_zc_cookies[0] remains non-zero while tx is on
445 * ksnp_zc_req_list. */
446 LASSERT (tx->tx_msg.ksm_type != KSOCK_MSG_NOOP);
447 LASSERT (tx->tx_zc_capable);
448
449 tx->tx_zc_checked = 1;
450
451 if (conn->ksnc_proto == &ksocknal_protocol_v1x ||
452 !conn->ksnc_zc_capable)
453 return;
454
455 /* assign cookie and queue tx to pending list, it will be released when
456 * a matching ack is received. See ksocknal_handle_zcack() */
457
458 ksocknal_tx_addref(tx);
459
460 spin_lock(&peer->ksnp_lock);
461
462 /* ZC_REQ is going to be pinned to the peer */
463 tx->tx_deadline =
464 cfs_time_shift(*ksocknal_tunables.ksnd_timeout);
465
466 LASSERT (tx->tx_msg.ksm_zc_cookies[0] == 0);
467
468 tx->tx_msg.ksm_zc_cookies[0] = peer->ksnp_zc_next_cookie++;
469
470 if (peer->ksnp_zc_next_cookie == 0)
471 peer->ksnp_zc_next_cookie = SOCKNAL_KEEPALIVE_PING + 1;
472
473 list_add_tail(&tx->tx_zc_list, &peer->ksnp_zc_req_list);
474
475 spin_unlock(&peer->ksnp_lock);
476}
477
478static void
479ksocknal_uncheck_zc_req(ksock_tx_t *tx)
480{
481 ksock_peer_t *peer = tx->tx_conn->ksnc_peer;
482
483 LASSERT(tx->tx_msg.ksm_type != KSOCK_MSG_NOOP);
484 LASSERT(tx->tx_zc_capable);
485
486 tx->tx_zc_checked = 0;
487
488 spin_lock(&peer->ksnp_lock);
489
490 if (tx->tx_msg.ksm_zc_cookies[0] == 0) {
491 /* Not waiting for an ACK */
492 spin_unlock(&peer->ksnp_lock);
493 return;
494 }
495
496 tx->tx_msg.ksm_zc_cookies[0] = 0;
497 list_del(&tx->tx_zc_list);
498
499 spin_unlock(&peer->ksnp_lock);
500
501 ksocknal_tx_decref(tx);
502}
503
504int
505ksocknal_process_transmit (ksock_conn_t *conn, ksock_tx_t *tx)
506{
507 int rc;
508
509 if (tx->tx_zc_capable && !tx->tx_zc_checked)
510 ksocknal_check_zc_req(tx);
511
512 rc = ksocknal_transmit (conn, tx);
513
514 CDEBUG (D_NET, "send(%d) %d\n", tx->tx_resid, rc);
515
516 if (tx->tx_resid == 0) {
517 /* Sent everything OK */
518 LASSERT (rc == 0);
519
520 return (0);
521 }
522
523 if (rc == -EAGAIN)
524 return (rc);
525
526 if (rc == -ENOMEM) {
527 static int counter;
528
529 counter++; /* exponential backoff warnings */
530 if ((counter & (-counter)) == counter)
531 CWARN("%u ENOMEM tx %p (%u allocated)\n",
532 counter, conn, atomic_read(&libcfs_kmemory));
533
534 /* Queue on ksnd_enomem_conns for retry after a timeout */
535 spin_lock_bh(&ksocknal_data.ksnd_reaper_lock);
536
537 /* enomem list takes over scheduler's ref... */
538 LASSERT (conn->ksnc_tx_scheduled);
539 list_add_tail(&conn->ksnc_tx_list,
540 &ksocknal_data.ksnd_enomem_conns);
541 if (!cfs_time_aftereq(cfs_time_add(cfs_time_current(),
542 SOCKNAL_ENOMEM_RETRY),
543 ksocknal_data.ksnd_reaper_waketime))
544 wake_up (&ksocknal_data.ksnd_reaper_waitq);
545
546 spin_unlock_bh(&ksocknal_data.ksnd_reaper_lock);
547 return (rc);
548 }
549
550 /* Actual error */
551 LASSERT (rc < 0);
552
553 if (!conn->ksnc_closing) {
554 switch (rc) {
555 case -ECONNRESET:
556 LCONSOLE_WARN("Host %u.%u.%u.%u reset our connection "
557 "while we were sending data; it may have "
558 "rebooted.\n",
559 HIPQUAD(conn->ksnc_ipaddr));
560 break;
561 default:
562 LCONSOLE_WARN("There was an unexpected network error "
563 "while writing to %u.%u.%u.%u: %d.\n",
564 HIPQUAD(conn->ksnc_ipaddr), rc);
565 break;
566 }
567 CDEBUG(D_NET, "[%p] Error %d on write to %s"
568 " ip %d.%d.%d.%d:%d\n", conn, rc,
569 libcfs_id2str(conn->ksnc_peer->ksnp_id),
570 HIPQUAD(conn->ksnc_ipaddr),
571 conn->ksnc_port);
572 }
573
574 if (tx->tx_zc_checked)
575 ksocknal_uncheck_zc_req(tx);
576
577 /* it's not an error if conn is being closed */
578 ksocknal_close_conn_and_siblings (conn,
579 (conn->ksnc_closing) ? 0 : rc);
580
581 return (rc);
582}
583
584void
585ksocknal_launch_connection_locked (ksock_route_t *route)
586{
587
588 /* called holding write lock on ksnd_global_lock */
589
590 LASSERT (!route->ksnr_scheduled);
591 LASSERT (!route->ksnr_connecting);
592 LASSERT ((ksocknal_route_mask() & ~route->ksnr_connected) != 0);
593
594 route->ksnr_scheduled = 1; /* scheduling conn for connd */
595 ksocknal_route_addref(route); /* extra ref for connd */
596
597 spin_lock_bh(&ksocknal_data.ksnd_connd_lock);
598
599 list_add_tail(&route->ksnr_connd_list,
600 &ksocknal_data.ksnd_connd_routes);
601 wake_up(&ksocknal_data.ksnd_connd_waitq);
602
603 spin_unlock_bh(&ksocknal_data.ksnd_connd_lock);
604}
605
606void
607ksocknal_launch_all_connections_locked (ksock_peer_t *peer)
608{
609 ksock_route_t *route;
610
611 /* called holding write lock on ksnd_global_lock */
612 for (;;) {
613 /* launch any/all connections that need it */
614 route = ksocknal_find_connectable_route_locked(peer);
615 if (route == NULL)
616 return;
617
618 ksocknal_launch_connection_locked(route);
619 }
620}
621
622ksock_conn_t *
623ksocknal_find_conn_locked(ksock_peer_t *peer, ksock_tx_t *tx, int nonblk)
624{
625 struct list_head *tmp;
626 ksock_conn_t *conn;
627 ksock_conn_t *typed = NULL;
628 ksock_conn_t *fallback = NULL;
629 int tnob = 0;
630 int fnob = 0;
631
632 list_for_each (tmp, &peer->ksnp_conns) {
633 ksock_conn_t *c = list_entry(tmp, ksock_conn_t, ksnc_list);
634 int nob = atomic_read(&c->ksnc_tx_nob) +
635 cfs_sock_wmem_queued(c->ksnc_sock);
636 int rc;
637
638 LASSERT (!c->ksnc_closing);
639 LASSERT (c->ksnc_proto != NULL &&
640 c->ksnc_proto->pro_match_tx != NULL);
641
642 rc = c->ksnc_proto->pro_match_tx(c, tx, nonblk);
643
644 switch (rc) {
645 default:
646 LBUG();
647 case SOCKNAL_MATCH_NO: /* protocol rejected the tx */
648 continue;
649
650 case SOCKNAL_MATCH_YES: /* typed connection */
651 if (typed == NULL || tnob > nob ||
652 (tnob == nob && *ksocknal_tunables.ksnd_round_robin &&
653 cfs_time_after(typed->ksnc_tx_last_post, c->ksnc_tx_last_post))) {
654 typed = c;
655 tnob = nob;
656 }
657 break;
658
659 case SOCKNAL_MATCH_MAY: /* fallback connection */
660 if (fallback == NULL || fnob > nob ||
661 (fnob == nob && *ksocknal_tunables.ksnd_round_robin &&
662 cfs_time_after(fallback->ksnc_tx_last_post, c->ksnc_tx_last_post))) {
663 fallback = c;
664 fnob = nob;
665 }
666 break;
667 }
668 }
669
670 /* prefer the typed selection */
671 conn = (typed != NULL) ? typed : fallback;
672
673 if (conn != NULL)
674 conn->ksnc_tx_last_post = cfs_time_current();
675
676 return conn;
677}
678
679void
680ksocknal_tx_prep(ksock_conn_t *conn, ksock_tx_t *tx)
681{
682 conn->ksnc_proto->pro_pack(tx);
683
684 atomic_add (tx->tx_nob, &conn->ksnc_tx_nob);
685 ksocknal_conn_addref(conn); /* +1 ref for tx */
686 tx->tx_conn = conn;
687}
688
689void
690ksocknal_queue_tx_locked (ksock_tx_t *tx, ksock_conn_t *conn)
691{
692 ksock_sched_t *sched = conn->ksnc_scheduler;
693 ksock_msg_t *msg = &tx->tx_msg;
694 ksock_tx_t *ztx = NULL;
695 int bufnob = 0;
696
697 /* called holding global lock (read or irq-write) and caller may
698 * not have dropped this lock between finding conn and calling me,
699 * so we don't need the {get,put}connsock dance to deref
700 * ksnc_sock... */
701 LASSERT(!conn->ksnc_closing);
702
703 CDEBUG (D_NET, "Sending to %s ip %d.%d.%d.%d:%d\n",
704 libcfs_id2str(conn->ksnc_peer->ksnp_id),
705 HIPQUAD(conn->ksnc_ipaddr),
706 conn->ksnc_port);
707
708 ksocknal_tx_prep(conn, tx);
709
710 /* Ensure the frags we've been given EXACTLY match the number of
711 * bytes we want to send. Many TCP/IP stacks disregard any total
712 * size parameters passed to them and just look at the frags.
713 *
714 * We always expect at least 1 mapped fragment containing the
715 * complete ksocknal message header. */
716 LASSERT (lnet_iov_nob (tx->tx_niov, tx->tx_iov) +
717 lnet_kiov_nob(tx->tx_nkiov, tx->tx_kiov) ==
718 (unsigned int)tx->tx_nob);
719 LASSERT (tx->tx_niov >= 1);
720 LASSERT (tx->tx_resid == tx->tx_nob);
721
722 CDEBUG (D_NET, "Packet %p type %d, nob %d niov %d nkiov %d\n",
723 tx, (tx->tx_lnetmsg != NULL) ? tx->tx_lnetmsg->msg_hdr.type:
724 KSOCK_MSG_NOOP,
725 tx->tx_nob, tx->tx_niov, tx->tx_nkiov);
726
727 /*
728 * FIXME: SOCK_WMEM_QUEUED and SOCK_ERROR could block in __DARWIN8__
729 * but they're used inside spinlocks a lot.
730 */
731 bufnob = cfs_sock_wmem_queued(conn->ksnc_sock);
732 spin_lock_bh(&sched->kss_lock);
733
734 if (list_empty(&conn->ksnc_tx_queue) && bufnob == 0) {
735 /* First packet starts the timeout */
736 conn->ksnc_tx_deadline =
737 cfs_time_shift(*ksocknal_tunables.ksnd_timeout);
738 if (conn->ksnc_tx_bufnob > 0) /* something got ACKed */
739 conn->ksnc_peer->ksnp_last_alive = cfs_time_current();
740 conn->ksnc_tx_bufnob = 0;
741 mb(); /* order with adding to tx_queue */
742 }
743
744 if (msg->ksm_type == KSOCK_MSG_NOOP) {
745 /* The packet is noop ZC ACK, try to piggyback the ack_cookie
746 * on a normal packet so I don't need to send it */
747 LASSERT (msg->ksm_zc_cookies[1] != 0);
748 LASSERT (conn->ksnc_proto->pro_queue_tx_zcack != NULL);
749
750 if (conn->ksnc_proto->pro_queue_tx_zcack(conn, tx, 0))
751 ztx = tx; /* ZC ACK piggybacked on ztx release tx later */
752
753 } else {
754 /* It's a normal packet - can it piggback a noop zc-ack that
755 * has been queued already? */
756 LASSERT (msg->ksm_zc_cookies[1] == 0);
757 LASSERT (conn->ksnc_proto->pro_queue_tx_msg != NULL);
758
759 ztx = conn->ksnc_proto->pro_queue_tx_msg(conn, tx);
760 /* ztx will be released later */
761 }
762
763 if (ztx != NULL) {
764 atomic_sub (ztx->tx_nob, &conn->ksnc_tx_nob);
765 list_add_tail(&ztx->tx_list, &sched->kss_zombie_noop_txs);
766 }
767
768 if (conn->ksnc_tx_ready && /* able to send */
769 !conn->ksnc_tx_scheduled) { /* not scheduled to send */
770 /* +1 ref for scheduler */
771 ksocknal_conn_addref(conn);
772 list_add_tail (&conn->ksnc_tx_list,
773 &sched->kss_tx_conns);
774 conn->ksnc_tx_scheduled = 1;
775 wake_up (&sched->kss_waitq);
776 }
777
778 spin_unlock_bh(&sched->kss_lock);
779}
780
781
782ksock_route_t *
783ksocknal_find_connectable_route_locked (ksock_peer_t *peer)
784{
785 cfs_time_t now = cfs_time_current();
786 struct list_head *tmp;
787 ksock_route_t *route;
788
789 list_for_each (tmp, &peer->ksnp_routes) {
790 route = list_entry (tmp, ksock_route_t, ksnr_list);
791
792 LASSERT (!route->ksnr_connecting || route->ksnr_scheduled);
793
794 if (route->ksnr_scheduled) /* connections being established */
795 continue;
796
797 /* all route types connected ? */
798 if ((ksocknal_route_mask() & ~route->ksnr_connected) == 0)
799 continue;
800
801 if (!(route->ksnr_retry_interval == 0 || /* first attempt */
802 cfs_time_aftereq(now, route->ksnr_timeout))) {
803 CDEBUG(D_NET,
804 "Too soon to retry route %u.%u.%u.%u "
805 "(cnted %d, interval %ld, %ld secs later)\n",
806 HIPQUAD(route->ksnr_ipaddr),
807 route->ksnr_connected,
808 route->ksnr_retry_interval,
809 cfs_duration_sec(route->ksnr_timeout - now));
810 continue;
811 }
812
813 return (route);
814 }
815
816 return (NULL);
817}
818
819ksock_route_t *
820ksocknal_find_connecting_route_locked (ksock_peer_t *peer)
821{
822 struct list_head *tmp;
823 ksock_route_t *route;
824
825 list_for_each (tmp, &peer->ksnp_routes) {
826 route = list_entry (tmp, ksock_route_t, ksnr_list);
827
828 LASSERT (!route->ksnr_connecting || route->ksnr_scheduled);
829
830 if (route->ksnr_scheduled)
831 return (route);
832 }
833
834 return (NULL);
835}
836
837int
838ksocknal_launch_packet (lnet_ni_t *ni, ksock_tx_t *tx, lnet_process_id_t id)
839{
840 ksock_peer_t *peer;
841 ksock_conn_t *conn;
842 rwlock_t *g_lock;
843 int retry;
844 int rc;
845
846 LASSERT (tx->tx_conn == NULL);
847
848 g_lock = &ksocknal_data.ksnd_global_lock;
849
850 for (retry = 0;; retry = 1) {
851 read_lock(g_lock);
852 peer = ksocknal_find_peer_locked(ni, id);
853 if (peer != NULL) {
854 if (ksocknal_find_connectable_route_locked(peer) == NULL) {
855 conn = ksocknal_find_conn_locked(peer, tx, tx->tx_nonblk);
856 if (conn != NULL) {
857 /* I've got no routes that need to be
858 * connecting and I do have an actual
859 * connection... */
860 ksocknal_queue_tx_locked (tx, conn);
861 read_unlock(g_lock);
862 return (0);
863 }
864 }
865 }
866
867 /* I'll need a write lock... */
868 read_unlock(g_lock);
869
870 write_lock_bh(g_lock);
871
872 peer = ksocknal_find_peer_locked(ni, id);
873 if (peer != NULL)
874 break;
875
876 write_unlock_bh(g_lock);
877
878 if ((id.pid & LNET_PID_USERFLAG) != 0) {
879 CERROR("Refusing to create a connection to "
880 "userspace process %s\n", libcfs_id2str(id));
881 return -EHOSTUNREACH;
882 }
883
884 if (retry) {
885 CERROR("Can't find peer %s\n", libcfs_id2str(id));
886 return -EHOSTUNREACH;
887 }
888
889 rc = ksocknal_add_peer(ni, id,
890 LNET_NIDADDR(id.nid),
891 lnet_acceptor_port());
892 if (rc != 0) {
893 CERROR("Can't add peer %s: %d\n",
894 libcfs_id2str(id), rc);
895 return rc;
896 }
897 }
898
899 ksocknal_launch_all_connections_locked(peer);
900
901 conn = ksocknal_find_conn_locked(peer, tx, tx->tx_nonblk);
902 if (conn != NULL) {
903 /* Connection exists; queue message on it */
904 ksocknal_queue_tx_locked (tx, conn);
905 write_unlock_bh(g_lock);
906 return (0);
907 }
908
909 if (peer->ksnp_accepting > 0 ||
910 ksocknal_find_connecting_route_locked (peer) != NULL) {
911 /* the message is going to be pinned to the peer */
912 tx->tx_deadline =
913 cfs_time_shift(*ksocknal_tunables.ksnd_timeout);
914
915 /* Queue the message until a connection is established */
916 list_add_tail (&tx->tx_list, &peer->ksnp_tx_queue);
917 write_unlock_bh(g_lock);
918 return 0;
919 }
920
921 write_unlock_bh(g_lock);
922
923 /* NB Routes may be ignored if connections to them failed recently */
924 CNETERR("No usable routes to %s\n", libcfs_id2str(id));
925 return (-EHOSTUNREACH);
926}
927
928int
929ksocknal_send(lnet_ni_t *ni, void *private, lnet_msg_t *lntmsg)
930{
931 int mpflag = 0;
932 int type = lntmsg->msg_type;
933 lnet_process_id_t target = lntmsg->msg_target;
934 unsigned int payload_niov = lntmsg->msg_niov;
935 struct iovec *payload_iov = lntmsg->msg_iov;
936 lnet_kiov_t *payload_kiov = lntmsg->msg_kiov;
937 unsigned int payload_offset = lntmsg->msg_offset;
938 unsigned int payload_nob = lntmsg->msg_len;
939 ksock_tx_t *tx;
940 int desc_size;
941 int rc;
942
943 /* NB 'private' is different depending on what we're sending.
944 * Just ignore it... */
945
946 CDEBUG(D_NET, "sending %u bytes in %d frags to %s\n",
947 payload_nob, payload_niov, libcfs_id2str(target));
948
949 LASSERT (payload_nob == 0 || payload_niov > 0);
950 LASSERT (payload_niov <= LNET_MAX_IOV);
951 /* payload is either all vaddrs or all pages */
952 LASSERT (!(payload_kiov != NULL && payload_iov != NULL));
953 LASSERT (!in_interrupt ());
954
955 if (payload_iov != NULL)
956 desc_size = offsetof(ksock_tx_t,
957 tx_frags.virt.iov[1 + payload_niov]);
958 else
959 desc_size = offsetof(ksock_tx_t,
960 tx_frags.paged.kiov[payload_niov]);
961
962 if (lntmsg->msg_vmflush)
963 mpflag = cfs_memory_pressure_get_and_set();
964 tx = ksocknal_alloc_tx(KSOCK_MSG_LNET, desc_size);
965 if (tx == NULL) {
966 CERROR("Can't allocate tx desc type %d size %d\n",
967 type, desc_size);
968 if (lntmsg->msg_vmflush)
969 cfs_memory_pressure_restore(mpflag);
970 return (-ENOMEM);
971 }
972
973 tx->tx_conn = NULL; /* set when assigned a conn */
974 tx->tx_lnetmsg = lntmsg;
975
976 if (payload_iov != NULL) {
977 tx->tx_kiov = NULL;
978 tx->tx_nkiov = 0;
979 tx->tx_iov = tx->tx_frags.virt.iov;
980 tx->tx_niov = 1 +
981 lnet_extract_iov(payload_niov, &tx->tx_iov[1],
982 payload_niov, payload_iov,
983 payload_offset, payload_nob);
984 } else {
985 tx->tx_niov = 1;
986 tx->tx_iov = &tx->tx_frags.paged.iov;
987 tx->tx_kiov = tx->tx_frags.paged.kiov;
988 tx->tx_nkiov = lnet_extract_kiov(payload_niov, tx->tx_kiov,
989 payload_niov, payload_kiov,
990 payload_offset, payload_nob);
991
992 if (payload_nob >= *ksocknal_tunables.ksnd_zc_min_payload)
993 tx->tx_zc_capable = 1;
994 }
995
996 socklnd_init_msg(&tx->tx_msg, KSOCK_MSG_LNET);
997
998 /* The first fragment will be set later in pro_pack */
999 rc = ksocknal_launch_packet(ni, tx, target);
1000 if (lntmsg->msg_vmflush)
1001 cfs_memory_pressure_restore(mpflag);
1002 if (rc == 0)
1003 return (0);
1004
1005 ksocknal_free_tx(tx);
1006 return (-EIO);
1007}
1008
1009int
1010ksocknal_thread_start(int (*fn)(void *arg), void *arg, char *name)
1011{
1012 task_t *task = kthread_run(fn, arg, name);
1013
1014 if (IS_ERR(task))
1015 return PTR_ERR(task);
1016
1017 write_lock_bh(&ksocknal_data.ksnd_global_lock);
1018 ksocknal_data.ksnd_nthreads++;
1019 write_unlock_bh(&ksocknal_data.ksnd_global_lock);
1020 return 0;
1021}
1022
1023void
1024ksocknal_thread_fini (void)
1025{
1026 write_lock_bh(&ksocknal_data.ksnd_global_lock);
1027 ksocknal_data.ksnd_nthreads--;
1028 write_unlock_bh(&ksocknal_data.ksnd_global_lock);
1029}
1030
1031int
1032ksocknal_new_packet (ksock_conn_t *conn, int nob_to_skip)
1033{
1034 static char ksocknal_slop_buffer[4096];
1035
1036 int nob;
1037 unsigned int niov;
1038 int skipped;
1039
1040 LASSERT(conn->ksnc_proto != NULL);
1041
1042 if ((*ksocknal_tunables.ksnd_eager_ack & conn->ksnc_type) != 0) {
1043 /* Remind the socket to ack eagerly... */
1044 ksocknal_lib_eager_ack(conn);
1045 }
1046
1047 if (nob_to_skip == 0) { /* right at next packet boundary now */
1048 conn->ksnc_rx_started = 0;
1049 mb(); /* racing with timeout thread */
1050
1051 switch (conn->ksnc_proto->pro_version) {
1052 case KSOCK_PROTO_V2:
1053 case KSOCK_PROTO_V3:
1054 conn->ksnc_rx_state = SOCKNAL_RX_KSM_HEADER;
1055 conn->ksnc_rx_iov = (struct iovec *)&conn->ksnc_rx_iov_space;
1056 conn->ksnc_rx_iov[0].iov_base = (char *)&conn->ksnc_msg;
1057
1058 conn->ksnc_rx_nob_wanted = offsetof(ksock_msg_t, ksm_u);
1059 conn->ksnc_rx_nob_left = offsetof(ksock_msg_t, ksm_u);
1060 conn->ksnc_rx_iov[0].iov_len = offsetof(ksock_msg_t, ksm_u);
1061 break;
1062
1063 case KSOCK_PROTO_V1:
1064 /* Receiving bare lnet_hdr_t */
1065 conn->ksnc_rx_state = SOCKNAL_RX_LNET_HEADER;
1066 conn->ksnc_rx_nob_wanted = sizeof(lnet_hdr_t);
1067 conn->ksnc_rx_nob_left = sizeof(lnet_hdr_t);
1068
1069 conn->ksnc_rx_iov = (struct iovec *)&conn->ksnc_rx_iov_space;
1070 conn->ksnc_rx_iov[0].iov_base = (char *)&conn->ksnc_msg.ksm_u.lnetmsg;
1071 conn->ksnc_rx_iov[0].iov_len = sizeof (lnet_hdr_t);
1072 break;
1073
1074 default:
1075 LBUG ();
1076 }
1077 conn->ksnc_rx_niov = 1;
1078
1079 conn->ksnc_rx_kiov = NULL;
1080 conn->ksnc_rx_nkiov = 0;
1081 conn->ksnc_rx_csum = ~0;
1082 return (1);
1083 }
1084
1085 /* Set up to skip as much as possible now. If there's more left
1086 * (ran out of iov entries) we'll get called again */
1087
1088 conn->ksnc_rx_state = SOCKNAL_RX_SLOP;
1089 conn->ksnc_rx_nob_left = nob_to_skip;
1090 conn->ksnc_rx_iov = (struct iovec *)&conn->ksnc_rx_iov_space;
1091 skipped = 0;
1092 niov = 0;
1093
1094 do {
1095 nob = MIN (nob_to_skip, sizeof (ksocknal_slop_buffer));
1096
1097 conn->ksnc_rx_iov[niov].iov_base = ksocknal_slop_buffer;
1098 conn->ksnc_rx_iov[niov].iov_len = nob;
1099 niov++;
1100 skipped += nob;
1101 nob_to_skip -=nob;
1102
1103 } while (nob_to_skip != 0 && /* mustn't overflow conn's rx iov */
1104 niov < sizeof(conn->ksnc_rx_iov_space) / sizeof (struct iovec));
1105
1106 conn->ksnc_rx_niov = niov;
1107 conn->ksnc_rx_kiov = NULL;
1108 conn->ksnc_rx_nkiov = 0;
1109 conn->ksnc_rx_nob_wanted = skipped;
1110 return (0);
1111}
1112
1113int
1114ksocknal_process_receive (ksock_conn_t *conn)
1115{
1116 lnet_hdr_t *lhdr;
1117 lnet_process_id_t *id;
1118 int rc;
1119
1120 LASSERT (atomic_read(&conn->ksnc_conn_refcount) > 0);
1121
1122 /* NB: sched lock NOT held */
1123 /* SOCKNAL_RX_LNET_HEADER is here for backward compatability */
1124 LASSERT (conn->ksnc_rx_state == SOCKNAL_RX_KSM_HEADER ||
1125 conn->ksnc_rx_state == SOCKNAL_RX_LNET_PAYLOAD ||
1126 conn->ksnc_rx_state == SOCKNAL_RX_LNET_HEADER ||
1127 conn->ksnc_rx_state == SOCKNAL_RX_SLOP);
1128 again:
1129 if (conn->ksnc_rx_nob_wanted != 0) {
1130 rc = ksocknal_receive(conn);
1131
1132 if (rc <= 0) {
1133 LASSERT (rc != -EAGAIN);
1134
1135 if (rc == 0)
1136 CDEBUG (D_NET, "[%p] EOF from %s"
1137 " ip %d.%d.%d.%d:%d\n", conn,
1138 libcfs_id2str(conn->ksnc_peer->ksnp_id),
1139 HIPQUAD(conn->ksnc_ipaddr),
1140 conn->ksnc_port);
1141 else if (!conn->ksnc_closing)
1142 CERROR ("[%p] Error %d on read from %s"
1143 " ip %d.%d.%d.%d:%d\n",
1144 conn, rc,
1145 libcfs_id2str(conn->ksnc_peer->ksnp_id),
1146 HIPQUAD(conn->ksnc_ipaddr),
1147 conn->ksnc_port);
1148
1149 /* it's not an error if conn is being closed */
1150 ksocknal_close_conn_and_siblings (conn,
1151 (conn->ksnc_closing) ? 0 : rc);
1152 return (rc == 0 ? -ESHUTDOWN : rc);
1153 }
1154
1155 if (conn->ksnc_rx_nob_wanted != 0) {
1156 /* short read */
1157 return (-EAGAIN);
1158 }
1159 }
1160 switch (conn->ksnc_rx_state) {
1161 case SOCKNAL_RX_KSM_HEADER:
1162 if (conn->ksnc_flip) {
1163 __swab32s(&conn->ksnc_msg.ksm_type);
1164 __swab32s(&conn->ksnc_msg.ksm_csum);
1165 __swab64s(&conn->ksnc_msg.ksm_zc_cookies[0]);
1166 __swab64s(&conn->ksnc_msg.ksm_zc_cookies[1]);
1167 }
1168
1169 if (conn->ksnc_msg.ksm_type != KSOCK_MSG_NOOP &&
1170 conn->ksnc_msg.ksm_type != KSOCK_MSG_LNET) {
1171 CERROR("%s: Unknown message type: %x\n",
1172 libcfs_id2str(conn->ksnc_peer->ksnp_id),
1173 conn->ksnc_msg.ksm_type);
1174 ksocknal_new_packet(conn, 0);
1175 ksocknal_close_conn_and_siblings(conn, -EPROTO);
1176 return (-EPROTO);
1177 }
1178
1179 if (conn->ksnc_msg.ksm_type == KSOCK_MSG_NOOP &&
1180 conn->ksnc_msg.ksm_csum != 0 && /* has checksum */
1181 conn->ksnc_msg.ksm_csum != conn->ksnc_rx_csum) {
1182 /* NOOP Checksum error */
1183 CERROR("%s: Checksum error, wire:0x%08X data:0x%08X\n",
1184 libcfs_id2str(conn->ksnc_peer->ksnp_id),
1185 conn->ksnc_msg.ksm_csum, conn->ksnc_rx_csum);
1186 ksocknal_new_packet(conn, 0);
1187 ksocknal_close_conn_and_siblings(conn, -EPROTO);
1188 return (-EIO);
1189 }
1190
1191 if (conn->ksnc_msg.ksm_zc_cookies[1] != 0) {
1192 __u64 cookie = 0;
1193
1194 LASSERT (conn->ksnc_proto != &ksocknal_protocol_v1x);
1195
1196 if (conn->ksnc_msg.ksm_type == KSOCK_MSG_NOOP)
1197 cookie = conn->ksnc_msg.ksm_zc_cookies[0];
1198
1199 rc = conn->ksnc_proto->pro_handle_zcack(conn, cookie,
1200 conn->ksnc_msg.ksm_zc_cookies[1]);
1201
1202 if (rc != 0) {
1203 CERROR("%s: Unknown ZC-ACK cookie: "LPU64", "LPU64"\n",
1204 libcfs_id2str(conn->ksnc_peer->ksnp_id),
1205 cookie, conn->ksnc_msg.ksm_zc_cookies[1]);
1206 ksocknal_new_packet(conn, 0);
1207 ksocknal_close_conn_and_siblings(conn, -EPROTO);
1208 return (rc);
1209 }
1210 }
1211
1212 if (conn->ksnc_msg.ksm_type == KSOCK_MSG_NOOP) {
1213 ksocknal_new_packet (conn, 0);
1214 return 0; /* NOOP is done and just return */
1215 }
1216
1217 conn->ksnc_rx_state = SOCKNAL_RX_LNET_HEADER;
1218 conn->ksnc_rx_nob_wanted = sizeof(ksock_lnet_msg_t);
1219 conn->ksnc_rx_nob_left = sizeof(ksock_lnet_msg_t);
1220
1221 conn->ksnc_rx_iov = (struct iovec *)&conn->ksnc_rx_iov_space;
1222 conn->ksnc_rx_iov[0].iov_base = (char *)&conn->ksnc_msg.ksm_u.lnetmsg;
1223 conn->ksnc_rx_iov[0].iov_len = sizeof(ksock_lnet_msg_t);
1224
1225 conn->ksnc_rx_niov = 1;
1226 conn->ksnc_rx_kiov = NULL;
1227 conn->ksnc_rx_nkiov = 0;
1228
1229 goto again; /* read lnet header now */
1230
1231 case SOCKNAL_RX_LNET_HEADER:
1232 /* unpack message header */
1233 conn->ksnc_proto->pro_unpack(&conn->ksnc_msg);
1234
1235 if ((conn->ksnc_peer->ksnp_id.pid & LNET_PID_USERFLAG) != 0) {
1236 /* Userspace peer */
1237 lhdr = &conn->ksnc_msg.ksm_u.lnetmsg.ksnm_hdr;
1238 id = &conn->ksnc_peer->ksnp_id;
1239
1240 /* Substitute process ID assigned at connection time */
1241 lhdr->src_pid = cpu_to_le32(id->pid);
1242 lhdr->src_nid = cpu_to_le64(id->nid);
1243 }
1244
1245 conn->ksnc_rx_state = SOCKNAL_RX_PARSE;
1246 ksocknal_conn_addref(conn); /* ++ref while parsing */
1247
1248 rc = lnet_parse(conn->ksnc_peer->ksnp_ni,
1249 &conn->ksnc_msg.ksm_u.lnetmsg.ksnm_hdr,
1250 conn->ksnc_peer->ksnp_id.nid, conn, 0);
1251 if (rc < 0) {
1252 /* I just received garbage: give up on this conn */
1253 ksocknal_new_packet(conn, 0);
1254 ksocknal_close_conn_and_siblings (conn, rc);
1255 ksocknal_conn_decref(conn);
1256 return (-EPROTO);
1257 }
1258
1259 /* I'm racing with ksocknal_recv() */
1260 LASSERT (conn->ksnc_rx_state == SOCKNAL_RX_PARSE ||
1261 conn->ksnc_rx_state == SOCKNAL_RX_LNET_PAYLOAD);
1262
1263 if (conn->ksnc_rx_state != SOCKNAL_RX_LNET_PAYLOAD)
1264 return 0;
1265
1266 /* ksocknal_recv() got called */
1267 goto again;
1268
1269 case SOCKNAL_RX_LNET_PAYLOAD:
1270 /* payload all received */
1271 rc = 0;
1272
1273 if (conn->ksnc_rx_nob_left == 0 && /* not truncating */
1274 conn->ksnc_msg.ksm_csum != 0 && /* has checksum */
1275 conn->ksnc_msg.ksm_csum != conn->ksnc_rx_csum) {
1276 CERROR("%s: Checksum error, wire:0x%08X data:0x%08X\n",
1277 libcfs_id2str(conn->ksnc_peer->ksnp_id),
1278 conn->ksnc_msg.ksm_csum, conn->ksnc_rx_csum);
1279 rc = -EIO;
1280 }
1281
1282 if (rc == 0 && conn->ksnc_msg.ksm_zc_cookies[0] != 0) {
1283 LASSERT(conn->ksnc_proto != &ksocknal_protocol_v1x);
1284
1285 lhdr = &conn->ksnc_msg.ksm_u.lnetmsg.ksnm_hdr;
1286 id = &conn->ksnc_peer->ksnp_id;
1287
1288 rc = conn->ksnc_proto->pro_handle_zcreq(conn,
1289 conn->ksnc_msg.ksm_zc_cookies[0],
1290 *ksocknal_tunables.ksnd_nonblk_zcack ||
1291 le64_to_cpu(lhdr->src_nid) != id->nid);
1292 }
1293
1294 lnet_finalize(conn->ksnc_peer->ksnp_ni, conn->ksnc_cookie, rc);
1295
1296 if (rc != 0) {
1297 ksocknal_new_packet(conn, 0);
1298 ksocknal_close_conn_and_siblings (conn, rc);
1299 return (-EPROTO);
1300 }
1301 /* Fall through */
1302
1303 case SOCKNAL_RX_SLOP:
1304 /* starting new packet? */
1305 if (ksocknal_new_packet (conn, conn->ksnc_rx_nob_left))
1306 return 0; /* come back later */
1307 goto again; /* try to finish reading slop now */
1308
1309 default:
1310 break;
1311 }
1312
1313 /* Not Reached */
1314 LBUG ();
1315 return (-EINVAL); /* keep gcc happy */
1316}
1317
1318int
1319ksocknal_recv (lnet_ni_t *ni, void *private, lnet_msg_t *msg, int delayed,
1320 unsigned int niov, struct iovec *iov, lnet_kiov_t *kiov,
1321 unsigned int offset, unsigned int mlen, unsigned int rlen)
1322{
1323 ksock_conn_t *conn = (ksock_conn_t *)private;
1324 ksock_sched_t *sched = conn->ksnc_scheduler;
1325
1326 LASSERT (mlen <= rlen);
1327 LASSERT (niov <= LNET_MAX_IOV);
1328
1329 conn->ksnc_cookie = msg;
1330 conn->ksnc_rx_nob_wanted = mlen;
1331 conn->ksnc_rx_nob_left = rlen;
1332
1333 if (mlen == 0 || iov != NULL) {
1334 conn->ksnc_rx_nkiov = 0;
1335 conn->ksnc_rx_kiov = NULL;
1336 conn->ksnc_rx_iov = conn->ksnc_rx_iov_space.iov;
1337 conn->ksnc_rx_niov =
1338 lnet_extract_iov(LNET_MAX_IOV, conn->ksnc_rx_iov,
1339 niov, iov, offset, mlen);
1340 } else {
1341 conn->ksnc_rx_niov = 0;
1342 conn->ksnc_rx_iov = NULL;
1343 conn->ksnc_rx_kiov = conn->ksnc_rx_iov_space.kiov;
1344 conn->ksnc_rx_nkiov =
1345 lnet_extract_kiov(LNET_MAX_IOV, conn->ksnc_rx_kiov,
1346 niov, kiov, offset, mlen);
1347 }
1348
1349 LASSERT (mlen ==
1350 lnet_iov_nob (conn->ksnc_rx_niov, conn->ksnc_rx_iov) +
1351 lnet_kiov_nob (conn->ksnc_rx_nkiov, conn->ksnc_rx_kiov));
1352
1353 LASSERT (conn->ksnc_rx_scheduled);
1354
1355 spin_lock_bh(&sched->kss_lock);
1356
1357 switch (conn->ksnc_rx_state) {
1358 case SOCKNAL_RX_PARSE_WAIT:
1359 list_add_tail(&conn->ksnc_rx_list, &sched->kss_rx_conns);
1360 wake_up (&sched->kss_waitq);
1361 LASSERT (conn->ksnc_rx_ready);
1362 break;
1363
1364 case SOCKNAL_RX_PARSE:
1365 /* scheduler hasn't noticed I'm parsing yet */
1366 break;
1367 }
1368
1369 conn->ksnc_rx_state = SOCKNAL_RX_LNET_PAYLOAD;
1370
1371 spin_unlock_bh(&sched->kss_lock);
1372 ksocknal_conn_decref(conn);
1373 return 0;
1374}
1375
1376static inline int
1377ksocknal_sched_cansleep(ksock_sched_t *sched)
1378{
1379 int rc;
1380
1381 spin_lock_bh(&sched->kss_lock);
1382
1383 rc = (!ksocknal_data.ksnd_shuttingdown &&
1384 list_empty(&sched->kss_rx_conns) &&
1385 list_empty(&sched->kss_tx_conns));
1386
1387 spin_unlock_bh(&sched->kss_lock);
1388 return rc;
1389}
1390
1391int ksocknal_scheduler(void *arg)
1392{
1393 struct ksock_sched_info *info;
1394 ksock_sched_t *sched;
1395 ksock_conn_t *conn;
1396 ksock_tx_t *tx;
1397 int rc;
1398 int nloops = 0;
1399 long id = (long)arg;
1400
1401 info = ksocknal_data.ksnd_sched_info[KSOCK_THREAD_CPT(id)];
1402 sched = &info->ksi_scheds[KSOCK_THREAD_SID(id)];
1403
1404 cfs_block_allsigs();
1405
1406 rc = cfs_cpt_bind(lnet_cpt_table(), info->ksi_cpt);
1407 if (rc != 0) {
1408 CERROR("Can't set CPT affinity to %d: %d\n",
1409 info->ksi_cpt, rc);
1410 }
1411
1412 spin_lock_bh(&sched->kss_lock);
1413
1414 while (!ksocknal_data.ksnd_shuttingdown) {
1415 int did_something = 0;
1416
1417 /* Ensure I progress everything semi-fairly */
1418
1419 if (!list_empty (&sched->kss_rx_conns)) {
1420 conn = list_entry(sched->kss_rx_conns.next,
1421 ksock_conn_t, ksnc_rx_list);
1422 list_del(&conn->ksnc_rx_list);
1423
1424 LASSERT(conn->ksnc_rx_scheduled);
1425 LASSERT(conn->ksnc_rx_ready);
1426
1427 /* clear rx_ready in case receive isn't complete.
1428 * Do it BEFORE we call process_recv, since
1429 * data_ready can set it any time after we release
1430 * kss_lock. */
1431 conn->ksnc_rx_ready = 0;
1432 spin_unlock_bh(&sched->kss_lock);
1433
1434 rc = ksocknal_process_receive(conn);
1435
1436 spin_lock_bh(&sched->kss_lock);
1437
1438 /* I'm the only one that can clear this flag */
1439 LASSERT(conn->ksnc_rx_scheduled);
1440
1441 /* Did process_receive get everything it wanted? */
1442 if (rc == 0)
1443 conn->ksnc_rx_ready = 1;
1444
1445 if (conn->ksnc_rx_state == SOCKNAL_RX_PARSE) {
1446 /* Conn blocked waiting for ksocknal_recv()
1447 * I change its state (under lock) to signal
1448 * it can be rescheduled */
1449 conn->ksnc_rx_state = SOCKNAL_RX_PARSE_WAIT;
1450 } else if (conn->ksnc_rx_ready) {
1451 /* reschedule for rx */
1452 list_add_tail (&conn->ksnc_rx_list,
1453 &sched->kss_rx_conns);
1454 } else {
1455 conn->ksnc_rx_scheduled = 0;
1456 /* drop my ref */
1457 ksocknal_conn_decref(conn);
1458 }
1459
1460 did_something = 1;
1461 }
1462
1463 if (!list_empty (&sched->kss_tx_conns)) {
1464 LIST_HEAD (zlist);
1465
1466 if (!list_empty(&sched->kss_zombie_noop_txs)) {
1467 list_add(&zlist,
1468 &sched->kss_zombie_noop_txs);
1469 list_del_init(&sched->kss_zombie_noop_txs);
1470 }
1471
1472 conn = list_entry(sched->kss_tx_conns.next,
1473 ksock_conn_t, ksnc_tx_list);
1474 list_del (&conn->ksnc_tx_list);
1475
1476 LASSERT(conn->ksnc_tx_scheduled);
1477 LASSERT(conn->ksnc_tx_ready);
1478 LASSERT(!list_empty(&conn->ksnc_tx_queue));
1479
1480 tx = list_entry(conn->ksnc_tx_queue.next,
1481 ksock_tx_t, tx_list);
1482
1483 if (conn->ksnc_tx_carrier == tx)
1484 ksocknal_next_tx_carrier(conn);
1485
1486 /* dequeue now so empty list => more to send */
1487 list_del(&tx->tx_list);
1488
1489 /* Clear tx_ready in case send isn't complete. Do
1490 * it BEFORE we call process_transmit, since
1491 * write_space can set it any time after we release
1492 * kss_lock. */
1493 conn->ksnc_tx_ready = 0;
1494 spin_unlock_bh(&sched->kss_lock);
1495
1496 if (!list_empty(&zlist)) {
1497 /* free zombie noop txs, it's fast because
1498 * noop txs are just put in freelist */
1499 ksocknal_txlist_done(NULL, &zlist, 0);
1500 }
1501
1502 rc = ksocknal_process_transmit(conn, tx);
1503
1504 if (rc == -ENOMEM || rc == -EAGAIN) {
1505 /* Incomplete send: replace tx on HEAD of tx_queue */
1506 spin_lock_bh(&sched->kss_lock);
1507 list_add(&tx->tx_list,
1508 &conn->ksnc_tx_queue);
1509 } else {
1510 /* Complete send; tx -ref */
1511 ksocknal_tx_decref(tx);
1512
1513 spin_lock_bh(&sched->kss_lock);
1514 /* assume space for more */
1515 conn->ksnc_tx_ready = 1;
1516 }
1517
1518 if (rc == -ENOMEM) {
1519 /* Do nothing; after a short timeout, this
1520 * conn will be reposted on kss_tx_conns. */
1521 } else if (conn->ksnc_tx_ready &&
1522 !list_empty (&conn->ksnc_tx_queue)) {
1523 /* reschedule for tx */
1524 list_add_tail (&conn->ksnc_tx_list,
1525 &sched->kss_tx_conns);
1526 } else {
1527 conn->ksnc_tx_scheduled = 0;
1528 /* drop my ref */
1529 ksocknal_conn_decref(conn);
1530 }
1531
1532 did_something = 1;
1533 }
1534 if (!did_something || /* nothing to do */
1535 ++nloops == SOCKNAL_RESCHED) { /* hogging CPU? */
1536 spin_unlock_bh(&sched->kss_lock);
1537
1538 nloops = 0;
1539
1540 if (!did_something) { /* wait for something to do */
1541 cfs_wait_event_interruptible_exclusive(
1542 sched->kss_waitq,
1543 !ksocknal_sched_cansleep(sched), rc);
1544 LASSERT (rc == 0);
1545 } else {
1546 cond_resched();
1547 }
1548
1549 spin_lock_bh(&sched->kss_lock);
1550 }
1551 }
1552
1553 spin_unlock_bh(&sched->kss_lock);
1554 ksocknal_thread_fini();
1555 return 0;
1556}
1557
1558/*
1559 * Add connection to kss_rx_conns of scheduler
1560 * and wakeup the scheduler.
1561 */
1562void ksocknal_read_callback (ksock_conn_t *conn)
1563{
1564 ksock_sched_t *sched;
1565 ENTRY;
1566
1567 sched = conn->ksnc_scheduler;
1568
1569 spin_lock_bh(&sched->kss_lock);
1570
1571 conn->ksnc_rx_ready = 1;
1572
1573 if (!conn->ksnc_rx_scheduled) { /* not being progressed */
1574 list_add_tail(&conn->ksnc_rx_list,
1575 &sched->kss_rx_conns);
1576 conn->ksnc_rx_scheduled = 1;
1577 /* extra ref for scheduler */
1578 ksocknal_conn_addref(conn);
1579
1580 wake_up (&sched->kss_waitq);
1581 }
1582 spin_unlock_bh(&sched->kss_lock);
1583
1584 EXIT;
1585}
1586
1587/*
1588 * Add connection to kss_tx_conns of scheduler
1589 * and wakeup the scheduler.
1590 */
1591void ksocknal_write_callback (ksock_conn_t *conn)
1592{
1593 ksock_sched_t *sched;
1594 ENTRY;
1595
1596 sched = conn->ksnc_scheduler;
1597
1598 spin_lock_bh(&sched->kss_lock);
1599
1600 conn->ksnc_tx_ready = 1;
1601
1602 if (!conn->ksnc_tx_scheduled && // not being progressed
1603 !list_empty(&conn->ksnc_tx_queue)){//packets to send
1604 list_add_tail (&conn->ksnc_tx_list,
1605 &sched->kss_tx_conns);
1606 conn->ksnc_tx_scheduled = 1;
1607 /* extra ref for scheduler */
1608 ksocknal_conn_addref(conn);
1609
1610 wake_up (&sched->kss_waitq);
1611 }
1612
1613 spin_unlock_bh(&sched->kss_lock);
1614
1615 EXIT;
1616}
1617
1618ksock_proto_t *
1619ksocknal_parse_proto_version (ksock_hello_msg_t *hello)
1620{
1621 __u32 version = 0;
1622
1623 if (hello->kshm_magic == LNET_PROTO_MAGIC)
1624 version = hello->kshm_version;
1625 else if (hello->kshm_magic == __swab32(LNET_PROTO_MAGIC))
1626 version = __swab32(hello->kshm_version);
1627
1628 if (version != 0) {
1629#if SOCKNAL_VERSION_DEBUG
1630 if (*ksocknal_tunables.ksnd_protocol == 1)
1631 return NULL;
1632
1633 if (*ksocknal_tunables.ksnd_protocol == 2 &&
1634 version == KSOCK_PROTO_V3)
1635 return NULL;
1636#endif
1637 if (version == KSOCK_PROTO_V2)
1638 return &ksocknal_protocol_v2x;
1639
1640 if (version == KSOCK_PROTO_V3)
1641 return &ksocknal_protocol_v3x;
1642
1643 return NULL;
1644 }
1645
1646 if (hello->kshm_magic == le32_to_cpu(LNET_PROTO_TCP_MAGIC)) {
1647 lnet_magicversion_t *hmv = (lnet_magicversion_t *)hello;
1648
1649 CLASSERT (sizeof (lnet_magicversion_t) ==
1650 offsetof (ksock_hello_msg_t, kshm_src_nid));
1651
1652 if (hmv->version_major == cpu_to_le16 (KSOCK_PROTO_V1_MAJOR) &&
1653 hmv->version_minor == cpu_to_le16 (KSOCK_PROTO_V1_MINOR))
1654 return &ksocknal_protocol_v1x;
1655 }
1656
1657 return NULL;
1658}
1659
1660int
1661ksocknal_send_hello (lnet_ni_t *ni, ksock_conn_t *conn,
1662 lnet_nid_t peer_nid, ksock_hello_msg_t *hello)
1663{
1664 /* CAVEAT EMPTOR: this byte flips 'ipaddrs' */
1665 ksock_net_t *net = (ksock_net_t *)ni->ni_data;
1666
1667 LASSERT (hello->kshm_nips <= LNET_MAX_INTERFACES);
1668
1669 /* rely on caller to hold a ref on socket so it wouldn't disappear */
1670 LASSERT (conn->ksnc_proto != NULL);
1671
1672 hello->kshm_src_nid = ni->ni_nid;
1673 hello->kshm_dst_nid = peer_nid;
1674 hello->kshm_src_pid = the_lnet.ln_pid;
1675
1676 hello->kshm_src_incarnation = net->ksnn_incarnation;
1677 hello->kshm_ctype = conn->ksnc_type;
1678
1679 return conn->ksnc_proto->pro_send_hello(conn, hello);
1680}
1681
1682int
1683ksocknal_invert_type(int type)
1684{
1685 switch (type)
1686 {
1687 case SOCKLND_CONN_ANY:
1688 case SOCKLND_CONN_CONTROL:
1689 return (type);
1690 case SOCKLND_CONN_BULK_IN:
1691 return SOCKLND_CONN_BULK_OUT;
1692 case SOCKLND_CONN_BULK_OUT:
1693 return SOCKLND_CONN_BULK_IN;
1694 default:
1695 return (SOCKLND_CONN_NONE);
1696 }
1697}
1698
1699int
1700ksocknal_recv_hello (lnet_ni_t *ni, ksock_conn_t *conn,
1701 ksock_hello_msg_t *hello, lnet_process_id_t *peerid,
1702 __u64 *incarnation)
1703{
1704 /* Return < 0 fatal error
1705 * 0 success
1706 * EALREADY lost connection race
1707 * EPROTO protocol version mismatch
1708 */
1709 socket_t *sock = conn->ksnc_sock;
1710 int active = (conn->ksnc_proto != NULL);
1711 int timeout;
1712 int proto_match;
1713 int rc;
1714 ksock_proto_t *proto;
1715 lnet_process_id_t recv_id;
1716
1717 /* socket type set on active connections - not set on passive */
1718 LASSERT (!active == !(conn->ksnc_type != SOCKLND_CONN_NONE));
1719
1720 timeout = active ? *ksocknal_tunables.ksnd_timeout :
1721 lnet_acceptor_timeout();
1722
1723 rc = libcfs_sock_read(sock, &hello->kshm_magic, sizeof (hello->kshm_magic), timeout);
1724 if (rc != 0) {
1725 CERROR ("Error %d reading HELLO from %u.%u.%u.%u\n",
1726 rc, HIPQUAD(conn->ksnc_ipaddr));
1727 LASSERT (rc < 0);
1728 return rc;
1729 }
1730
1731 if (hello->kshm_magic != LNET_PROTO_MAGIC &&
1732 hello->kshm_magic != __swab32(LNET_PROTO_MAGIC) &&
1733 hello->kshm_magic != le32_to_cpu (LNET_PROTO_TCP_MAGIC)) {
1734 /* Unexpected magic! */
1735 CERROR ("Bad magic(1) %#08x (%#08x expected) from "
1736 "%u.%u.%u.%u\n", __cpu_to_le32 (hello->kshm_magic),
1737 LNET_PROTO_TCP_MAGIC,
1738 HIPQUAD(conn->ksnc_ipaddr));
1739 return -EPROTO;
1740 }
1741
1742 rc = libcfs_sock_read(sock, &hello->kshm_version,
1743 sizeof(hello->kshm_version), timeout);
1744 if (rc != 0) {
1745 CERROR ("Error %d reading HELLO from %u.%u.%u.%u\n",
1746 rc, HIPQUAD(conn->ksnc_ipaddr));
1747 LASSERT (rc < 0);
1748 return rc;
1749 }
1750
1751 proto = ksocknal_parse_proto_version(hello);
1752 if (proto == NULL) {
1753 if (!active) {
1754 /* unknown protocol from peer, tell peer my protocol */
1755 conn->ksnc_proto = &ksocknal_protocol_v3x;
1756#if SOCKNAL_VERSION_DEBUG
1757 if (*ksocknal_tunables.ksnd_protocol == 2)
1758 conn->ksnc_proto = &ksocknal_protocol_v2x;
1759 else if (*ksocknal_tunables.ksnd_protocol == 1)
1760 conn->ksnc_proto = &ksocknal_protocol_v1x;
1761#endif
1762 hello->kshm_nips = 0;
1763 ksocknal_send_hello(ni, conn, ni->ni_nid, hello);
1764 }
1765
1766 CERROR ("Unknown protocol version (%d.x expected)"
1767 " from %u.%u.%u.%u\n",
1768 conn->ksnc_proto->pro_version,
1769 HIPQUAD(conn->ksnc_ipaddr));
1770
1771 return -EPROTO;
1772 }
1773
1774 proto_match = (conn->ksnc_proto == proto);
1775 conn->ksnc_proto = proto;
1776
1777 /* receive the rest of hello message anyway */
1778 rc = conn->ksnc_proto->pro_recv_hello(conn, hello, timeout);
1779 if (rc != 0) {
1780 CERROR("Error %d reading or checking hello from from %u.%u.%u.%u\n",
1781 rc, HIPQUAD(conn->ksnc_ipaddr));
1782 LASSERT (rc < 0);
1783 return rc;
1784 }
1785
1786 *incarnation = hello->kshm_src_incarnation;
1787
1788 if (hello->kshm_src_nid == LNET_NID_ANY) {
1789 CERROR("Expecting a HELLO hdr with a NID, but got LNET_NID_ANY"
1790 "from %u.%u.%u.%u\n", HIPQUAD(conn->ksnc_ipaddr));
1791 return -EPROTO;
1792 }
1793
1794 if (!active &&
1795 conn->ksnc_port > LNET_ACCEPTOR_MAX_RESERVED_PORT) {
1796 /* Userspace NAL assigns peer process ID from socket */
1797 recv_id.pid = conn->ksnc_port | LNET_PID_USERFLAG;
1798 recv_id.nid = LNET_MKNID(LNET_NIDNET(ni->ni_nid), conn->ksnc_ipaddr);
1799 } else {
1800 recv_id.nid = hello->kshm_src_nid;
1801 recv_id.pid = hello->kshm_src_pid;
1802 }
1803
1804 if (!active) {
1805 *peerid = recv_id;
1806
1807 /* peer determines type */
1808 conn->ksnc_type = ksocknal_invert_type(hello->kshm_ctype);
1809 if (conn->ksnc_type == SOCKLND_CONN_NONE) {
1810 CERROR ("Unexpected type %d from %s ip %u.%u.%u.%u\n",
1811 hello->kshm_ctype, libcfs_id2str(*peerid),
1812 HIPQUAD(conn->ksnc_ipaddr));
1813 return -EPROTO;
1814 }
1815
1816 return 0;
1817 }
1818
1819 if (peerid->pid != recv_id.pid ||
1820 peerid->nid != recv_id.nid) {
1821 LCONSOLE_ERROR_MSG(0x130, "Connected successfully to %s on host"
1822 " %u.%u.%u.%u, but they claimed they were "
1823 "%s; please check your Lustre "
1824 "configuration.\n",
1825 libcfs_id2str(*peerid),
1826 HIPQUAD(conn->ksnc_ipaddr),
1827 libcfs_id2str(recv_id));
1828 return -EPROTO;
1829 }
1830
1831 if (hello->kshm_ctype == SOCKLND_CONN_NONE) {
1832 /* Possible protocol mismatch or I lost the connection race */
1833 return proto_match ? EALREADY : EPROTO;
1834 }
1835
1836 if (ksocknal_invert_type(hello->kshm_ctype) != conn->ksnc_type) {
1837 CERROR ("Mismatched types: me %d, %s ip %u.%u.%u.%u %d\n",
1838 conn->ksnc_type, libcfs_id2str(*peerid),
1839 HIPQUAD(conn->ksnc_ipaddr),
1840 hello->kshm_ctype);
1841 return -EPROTO;
1842 }
1843
1844 return 0;
1845}
1846
1847int
1848ksocknal_connect (ksock_route_t *route)
1849{
1850 LIST_HEAD (zombies);
1851 ksock_peer_t *peer = route->ksnr_peer;
1852 int type;
1853 int wanted;
1854 socket_t *sock;
1855 cfs_time_t deadline;
1856 int retry_later = 0;
1857 int rc = 0;
1858
1859 deadline = cfs_time_add(cfs_time_current(),
1860 cfs_time_seconds(*ksocknal_tunables.ksnd_timeout));
1861
1862 write_lock_bh(&ksocknal_data.ksnd_global_lock);
1863
1864 LASSERT (route->ksnr_scheduled);
1865 LASSERT (!route->ksnr_connecting);
1866
1867 route->ksnr_connecting = 1;
1868
1869 for (;;) {
1870 wanted = ksocknal_route_mask() & ~route->ksnr_connected;
1871
1872 /* stop connecting if peer/route got closed under me, or
1873 * route got connected while queued */
1874 if (peer->ksnp_closing || route->ksnr_deleted ||
1875 wanted == 0) {
1876 retry_later = 0;
1877 break;
1878 }
1879
1880 /* reschedule if peer is connecting to me */
1881 if (peer->ksnp_accepting > 0) {
1882 CDEBUG(D_NET,
1883 "peer %s(%d) already connecting to me, retry later.\n",
1884 libcfs_nid2str(peer->ksnp_id.nid), peer->ksnp_accepting);
1885 retry_later = 1;
1886 }
1887
1888 if (retry_later) /* needs reschedule */
1889 break;
1890
1891 if ((wanted & (1 << SOCKLND_CONN_ANY)) != 0) {
1892 type = SOCKLND_CONN_ANY;
1893 } else if ((wanted & (1 << SOCKLND_CONN_CONTROL)) != 0) {
1894 type = SOCKLND_CONN_CONTROL;
1895 } else if ((wanted & (1 << SOCKLND_CONN_BULK_IN)) != 0) {
1896 type = SOCKLND_CONN_BULK_IN;
1897 } else {
1898 LASSERT ((wanted & (1 << SOCKLND_CONN_BULK_OUT)) != 0);
1899 type = SOCKLND_CONN_BULK_OUT;
1900 }
1901
1902 write_unlock_bh(&ksocknal_data.ksnd_global_lock);
1903
1904 if (cfs_time_aftereq(cfs_time_current(), deadline)) {
1905 rc = -ETIMEDOUT;
1906 lnet_connect_console_error(rc, peer->ksnp_id.nid,
1907 route->ksnr_ipaddr,
1908 route->ksnr_port);
1909 goto failed;
1910 }
1911
1912 rc = lnet_connect(&sock, peer->ksnp_id.nid,
1913 route->ksnr_myipaddr,
1914 route->ksnr_ipaddr, route->ksnr_port);
1915 if (rc != 0)
1916 goto failed;
1917
1918 rc = ksocknal_create_conn(peer->ksnp_ni, route, sock, type);
1919 if (rc < 0) {
1920 lnet_connect_console_error(rc, peer->ksnp_id.nid,
1921 route->ksnr_ipaddr,
1922 route->ksnr_port);
1923 goto failed;
1924 }
1925
1926 /* A +ve RC means I have to retry because I lost the connection
1927 * race or I have to renegotiate protocol version */
1928 retry_later = (rc != 0);
1929 if (retry_later)
1930 CDEBUG(D_NET, "peer %s: conn race, retry later.\n",
1931 libcfs_nid2str(peer->ksnp_id.nid));
1932
1933 write_lock_bh(&ksocknal_data.ksnd_global_lock);
1934 }
1935
1936 route->ksnr_scheduled = 0;
1937 route->ksnr_connecting = 0;
1938
1939 if (retry_later) {
1940 /* re-queue for attention; this frees me up to handle
1941 * the peer's incoming connection request */
1942
1943 if (rc == EALREADY ||
1944 (rc == 0 && peer->ksnp_accepting > 0)) {
1945 /* We want to introduce a delay before next
1946 * attempt to connect if we lost conn race,
1947 * but the race is resolved quickly usually,
1948 * so min_reconnectms should be good heuristic */
1949 route->ksnr_retry_interval =
1950 cfs_time_seconds(*ksocknal_tunables.ksnd_min_reconnectms)/1000;
1951 route->ksnr_timeout = cfs_time_add(cfs_time_current(),
1952 route->ksnr_retry_interval);
1953 }
1954
1955 ksocknal_launch_connection_locked(route);
1956 }
1957
1958 write_unlock_bh(&ksocknal_data.ksnd_global_lock);
1959 return retry_later;
1960
1961 failed:
1962 write_lock_bh(&ksocknal_data.ksnd_global_lock);
1963
1964 route->ksnr_scheduled = 0;
1965 route->ksnr_connecting = 0;
1966
1967 /* This is a retry rather than a new connection */
1968 route->ksnr_retry_interval *= 2;
1969 route->ksnr_retry_interval =
1970 MAX(route->ksnr_retry_interval,
1971 cfs_time_seconds(*ksocknal_tunables.ksnd_min_reconnectms)/1000);
1972 route->ksnr_retry_interval =
1973 MIN(route->ksnr_retry_interval,
1974 cfs_time_seconds(*ksocknal_tunables.ksnd_max_reconnectms)/1000);
1975
1976 LASSERT (route->ksnr_retry_interval != 0);
1977 route->ksnr_timeout = cfs_time_add(cfs_time_current(),
1978 route->ksnr_retry_interval);
1979
1980 if (!list_empty(&peer->ksnp_tx_queue) &&
1981 peer->ksnp_accepting == 0 &&
1982 ksocknal_find_connecting_route_locked(peer) == NULL) {
1983 ksock_conn_t *conn;
1984
1985 /* ksnp_tx_queue is queued on a conn on successful
1986 * connection for V1.x and V2.x */
1987 if (!list_empty (&peer->ksnp_conns)) {
1988 conn = list_entry(peer->ksnp_conns.next,
1989 ksock_conn_t, ksnc_list);
1990 LASSERT (conn->ksnc_proto == &ksocknal_protocol_v3x);
1991 }
1992
1993 /* take all the blocked packets while I've got the lock and
1994 * complete below... */
1995 list_splice_init(&peer->ksnp_tx_queue, &zombies);
1996 }
1997
1998#if 0 /* irrelevent with only eager routes */
1999 if (!route->ksnr_deleted) {
2000 /* make this route least-favourite for re-selection */
2001 list_del(&route->ksnr_list);
2002 list_add_tail(&route->ksnr_list, &peer->ksnp_routes);
2003 }
2004#endif
2005 write_unlock_bh(&ksocknal_data.ksnd_global_lock);
2006
2007 ksocknal_peer_failed(peer);
2008 ksocknal_txlist_done(peer->ksnp_ni, &zombies, 1);
2009 return 0;
2010}
2011
2012/*
2013 * check whether we need to create more connds.
2014 * It will try to create new thread if it's necessary, @timeout can
2015 * be updated if failed to create, so caller wouldn't keep try while
2016 * running out of resource.
2017 */
2018static int
2019ksocknal_connd_check_start(long sec, long *timeout)
2020{
2021 char name[16];
2022 int rc;
2023 int total = ksocknal_data.ksnd_connd_starting +
2024 ksocknal_data.ksnd_connd_running;
2025
2026 if (unlikely(ksocknal_data.ksnd_init < SOCKNAL_INIT_ALL)) {
2027 /* still in initializing */
2028 return 0;
2029 }
2030
2031 if (total >= *ksocknal_tunables.ksnd_nconnds_max ||
2032 total > ksocknal_data.ksnd_connd_connecting + SOCKNAL_CONND_RESV) {
2033 /* can't create more connd, or still have enough
2034 * threads to handle more connecting */
2035 return 0;
2036 }
2037
2038 if (list_empty(&ksocknal_data.ksnd_connd_routes)) {
2039 /* no pending connecting request */
2040 return 0;
2041 }
2042
2043 if (sec - ksocknal_data.ksnd_connd_failed_stamp <= 1) {
2044 /* may run out of resource, retry later */
2045 *timeout = cfs_time_seconds(1);
2046 return 0;
2047 }
2048
2049 if (ksocknal_data.ksnd_connd_starting > 0) {
2050 /* serialize starting to avoid flood */
2051 return 0;
2052 }
2053
2054 ksocknal_data.ksnd_connd_starting_stamp = sec;
2055 ksocknal_data.ksnd_connd_starting++;
2056 spin_unlock_bh(&ksocknal_data.ksnd_connd_lock);
2057
2058 /* NB: total is the next id */
2059 snprintf(name, sizeof(name), "socknal_cd%02d", total);
2060 rc = ksocknal_thread_start(ksocknal_connd, NULL, name);
2061
2062 spin_lock_bh(&ksocknal_data.ksnd_connd_lock);
2063 if (rc == 0)
2064 return 1;
2065
2066 /* we tried ... */
2067 LASSERT(ksocknal_data.ksnd_connd_starting > 0);
2068 ksocknal_data.ksnd_connd_starting--;
2069 ksocknal_data.ksnd_connd_failed_stamp = cfs_time_current_sec();
2070
2071 return 1;
2072}
2073
2074/*
2075 * check whether current thread can exit, it will return 1 if there are too
2076 * many threads and no creating in past 120 seconds.
2077 * Also, this function may update @timeout to make caller come back
2078 * again to recheck these conditions.
2079 */
2080static int
2081ksocknal_connd_check_stop(long sec, long *timeout)
2082{
2083 int val;
2084
2085 if (unlikely(ksocknal_data.ksnd_init < SOCKNAL_INIT_ALL)) {
2086 /* still in initializing */
2087 return 0;
2088 }
2089
2090 if (ksocknal_data.ksnd_connd_starting > 0) {
2091 /* in progress of starting new thread */
2092 return 0;
2093 }
2094
2095 if (ksocknal_data.ksnd_connd_running <=
2096 *ksocknal_tunables.ksnd_nconnds) { /* can't shrink */
2097 return 0;
2098 }
2099
2100 /* created thread in past 120 seconds? */
2101 val = (int)(ksocknal_data.ksnd_connd_starting_stamp +
2102 SOCKNAL_CONND_TIMEOUT - sec);
2103
2104 *timeout = (val > 0) ? cfs_time_seconds(val) :
2105 cfs_time_seconds(SOCKNAL_CONND_TIMEOUT);
2106 if (val > 0)
2107 return 0;
2108
2109 /* no creating in past 120 seconds */
2110
2111 return ksocknal_data.ksnd_connd_running >
2112 ksocknal_data.ksnd_connd_connecting + SOCKNAL_CONND_RESV;
2113}
2114
2115/* Go through connd_routes queue looking for a route that we can process
2116 * right now, @timeout_p can be updated if we need to come back later */
2117static ksock_route_t *
2118ksocknal_connd_get_route_locked(signed long *timeout_p)
2119{
2120 ksock_route_t *route;
2121 cfs_time_t now;
2122
2123 now = cfs_time_current();
2124
2125 /* connd_routes can contain both pending and ordinary routes */
2126 list_for_each_entry (route, &ksocknal_data.ksnd_connd_routes,
2127 ksnr_connd_list) {
2128
2129 if (route->ksnr_retry_interval == 0 ||
2130 cfs_time_aftereq(now, route->ksnr_timeout))
2131 return route;
2132
2133 if (*timeout_p == MAX_SCHEDULE_TIMEOUT ||
2134 (int)*timeout_p > (int)(route->ksnr_timeout - now))
2135 *timeout_p = (int)(route->ksnr_timeout - now);
2136 }
2137
2138 return NULL;
2139}
2140
2141int
2142ksocknal_connd (void *arg)
2143{
2144 spinlock_t *connd_lock = &ksocknal_data.ksnd_connd_lock;
2145 ksock_connreq_t *cr;
2146 wait_queue_t wait;
2147 int nloops = 0;
2148 int cons_retry = 0;
2149
2150 cfs_block_allsigs ();
2151
2152 init_waitqueue_entry_current (&wait);
2153
2154 spin_lock_bh(connd_lock);
2155
2156 LASSERT(ksocknal_data.ksnd_connd_starting > 0);
2157 ksocknal_data.ksnd_connd_starting--;
2158 ksocknal_data.ksnd_connd_running++;
2159
2160 while (!ksocknal_data.ksnd_shuttingdown) {
2161 ksock_route_t *route = NULL;
2162 long sec = cfs_time_current_sec();
2163 long timeout = MAX_SCHEDULE_TIMEOUT;
2164 int dropped_lock = 0;
2165
2166 if (ksocknal_connd_check_stop(sec, &timeout)) {
2167 /* wakeup another one to check stop */
2168 wake_up(&ksocknal_data.ksnd_connd_waitq);
2169 break;
2170 }
2171
2172 if (ksocknal_connd_check_start(sec, &timeout)) {
2173 /* created new thread */
2174 dropped_lock = 1;
2175 }
2176
2177 if (!list_empty(&ksocknal_data.ksnd_connd_connreqs)) {
2178 /* Connection accepted by the listener */
2179 cr = list_entry(ksocknal_data.ksnd_connd_connreqs. \
2180 next, ksock_connreq_t, ksncr_list);
2181
2182 list_del(&cr->ksncr_list);
2183 spin_unlock_bh(connd_lock);
2184 dropped_lock = 1;
2185
2186 ksocknal_create_conn(cr->ksncr_ni, NULL,
2187 cr->ksncr_sock, SOCKLND_CONN_NONE);
2188 lnet_ni_decref(cr->ksncr_ni);
2189 LIBCFS_FREE(cr, sizeof(*cr));
2190
2191 spin_lock_bh(connd_lock);
2192 }
2193
2194 /* Only handle an outgoing connection request if there
2195 * is a thread left to handle incoming connections and
2196 * create new connd */
2197 if (ksocknal_data.ksnd_connd_connecting + SOCKNAL_CONND_RESV <
2198 ksocknal_data.ksnd_connd_running) {
2199 route = ksocknal_connd_get_route_locked(&timeout);
2200 }
2201 if (route != NULL) {
2202 list_del (&route->ksnr_connd_list);
2203 ksocknal_data.ksnd_connd_connecting++;
2204 spin_unlock_bh(connd_lock);
2205 dropped_lock = 1;
2206
2207 if (ksocknal_connect(route)) {
2208 /* consecutive retry */
2209 if (cons_retry++ > SOCKNAL_INSANITY_RECONN) {
2210 CWARN("massive consecutive "
2211 "re-connecting to %u.%u.%u.%u\n",
2212 HIPQUAD(route->ksnr_ipaddr));
2213 cons_retry = 0;
2214 }
2215 } else {
2216 cons_retry = 0;
2217 }
2218
2219 ksocknal_route_decref(route);
2220
2221 spin_lock_bh(connd_lock);
2222 ksocknal_data.ksnd_connd_connecting--;
2223 }
2224
2225 if (dropped_lock) {
2226 if (++nloops < SOCKNAL_RESCHED)
2227 continue;
2228 spin_unlock_bh(connd_lock);
2229 nloops = 0;
2230 cond_resched();
2231 spin_lock_bh(connd_lock);
2232 continue;
2233 }
2234
2235 /* Nothing to do for 'timeout' */
2236 set_current_state(TASK_INTERRUPTIBLE);
2237 add_wait_queue_exclusive(&ksocknal_data.ksnd_connd_waitq, &wait);
2238 spin_unlock_bh(connd_lock);
2239
2240 nloops = 0;
2241 waitq_timedwait(&wait, TASK_INTERRUPTIBLE, timeout);
2242
2243 set_current_state(TASK_RUNNING);
2244 remove_wait_queue(&ksocknal_data.ksnd_connd_waitq, &wait);
2245 spin_lock_bh(connd_lock);
2246 }
2247 ksocknal_data.ksnd_connd_running--;
2248 spin_unlock_bh(connd_lock);
2249
2250 ksocknal_thread_fini();
2251 return 0;
2252}
2253
2254ksock_conn_t *
2255ksocknal_find_timed_out_conn (ksock_peer_t *peer)
2256{
2257 /* We're called with a shared lock on ksnd_global_lock */
2258 ksock_conn_t *conn;
2259 struct list_head *ctmp;
2260
2261 list_for_each (ctmp, &peer->ksnp_conns) {
2262 int error;
2263 conn = list_entry (ctmp, ksock_conn_t, ksnc_list);
2264
2265 /* Don't need the {get,put}connsock dance to deref ksnc_sock */
2266 LASSERT (!conn->ksnc_closing);
2267
2268 /* SOCK_ERROR will reset error code of socket in
2269 * some platform (like Darwin8.x) */
2270 error = cfs_sock_error(conn->ksnc_sock);
2271 if (error != 0) {
2272 ksocknal_conn_addref(conn);
2273
2274 switch (error) {
2275 case ECONNRESET:
2276 CNETERR("A connection with %s "
2277 "(%u.%u.%u.%u:%d) was reset; "
2278 "it may have rebooted.\n",
2279 libcfs_id2str(peer->ksnp_id),
2280 HIPQUAD(conn->ksnc_ipaddr),
2281 conn->ksnc_port);
2282 break;
2283 case ETIMEDOUT:
2284 CNETERR("A connection with %s "
2285 "(%u.%u.%u.%u:%d) timed out; the "
2286 "network or node may be down.\n",
2287 libcfs_id2str(peer->ksnp_id),
2288 HIPQUAD(conn->ksnc_ipaddr),
2289 conn->ksnc_port);
2290 break;
2291 default:
2292 CNETERR("An unexpected network error %d "
2293 "occurred with %s "
2294 "(%u.%u.%u.%u:%d\n", error,
2295 libcfs_id2str(peer->ksnp_id),
2296 HIPQUAD(conn->ksnc_ipaddr),
2297 conn->ksnc_port);
2298 break;
2299 }
2300
2301 return (conn);
2302 }
2303
2304 if (conn->ksnc_rx_started &&
2305 cfs_time_aftereq(cfs_time_current(),
2306 conn->ksnc_rx_deadline)) {
2307 /* Timed out incomplete incoming message */
2308 ksocknal_conn_addref(conn);
2309 CNETERR("Timeout receiving from %s (%u.%u.%u.%u:%d), "
2310 "state %d wanted %d left %d\n",
2311 libcfs_id2str(peer->ksnp_id),
2312 HIPQUAD(conn->ksnc_ipaddr),
2313 conn->ksnc_port,
2314 conn->ksnc_rx_state,
2315 conn->ksnc_rx_nob_wanted,
2316 conn->ksnc_rx_nob_left);
2317 return (conn);
2318 }
2319
2320 if ((!list_empty(&conn->ksnc_tx_queue) ||
2321 cfs_sock_wmem_queued(conn->ksnc_sock) != 0) &&
2322 cfs_time_aftereq(cfs_time_current(),
2323 conn->ksnc_tx_deadline)) {
2324 /* Timed out messages queued for sending or
2325 * buffered in the socket's send buffer */
2326 ksocknal_conn_addref(conn);
2327 CNETERR("Timeout sending data to %s (%u.%u.%u.%u:%d) "
2328 "the network or that node may be down.\n",
2329 libcfs_id2str(peer->ksnp_id),
2330 HIPQUAD(conn->ksnc_ipaddr),
2331 conn->ksnc_port);
2332 return (conn);
2333 }
2334 }
2335
2336 return (NULL);
2337}
2338
2339static inline void
2340ksocknal_flush_stale_txs(ksock_peer_t *peer)
2341{
2342 ksock_tx_t *tx;
2343 LIST_HEAD (stale_txs);
2344
2345 write_lock_bh(&ksocknal_data.ksnd_global_lock);
2346
2347 while (!list_empty (&peer->ksnp_tx_queue)) {
2348 tx = list_entry (peer->ksnp_tx_queue.next,
2349 ksock_tx_t, tx_list);
2350
2351 if (!cfs_time_aftereq(cfs_time_current(),
2352 tx->tx_deadline))
2353 break;
2354
2355 list_del (&tx->tx_list);
2356 list_add_tail (&tx->tx_list, &stale_txs);
2357 }
2358
2359 write_unlock_bh(&ksocknal_data.ksnd_global_lock);
2360
2361 ksocknal_txlist_done(peer->ksnp_ni, &stale_txs, 1);
2362}
2363
2364int
2365ksocknal_send_keepalive_locked(ksock_peer_t *peer)
2366{
2367 ksock_sched_t *sched;
2368 ksock_conn_t *conn;
2369 ksock_tx_t *tx;
2370
2371 if (list_empty(&peer->ksnp_conns)) /* last_alive will be updated by create_conn */
2372 return 0;
2373
2374 if (peer->ksnp_proto != &ksocknal_protocol_v3x)
2375 return 0;
2376
2377 if (*ksocknal_tunables.ksnd_keepalive <= 0 ||
2378 cfs_time_before(cfs_time_current(),
2379 cfs_time_add(peer->ksnp_last_alive,
2380 cfs_time_seconds(*ksocknal_tunables.ksnd_keepalive))))
2381 return 0;
2382
2383 if (cfs_time_before(cfs_time_current(),
2384 peer->ksnp_send_keepalive))
2385 return 0;
2386
2387 /* retry 10 secs later, so we wouldn't put pressure
2388 * on this peer if we failed to send keepalive this time */
2389 peer->ksnp_send_keepalive = cfs_time_shift(10);
2390
2391 conn = ksocknal_find_conn_locked(peer, NULL, 1);
2392 if (conn != NULL) {
2393 sched = conn->ksnc_scheduler;
2394
2395 spin_lock_bh(&sched->kss_lock);
2396 if (!list_empty(&conn->ksnc_tx_queue)) {
2397 spin_unlock_bh(&sched->kss_lock);
2398 /* there is an queued ACK, don't need keepalive */
2399 return 0;
2400 }
2401
2402 spin_unlock_bh(&sched->kss_lock);
2403 }
2404
2405 read_unlock(&ksocknal_data.ksnd_global_lock);
2406
2407 /* cookie = 1 is reserved for keepalive PING */
2408 tx = ksocknal_alloc_tx_noop(1, 1);
2409 if (tx == NULL) {
2410 read_lock(&ksocknal_data.ksnd_global_lock);
2411 return -ENOMEM;
2412 }
2413
2414 if (ksocknal_launch_packet(peer->ksnp_ni, tx, peer->ksnp_id) == 0) {
2415 read_lock(&ksocknal_data.ksnd_global_lock);
2416 return 1;
2417 }
2418
2419 ksocknal_free_tx(tx);
2420 read_lock(&ksocknal_data.ksnd_global_lock);
2421
2422 return -EIO;
2423}
2424
2425
2426void
2427ksocknal_check_peer_timeouts (int idx)
2428{
2429 struct list_head *peers = &ksocknal_data.ksnd_peers[idx];
2430 ksock_peer_t *peer;
2431 ksock_conn_t *conn;
2432 ksock_tx_t *tx;
2433
2434 again:
2435 /* NB. We expect to have a look at all the peers and not find any
2436 * connections to time out, so we just use a shared lock while we
2437 * take a look... */
2438 read_lock(&ksocknal_data.ksnd_global_lock);
2439
2440 list_for_each_entry(peer, peers, ksnp_list) {
2441 cfs_time_t deadline = 0;
2442 int resid = 0;
2443 int n = 0;
2444
2445 if (ksocknal_send_keepalive_locked(peer) != 0) {
2446 read_unlock(&ksocknal_data.ksnd_global_lock);
2447 goto again;
2448 }
2449
2450 conn = ksocknal_find_timed_out_conn (peer);
2451
2452 if (conn != NULL) {
2453 read_unlock(&ksocknal_data.ksnd_global_lock);
2454
2455 ksocknal_close_conn_and_siblings (conn, -ETIMEDOUT);
2456
2457 /* NB we won't find this one again, but we can't
2458 * just proceed with the next peer, since we dropped
2459 * ksnd_global_lock and it might be dead already! */
2460 ksocknal_conn_decref(conn);
2461 goto again;
2462 }
2463
2464 /* we can't process stale txs right here because we're
2465 * holding only shared lock */
2466 if (!list_empty (&peer->ksnp_tx_queue)) {
2467 ksock_tx_t *tx =
2468 list_entry (peer->ksnp_tx_queue.next,
2469 ksock_tx_t, tx_list);
2470
2471 if (cfs_time_aftereq(cfs_time_current(),
2472 tx->tx_deadline)) {
2473
2474 ksocknal_peer_addref(peer);
2475 read_unlock(&ksocknal_data.ksnd_global_lock);
2476
2477 ksocknal_flush_stale_txs(peer);
2478
2479 ksocknal_peer_decref(peer);
2480 goto again;
2481 }
2482 }
2483
2484 if (list_empty(&peer->ksnp_zc_req_list))
2485 continue;
2486
2487 spin_lock(&peer->ksnp_lock);
2488 list_for_each_entry(tx, &peer->ksnp_zc_req_list, tx_zc_list) {
2489 if (!cfs_time_aftereq(cfs_time_current(),
2490 tx->tx_deadline))
2491 break;
2492 /* ignore the TX if connection is being closed */
2493 if (tx->tx_conn->ksnc_closing)
2494 continue;
2495 n++;
2496 }
2497
2498 if (n == 0) {
2499 spin_unlock(&peer->ksnp_lock);
2500 continue;
2501 }
2502
2503 tx = list_entry(peer->ksnp_zc_req_list.next,
2504 ksock_tx_t, tx_zc_list);
2505 deadline = tx->tx_deadline;
2506 resid = tx->tx_resid;
2507 conn = tx->tx_conn;
2508 ksocknal_conn_addref(conn);
2509
2510 spin_unlock(&peer->ksnp_lock);
2511 read_unlock(&ksocknal_data.ksnd_global_lock);
2512
2513 CERROR("Total %d stale ZC_REQs for peer %s detected; the "
2514 "oldest(%p) timed out %ld secs ago, "
2515 "resid: %d, wmem: %d\n",
2516 n, libcfs_nid2str(peer->ksnp_id.nid), tx,
2517 cfs_duration_sec(cfs_time_current() - deadline),
2518 resid, cfs_sock_wmem_queued(conn->ksnc_sock));
2519
2520 ksocknal_close_conn_and_siblings (conn, -ETIMEDOUT);
2521 ksocknal_conn_decref(conn);
2522 goto again;
2523 }
2524
2525 read_unlock(&ksocknal_data.ksnd_global_lock);
2526}
2527
2528int
2529ksocknal_reaper (void *arg)
2530{
2531 wait_queue_t wait;
2532 ksock_conn_t *conn;
2533 ksock_sched_t *sched;
2534 struct list_head enomem_conns;
2535 int nenomem_conns;
2536 cfs_duration_t timeout;
2537 int i;
2538 int peer_index = 0;
2539 cfs_time_t deadline = cfs_time_current();
2540
2541 cfs_block_allsigs ();
2542
2543 INIT_LIST_HEAD(&enomem_conns);
2544 init_waitqueue_entry_current (&wait);
2545
2546 spin_lock_bh(&ksocknal_data.ksnd_reaper_lock);
2547
2548 while (!ksocknal_data.ksnd_shuttingdown) {
2549
2550 if (!list_empty (&ksocknal_data.ksnd_deathrow_conns)) {
2551 conn = list_entry (ksocknal_data. \
2552 ksnd_deathrow_conns.next,
2553 ksock_conn_t, ksnc_list);
2554 list_del (&conn->ksnc_list);
2555
2556 spin_unlock_bh(&ksocknal_data.ksnd_reaper_lock);
2557
2558 ksocknal_terminate_conn(conn);
2559 ksocknal_conn_decref(conn);
2560
2561 spin_lock_bh(&ksocknal_data.ksnd_reaper_lock);
2562 continue;
2563 }
2564
2565 if (!list_empty (&ksocknal_data.ksnd_zombie_conns)) {
2566 conn = list_entry (ksocknal_data.ksnd_zombie_conns.\
2567 next, ksock_conn_t, ksnc_list);
2568 list_del (&conn->ksnc_list);
2569
2570 spin_unlock_bh(&ksocknal_data.ksnd_reaper_lock);
2571
2572 ksocknal_destroy_conn(conn);
2573
2574 spin_lock_bh(&ksocknal_data.ksnd_reaper_lock);
2575 continue;
2576 }
2577
2578 if (!list_empty (&ksocknal_data.ksnd_enomem_conns)) {
2579 list_add(&enomem_conns,
2580 &ksocknal_data.ksnd_enomem_conns);
2581 list_del_init(&ksocknal_data.ksnd_enomem_conns);
2582 }
2583
2584 spin_unlock_bh(&ksocknal_data.ksnd_reaper_lock);
2585
2586 /* reschedule all the connections that stalled with ENOMEM... */
2587 nenomem_conns = 0;
2588 while (!list_empty (&enomem_conns)) {
2589 conn = list_entry (enomem_conns.next,
2590 ksock_conn_t, ksnc_tx_list);
2591 list_del (&conn->ksnc_tx_list);
2592
2593 sched = conn->ksnc_scheduler;
2594
2595 spin_lock_bh(&sched->kss_lock);
2596
2597 LASSERT(conn->ksnc_tx_scheduled);
2598 conn->ksnc_tx_ready = 1;
2599 list_add_tail(&conn->ksnc_tx_list,
2600 &sched->kss_tx_conns);
2601 wake_up(&sched->kss_waitq);
2602
2603 spin_unlock_bh(&sched->kss_lock);
2604 nenomem_conns++;
2605 }
2606
2607 /* careful with the jiffy wrap... */
2608 while ((timeout = cfs_time_sub(deadline,
2609 cfs_time_current())) <= 0) {
2610 const int n = 4;
2611 const int p = 1;
2612 int chunk = ksocknal_data.ksnd_peer_hash_size;
2613
2614 /* Time to check for timeouts on a few more peers: I do
2615 * checks every 'p' seconds on a proportion of the peer
2616 * table and I need to check every connection 'n' times
2617 * within a timeout interval, to ensure I detect a
2618 * timeout on any connection within (n+1)/n times the
2619 * timeout interval. */
2620
2621 if (*ksocknal_tunables.ksnd_timeout > n * p)
2622 chunk = (chunk * n * p) /
2623 *ksocknal_tunables.ksnd_timeout;
2624 if (chunk == 0)
2625 chunk = 1;
2626
2627 for (i = 0; i < chunk; i++) {
2628 ksocknal_check_peer_timeouts (peer_index);
2629 peer_index = (peer_index + 1) %
2630 ksocknal_data.ksnd_peer_hash_size;
2631 }
2632
2633 deadline = cfs_time_add(deadline, cfs_time_seconds(p));
2634 }
2635
2636 if (nenomem_conns != 0) {
2637 /* Reduce my timeout if I rescheduled ENOMEM conns.
2638 * This also prevents me getting woken immediately
2639 * if any go back on my enomem list. */
2640 timeout = SOCKNAL_ENOMEM_RETRY;
2641 }
2642 ksocknal_data.ksnd_reaper_waketime =
2643 cfs_time_add(cfs_time_current(), timeout);
2644
2645 set_current_state (TASK_INTERRUPTIBLE);
2646 add_wait_queue (&ksocknal_data.ksnd_reaper_waitq, &wait);
2647
2648 if (!ksocknal_data.ksnd_shuttingdown &&
2649 list_empty (&ksocknal_data.ksnd_deathrow_conns) &&
2650 list_empty (&ksocknal_data.ksnd_zombie_conns))
2651 waitq_timedwait (&wait, TASK_INTERRUPTIBLE,
2652 timeout);
2653
2654 set_current_state (TASK_RUNNING);
2655 remove_wait_queue (&ksocknal_data.ksnd_reaper_waitq, &wait);
2656
2657 spin_lock_bh(&ksocknal_data.ksnd_reaper_lock);
2658 }
2659
2660 spin_unlock_bh(&ksocknal_data.ksnd_reaper_lock);
2661
2662 ksocknal_thread_fini();
2663 return 0;
2664}
Linux kernel - Deliverables
This page took 0.150261 seconds and 5 git commands to generate.