1 /*
2 * Copyright (c) 2003, 2004 Jeffrey M. Hsu. All rights reserved.
3 * Copyright (c) 2003, 2004 The DragonFly Project. All rights reserved.
4 *
5 * This code is derived from software contributed to The DragonFly Project
6 * by Jeffrey M. Hsu.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 * notice, this list of conditions and the following disclaimer in the
15 * documentation and/or other materials provided with the distribution.
16 * 3. Neither the name of The DragonFly Project nor the names of its
17 * contributors may be used to endorse or promote products derived
18 * from this software without specific, prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
22 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
23 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
24 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
25 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
26 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
27 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
28 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
29 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
30 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 */
33
34 #include <sys/param.h>
35 #include <sys/systm.h>
36 #include <sys/kernel.h>
37 #include <sys/malloc.h>
38 #include <sys/msgport.h>
39 #include <sys/protosw.h>
40 #include <sys/socket.h>
41 #include <sys/socketvar.h>
42 #include <sys/socketops.h>
43 #include <sys/thread.h>
44 #include <sys/msgport2.h>
45 #include <sys/spinlock2.h>
46 #include <sys/sysctl.h>
47 #include <sys/mbuf.h>
48 #include <vm/pmap.h>
49
50 #include <net/netmsg2.h>
51 #include <net/netisr2.h>
52 #include <sys/socketvar2.h>
53
54 #include <net/netisr.h>
55 #include <net/netmsg.h>
56
57 static int async_rcvd_drop_race = 0;
58 SYSCTL_INT(_kern_ipc, OID_AUTO, async_rcvd_drop_race, CTLFLAG_RW,
59 &async_rcvd_drop_race, 0, "# of asynchronized pru_rcvd msg drop races");
60
61 /*
62 * Abort a socket and free it, asynchronously. Called from
63 * soabort_async() only. soabort_async() got a ref on the
64 * socket which we must free on reply.
65 */
66 void
so_pru_abort_async(struct socket * so)67 so_pru_abort_async(struct socket *so)
68 {
69 struct netmsg_pru_abort *msg;
70
71 msg = kmalloc(sizeof(*msg), M_LWKTMSG, M_WAITOK | M_ZERO);
72 netmsg_init(&msg->base, so, &netisr_afree_free_so_rport,
73 0, so->so_proto->pr_usrreqs->pru_abort);
74 lwkt_sendmsg(so->so_port, &msg->base.lmsg);
75 }
76
77 /*
78 * Abort a socket and free it. Called from soabort_direct() only.
79 * Caller must make sure that the current CPU is inpcb's owner CPU.
80 * soabort_direct() got a ref on the socket which we must free.
81 */
82 void
so_pru_abort_direct(struct socket * so)83 so_pru_abort_direct(struct socket *so)
84 {
85 struct netmsg_pru_abort msg;
86 netisr_fn_t func = so->so_proto->pr_usrreqs->pru_abort;
87
88 netmsg_init(&msg.base, so, &netisr_adone_rport, 0, func);
89 msg.base.lmsg.ms_flags &= ~(MSGF_REPLY | MSGF_DONE);
90 msg.base.lmsg.ms_flags |= MSGF_SYNC;
91 func((netmsg_t)&msg);
92 KKASSERT(msg.base.lmsg.ms_flags & MSGF_DONE);
93 sofree(msg.base.nm_so);
94 }
95
96 int
so_pru_accept(struct socket * so,struct sockaddr ** nam)97 so_pru_accept(struct socket *so, struct sockaddr **nam)
98 {
99 struct netmsg_pru_accept msg;
100
101 netmsg_init(&msg.base, so, &curthread->td_msgport,
102 0, so->so_proto->pr_usrreqs->pru_accept);
103 msg.nm_nam = nam;
104
105 return lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
106 }
107
108 int
so_pru_attach(struct socket * so,int proto,struct pru_attach_info * ai)109 so_pru_attach(struct socket *so, int proto, struct pru_attach_info *ai)
110 {
111 struct netmsg_pru_attach msg;
112 int error;
113
114 netmsg_init(&msg.base, so, &curthread->td_msgport,
115 0, so->so_proto->pr_usrreqs->pru_attach);
116 msg.nm_proto = proto;
117 msg.nm_ai = ai;
118 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
119 return (error);
120 }
121
122 int
so_pru_attach_direct(struct socket * so,int proto,struct pru_attach_info * ai)123 so_pru_attach_direct(struct socket *so, int proto, struct pru_attach_info *ai)
124 {
125 struct netmsg_pru_attach msg;
126 netisr_fn_t func = so->so_proto->pr_usrreqs->pru_attach;
127
128 netmsg_init(&msg.base, so, &netisr_adone_rport, 0, func);
129 msg.base.lmsg.ms_flags &= ~(MSGF_REPLY | MSGF_DONE);
130 msg.base.lmsg.ms_flags |= MSGF_SYNC;
131 msg.nm_proto = proto;
132 msg.nm_ai = ai;
133 func((netmsg_t)&msg);
134 KKASSERT(msg.base.lmsg.ms_flags & MSGF_DONE);
135 return(msg.base.lmsg.ms_error);
136 }
137
138 int
so_pru_attach_fast(struct socket * so,int proto,struct pru_attach_info * ai)139 so_pru_attach_fast(struct socket *so, int proto, struct pru_attach_info *ai)
140 {
141 struct netmsg_pru_attach *msg;
142 int error;
143
144 error = so->so_proto->pr_usrreqs->pru_preattach(so, proto, ai);
145 if (error)
146 return error;
147
148 msg = kmalloc(sizeof(*msg), M_LWKTMSG, M_WAITOK | M_NULLOK);
149 if (msg == NULL) {
150 /*
151 * Fail to allocate message; fallback to
152 * synchronized pru_attach.
153 */
154 return so_pru_attach(so, proto, NULL /* postattach */);
155 }
156
157 netmsg_init(&msg->base, so, &netisr_afree_rport, 0,
158 so->so_proto->pr_usrreqs->pru_attach);
159 msg->nm_proto = proto;
160 msg->nm_ai = NULL; /* postattach */
161 if (so->so_port == netisr_curport())
162 lwkt_sendmsg_oncpu(so->so_port, &msg->base.lmsg);
163 else
164 lwkt_sendmsg(so->so_port, &msg->base.lmsg);
165
166 return 0;
167 }
168
169 /*
170 * NOTE: If the target port changes the bind operation will deal with it.
171 */
172 int
so_pru_bind(struct socket * so,struct sockaddr * nam,struct thread * td)173 so_pru_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
174 {
175 struct netmsg_pru_bind msg;
176 int error;
177
178 netmsg_init(&msg.base, so, &curthread->td_msgport,
179 0, so->so_proto->pr_usrreqs->pru_bind);
180 msg.nm_nam = nam;
181 msg.nm_td = td; /* used only for prison_ip() */
182 msg.nm_flags = 0;
183 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
184 return (error);
185 }
186
187 int
so_pru_connect(struct socket * so,struct sockaddr * nam,struct thread * td)188 so_pru_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
189 {
190 struct netmsg_pru_connect msg;
191 int error;
192
193 netmsg_init(&msg.base, so, &curthread->td_msgport,
194 0, so->so_proto->pr_usrreqs->pru_connect);
195 msg.nm_nam = nam;
196 msg.nm_td = td;
197 msg.nm_m = NULL;
198 msg.nm_sndflags = 0;
199 msg.nm_flags = 0;
200 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
201 return (error);
202 }
203
204 int
so_pru_connect_async(struct socket * so,struct sockaddr * nam,struct thread * td)205 so_pru_connect_async(struct socket *so, struct sockaddr *nam, struct thread *td)
206 {
207 struct netmsg_pru_connect *msg;
208 int error, flags;
209
210 KASSERT(so->so_proto->pr_usrreqs->pru_preconnect != NULL,
211 ("async pru_connect is not supported"));
212
213 /* NOTE: sockaddr immediately follows netmsg */
214 msg = kmalloc(sizeof(*msg) + nam->sa_len, M_LWKTMSG,
215 M_WAITOK | M_NULLOK);
216 if (msg == NULL) {
217 /*
218 * Fail to allocate message; fallback to
219 * synchronized pru_connect.
220 */
221 return so_pru_connect(so, nam, td);
222 }
223
224 error = so->so_proto->pr_usrreqs->pru_preconnect(so, nam, td);
225 if (error) {
226 kfree(msg, M_LWKTMSG);
227 return error;
228 }
229
230 flags = PRUC_ASYNC;
231 if (td != NULL && (so->so_proto->pr_flags & PR_ACONN_HOLDTD)) {
232 lwkt_hold(td);
233 flags |= PRUC_HELDTD;
234 }
235
236 netmsg_init(&msg->base, so, &netisr_afree_rport, 0,
237 so->so_proto->pr_usrreqs->pru_connect);
238 msg->nm_nam = (struct sockaddr *)(msg + 1);
239 memcpy(msg->nm_nam, nam, nam->sa_len);
240 msg->nm_td = td;
241 msg->nm_m = NULL;
242 msg->nm_sndflags = 0;
243 msg->nm_flags = flags;
244 if (so->so_port == netisr_curport())
245 lwkt_sendmsg_oncpu(so->so_port, &msg->base.lmsg);
246 else
247 lwkt_sendmsg(so->so_port, &msg->base.lmsg);
248 return 0;
249 }
250
251 int
so_pru_connect2(struct socket * so1,struct socket * so2,struct ucred * cred)252 so_pru_connect2(struct socket *so1, struct socket *so2, struct ucred *cred)
253 {
254 struct netmsg_pru_connect2 msg;
255 int error;
256
257 netmsg_init(&msg.base, so1, &curthread->td_msgport,
258 0, so1->so_proto->pr_usrreqs->pru_connect2);
259 msg.nm_so1 = so1;
260 msg.nm_so2 = so2;
261 msg.nm_cred = cred;
262 error = lwkt_domsg(so1->so_port, &msg.base.lmsg, 0);
263 return (error);
264 }
265
266 /*
267 * WARNING! Synchronous call from user context. Control function may do
268 * copyin/copyout.
269 */
270 int
so_pru_control_direct(struct socket * so,u_long cmd,caddr_t data,struct ifnet * ifp)271 so_pru_control_direct(struct socket *so, u_long cmd, caddr_t data,
272 struct ifnet *ifp)
273 {
274 struct netmsg_pru_control msg;
275 netisr_fn_t func = so->so_proto->pr_usrreqs->pru_control;
276
277 netmsg_init(&msg.base, so, &netisr_adone_rport, 0, func);
278 msg.base.lmsg.ms_flags &= ~(MSGF_REPLY | MSGF_DONE);
279 msg.base.lmsg.ms_flags |= MSGF_SYNC;
280 msg.nm_cmd = cmd;
281 msg.nm_data = data;
282 msg.nm_ifp = ifp;
283 msg.nm_td = curthread;
284 func((netmsg_t)&msg);
285 KKASSERT(msg.base.lmsg.ms_flags & MSGF_DONE);
286 return(msg.base.lmsg.ms_error);
287 }
288
289 int
so_pru_detach(struct socket * so)290 so_pru_detach(struct socket *so)
291 {
292 struct netmsg_pru_detach msg;
293 int error;
294
295 netmsg_init(&msg.base, so, &curthread->td_msgport,
296 0, so->so_proto->pr_usrreqs->pru_detach);
297 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
298 return (error);
299 }
300
301 int
so_pru_detach_direct(struct socket * so)302 so_pru_detach_direct(struct socket *so)
303 {
304 struct netmsg_pru_detach msg;
305 netisr_fn_t func = so->so_proto->pr_usrreqs->pru_detach;
306
307 netmsg_init(&msg.base, so, &netisr_adone_rport, 0, func);
308 msg.base.lmsg.ms_flags &= ~(MSGF_REPLY | MSGF_DONE);
309 msg.base.lmsg.ms_flags |= MSGF_SYNC;
310 func((netmsg_t)&msg);
311 KKASSERT(msg.base.lmsg.ms_flags & MSGF_DONE);
312 return(msg.base.lmsg.ms_error);
313 }
314
315 int
so_pru_disconnect(struct socket * so)316 so_pru_disconnect(struct socket *so)
317 {
318 struct netmsg_pru_disconnect msg;
319 int error;
320
321 netmsg_init(&msg.base, so, &curthread->td_msgport,
322 0, so->so_proto->pr_usrreqs->pru_disconnect);
323 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
324 return (error);
325 }
326
327 void
so_pru_disconnect_direct(struct socket * so)328 so_pru_disconnect_direct(struct socket *so)
329 {
330 struct netmsg_pru_disconnect msg;
331 netisr_fn_t func = so->so_proto->pr_usrreqs->pru_disconnect;
332
333 netmsg_init(&msg.base, so, &netisr_adone_rport, 0, func);
334 msg.base.lmsg.ms_flags &= ~(MSGF_REPLY | MSGF_DONE);
335 msg.base.lmsg.ms_flags |= MSGF_SYNC;
336 func((netmsg_t)&msg);
337 KKASSERT(msg.base.lmsg.ms_flags & MSGF_DONE);
338 }
339
340 int
so_pru_listen(struct socket * so,struct thread * td)341 so_pru_listen(struct socket *so, struct thread *td)
342 {
343 struct netmsg_pru_listen msg;
344 int error;
345
346 netmsg_init(&msg.base, so, &curthread->td_msgport,
347 0, so->so_proto->pr_usrreqs->pru_listen);
348 msg.nm_td = td; /* used only for prison_ip() XXX JH */
349 msg.nm_flags = 0;
350 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
351 return (error);
352 }
353
354 int
so_pru_peeraddr(struct socket * so,struct sockaddr ** nam)355 so_pru_peeraddr(struct socket *so, struct sockaddr **nam)
356 {
357 struct netmsg_pru_peeraddr msg;
358 int error;
359
360 netmsg_init(&msg.base, so, &curthread->td_msgport,
361 0, so->so_proto->pr_usrreqs->pru_peeraddr);
362 msg.nm_nam = nam;
363 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
364 return (error);
365 }
366
367 int
so_pru_rcvd(struct socket * so,int flags)368 so_pru_rcvd(struct socket *so, int flags)
369 {
370 struct netmsg_pru_rcvd msg;
371 int error;
372
373 netmsg_init(&msg.base, so, &curthread->td_msgport,
374 0, so->so_proto->pr_usrreqs->pru_rcvd);
375 msg.nm_flags = flags;
376 msg.nm_pru_flags = 0;
377 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
378 return (error);
379 }
380
381 void
so_pru_rcvd_async(struct socket * so)382 so_pru_rcvd_async(struct socket *so)
383 {
384 lwkt_msg_t lmsg = &so->so_rcvd_msg.base.lmsg;
385
386 KASSERT(so->so_proto->pr_flags & PR_ASYNC_RCVD,
387 ("async pru_rcvd is not supported"));
388
389 /*
390 * WARNING! Spinlock is a bit dodgy, use hacked up sendmsg
391 * to avoid deadlocking.
392 */
393 spin_lock(&so->so_rcvd_spin);
394 if ((so->so_rcvd_msg.nm_pru_flags & PRUR_DEAD) == 0) {
395 if (lmsg->ms_flags & MSGF_DONE) {
396 lwkt_sendmsg_prepare(so->so_port, lmsg);
397 spin_unlock(&so->so_rcvd_spin);
398 if (so->so_port == netisr_curport())
399 lwkt_sendmsg_start_oncpu(so->so_port, lmsg);
400 else
401 lwkt_sendmsg_start(so->so_port, lmsg);
402 } else {
403 spin_unlock(&so->so_rcvd_spin);
404 }
405 } else {
406 spin_unlock(&so->so_rcvd_spin);
407 }
408 }
409
410 int
so_pru_rcvoob(struct socket * so,struct mbuf * m,int flags)411 so_pru_rcvoob(struct socket *so, struct mbuf *m, int flags)
412 {
413 struct netmsg_pru_rcvoob msg;
414 int error;
415
416 netmsg_init(&msg.base, so, &curthread->td_msgport,
417 0, so->so_proto->pr_usrreqs->pru_rcvoob);
418 msg.nm_m = m;
419 msg.nm_flags = flags;
420 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
421 return (error);
422 }
423
424 /*
425 * NOTE: If the target port changes the implied connect will deal with it.
426 */
427 int
so_pru_send(struct socket * so,int flags,struct mbuf * m,struct sockaddr * addr,struct mbuf * control,struct thread * td)428 so_pru_send(struct socket *so, int flags, struct mbuf *m,
429 struct sockaddr *addr, struct mbuf *control, struct thread *td)
430 {
431 struct netmsg_pru_send msg;
432 int error;
433
434 netmsg_init(&msg.base, so, &curthread->td_msgport,
435 0, so->so_proto->pr_usrreqs->pru_send);
436 msg.nm_flags = flags;
437 msg.nm_m = m;
438 msg.nm_addr = addr;
439 msg.nm_control = control;
440 msg.nm_td = td;
441 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
442 return (error);
443 }
444
445 void
so_pru_sync(struct socket * so)446 so_pru_sync(struct socket *so)
447 {
448 struct netmsg_base msg;
449
450 netmsg_init(&msg, so, &curthread->td_msgport, 0,
451 netmsg_sync_handler);
452 lwkt_domsg(so->so_port, &msg.lmsg, 0);
453 }
454
455 void
so_pru_send_async(struct socket * so,int flags,struct mbuf * m,struct sockaddr * addr0,struct mbuf * control,struct thread * td)456 so_pru_send_async(struct socket *so, int flags, struct mbuf *m,
457 struct sockaddr *addr0, struct mbuf *control, struct thread *td)
458 {
459 struct netmsg_pru_send *msg;
460 struct sockaddr *addr = NULL;
461
462 KASSERT(so->so_proto->pr_flags & PR_ASYNC_SEND,
463 ("async pru_send is not supported"));
464
465 if (addr0 != NULL) {
466 addr = kmalloc(addr0->sa_len, M_SONAME, M_WAITOK | M_NULLOK);
467 if (addr == NULL) {
468 /*
469 * Fail to allocate address; fallback to
470 * synchronized pru_send.
471 */
472 so_pru_send(so, flags, m, addr0, control, td);
473 return;
474 }
475 memcpy(addr, addr0, addr0->sa_len);
476 flags |= PRUS_FREEADDR;
477 }
478 flags |= PRUS_NOREPLY;
479
480 if (td != NULL && (so->so_proto->pr_flags & PR_ASEND_HOLDTD)) {
481 lwkt_hold(td);
482 flags |= PRUS_HELDTD;
483 }
484
485 msg = &m->m_hdr.mh_sndmsg;
486 netmsg_init(&msg->base, so, &netisr_apanic_rport,
487 0, so->so_proto->pr_usrreqs->pru_send);
488 msg->nm_flags = flags;
489 msg->nm_m = m;
490 msg->nm_addr = addr;
491 msg->nm_control = control;
492 msg->nm_td = td;
493 if (so->so_port == netisr_curport())
494 lwkt_sendmsg_oncpu(so->so_port, &msg->base.lmsg);
495 else
496 lwkt_sendmsg(so->so_port, &msg->base.lmsg);
497 }
498
499 int
so_pru_sense(struct socket * so,struct stat * sb)500 so_pru_sense(struct socket *so, struct stat *sb)
501 {
502 struct netmsg_pru_sense msg;
503 int error;
504
505 netmsg_init(&msg.base, so, &curthread->td_msgport,
506 0, so->so_proto->pr_usrreqs->pru_sense);
507 msg.nm_stat = sb;
508 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
509 return (error);
510 }
511
512 int
so_pru_shutdown(struct socket * so)513 so_pru_shutdown(struct socket *so)
514 {
515 struct netmsg_pru_shutdown msg;
516 int error;
517
518 netmsg_init(&msg.base, so, &curthread->td_msgport,
519 0, so->so_proto->pr_usrreqs->pru_shutdown);
520 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
521 return (error);
522 }
523
524 int
so_pru_sockaddr(struct socket * so,struct sockaddr ** nam)525 so_pru_sockaddr(struct socket *so, struct sockaddr **nam)
526 {
527 struct netmsg_pru_sockaddr msg;
528 int error;
529
530 netmsg_init(&msg.base, so, &curthread->td_msgport,
531 0, so->so_proto->pr_usrreqs->pru_sockaddr);
532 msg.nm_nam = nam;
533 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
534 return (error);
535 }
536
537 int
so_pr_ctloutput(struct socket * so,struct sockopt * sopt)538 so_pr_ctloutput(struct socket *so, struct sockopt *sopt)
539 {
540 struct netmsg_pr_ctloutput msg;
541 int error;
542
543 KKASSERT(!sopt->sopt_val || kva_p(sopt->sopt_val));
544
545 if (sopt->sopt_dir == SOPT_SET && so->so_proto->pr_ctloutmsg != NULL) {
546 struct netmsg_pr_ctloutput *amsg;
547
548 /* Fast path: asynchronous pr_ctloutput */
549 amsg = so->so_proto->pr_ctloutmsg(sopt);
550 if (amsg != NULL) {
551 netmsg_init(&amsg->base, so, &netisr_afree_rport, 0,
552 so->so_proto->pr_ctloutput);
553 /* nm_flags and nm_sopt are setup by pr_ctloutmsg */
554 if (so->so_port == netisr_curport()) {
555 lwkt_sendmsg_oncpu(so->so_port,
556 &amsg->base.lmsg);
557 } else {
558 lwkt_sendmsg(so->so_port, &amsg->base.lmsg);
559 }
560 return 0;
561 }
562 /* FALLTHROUGH */
563 }
564
565 netmsg_init(&msg.base, so, &curthread->td_msgport,
566 0, so->so_proto->pr_ctloutput);
567 msg.nm_flags = 0;
568 msg.nm_sopt = sopt;
569 error = lwkt_domsg(so->so_port, &msg.base.lmsg, 0);
570 return (error);
571 }
572
573 struct lwkt_port *
so_pr_ctlport(struct protosw * pr,int cmd,struct sockaddr * arg,void * extra,int * cpuid)574 so_pr_ctlport(struct protosw *pr, int cmd, struct sockaddr *arg,
575 void *extra, int *cpuid)
576 {
577 if (pr->pr_ctlport == NULL)
578 return NULL;
579 KKASSERT(pr->pr_ctlinput != NULL);
580
581 return pr->pr_ctlport(cmd, arg, extra, cpuid);
582 }
583
584 /*
585 * Protocol control input, typically via icmp.
586 *
587 * If the protocol pr_ctlport is not NULL we call it to figure out the
588 * protocol port. If NULL is returned we can just return, otherwise
589 * we issue a netmsg to call pr_ctlinput in the proper thread.
590 *
591 * This must be done synchronously as arg and/or extra may point to
592 * temporary data.
593 */
594 void
so_pr_ctlinput(struct protosw * pr,int cmd,struct sockaddr * arg,void * extra)595 so_pr_ctlinput(struct protosw *pr, int cmd, struct sockaddr *arg, void *extra)
596 {
597 struct netmsg_pr_ctlinput msg;
598 lwkt_port_t port;
599 int cpuid;
600
601 port = so_pr_ctlport(pr, cmd, arg, extra, &cpuid);
602 if (port == NULL)
603 return;
604 netmsg_init(&msg.base, NULL, &curthread->td_msgport,
605 0, pr->pr_ctlinput);
606 msg.nm_cmd = cmd;
607 msg.nm_direct = 0;
608 msg.nm_arg = arg;
609 msg.nm_extra = extra;
610 lwkt_domsg(port, &msg.base.lmsg, 0);
611 }
612
613 void
so_pr_ctlinput_direct(struct protosw * pr,int cmd,struct sockaddr * arg,void * extra)614 so_pr_ctlinput_direct(struct protosw *pr, int cmd, struct sockaddr *arg,
615 void *extra)
616 {
617 struct netmsg_pr_ctlinput msg;
618 netisr_fn_t func;
619 lwkt_port_t port;
620 int cpuid;
621
622 port = so_pr_ctlport(pr, cmd, arg, extra, &cpuid);
623 if (port == NULL)
624 return;
625 if (cpuid != netisr_ncpus && cpuid != mycpuid)
626 return;
627
628 func = pr->pr_ctlinput;
629 netmsg_init(&msg.base, NULL, &netisr_adone_rport, 0, func);
630 msg.base.lmsg.ms_flags &= ~(MSGF_REPLY | MSGF_DONE);
631 msg.base.lmsg.ms_flags |= MSGF_SYNC;
632 msg.nm_cmd = cmd;
633 msg.nm_direct = 1;
634 msg.nm_arg = arg;
635 msg.nm_extra = extra;
636 func((netmsg_t)&msg);
637 KKASSERT(msg.base.lmsg.ms_flags & MSGF_DONE);
638 }
639
640 /*
641 * If we convert all the protosw pr_ functions for all the protocols
642 * to take a message directly, this layer can go away. For the moment
643 * our dispatcher ignores the return value, but since we are handling
644 * the replymsg ourselves we return EASYNC by convention.
645 */
646
647 /*
648 * Handle a predicate event request. This function is only called once
649 * when the predicate message queueing request is received.
650 */
651 void
netmsg_so_notify(netmsg_t msg)652 netmsg_so_notify(netmsg_t msg)
653 {
654 struct socket *so = msg->base.nm_so;
655 struct signalsockbuf *ssb;
656
657 ssb = (msg->notify.nm_etype & NM_REVENT) ? &so->so_rcv : &so->so_snd;
658
659 /*
660 * Reply immediately if the event has occured, otherwise queue the
661 * request.
662 *
663 * NOTE: Socket can change if this is an accept predicate so cache
664 * the token.
665 */
666 lwkt_getpooltoken(so);
667 atomic_set_int(&ssb->ssb_flags, SSB_MEVENT);
668 if (msg->notify.nm_predicate(&msg->notify)) {
669 if (TAILQ_EMPTY(&ssb->ssb_mlist))
670 atomic_clear_int(&ssb->ssb_flags, SSB_MEVENT);
671 lwkt_relpooltoken(so);
672 lwkt_replymsg(&msg->base.lmsg,
673 msg->base.lmsg.ms_error);
674 } else {
675 TAILQ_INSERT_TAIL(&ssb->ssb_mlist, &msg->notify, nm_list);
676 /*
677 * NOTE:
678 * If predict ever blocks, 'tok' will be released, so
679 * SSB_MEVENT set beforehand could have been cleared
680 * when we reach here. In case that happens, we set
681 * SSB_MEVENT again, after the notify has been queued.
682 */
683 atomic_set_int(&ssb->ssb_flags, SSB_MEVENT);
684 lwkt_relpooltoken(so);
685 }
686 }
687
688 /*
689 * Called by doio when trying to abort a netmsg_so_notify message.
690 * Unlike the other functions this one is dispatched directly by
691 * the LWKT subsystem, so it takes a lwkt_msg_t as an argument.
692 *
693 * The original message, lmsg, is under the control of the caller and
694 * will not be destroyed until we return so we can safely reference it
695 * in our synchronous abort request.
696 *
697 * This part of the abort request occurs on the originating cpu which
698 * means we may race the message flags and the original message may
699 * not even have been processed by the target cpu yet.
700 */
701 void
netmsg_so_notify_doabort(lwkt_msg_t lmsg)702 netmsg_so_notify_doabort(lwkt_msg_t lmsg)
703 {
704 struct netmsg_so_notify_abort msg;
705
706 if ((lmsg->ms_flags & (MSGF_DONE | MSGF_REPLY)) == 0) {
707 const struct netmsg_base *nmsg =
708 (const struct netmsg_base *)lmsg;
709
710 netmsg_init(&msg.base, nmsg->nm_so, &curthread->td_msgport,
711 0, netmsg_so_notify_abort);
712 msg.nm_notifymsg = (void *)lmsg;
713 lwkt_domsg(lmsg->ms_target_port, &msg.base.lmsg, 0);
714 }
715 }
716
717 /*
718 * Predicate requests can be aborted. This function is only called once
719 * and will interlock against processing/reply races (since such races
720 * occur on the same thread that controls the port where the abort is
721 * requeued).
722 *
723 * This part of the abort request occurs on the target cpu. The message
724 * flags must be tested again in case the test that we did on the
725 * originating cpu raced. Since messages are handled in sequence, the
726 * original message will have already been handled by the loop and either
727 * replied to or queued.
728 *
729 * We really only need to interlock with MSGF_REPLY (a bit that is set on
730 * our cpu when we reply). Note that MSGF_DONE is not set until the
731 * reply reaches the originating cpu. Test both bits anyway.
732 */
733 void
netmsg_so_notify_abort(netmsg_t msg)734 netmsg_so_notify_abort(netmsg_t msg)
735 {
736 struct netmsg_so_notify_abort *abrtmsg = &msg->notify_abort;
737 struct netmsg_so_notify *nmsg = abrtmsg->nm_notifymsg;
738 struct signalsockbuf *ssb;
739
740 /*
741 * The original notify message is not destroyed until after the
742 * abort request is returned, so we can check its state.
743 */
744 lwkt_getpooltoken(nmsg->base.nm_so);
745 if ((nmsg->base.lmsg.ms_flags & (MSGF_DONE | MSGF_REPLY)) == 0) {
746 ssb = (nmsg->nm_etype & NM_REVENT) ?
747 &nmsg->base.nm_so->so_rcv :
748 &nmsg->base.nm_so->so_snd;
749 TAILQ_REMOVE(&ssb->ssb_mlist, nmsg, nm_list);
750 lwkt_relpooltoken(nmsg->base.nm_so);
751 lwkt_replymsg(&nmsg->base.lmsg, EINTR);
752 } else {
753 lwkt_relpooltoken(nmsg->base.nm_so);
754 }
755
756 /*
757 * Reply to the abort message
758 */
759 lwkt_replymsg(&abrtmsg->base.lmsg, 0);
760 }
761
762 void
so_async_rcvd_reply(struct socket * so)763 so_async_rcvd_reply(struct socket *so)
764 {
765 /*
766 * Spinlock safe, reply runs to degenerate lwkt_null_replyport()
767 */
768 spin_lock(&so->so_rcvd_spin);
769 lwkt_replymsg(&so->so_rcvd_msg.base.lmsg, 0);
770 spin_unlock(&so->so_rcvd_spin);
771 }
772
773 void
so_async_rcvd_drop(struct socket * so)774 so_async_rcvd_drop(struct socket *so)
775 {
776 lwkt_msg_t lmsg = &so->so_rcvd_msg.base.lmsg;
777
778 /*
779 * Spinlock safe, drop runs to degenerate lwkt_spin_dropmsg()
780 */
781 spin_lock(&so->so_rcvd_spin);
782 so->so_rcvd_msg.nm_pru_flags |= PRUR_DEAD;
783 again:
784 lwkt_dropmsg(lmsg);
785 if ((lmsg->ms_flags & MSGF_DONE) == 0) {
786 ++async_rcvd_drop_race;
787 ssleep(so, &so->so_rcvd_spin, 0, "soadrop", 1);
788 goto again;
789 }
790 spin_unlock(&so->so_rcvd_spin);
791 }
792