1 /*
2 * GPL HEADER START
3 *
4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
9 *
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
15 *
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
19 *
20 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
21 * CA 95054 USA or visit www.sun.com if you need additional information or
22 * have any questions.
23 *
24 * GPL HEADER END
25 */
26 /*
27 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
29 *
30 * Copyright (c) 2011, 2012, Intel Corporation.
31 */
32 /*
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
35 */
36
37 #include "socklnd.h"
38
39 int
ksocknal_lib_get_conn_addrs(ksock_conn_t * conn)40 ksocknal_lib_get_conn_addrs(ksock_conn_t *conn)
41 {
42 int rc = libcfs_sock_getaddr(conn->ksnc_sock, 1,
43 &conn->ksnc_ipaddr,
44 &conn->ksnc_port);
45
46 /* Didn't need the {get,put}connsock dance to deref ksnc_sock... */
47 LASSERT(!conn->ksnc_closing);
48
49 if (rc != 0) {
50 CERROR("Error %d getting sock peer IP\n", rc);
51 return rc;
52 }
53
54 rc = libcfs_sock_getaddr(conn->ksnc_sock, 0,
55 &conn->ksnc_myipaddr, NULL);
56 if (rc != 0) {
57 CERROR("Error %d getting sock local IP\n", rc);
58 return rc;
59 }
60
61 return 0;
62 }
63
64 int
ksocknal_lib_zc_capable(ksock_conn_t * conn)65 ksocknal_lib_zc_capable(ksock_conn_t *conn)
66 {
67 int caps = conn->ksnc_sock->sk->sk_route_caps;
68
69 if (conn->ksnc_proto == &ksocknal_protocol_v1x)
70 return 0;
71
72 /* ZC if the socket supports scatter/gather and doesn't need software
73 * checksums */
74 return ((caps & NETIF_F_SG) != 0 && (caps & NETIF_F_ALL_CSUM) != 0);
75 }
76
77 int
ksocknal_lib_send_iov(ksock_conn_t * conn,ksock_tx_t * tx)78 ksocknal_lib_send_iov(ksock_conn_t *conn, ksock_tx_t *tx)
79 {
80 struct socket *sock = conn->ksnc_sock;
81 int nob;
82 int rc;
83
84 if (*ksocknal_tunables.ksnd_enable_csum && /* checksum enabled */
85 conn->ksnc_proto == &ksocknal_protocol_v2x && /* V2.x connection */
86 tx->tx_nob == tx->tx_resid && /* frist sending */
87 tx->tx_msg.ksm_csum == 0) /* not checksummed */
88 ksocknal_lib_csum_tx(tx);
89
90 /* NB we can't trust socket ops to either consume our iovs
91 * or leave them alone. */
92
93 {
94 #if SOCKNAL_SINGLE_FRAG_TX
95 struct kvec scratch;
96 struct kvec *scratchiov = &scratch;
97 unsigned int niov = 1;
98 #else
99 struct kvec *scratchiov = conn->ksnc_scheduler->kss_scratch_iov;
100 unsigned int niov = tx->tx_niov;
101 #endif
102 struct msghdr msg = {.msg_flags = MSG_DONTWAIT};
103 int i;
104
105 for (nob = i = 0; i < niov; i++) {
106 scratchiov[i] = tx->tx_iov[i];
107 nob += scratchiov[i].iov_len;
108 }
109
110 if (!list_empty(&conn->ksnc_tx_queue) ||
111 nob < tx->tx_resid)
112 msg.msg_flags |= MSG_MORE;
113
114 rc = kernel_sendmsg(sock, &msg, scratchiov, niov, nob);
115 }
116 return rc;
117 }
118
119 int
ksocknal_lib_send_kiov(ksock_conn_t * conn,ksock_tx_t * tx)120 ksocknal_lib_send_kiov(ksock_conn_t *conn, ksock_tx_t *tx)
121 {
122 struct socket *sock = conn->ksnc_sock;
123 lnet_kiov_t *kiov = tx->tx_kiov;
124 int rc;
125 int nob;
126
127 /* Not NOOP message */
128 LASSERT(tx->tx_lnetmsg != NULL);
129
130 /* NB we can't trust socket ops to either consume our iovs
131 * or leave them alone. */
132 if (tx->tx_msg.ksm_zc_cookies[0] != 0) {
133 /* Zero copy is enabled */
134 struct sock *sk = sock->sk;
135 struct page *page = kiov->kiov_page;
136 int offset = kiov->kiov_offset;
137 int fragsize = kiov->kiov_len;
138 int msgflg = MSG_DONTWAIT;
139
140 CDEBUG(D_NET, "page %p + offset %x for %d\n",
141 page, offset, kiov->kiov_len);
142
143 if (!list_empty(&conn->ksnc_tx_queue) ||
144 fragsize < tx->tx_resid)
145 msgflg |= MSG_MORE;
146
147 if (sk->sk_prot->sendpage != NULL) {
148 rc = sk->sk_prot->sendpage(sk, page,
149 offset, fragsize, msgflg);
150 } else {
151 rc = cfs_tcp_sendpage(sk, page, offset, fragsize,
152 msgflg);
153 }
154 } else {
155 #if SOCKNAL_SINGLE_FRAG_TX || !SOCKNAL_RISK_KMAP_DEADLOCK
156 struct kvec scratch;
157 struct kvec *scratchiov = &scratch;
158 unsigned int niov = 1;
159 #else
160 #ifdef CONFIG_HIGHMEM
161 #warning "XXX risk of kmap deadlock on multiple frags..."
162 #endif
163 struct kvec *scratchiov = conn->ksnc_scheduler->kss_scratch_iov;
164 unsigned int niov = tx->tx_nkiov;
165 #endif
166 struct msghdr msg = {.msg_flags = MSG_DONTWAIT};
167 int i;
168
169 for (nob = i = 0; i < niov; i++) {
170 scratchiov[i].iov_base = kmap(kiov[i].kiov_page) +
171 kiov[i].kiov_offset;
172 nob += scratchiov[i].iov_len = kiov[i].kiov_len;
173 }
174
175 if (!list_empty(&conn->ksnc_tx_queue) ||
176 nob < tx->tx_resid)
177 msg.msg_flags |= MSG_MORE;
178
179 rc = kernel_sendmsg(sock, &msg, (struct kvec *)scratchiov, niov, nob);
180
181 for (i = 0; i < niov; i++)
182 kunmap(kiov[i].kiov_page);
183 }
184 return rc;
185 }
186
187 void
ksocknal_lib_eager_ack(ksock_conn_t * conn)188 ksocknal_lib_eager_ack(ksock_conn_t *conn)
189 {
190 int opt = 1;
191 struct socket *sock = conn->ksnc_sock;
192
193 /* Remind the socket to ACK eagerly. If I don't, the socket might
194 * think I'm about to send something it could piggy-back the ACK
195 * on, introducing delay in completing zero-copy sends in my
196 * peer. */
197
198 kernel_setsockopt(sock, SOL_TCP, TCP_QUICKACK,
199 (char *)&opt, sizeof(opt));
200 }
201
202 int
ksocknal_lib_recv_iov(ksock_conn_t * conn)203 ksocknal_lib_recv_iov(ksock_conn_t *conn)
204 {
205 #if SOCKNAL_SINGLE_FRAG_RX
206 struct kvec scratch;
207 struct kvec *scratchiov = &scratch;
208 unsigned int niov = 1;
209 #else
210 struct kvec *scratchiov = conn->ksnc_scheduler->kss_scratch_iov;
211 unsigned int niov = conn->ksnc_rx_niov;
212 #endif
213 struct kvec *iov = conn->ksnc_rx_iov;
214 struct msghdr msg = {
215 .msg_flags = 0
216 };
217 int nob;
218 int i;
219 int rc;
220 int fragnob;
221 int sum;
222 __u32 saved_csum;
223
224 /* NB we can't trust socket ops to either consume our iovs
225 * or leave them alone. */
226 LASSERT(niov > 0);
227
228 for (nob = i = 0; i < niov; i++) {
229 scratchiov[i] = iov[i];
230 nob += scratchiov[i].iov_len;
231 }
232 LASSERT(nob <= conn->ksnc_rx_nob_wanted);
233
234 rc = kernel_recvmsg(conn->ksnc_sock, &msg,
235 scratchiov, niov, nob, MSG_DONTWAIT);
236
237 saved_csum = 0;
238 if (conn->ksnc_proto == &ksocknal_protocol_v2x) {
239 saved_csum = conn->ksnc_msg.ksm_csum;
240 conn->ksnc_msg.ksm_csum = 0;
241 }
242
243 if (saved_csum != 0) {
244 /* accumulate checksum */
245 for (i = 0, sum = rc; sum > 0; i++, sum -= fragnob) {
246 LASSERT(i < niov);
247
248 fragnob = iov[i].iov_len;
249 if (fragnob > sum)
250 fragnob = sum;
251
252 conn->ksnc_rx_csum = ksocknal_csum(conn->ksnc_rx_csum,
253 iov[i].iov_base, fragnob);
254 }
255 conn->ksnc_msg.ksm_csum = saved_csum;
256 }
257
258 return rc;
259 }
260
261 static void
ksocknal_lib_kiov_vunmap(void * addr)262 ksocknal_lib_kiov_vunmap(void *addr)
263 {
264 if (addr == NULL)
265 return;
266
267 vunmap(addr);
268 }
269
270 static void *
ksocknal_lib_kiov_vmap(lnet_kiov_t * kiov,int niov,struct kvec * iov,struct page ** pages)271 ksocknal_lib_kiov_vmap(lnet_kiov_t *kiov, int niov,
272 struct kvec *iov, struct page **pages)
273 {
274 void *addr;
275 int nob;
276 int i;
277
278 if (!*ksocknal_tunables.ksnd_zc_recv || pages == NULL)
279 return NULL;
280
281 LASSERT(niov <= LNET_MAX_IOV);
282
283 if (niov < 2 ||
284 niov < *ksocknal_tunables.ksnd_zc_recv_min_nfrags)
285 return NULL;
286
287 for (nob = i = 0; i < niov; i++) {
288 if ((kiov[i].kiov_offset != 0 && i > 0) ||
289 (kiov[i].kiov_offset + kiov[i].kiov_len != PAGE_CACHE_SIZE && i < niov - 1))
290 return NULL;
291
292 pages[i] = kiov[i].kiov_page;
293 nob += kiov[i].kiov_len;
294 }
295
296 addr = vmap(pages, niov, VM_MAP, PAGE_KERNEL);
297 if (addr == NULL)
298 return NULL;
299
300 iov->iov_base = addr + kiov[0].kiov_offset;
301 iov->iov_len = nob;
302
303 return addr;
304 }
305
306 int
ksocknal_lib_recv_kiov(ksock_conn_t * conn)307 ksocknal_lib_recv_kiov(ksock_conn_t *conn)
308 {
309 #if SOCKNAL_SINGLE_FRAG_RX || !SOCKNAL_RISK_KMAP_DEADLOCK
310 struct kvec scratch;
311 struct kvec *scratchiov = &scratch;
312 struct page **pages = NULL;
313 unsigned int niov = 1;
314 #else
315 #ifdef CONFIG_HIGHMEM
316 #warning "XXX risk of kmap deadlock on multiple frags..."
317 #endif
318 struct kvec *scratchiov = conn->ksnc_scheduler->kss_scratch_iov;
319 struct page **pages = conn->ksnc_scheduler->kss_rx_scratch_pgs;
320 unsigned int niov = conn->ksnc_rx_nkiov;
321 #endif
322 lnet_kiov_t *kiov = conn->ksnc_rx_kiov;
323 struct msghdr msg = {
324 .msg_flags = 0
325 };
326 int nob;
327 int i;
328 int rc;
329 void *base;
330 void *addr;
331 int sum;
332 int fragnob;
333 int n;
334
335 /* NB we can't trust socket ops to either consume our iovs
336 * or leave them alone. */
337 addr = ksocknal_lib_kiov_vmap(kiov, niov, scratchiov, pages);
338 if (addr != NULL) {
339 nob = scratchiov[0].iov_len;
340 n = 1;
341
342 } else {
343 for (nob = i = 0; i < niov; i++) {
344 nob += scratchiov[i].iov_len = kiov[i].kiov_len;
345 scratchiov[i].iov_base = kmap(kiov[i].kiov_page) +
346 kiov[i].kiov_offset;
347 }
348 n = niov;
349 }
350
351 LASSERT(nob <= conn->ksnc_rx_nob_wanted);
352
353 rc = kernel_recvmsg(conn->ksnc_sock, &msg,
354 (struct kvec *)scratchiov, n, nob, MSG_DONTWAIT);
355
356 if (conn->ksnc_msg.ksm_csum != 0) {
357 for (i = 0, sum = rc; sum > 0; i++, sum -= fragnob) {
358 LASSERT(i < niov);
359
360 /* Dang! have to kmap again because I have nowhere to stash the
361 * mapped address. But by doing it while the page is still
362 * mapped, the kernel just bumps the map count and returns me
363 * the address it stashed. */
364 base = kmap(kiov[i].kiov_page) + kiov[i].kiov_offset;
365 fragnob = kiov[i].kiov_len;
366 if (fragnob > sum)
367 fragnob = sum;
368
369 conn->ksnc_rx_csum = ksocknal_csum(conn->ksnc_rx_csum,
370 base, fragnob);
371
372 kunmap(kiov[i].kiov_page);
373 }
374 }
375
376 if (addr != NULL) {
377 ksocknal_lib_kiov_vunmap(addr);
378 } else {
379 for (i = 0; i < niov; i++)
380 kunmap(kiov[i].kiov_page);
381 }
382
383 return rc;
384 }
385
386 void
ksocknal_lib_csum_tx(ksock_tx_t * tx)387 ksocknal_lib_csum_tx(ksock_tx_t *tx)
388 {
389 int i;
390 __u32 csum;
391 void *base;
392
393 LASSERT(tx->tx_iov[0].iov_base == &tx->tx_msg);
394 LASSERT(tx->tx_conn != NULL);
395 LASSERT(tx->tx_conn->ksnc_proto == &ksocknal_protocol_v2x);
396
397 tx->tx_msg.ksm_csum = 0;
398
399 csum = ksocknal_csum(~0, tx->tx_iov[0].iov_base,
400 tx->tx_iov[0].iov_len);
401
402 if (tx->tx_kiov != NULL) {
403 for (i = 0; i < tx->tx_nkiov; i++) {
404 base = kmap(tx->tx_kiov[i].kiov_page) +
405 tx->tx_kiov[i].kiov_offset;
406
407 csum = ksocknal_csum(csum, base, tx->tx_kiov[i].kiov_len);
408
409 kunmap(tx->tx_kiov[i].kiov_page);
410 }
411 } else {
412 for (i = 1; i < tx->tx_niov; i++)
413 csum = ksocknal_csum(csum, tx->tx_iov[i].iov_base,
414 tx->tx_iov[i].iov_len);
415 }
416
417 if (*ksocknal_tunables.ksnd_inject_csum_error) {
418 csum++;
419 *ksocknal_tunables.ksnd_inject_csum_error = 0;
420 }
421
422 tx->tx_msg.ksm_csum = csum;
423 }
424
425 int
ksocknal_lib_get_conn_tunables(ksock_conn_t * conn,int * txmem,int * rxmem,int * nagle)426 ksocknal_lib_get_conn_tunables(ksock_conn_t *conn, int *txmem, int *rxmem, int *nagle)
427 {
428 struct socket *sock = conn->ksnc_sock;
429 int len;
430 int rc;
431
432 rc = ksocknal_connsock_addref(conn);
433 if (rc != 0) {
434 LASSERT(conn->ksnc_closing);
435 *txmem = *rxmem = *nagle = 0;
436 return -ESHUTDOWN;
437 }
438
439 rc = libcfs_sock_getbuf(sock, txmem, rxmem);
440 if (rc == 0) {
441 len = sizeof(*nagle);
442 rc = kernel_getsockopt(sock, SOL_TCP, TCP_NODELAY,
443 (char *)nagle, &len);
444 }
445
446 ksocknal_connsock_decref(conn);
447
448 if (rc == 0)
449 *nagle = !*nagle;
450 else
451 *txmem = *rxmem = *nagle = 0;
452
453 return rc;
454 }
455
456 int
ksocknal_lib_setup_sock(struct socket * sock)457 ksocknal_lib_setup_sock(struct socket *sock)
458 {
459 int rc;
460 int option;
461 int keep_idle;
462 int keep_intvl;
463 int keep_count;
464 int do_keepalive;
465 struct linger linger;
466
467 sock->sk->sk_allocation = GFP_NOFS;
468
469 /* Ensure this socket aborts active sends immediately when we close
470 * it. */
471
472 linger.l_onoff = 0;
473 linger.l_linger = 0;
474
475 rc = kernel_setsockopt(sock, SOL_SOCKET, SO_LINGER,
476 (char *)&linger, sizeof(linger));
477 if (rc != 0) {
478 CERROR("Can't set SO_LINGER: %d\n", rc);
479 return rc;
480 }
481
482 option = -1;
483 rc = kernel_setsockopt(sock, SOL_TCP, TCP_LINGER2,
484 (char *)&option, sizeof(option));
485 if (rc != 0) {
486 CERROR("Can't set SO_LINGER2: %d\n", rc);
487 return rc;
488 }
489
490 if (!*ksocknal_tunables.ksnd_nagle) {
491 option = 1;
492
493 rc = kernel_setsockopt(sock, SOL_TCP, TCP_NODELAY,
494 (char *)&option, sizeof(option));
495 if (rc != 0) {
496 CERROR("Can't disable nagle: %d\n", rc);
497 return rc;
498 }
499 }
500
501 rc = libcfs_sock_setbuf(sock,
502 *ksocknal_tunables.ksnd_tx_buffer_size,
503 *ksocknal_tunables.ksnd_rx_buffer_size);
504 if (rc != 0) {
505 CERROR("Can't set buffer tx %d, rx %d buffers: %d\n",
506 *ksocknal_tunables.ksnd_tx_buffer_size,
507 *ksocknal_tunables.ksnd_rx_buffer_size, rc);
508 return rc;
509 }
510
511 /* TCP_BACKOFF_* sockopt tunables unsupported in stock kernels */
512
513 /* snapshot tunables */
514 keep_idle = *ksocknal_tunables.ksnd_keepalive_idle;
515 keep_count = *ksocknal_tunables.ksnd_keepalive_count;
516 keep_intvl = *ksocknal_tunables.ksnd_keepalive_intvl;
517
518 do_keepalive = (keep_idle > 0 && keep_count > 0 && keep_intvl > 0);
519
520 option = (do_keepalive ? 1 : 0);
521 rc = kernel_setsockopt(sock, SOL_SOCKET, SO_KEEPALIVE,
522 (char *)&option, sizeof(option));
523 if (rc != 0) {
524 CERROR("Can't set SO_KEEPALIVE: %d\n", rc);
525 return rc;
526 }
527
528 if (!do_keepalive)
529 return 0;
530
531 rc = kernel_setsockopt(sock, SOL_TCP, TCP_KEEPIDLE,
532 (char *)&keep_idle, sizeof(keep_idle));
533 if (rc != 0) {
534 CERROR("Can't set TCP_KEEPIDLE: %d\n", rc);
535 return rc;
536 }
537
538 rc = kernel_setsockopt(sock, SOL_TCP, TCP_KEEPINTVL,
539 (char *)&keep_intvl, sizeof(keep_intvl));
540 if (rc != 0) {
541 CERROR("Can't set TCP_KEEPINTVL: %d\n", rc);
542 return rc;
543 }
544
545 rc = kernel_setsockopt(sock, SOL_TCP, TCP_KEEPCNT,
546 (char *)&keep_count, sizeof(keep_count));
547 if (rc != 0) {
548 CERROR("Can't set TCP_KEEPCNT: %d\n", rc);
549 return rc;
550 }
551
552 return 0;
553 }
554
555 void
ksocknal_lib_push_conn(ksock_conn_t * conn)556 ksocknal_lib_push_conn(ksock_conn_t *conn)
557 {
558 struct sock *sk;
559 struct tcp_sock *tp;
560 int nonagle;
561 int val = 1;
562 int rc;
563
564 rc = ksocknal_connsock_addref(conn);
565 if (rc != 0) /* being shut down */
566 return;
567
568 sk = conn->ksnc_sock->sk;
569 tp = tcp_sk(sk);
570
571 lock_sock(sk);
572 nonagle = tp->nonagle;
573 tp->nonagle = 1;
574 release_sock(sk);
575
576 rc = kernel_setsockopt(conn->ksnc_sock, SOL_TCP, TCP_NODELAY,
577 (char *)&val, sizeof(val));
578 LASSERT(rc == 0);
579
580 lock_sock(sk);
581 tp->nonagle = nonagle;
582 release_sock(sk);
583
584 ksocknal_connsock_decref(conn);
585 }
586
587 extern void ksocknal_read_callback(ksock_conn_t *conn);
588 extern void ksocknal_write_callback(ksock_conn_t *conn);
589 /*
590 * socket call back in Linux
591 */
592 static void
ksocknal_data_ready(struct sock * sk)593 ksocknal_data_ready(struct sock *sk)
594 {
595 ksock_conn_t *conn;
596
597 /* interleave correctly with closing sockets... */
598 LASSERT(!in_irq());
599 read_lock(&ksocknal_data.ksnd_global_lock);
600
601 conn = sk->sk_user_data;
602 if (conn == NULL) { /* raced with ksocknal_terminate_conn */
603 LASSERT(sk->sk_data_ready != &ksocknal_data_ready);
604 sk->sk_data_ready(sk);
605 } else
606 ksocknal_read_callback(conn);
607
608 read_unlock(&ksocknal_data.ksnd_global_lock);
609 }
610
611 static void
ksocknal_write_space(struct sock * sk)612 ksocknal_write_space(struct sock *sk)
613 {
614 ksock_conn_t *conn;
615 int wspace;
616 int min_wpace;
617
618 /* interleave correctly with closing sockets... */
619 LASSERT(!in_irq());
620 read_lock(&ksocknal_data.ksnd_global_lock);
621
622 conn = sk->sk_user_data;
623 wspace = SOCKNAL_WSPACE(sk);
624 min_wpace = SOCKNAL_MIN_WSPACE(sk);
625
626 CDEBUG(D_NET, "sk %p wspace %d low water %d conn %p%s%s%s\n",
627 sk, wspace, min_wpace, conn,
628 (conn == NULL) ? "" : (conn->ksnc_tx_ready ?
629 " ready" : " blocked"),
630 (conn == NULL) ? "" : (conn->ksnc_tx_scheduled ?
631 " scheduled" : " idle"),
632 (conn == NULL) ? "" : (list_empty(&conn->ksnc_tx_queue) ?
633 " empty" : " queued"));
634
635 if (conn == NULL) { /* raced with ksocknal_terminate_conn */
636 LASSERT(sk->sk_write_space != &ksocknal_write_space);
637 sk->sk_write_space(sk);
638
639 read_unlock(&ksocknal_data.ksnd_global_lock);
640 return;
641 }
642
643 if (wspace >= min_wpace) { /* got enough space */
644 ksocknal_write_callback(conn);
645
646 /* Clear SOCK_NOSPACE _after_ ksocknal_write_callback so the
647 * ENOMEM check in ksocknal_transmit is race-free (think about
648 * it). */
649
650 clear_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
651 }
652
653 read_unlock(&ksocknal_data.ksnd_global_lock);
654 }
655
656 void
ksocknal_lib_save_callback(struct socket * sock,ksock_conn_t * conn)657 ksocknal_lib_save_callback(struct socket *sock, ksock_conn_t *conn)
658 {
659 conn->ksnc_saved_data_ready = sock->sk->sk_data_ready;
660 conn->ksnc_saved_write_space = sock->sk->sk_write_space;
661 }
662
663 void
ksocknal_lib_set_callback(struct socket * sock,ksock_conn_t * conn)664 ksocknal_lib_set_callback(struct socket *sock, ksock_conn_t *conn)
665 {
666 sock->sk->sk_user_data = conn;
667 sock->sk->sk_data_ready = ksocknal_data_ready;
668 sock->sk->sk_write_space = ksocknal_write_space;
669 return;
670 }
671
672 void
ksocknal_lib_reset_callback(struct socket * sock,ksock_conn_t * conn)673 ksocknal_lib_reset_callback(struct socket *sock, ksock_conn_t *conn)
674 {
675 /* Remove conn's network callbacks.
676 * NB I _have_ to restore the callback, rather than storing a noop,
677 * since the socket could survive past this module being unloaded!! */
678 sock->sk->sk_data_ready = conn->ksnc_saved_data_ready;
679 sock->sk->sk_write_space = conn->ksnc_saved_write_space;
680
681 /* A callback could be in progress already; they hold a read lock
682 * on ksnd_global_lock (to serialise with me) and NOOP if
683 * sk_user_data is NULL. */
684 sock->sk->sk_user_data = NULL;
685
686 return ;
687 }
688
689 int
ksocknal_lib_memory_pressure(ksock_conn_t * conn)690 ksocknal_lib_memory_pressure(ksock_conn_t *conn)
691 {
692 int rc = 0;
693 ksock_sched_t *sched;
694
695 sched = conn->ksnc_scheduler;
696 spin_lock_bh(&sched->kss_lock);
697
698 if (!test_bit(SOCK_NOSPACE, &conn->ksnc_sock->flags) &&
699 !conn->ksnc_tx_ready) {
700 /* SOCK_NOSPACE is set when the socket fills
701 * and cleared in the write_space callback
702 * (which also sets ksnc_tx_ready). If
703 * SOCK_NOSPACE and ksnc_tx_ready are BOTH
704 * zero, I didn't fill the socket and
705 * write_space won't reschedule me, so I
706 * return -ENOMEM to get my caller to retry
707 * after a timeout */
708 rc = -ENOMEM;
709 }
710
711 spin_unlock_bh(&sched->kss_lock);
712
713 return rc;
714 }
715