?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
writeback: split writeback_inodes_wb
[linux-2.6/next.git] / net / rxrpc / ar-ack.c
blob2714da167fb8980fd7a43f3485e25e2bdcfc7feb
1 /* Management of Tx window, Tx resend, ACKs and out-of-sequence reception
2 *
3 * Copyright (C) 2007 Red Hat, Inc. All Rights Reserved.
4 * Written by David Howells (dhowells@redhat.com)
5 *
6 * This program is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU General Public License
8 * as published by the Free Software Foundation; either version
9 * 2 of the License, or (at your option) any later version.
10 */
11
12 #include <linux/module.h>
13 #include <linux/circ_buf.h>
14 #include <linux/net.h>
15 #include <linux/skbuff.h>
16 #include <linux/slab.h>
17 #include <linux/udp.h>
18 #include <net/sock.h>
19 #include <net/af_rxrpc.h>
20 #include "ar-internal.h"
21
22 static unsigned rxrpc_ack_defer = 1;
23
24 static const char *const rxrpc_acks[] = {
25 "---", "REQ", "DUP", "OOS", "WIN", "MEM", "PNG", "PNR", "DLY", "IDL",
26 "-?-"
27 };
28
29 static const s8 rxrpc_ack_priority[] = {
30 [0] = 0,
31 [RXRPC_ACK_DELAY] = 1,
32 [RXRPC_ACK_REQUESTED] = 2,
33 [RXRPC_ACK_IDLE] = 3,
34 [RXRPC_ACK_PING_RESPONSE] = 4,
35 [RXRPC_ACK_DUPLICATE] = 5,
36 [RXRPC_ACK_OUT_OF_SEQUENCE] = 6,
37 [RXRPC_ACK_EXCEEDS_WINDOW] = 7,
38 [RXRPC_ACK_NOSPACE] = 8,
39 };
40
41 /*
42 * propose an ACK be sent
43 */
44 void __rxrpc_propose_ACK(struct rxrpc_call *call, u8 ack_reason,
45 __be32 serial, bool immediate)
46 {
47 unsigned long expiry;
48 s8 prior = rxrpc_ack_priority[ack_reason];
49
50 ASSERTCMP(prior, >, 0);
51
52 _enter("{%d},%s,%%%x,%u",
53 call->debug_id, rxrpc_acks[ack_reason], ntohl(serial),
54 immediate);
55
56 if (prior < rxrpc_ack_priority[call->ackr_reason]) {
57 if (immediate)
58 goto cancel_timer;
59 return;
60 }
61
62 /* update DELAY, IDLE, REQUESTED and PING_RESPONSE ACK serial
63 * numbers */
64 if (prior == rxrpc_ack_priority[call->ackr_reason]) {
65 if (prior <= 4)
66 call->ackr_serial = serial;
67 if (immediate)
68 goto cancel_timer;
69 return;
70 }
71
72 call->ackr_reason = ack_reason;
73 call->ackr_serial = serial;
74
75 switch (ack_reason) {
76 case RXRPC_ACK_DELAY:
77 _debug("run delay timer");
78 call->ack_timer.expires = jiffies + rxrpc_ack_timeout * HZ;
79 add_timer(&call->ack_timer);
80 return;
81
82 case RXRPC_ACK_IDLE:
83 if (!immediate) {
84 _debug("run defer timer");
85 expiry = 1;
86 goto run_timer;
87 }
88 goto cancel_timer;
89
90 case RXRPC_ACK_REQUESTED:
91 if (!rxrpc_ack_defer)
92 goto cancel_timer;
93 if (!immediate || serial == cpu_to_be32(1)) {
94 _debug("run defer timer");
95 expiry = rxrpc_ack_defer;
96 goto run_timer;
97 }
98
99 default:
100 _debug("immediate ACK");
101 goto cancel_timer;
102 }
103
104 run_timer:
105 expiry += jiffies;
106 if (!timer_pending(&call->ack_timer) ||
107 time_after(call->ack_timer.expires, expiry))
108 mod_timer(&call->ack_timer, expiry);
109 return;
110
111 cancel_timer:
112 _debug("cancel timer %%%u", ntohl(serial));
113 try_to_del_timer_sync(&call->ack_timer);
114 read_lock_bh(&call->state_lock);
115 if (call->state <= RXRPC_CALL_COMPLETE &&
116 !test_and_set_bit(RXRPC_CALL_ACK, &call->events))
117 rxrpc_queue_call(call);
118 read_unlock_bh(&call->state_lock);
119 }
120
121 /*
122 * propose an ACK be sent, locking the call structure
123 */
124 void rxrpc_propose_ACK(struct rxrpc_call *call, u8 ack_reason,
125 __be32 serial, bool immediate)
126 {
127 s8 prior = rxrpc_ack_priority[ack_reason];
128
129 if (prior > rxrpc_ack_priority[call->ackr_reason]) {
130 spin_lock_bh(&call->lock);
131 __rxrpc_propose_ACK(call, ack_reason, serial, immediate);
132 spin_unlock_bh(&call->lock);
133 }
134 }
135
136 /*
137 * set the resend timer
138 */
139 static void rxrpc_set_resend(struct rxrpc_call *call, u8 resend,
140 unsigned long resend_at)
141 {
142 read_lock_bh(&call->state_lock);
143 if (call->state >= RXRPC_CALL_COMPLETE)
144 resend = 0;
145
146 if (resend & 1) {
147 _debug("SET RESEND");
148 set_bit(RXRPC_CALL_RESEND, &call->events);
149 }
150
151 if (resend & 2) {
152 _debug("MODIFY RESEND TIMER");
153 set_bit(RXRPC_CALL_RUN_RTIMER, &call->flags);
154 mod_timer(&call->resend_timer, resend_at);
155 } else {
156 _debug("KILL RESEND TIMER");
157 del_timer_sync(&call->resend_timer);
158 clear_bit(RXRPC_CALL_RESEND_TIMER, &call->events);
159 clear_bit(RXRPC_CALL_RUN_RTIMER, &call->flags);
160 }
161 read_unlock_bh(&call->state_lock);
162 }
163
164 /*
165 * resend packets
166 */
167 static void rxrpc_resend(struct rxrpc_call *call)
168 {
169 struct rxrpc_skb_priv *sp;
170 struct rxrpc_header *hdr;
171 struct sk_buff *txb;
172 unsigned long *p_txb, resend_at;
173 int loop, stop;
174 u8 resend;
175
176 _enter("{%d,%d,%d,%d},",
177 call->acks_hard, call->acks_unacked,
178 atomic_read(&call->sequence),
179 CIRC_CNT(call->acks_head, call->acks_tail, call->acks_winsz));
180
181 stop = 0;
182 resend = 0;
183 resend_at = 0;
184
185 for (loop = call->acks_tail;
186 loop != call->acks_head || stop;
187 loop = (loop + 1) & (call->acks_winsz - 1)
188 ) {
189 p_txb = call->acks_window + loop;
190 smp_read_barrier_depends();
191 if (*p_txb & 1)
192 continue;
193
194 txb = (struct sk_buff *) *p_txb;
195 sp = rxrpc_skb(txb);
196
197 if (sp->need_resend) {
198 sp->need_resend = 0;
199
200 /* each Tx packet has a new serial number */
201 sp->hdr.serial =
202 htonl(atomic_inc_return(&call->conn->serial));
203
204 hdr = (struct rxrpc_header *) txb->head;
205 hdr->serial = sp->hdr.serial;
206
207 _proto("Tx DATA %%%u { #%d }",
208 ntohl(sp->hdr.serial), ntohl(sp->hdr.seq));
209 if (rxrpc_send_packet(call->conn->trans, txb) < 0) {
210 stop = 0;
211 sp->resend_at = jiffies + 3;
212 } else {
213 sp->resend_at =
214 jiffies + rxrpc_resend_timeout * HZ;
215 }
216 }
217
218 if (time_after_eq(jiffies + 1, sp->resend_at)) {
219 sp->need_resend = 1;
220 resend |= 1;
221 } else if (resend & 2) {
222 if (time_before(sp->resend_at, resend_at))
223 resend_at = sp->resend_at;
224 } else {
225 resend_at = sp->resend_at;
226 resend |= 2;
227 }
228 }
229
230 rxrpc_set_resend(call, resend, resend_at);
231 _leave("");
232 }
233
234 /*
235 * handle resend timer expiry
236 */
237 static void rxrpc_resend_timer(struct rxrpc_call *call)
238 {
239 struct rxrpc_skb_priv *sp;
240 struct sk_buff *txb;
241 unsigned long *p_txb, resend_at;
242 int loop;
243 u8 resend;
244
245 _enter("%d,%d,%d",
246 call->acks_tail, call->acks_unacked, call->acks_head);
247
248 resend = 0;
249 resend_at = 0;
250
251 for (loop = call->acks_unacked;
252 loop != call->acks_head;
253 loop = (loop + 1) & (call->acks_winsz - 1)
254 ) {
255 p_txb = call->acks_window + loop;
256 smp_read_barrier_depends();
257 txb = (struct sk_buff *) (*p_txb & ~1);
258 sp = rxrpc_skb(txb);
259
260 ASSERT(!(*p_txb & 1));
261
262 if (sp->need_resend) {
263 ;
264 } else if (time_after_eq(jiffies + 1, sp->resend_at)) {
265 sp->need_resend = 1;
266 resend |= 1;
267 } else if (resend & 2) {
268 if (time_before(sp->resend_at, resend_at))
269 resend_at = sp->resend_at;
270 } else {
271 resend_at = sp->resend_at;
272 resend |= 2;
273 }
274 }
275
276 rxrpc_set_resend(call, resend, resend_at);
277 _leave("");
278 }
279
280 /*
281 * process soft ACKs of our transmitted packets
282 * - these indicate packets the peer has or has not received, but hasn't yet
283 * given to the consumer, and so can still be discarded and re-requested
284 */
285 static int rxrpc_process_soft_ACKs(struct rxrpc_call *call,
286 struct rxrpc_ackpacket *ack,
287 struct sk_buff *skb)
288 {
289 struct rxrpc_skb_priv *sp;
290 struct sk_buff *txb;
291 unsigned long *p_txb, resend_at;
292 int loop;
293 u8 sacks[RXRPC_MAXACKS], resend;
294
295 _enter("{%d,%d},{%d},",
296 call->acks_hard,
297 CIRC_CNT(call->acks_head, call->acks_tail, call->acks_winsz),
298 ack->nAcks);
299
300 if (skb_copy_bits(skb, 0, sacks, ack->nAcks) < 0)
301 goto protocol_error;
302
303 resend = 0;
304 resend_at = 0;
305 for (loop = 0; loop < ack->nAcks; loop++) {
306 p_txb = call->acks_window;
307 p_txb += (call->acks_tail + loop) & (call->acks_winsz - 1);
308 smp_read_barrier_depends();
309 txb = (struct sk_buff *) (*p_txb & ~1);
310 sp = rxrpc_skb(txb);
311
312 switch (sacks[loop]) {
313 case RXRPC_ACK_TYPE_ACK:
314 sp->need_resend = 0;
315 *p_txb |= 1;
316 break;
317 case RXRPC_ACK_TYPE_NACK:
318 sp->need_resend = 1;
319 *p_txb &= ~1;
320 resend = 1;
321 break;
322 default:
323 _debug("Unsupported ACK type %d", sacks[loop]);
324 goto protocol_error;
325 }
326 }
327
328 smp_mb();
329 call->acks_unacked = (call->acks_tail + loop) & (call->acks_winsz - 1);
330
331 /* anything not explicitly ACK'd is implicitly NACK'd, but may just not
332 * have been received or processed yet by the far end */
333 for (loop = call->acks_unacked;
334 loop != call->acks_head;
335 loop = (loop + 1) & (call->acks_winsz - 1)
336 ) {
337 p_txb = call->acks_window + loop;
338 smp_read_barrier_depends();
339 txb = (struct sk_buff *) (*p_txb & ~1);
340 sp = rxrpc_skb(txb);
341
342 if (*p_txb & 1) {
343 /* packet must have been discarded */
344 sp->need_resend = 1;
345 *p_txb &= ~1;
346 resend |= 1;
347 } else if (sp->need_resend) {
348 ;
349 } else if (time_after_eq(jiffies + 1, sp->resend_at)) {
350 sp->need_resend = 1;
351 resend |= 1;
352 } else if (resend & 2) {
353 if (time_before(sp->resend_at, resend_at))
354 resend_at = sp->resend_at;
355 } else {
356 resend_at = sp->resend_at;
357 resend |= 2;
358 }
359 }
360
361 rxrpc_set_resend(call, resend, resend_at);
362 _leave(" = 0");
363 return 0;
364
365 protocol_error:
366 _leave(" = -EPROTO");
367 return -EPROTO;
368 }
369
370 /*
371 * discard hard-ACK'd packets from the Tx window
372 */
373 static void rxrpc_rotate_tx_window(struct rxrpc_call *call, u32 hard)
374 {
375 struct rxrpc_skb_priv *sp;
376 unsigned long _skb;
377 int tail = call->acks_tail, old_tail;
378 int win = CIRC_CNT(call->acks_head, tail, call->acks_winsz);
379
380 _enter("{%u,%u},%u", call->acks_hard, win, hard);
381
382 ASSERTCMP(hard - call->acks_hard, <=, win);
383
384 while (call->acks_hard < hard) {
385 smp_read_barrier_depends();
386 _skb = call->acks_window[tail] & ~1;
387 sp = rxrpc_skb((struct sk_buff *) _skb);
388 rxrpc_free_skb((struct sk_buff *) _skb);
389 old_tail = tail;
390 tail = (tail + 1) & (call->acks_winsz - 1);
391 call->acks_tail = tail;
392 if (call->acks_unacked == old_tail)
393 call->acks_unacked = tail;
394 call->acks_hard++;
395 }
396
397 wake_up(&call->tx_waitq);
398 }
399
400 /*
401 * clear the Tx window in the event of a failure
402 */
403 static void rxrpc_clear_tx_window(struct rxrpc_call *call)
404 {
405 rxrpc_rotate_tx_window(call, atomic_read(&call->sequence));
406 }
407
408 /*
409 * drain the out of sequence received packet queue into the packet Rx queue
410 */
411 static int rxrpc_drain_rx_oos_queue(struct rxrpc_call *call)
412 {
413 struct rxrpc_skb_priv *sp;
414 struct sk_buff *skb;
415 bool terminal;
416 int ret;
417
418 _enter("{%d,%d}", call->rx_data_post, call->rx_first_oos);
419
420 spin_lock_bh(&call->lock);
421
422 ret = -ECONNRESET;
423 if (test_bit(RXRPC_CALL_RELEASED, &call->flags))
424 goto socket_unavailable;
425
426 skb = skb_dequeue(&call->rx_oos_queue);
427 if (skb) {
428 sp = rxrpc_skb(skb);
429
430 _debug("drain OOS packet %d [%d]",
431 ntohl(sp->hdr.seq), call->rx_first_oos);
432
433 if (ntohl(sp->hdr.seq) != call->rx_first_oos) {
434 skb_queue_head(&call->rx_oos_queue, skb);
435 call->rx_first_oos = ntohl(rxrpc_skb(skb)->hdr.seq);
436 _debug("requeue %p {%u}", skb, call->rx_first_oos);
437 } else {
438 skb->mark = RXRPC_SKB_MARK_DATA;
439 terminal = ((sp->hdr.flags & RXRPC_LAST_PACKET) &&
440 !(sp->hdr.flags & RXRPC_CLIENT_INITIATED));
441 ret = rxrpc_queue_rcv_skb(call, skb, true, terminal);
442 BUG_ON(ret < 0);
443 _debug("drain #%u", call->rx_data_post);
444 call->rx_data_post++;
445
446 /* find out what the next packet is */
447 skb = skb_peek(&call->rx_oos_queue);
448 if (skb)
449 call->rx_first_oos =
450 ntohl(rxrpc_skb(skb)->hdr.seq);
451 else
452 call->rx_first_oos = 0;
453 _debug("peek %p {%u}", skb, call->rx_first_oos);
454 }
455 }
456
457 ret = 0;
458 socket_unavailable:
459 spin_unlock_bh(&call->lock);
460 _leave(" = %d", ret);
461 return ret;
462 }
463
464 /*
465 * insert an out of sequence packet into the buffer
466 */
467 static void rxrpc_insert_oos_packet(struct rxrpc_call *call,
468 struct sk_buff *skb)
469 {
470 struct rxrpc_skb_priv *sp, *psp;
471 struct sk_buff *p;
472 u32 seq;
473
474 sp = rxrpc_skb(skb);
475 seq = ntohl(sp->hdr.seq);
476 _enter(",,{%u}", seq);
477
478 skb->destructor = rxrpc_packet_destructor;
479 ASSERTCMP(sp->call, ==, NULL);
480 sp->call = call;
481 rxrpc_get_call(call);
482
483 /* insert into the buffer in sequence order */
484 spin_lock_bh(&call->lock);
485
486 skb_queue_walk(&call->rx_oos_queue, p) {
487 psp = rxrpc_skb(p);
488 if (ntohl(psp->hdr.seq) > seq) {
489 _debug("insert oos #%u before #%u",
490 seq, ntohl(psp->hdr.seq));
491 skb_insert(p, skb, &call->rx_oos_queue);
492 goto inserted;
493 }
494 }
495
496 _debug("append oos #%u", seq);
497 skb_queue_tail(&call->rx_oos_queue, skb);
498 inserted:
499
500 /* we might now have a new front to the queue */
501 if (call->rx_first_oos == 0 || seq < call->rx_first_oos)
502 call->rx_first_oos = seq;
503
504 read_lock(&call->state_lock);
505 if (call->state < RXRPC_CALL_COMPLETE &&
506 call->rx_data_post == call->rx_first_oos) {
507 _debug("drain rx oos now");
508 set_bit(RXRPC_CALL_DRAIN_RX_OOS, &call->events);
509 }
510 read_unlock(&call->state_lock);
511
512 spin_unlock_bh(&call->lock);
513 _leave(" [stored #%u]", call->rx_first_oos);
514 }
515
516 /*
517 * clear the Tx window on final ACK reception
518 */
519 static void rxrpc_zap_tx_window(struct rxrpc_call *call)
520 {
521 struct rxrpc_skb_priv *sp;
522 struct sk_buff *skb;
523 unsigned long _skb, *acks_window;
524 u8 winsz = call->acks_winsz;
525 int tail;
526
527 acks_window = call->acks_window;
528 call->acks_window = NULL;
529
530 while (CIRC_CNT(call->acks_head, call->acks_tail, winsz) > 0) {
531 tail = call->acks_tail;
532 smp_read_barrier_depends();
533 _skb = acks_window[tail] & ~1;
534 smp_mb();
535 call->acks_tail = (call->acks_tail + 1) & (winsz - 1);
536
537 skb = (struct sk_buff *) _skb;
538 sp = rxrpc_skb(skb);
539 _debug("+++ clear Tx %u", ntohl(sp->hdr.seq));
540 rxrpc_free_skb(skb);
541 }
542
543 kfree(acks_window);
544 }
545
546 /*
547 * process the extra information that may be appended to an ACK packet
548 */
549 static void rxrpc_extract_ackinfo(struct rxrpc_call *call, struct sk_buff *skb,
550 unsigned latest, int nAcks)
551 {
552 struct rxrpc_ackinfo ackinfo;
553 struct rxrpc_peer *peer;
554 unsigned mtu;
555
556 if (skb_copy_bits(skb, nAcks + 3, &ackinfo, sizeof(ackinfo)) < 0) {
557 _leave(" [no ackinfo]");
558 return;
559 }
560
561 _proto("Rx ACK %%%u Info { rx=%u max=%u rwin=%u jm=%u }",
562 latest,
563 ntohl(ackinfo.rxMTU), ntohl(ackinfo.maxMTU),
564 ntohl(ackinfo.rwind), ntohl(ackinfo.jumbo_max));
565
566 mtu = min(ntohl(ackinfo.rxMTU), ntohl(ackinfo.maxMTU));
567
568 peer = call->conn->trans->peer;
569 if (mtu < peer->maxdata) {
570 spin_lock_bh(&peer->lock);
571 peer->maxdata = mtu;
572 peer->mtu = mtu + peer->hdrsize;
573 spin_unlock_bh(&peer->lock);
574 _net("Net MTU %u (maxdata %u)", peer->mtu, peer->maxdata);
575 }
576 }
577
578 /*
579 * process packets in the reception queue
580 */
581 static int rxrpc_process_rx_queue(struct rxrpc_call *call,
582 u32 *_abort_code)
583 {
584 struct rxrpc_ackpacket ack;
585 struct rxrpc_skb_priv *sp;
586 struct sk_buff *skb;
587 bool post_ACK;
588 int latest;
589 u32 hard, tx;
590
591 _enter("");
592
593 process_further:
594 skb = skb_dequeue(&call->rx_queue);
595 if (!skb)
596 return -EAGAIN;
597
598 _net("deferred skb %p", skb);
599
600 sp = rxrpc_skb(skb);
601
602 _debug("process %s [st %d]", rxrpc_pkts[sp->hdr.type], call->state);
603
604 post_ACK = false;
605
606 switch (sp->hdr.type) {
607 /* data packets that wind up here have been received out of
608 * order, need security processing or are jumbo packets */
609 case RXRPC_PACKET_TYPE_DATA:
610 _proto("OOSQ DATA %%%u { #%u }",
611 ntohl(sp->hdr.serial), ntohl(sp->hdr.seq));
612
613 /* secured packets must be verified and possibly decrypted */
614 if (rxrpc_verify_packet(call, skb, _abort_code) < 0)
615 goto protocol_error;
616
617 rxrpc_insert_oos_packet(call, skb);
618 goto process_further;
619
620 /* partial ACK to process */
621 case RXRPC_PACKET_TYPE_ACK:
622 if (skb_copy_bits(skb, 0, &ack, sizeof(ack)) < 0) {
623 _debug("extraction failure");
624 goto protocol_error;
625 }
626 if (!skb_pull(skb, sizeof(ack)))
627 BUG();
628
629 latest = ntohl(sp->hdr.serial);
630 hard = ntohl(ack.firstPacket);
631 tx = atomic_read(&call->sequence);
632
633 _proto("Rx ACK %%%u { m=%hu f=#%u p=#%u s=%%%u r=%s n=%u }",
634 latest,
635 ntohs(ack.maxSkew),
636 hard,
637 ntohl(ack.previousPacket),
638 ntohl(ack.serial),
639 rxrpc_acks[ack.reason],
640 ack.nAcks);
641
642 rxrpc_extract_ackinfo(call, skb, latest, ack.nAcks);
643
644 if (ack.reason == RXRPC_ACK_PING) {
645 _proto("Rx ACK %%%u PING Request", latest);
646 rxrpc_propose_ACK(call, RXRPC_ACK_PING_RESPONSE,
647 sp->hdr.serial, true);
648 }
649
650 /* discard any out-of-order or duplicate ACKs */
651 if (latest - call->acks_latest <= 0) {
652 _debug("discard ACK %d <= %d",
653 latest, call->acks_latest);
654 goto discard;
655 }
656 call->acks_latest = latest;
657
658 if (call->state != RXRPC_CALL_CLIENT_SEND_REQUEST &&
659 call->state != RXRPC_CALL_CLIENT_AWAIT_REPLY &&
660 call->state != RXRPC_CALL_SERVER_SEND_REPLY &&
661 call->state != RXRPC_CALL_SERVER_AWAIT_ACK)
662 goto discard;
663
664 _debug("Tx=%d H=%u S=%d", tx, call->acks_hard, call->state);
665
666 if (hard > 0) {
667 if (hard - 1 > tx) {
668 _debug("hard-ACK'd packet %d not transmitted"
669 " (%d top)",
670 hard - 1, tx);
671 goto protocol_error;
672 }
673
674 if ((call->state == RXRPC_CALL_CLIENT_AWAIT_REPLY ||
675 call->state == RXRPC_CALL_SERVER_AWAIT_ACK) &&
676 hard > tx)
677 goto all_acked;
678
679 smp_rmb();
680 rxrpc_rotate_tx_window(call, hard - 1);
681 }
682
683 if (ack.nAcks > 0) {
684 if (hard - 1 + ack.nAcks > tx) {
685 _debug("soft-ACK'd packet %d+%d not"
686 " transmitted (%d top)",
687 hard - 1, ack.nAcks, tx);
688 goto protocol_error;
689 }
690
691 if (rxrpc_process_soft_ACKs(call, &ack, skb) < 0)
692 goto protocol_error;
693 }
694 goto discard;
695
696 /* complete ACK to process */
697 case RXRPC_PACKET_TYPE_ACKALL:
698 goto all_acked;
699
700 /* abort and busy are handled elsewhere */
701 case RXRPC_PACKET_TYPE_BUSY:
702 case RXRPC_PACKET_TYPE_ABORT:
703 BUG();
704
705 /* connection level events - also handled elsewhere */
706 case RXRPC_PACKET_TYPE_CHALLENGE:
707 case RXRPC_PACKET_TYPE_RESPONSE:
708 case RXRPC_PACKET_TYPE_DEBUG:
709 BUG();
710 }
711
712 /* if we've had a hard ACK that covers all the packets we've sent, then
713 * that ends that phase of the operation */
714 all_acked:
715 write_lock_bh(&call->state_lock);
716 _debug("ack all %d", call->state);
717
718 switch (call->state) {
719 case RXRPC_CALL_CLIENT_AWAIT_REPLY:
720 call->state = RXRPC_CALL_CLIENT_RECV_REPLY;
721 break;
722 case RXRPC_CALL_SERVER_AWAIT_ACK:
723 _debug("srv complete");
724 call->state = RXRPC_CALL_COMPLETE;
725 post_ACK = true;
726 break;
727 case RXRPC_CALL_CLIENT_SEND_REQUEST:
728 case RXRPC_CALL_SERVER_RECV_REQUEST:
729 goto protocol_error_unlock; /* can't occur yet */
730 default:
731 write_unlock_bh(&call->state_lock);
732 goto discard; /* assume packet left over from earlier phase */
733 }
734
735 write_unlock_bh(&call->state_lock);
736
737 /* if all the packets we sent are hard-ACK'd, then we can discard
738 * whatever we've got left */
739 _debug("clear Tx %d",
740 CIRC_CNT(call->acks_head, call->acks_tail, call->acks_winsz));
741
742 del_timer_sync(&call->resend_timer);
743 clear_bit(RXRPC_CALL_RUN_RTIMER, &call->flags);
744 clear_bit(RXRPC_CALL_RESEND_TIMER, &call->events);
745
746 if (call->acks_window)
747 rxrpc_zap_tx_window(call);
748
749 if (post_ACK) {
750 /* post the final ACK message for userspace to pick up */
751 _debug("post ACK");
752 skb->mark = RXRPC_SKB_MARK_FINAL_ACK;
753 sp->call = call;
754 rxrpc_get_call(call);
755 spin_lock_bh(&call->lock);
756 if (rxrpc_queue_rcv_skb(call, skb, true, true) < 0)
757 BUG();
758 spin_unlock_bh(&call->lock);
759 goto process_further;
760 }
761
762 discard:
763 rxrpc_free_skb(skb);
764 goto process_further;
765
766 protocol_error_unlock:
767 write_unlock_bh(&call->state_lock);
768 protocol_error:
769 rxrpc_free_skb(skb);
770 _leave(" = -EPROTO");
771 return -EPROTO;
772 }
773
774 /*
775 * post a message to the socket Rx queue for recvmsg() to pick up
776 */
777 static int rxrpc_post_message(struct rxrpc_call *call, u32 mark, u32 error,
778 bool fatal)
779 {
780 struct rxrpc_skb_priv *sp;
781 struct sk_buff *skb;
782 int ret;
783
784 _enter("{%d,%lx},%u,%u,%d",
785 call->debug_id, call->flags, mark, error, fatal);
786
787 /* remove timers and things for fatal messages */
788 if (fatal) {
789 del_timer_sync(&call->resend_timer);
790 del_timer_sync(&call->ack_timer);
791 clear_bit(RXRPC_CALL_RUN_RTIMER, &call->flags);
792 }
793
794 if (mark != RXRPC_SKB_MARK_NEW_CALL &&
795 !test_bit(RXRPC_CALL_HAS_USERID, &call->flags)) {
796 _leave("[no userid]");
797 return 0;
798 }
799
800 if (!test_bit(RXRPC_CALL_TERMINAL_MSG, &call->flags)) {
801 skb = alloc_skb(0, GFP_NOFS);
802 if (!skb)
803 return -ENOMEM;
804
805 rxrpc_new_skb(skb);
806
807 skb->mark = mark;
808
809 sp = rxrpc_skb(skb);
810 memset(sp, 0, sizeof(*sp));
811 sp->error = error;
812 sp->call = call;
813 rxrpc_get_call(call);
814
815 spin_lock_bh(&call->lock);
816 ret = rxrpc_queue_rcv_skb(call, skb, true, fatal);
817 spin_unlock_bh(&call->lock);
818 BUG_ON(ret < 0);
819 }
820
821 return 0;
822 }
823
824 /*
825 * handle background processing of incoming call packets and ACK / abort
826 * generation
827 */
828 void rxrpc_process_call(struct work_struct *work)
829 {
830 struct rxrpc_call *call =
831 container_of(work, struct rxrpc_call, processor);
832 struct rxrpc_ackpacket ack;
833 struct rxrpc_ackinfo ackinfo;
834 struct rxrpc_header hdr;
835 struct msghdr msg;
836 struct kvec iov[5];
837 unsigned long bits;
838 __be32 data, pad;
839 size_t len;
840 int genbit, loop, nbit, ioc, ret, mtu;
841 u32 abort_code = RX_PROTOCOL_ERROR;
842 u8 *acks = NULL;
843
844 //printk("\n--------------------\n");
845 _enter("{%d,%s,%lx} [%lu]",
846 call->debug_id, rxrpc_call_states[call->state], call->events,
847 (jiffies - call->creation_jif) / (HZ / 10));
848
849 if (test_and_set_bit(RXRPC_CALL_PROC_BUSY, &call->flags)) {
850 _debug("XXXXXXXXXXXXX RUNNING ON MULTIPLE CPUS XXXXXXXXXXXXX");
851 return;
852 }
853
854 /* there's a good chance we're going to have to send a message, so set
855 * one up in advance */
856 msg.msg_name = &call->conn->trans->peer->srx.transport.sin;
857 msg.msg_namelen = sizeof(call->conn->trans->peer->srx.transport.sin);
858 msg.msg_control = NULL;
859 msg.msg_controllen = 0;
860 msg.msg_flags = 0;
861
862 hdr.epoch = call->conn->epoch;
863 hdr.cid = call->cid;
864 hdr.callNumber = call->call_id;
865 hdr.seq = 0;
866 hdr.type = RXRPC_PACKET_TYPE_ACK;
867 hdr.flags = call->conn->out_clientflag;
868 hdr.userStatus = 0;
869 hdr.securityIndex = call->conn->security_ix;
870 hdr._rsvd = 0;
871 hdr.serviceId = call->conn->service_id;
872
873 memset(iov, 0, sizeof(iov));
874 iov[0].iov_base = &hdr;
875 iov[0].iov_len = sizeof(hdr);
876
877 /* deal with events of a final nature */
878 if (test_bit(RXRPC_CALL_RELEASE, &call->events)) {
879 rxrpc_release_call(call);
880 clear_bit(RXRPC_CALL_RELEASE, &call->events);
881 }
882
883 if (test_bit(RXRPC_CALL_RCVD_ERROR, &call->events)) {
884 int error;
885
886 clear_bit(RXRPC_CALL_CONN_ABORT, &call->events);
887 clear_bit(RXRPC_CALL_REJECT_BUSY, &call->events);
888 clear_bit(RXRPC_CALL_ABORT, &call->events);
889
890 error = call->conn->trans->peer->net_error;
891 _debug("post net error %d", error);
892
893 if (rxrpc_post_message(call, RXRPC_SKB_MARK_NET_ERROR,
894 error, true) < 0)
895 goto no_mem;
896 clear_bit(RXRPC_CALL_RCVD_ERROR, &call->events);
897 goto kill_ACKs;
898 }
899
900 if (test_bit(RXRPC_CALL_CONN_ABORT, &call->events)) {
901 ASSERTCMP(call->state, >, RXRPC_CALL_COMPLETE);
902
903 clear_bit(RXRPC_CALL_REJECT_BUSY, &call->events);
904 clear_bit(RXRPC_CALL_ABORT, &call->events);
905
906 _debug("post conn abort");
907
908 if (rxrpc_post_message(call, RXRPC_SKB_MARK_LOCAL_ERROR,
909 call->conn->error, true) < 0)
910 goto no_mem;
911 clear_bit(RXRPC_CALL_CONN_ABORT, &call->events);
912 goto kill_ACKs;
913 }
914
915 if (test_bit(RXRPC_CALL_REJECT_BUSY, &call->events)) {
916 hdr.type = RXRPC_PACKET_TYPE_BUSY;
917 genbit = RXRPC_CALL_REJECT_BUSY;
918 goto send_message;
919 }
920
921 if (test_bit(RXRPC_CALL_ABORT, &call->events)) {
922 ASSERTCMP(call->state, >, RXRPC_CALL_COMPLETE);
923
924 if (rxrpc_post_message(call, RXRPC_SKB_MARK_LOCAL_ERROR,
925 ECONNABORTED, true) < 0)
926 goto no_mem;
927 hdr.type = RXRPC_PACKET_TYPE_ABORT;
928 data = htonl(call->abort_code);
929 iov[1].iov_base = &data;
930 iov[1].iov_len = sizeof(data);
931 genbit = RXRPC_CALL_ABORT;
932 goto send_message;
933 }
934
935 if (test_bit(RXRPC_CALL_ACK_FINAL, &call->events)) {
936 genbit = RXRPC_CALL_ACK_FINAL;
937
938 ack.bufferSpace = htons(8);
939 ack.maxSkew = 0;
940 ack.serial = 0;
941 ack.reason = RXRPC_ACK_IDLE;
942 ack.nAcks = 0;
943 call->ackr_reason = 0;
944
945 spin_lock_bh(&call->lock);
946 ack.serial = call->ackr_serial;
947 ack.previousPacket = call->ackr_prev_seq;
948 ack.firstPacket = htonl(call->rx_data_eaten + 1);
949 spin_unlock_bh(&call->lock);
950
951 pad = 0;
952
953 iov[1].iov_base = &ack;
954 iov[1].iov_len = sizeof(ack);
955 iov[2].iov_base = &pad;
956 iov[2].iov_len = 3;
957 iov[3].iov_base = &ackinfo;
958 iov[3].iov_len = sizeof(ackinfo);
959 goto send_ACK;
960 }
961
962 if (call->events & ((1 << RXRPC_CALL_RCVD_BUSY) |
963 (1 << RXRPC_CALL_RCVD_ABORT))
964 ) {
965 u32 mark;
966
967 if (test_bit(RXRPC_CALL_RCVD_ABORT, &call->events))
968 mark = RXRPC_SKB_MARK_REMOTE_ABORT;
969 else
970 mark = RXRPC_SKB_MARK_BUSY;
971
972 _debug("post abort/busy");
973 rxrpc_clear_tx_window(call);
974 if (rxrpc_post_message(call, mark, ECONNABORTED, true) < 0)
975 goto no_mem;
976
977 clear_bit(RXRPC_CALL_RCVD_BUSY, &call->events);
978 clear_bit(RXRPC_CALL_RCVD_ABORT, &call->events);
979 goto kill_ACKs;
980 }
981
982 if (test_and_clear_bit(RXRPC_CALL_RCVD_ACKALL, &call->events)) {
983 _debug("do implicit ackall");
984 rxrpc_clear_tx_window(call);
985 }
986
987 if (test_bit(RXRPC_CALL_LIFE_TIMER, &call->events)) {
988 write_lock_bh(&call->state_lock);
989 if (call->state <= RXRPC_CALL_COMPLETE) {
990 call->state = RXRPC_CALL_LOCALLY_ABORTED;
991 call->abort_code = RX_CALL_TIMEOUT;
992 set_bit(RXRPC_CALL_ABORT, &call->events);
993 }
994 write_unlock_bh(&call->state_lock);
995
996 _debug("post timeout");
997 if (rxrpc_post_message(call, RXRPC_SKB_MARK_LOCAL_ERROR,
998 ETIME, true) < 0)
999 goto no_mem;
1000
1001 clear_bit(RXRPC_CALL_LIFE_TIMER, &call->events);
1002 goto kill_ACKs;
1003 }
1004
1005 /* deal with assorted inbound messages */
1006 if (!skb_queue_empty(&call->rx_queue)) {
1007 switch (rxrpc_process_rx_queue(call, &abort_code)) {
1008 case 0:
1009 case -EAGAIN:
1010 break;
1011 case -ENOMEM:
1012 goto no_mem;
1013 case -EKEYEXPIRED:
1014 case -EKEYREJECTED:
1015 case -EPROTO:
1016 rxrpc_abort_call(call, abort_code);
1017 goto kill_ACKs;
1018 }
1019 }
1020
1021 /* handle resending */
1022 if (test_and_clear_bit(RXRPC_CALL_RESEND_TIMER, &call->events))
1023 rxrpc_resend_timer(call);
1024 if (test_and_clear_bit(RXRPC_CALL_RESEND, &call->events))
1025 rxrpc_resend(call);
1026
1027 /* consider sending an ordinary ACK */
1028 if (test_bit(RXRPC_CALL_ACK, &call->events)) {
1029 _debug("send ACK: window: %d - %d { %lx }",
1030 call->rx_data_eaten, call->ackr_win_top,
1031 call->ackr_window[0]);
1032
1033 if (call->state > RXRPC_CALL_SERVER_ACK_REQUEST &&
1034 call->ackr_reason != RXRPC_ACK_PING_RESPONSE) {
1035 /* ACK by sending reply DATA packet in this state */
1036 clear_bit(RXRPC_CALL_ACK, &call->events);
1037 goto maybe_reschedule;
1038 }
1039
1040 genbit = RXRPC_CALL_ACK;
1041
1042 acks = kzalloc(call->ackr_win_top - call->rx_data_eaten,
1043 GFP_NOFS);
1044 if (!acks)
1045 goto no_mem;
1046
1047 //hdr.flags = RXRPC_SLOW_START_OK;
1048 ack.bufferSpace = htons(8);
1049 ack.maxSkew = 0;
1050 ack.serial = 0;
1051 ack.reason = 0;
1052
1053 spin_lock_bh(&call->lock);
1054 ack.reason = call->ackr_reason;
1055 ack.serial = call->ackr_serial;
1056 ack.previousPacket = call->ackr_prev_seq;
1057 ack.firstPacket = htonl(call->rx_data_eaten + 1);
1058
1059 ack.nAcks = 0;
1060 for (loop = 0; loop < RXRPC_ACKR_WINDOW_ASZ; loop++) {
1061 nbit = loop * BITS_PER_LONG;
1062 for (bits = call->ackr_window[loop]; bits; bits >>= 1
1063 ) {
1064 _debug("- l=%d n=%d b=%lx", loop, nbit, bits);
1065 if (bits & 1) {
1066 acks[nbit] = RXRPC_ACK_TYPE_ACK;
1067 ack.nAcks = nbit + 1;
1068 }
1069 nbit++;
1070 }
1071 }
1072 call->ackr_reason = 0;
1073 spin_unlock_bh(&call->lock);
1074
1075 pad = 0;
1076
1077 iov[1].iov_base = &ack;
1078 iov[1].iov_len = sizeof(ack);
1079 iov[2].iov_base = acks;
1080 iov[2].iov_len = ack.nAcks;
1081 iov[3].iov_base = &pad;
1082 iov[3].iov_len = 3;
1083 iov[4].iov_base = &ackinfo;
1084 iov[4].iov_len = sizeof(ackinfo);
1085
1086 switch (ack.reason) {
1087 case RXRPC_ACK_REQUESTED:
1088 case RXRPC_ACK_DUPLICATE:
1089 case RXRPC_ACK_OUT_OF_SEQUENCE:
1090 case RXRPC_ACK_EXCEEDS_WINDOW:
1091 case RXRPC_ACK_NOSPACE:
1092 case RXRPC_ACK_PING:
1093 case RXRPC_ACK_PING_RESPONSE:
1094 goto send_ACK_with_skew;
1095 case RXRPC_ACK_DELAY:
1096 case RXRPC_ACK_IDLE:
1097 goto send_ACK;
1098 }
1099 }
1100
1101 /* handle completion of security negotiations on an incoming
1102 * connection */
1103 if (test_and_clear_bit(RXRPC_CALL_SECURED, &call->events)) {
1104 _debug("secured");
1105 spin_lock_bh(&call->lock);
1106
1107 if (call->state == RXRPC_CALL_SERVER_SECURING) {
1108 _debug("securing");
1109 write_lock(&call->conn->lock);
1110 if (!test_bit(RXRPC_CALL_RELEASED, &call->flags) &&
1111 !test_bit(RXRPC_CALL_RELEASE, &call->events)) {
1112 _debug("not released");
1113 call->state = RXRPC_CALL_SERVER_ACCEPTING;
1114 list_move_tail(&call->accept_link,
1115 &call->socket->acceptq);
1116 }
1117 write_unlock(&call->conn->lock);
1118 read_lock(&call->state_lock);
1119 if (call->state < RXRPC_CALL_COMPLETE)
1120 set_bit(RXRPC_CALL_POST_ACCEPT, &call->events);
1121 read_unlock(&call->state_lock);
1122 }
1123
1124 spin_unlock_bh(&call->lock);
1125 if (!test_bit(RXRPC_CALL_POST_ACCEPT, &call->events))
1126 goto maybe_reschedule;
1127 }
1128
1129 /* post a notification of an acceptable connection to the app */
1130 if (test_bit(RXRPC_CALL_POST_ACCEPT, &call->events)) {
1131 _debug("post accept");
1132 if (rxrpc_post_message(call, RXRPC_SKB_MARK_NEW_CALL,
1133 0, false) < 0)
1134 goto no_mem;
1135 clear_bit(RXRPC_CALL_POST_ACCEPT, &call->events);
1136 goto maybe_reschedule;
1137 }
1138
1139 /* handle incoming call acceptance */
1140 if (test_and_clear_bit(RXRPC_CALL_ACCEPTED, &call->events)) {
1141 _debug("accepted");
1142 ASSERTCMP(call->rx_data_post, ==, 0);
1143 call->rx_data_post = 1;
1144 read_lock_bh(&call->state_lock);
1145 if (call->state < RXRPC_CALL_COMPLETE)
1146 set_bit(RXRPC_CALL_DRAIN_RX_OOS, &call->events);
1147 read_unlock_bh(&call->state_lock);
1148 }
1149
1150 /* drain the out of sequence received packet queue into the packet Rx
1151 * queue */
1152 if (test_and_clear_bit(RXRPC_CALL_DRAIN_RX_OOS, &call->events)) {
1153 while (call->rx_data_post == call->rx_first_oos)
1154 if (rxrpc_drain_rx_oos_queue(call) < 0)
1155 break;
1156 goto maybe_reschedule;
1157 }
1158
1159 /* other events may have been raised since we started checking */
1160 goto maybe_reschedule;
1161
1162 send_ACK_with_skew:
1163 ack.maxSkew = htons(atomic_read(&call->conn->hi_serial) -
1164 ntohl(ack.serial));
1165 send_ACK:
1166 mtu = call->conn->trans->peer->if_mtu;
1167 mtu -= call->conn->trans->peer->hdrsize;
1168 ackinfo.maxMTU = htonl(mtu);
1169 ackinfo.rwind = htonl(32);
1170
1171 /* permit the peer to send us jumbo packets if it wants to */
1172 ackinfo.rxMTU = htonl(5692);
1173 ackinfo.jumbo_max = htonl(4);
1174
1175 hdr.serial = htonl(atomic_inc_return(&call->conn->serial));
1176 _proto("Tx ACK %%%u { m=%hu f=#%u p=#%u s=%%%u r=%s n=%u }",
1177 ntohl(hdr.serial),
1178 ntohs(ack.maxSkew),
1179 ntohl(ack.firstPacket),
1180 ntohl(ack.previousPacket),
1181 ntohl(ack.serial),
1182 rxrpc_acks[ack.reason],
1183 ack.nAcks);
1184
1185 del_timer_sync(&call->ack_timer);
1186 if (ack.nAcks > 0)
1187 set_bit(RXRPC_CALL_TX_SOFT_ACK, &call->flags);
1188 goto send_message_2;
1189
1190 send_message:
1191 _debug("send message");
1192
1193 hdr.serial = htonl(atomic_inc_return(&call->conn->serial));
1194 _proto("Tx %s %%%u", rxrpc_pkts[hdr.type], ntohl(hdr.serial));
1195 send_message_2:
1196
1197 len = iov[0].iov_len;
1198 ioc = 1;
1199 if (iov[4].iov_len) {
1200 ioc = 5;
1201 len += iov[4].iov_len;
1202 len += iov[3].iov_len;
1203 len += iov[2].iov_len;
1204 len += iov[1].iov_len;
1205 } else if (iov[3].iov_len) {
1206 ioc = 4;
1207 len += iov[3].iov_len;
1208 len += iov[2].iov_len;
1209 len += iov[1].iov_len;
1210 } else if (iov[2].iov_len) {
1211 ioc = 3;
1212 len += iov[2].iov_len;
1213 len += iov[1].iov_len;
1214 } else if (iov[1].iov_len) {
1215 ioc = 2;
1216 len += iov[1].iov_len;
1217 }
1218
1219 ret = kernel_sendmsg(call->conn->trans->local->socket,
1220 &msg, iov, ioc, len);
1221 if (ret < 0) {
1222 _debug("sendmsg failed: %d", ret);
1223 read_lock_bh(&call->state_lock);
1224 if (call->state < RXRPC_CALL_DEAD)
1225 rxrpc_queue_call(call);
1226 read_unlock_bh(&call->state_lock);
1227 goto error;
1228 }
1229
1230 switch (genbit) {
1231 case RXRPC_CALL_ABORT:
1232 clear_bit(genbit, &call->events);
1233 clear_bit(RXRPC_CALL_RCVD_ABORT, &call->events);
1234 goto kill_ACKs;
1235
1236 case RXRPC_CALL_ACK_FINAL:
1237 write_lock_bh(&call->state_lock);
1238 if (call->state == RXRPC_CALL_CLIENT_FINAL_ACK)
1239 call->state = RXRPC_CALL_COMPLETE;
1240 write_unlock_bh(&call->state_lock);
1241 goto kill_ACKs;
1242
1243 default:
1244 clear_bit(genbit, &call->events);
1245 switch (call->state) {
1246 case RXRPC_CALL_CLIENT_AWAIT_REPLY:
1247 case RXRPC_CALL_CLIENT_RECV_REPLY:
1248 case RXRPC_CALL_SERVER_RECV_REQUEST:
1249 case RXRPC_CALL_SERVER_ACK_REQUEST:
1250 _debug("start ACK timer");
1251 rxrpc_propose_ACK(call, RXRPC_ACK_DELAY,
1252 call->ackr_serial, false);
1253 default:
1254 break;
1255 }
1256 goto maybe_reschedule;
1257 }
1258
1259 kill_ACKs:
1260 del_timer_sync(&call->ack_timer);
1261 if (test_and_clear_bit(RXRPC_CALL_ACK_FINAL, &call->events))
1262 rxrpc_put_call(call);
1263 clear_bit(RXRPC_CALL_ACK, &call->events);
1264
1265 maybe_reschedule:
1266 if (call->events || !skb_queue_empty(&call->rx_queue)) {
1267 read_lock_bh(&call->state_lock);
1268 if (call->state < RXRPC_CALL_DEAD)
1269 rxrpc_queue_call(call);
1270 read_unlock_bh(&call->state_lock);
1271 }
1272
1273 /* don't leave aborted connections on the accept queue */
1274 if (call->state >= RXRPC_CALL_COMPLETE &&
1275 !list_empty(&call->accept_link)) {
1276 _debug("X unlinking once-pending call %p { e=%lx f=%lx c=%x }",
1277 call, call->events, call->flags,
1278 ntohl(call->conn->cid));
1279
1280 read_lock_bh(&call->state_lock);
1281 if (!test_bit(RXRPC_CALL_RELEASED, &call->flags) &&
1282 !test_and_set_bit(RXRPC_CALL_RELEASE, &call->events))
1283 rxrpc_queue_call(call);
1284 read_unlock_bh(&call->state_lock);
1285 }
1286
1287 error:
1288 clear_bit(RXRPC_CALL_PROC_BUSY, &call->flags);
1289 kfree(acks);
1290
1291 /* because we don't want two CPUs both processing the work item for one
1292 * call at the same time, we use a flag to note when it's busy; however
1293 * this means there's a race between clearing the flag and setting the
1294 * work pending bit and the work item being processed again */
1295 if (call->events && !work_pending(&call->processor)) {
1296 _debug("jumpstart %x", ntohl(call->conn->cid));
1297 rxrpc_queue_call(call);
1298 }
1299
1300 _leave("");
1301 return;
1302
1303 no_mem:
1304 _debug("out of memory");
1305 goto maybe_reschedule;
1306 }
linux-next