1 /*- 2 * Copyright (c) 1982, 1986, 1990, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 3. Neither the name of the University nor the names of its contributors 14 * may be used to endorse or promote products derived from this software 15 * without specific prior written permission. 16 * 17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 27 * SUCH DAMAGE. 28 * 29 * @(#)sys_socket.c 8.1 (Berkeley) 6/10/93 30 */ 31 32 #include <sys/cdefs.h> 33 __FBSDID("$FreeBSD$"); 34 35 #include <sys/param.h> 36 #include <sys/systm.h> 37 #include <sys/aio.h> 38 #include <sys/domain.h> 39 #include <sys/file.h> 40 #include <sys/filedesc.h> 41 #include <sys/kernel.h> 42 #include <sys/kthread.h> 43 #include <sys/malloc.h> 44 #include <sys/proc.h> 45 #include <sys/protosw.h> 46 #include <sys/sigio.h> 47 #include <sys/signal.h> 48 #include <sys/signalvar.h> 49 #include <sys/socket.h> 50 #include <sys/socketvar.h> 51 #include <sys/filio.h> /* XXX */ 52 #include <sys/sockio.h> 53 #include <sys/stat.h> 54 #include <sys/sysctl.h> 55 #include <sys/sysproto.h> 56 #include <sys/taskqueue.h> 57 #include <sys/uio.h> 58 #include <sys/ucred.h> 59 #include <sys/un.h> 60 #include <sys/unpcb.h> 61 #include <sys/user.h> 62 63 #include <net/if.h> 64 #include <net/if_var.h> 65 #include <net/route.h> 66 #include <net/vnet.h> 67 68 #include <netinet/in.h> 69 #include <netinet/in_pcb.h> 70 71 #include <security/mac/mac_framework.h> 72 73 #include <vm/vm.h> 74 #include <vm/pmap.h> 75 #include <vm/vm_extern.h> 76 #include <vm/vm_map.h> 77 78 static SYSCTL_NODE(_kern_ipc, OID_AUTO, aio, CTLFLAG_RD, NULL, 79 "socket AIO stats"); 80 81 static int empty_results; 82 SYSCTL_INT(_kern_ipc_aio, OID_AUTO, empty_results, CTLFLAG_RD, &empty_results, 83 0, "socket operation returned EAGAIN"); 84 85 static int empty_retries; 86 SYSCTL_INT(_kern_ipc_aio, OID_AUTO, empty_retries, CTLFLAG_RD, &empty_retries, 87 0, "socket operation retries"); 88 89 static fo_rdwr_t soo_read; 90 static fo_rdwr_t soo_write; 91 static fo_ioctl_t soo_ioctl; 92 static fo_poll_t soo_poll; 93 extern fo_kqfilter_t soo_kqfilter; 94 static fo_stat_t soo_stat; 95 static fo_close_t soo_close; 96 static fo_fill_kinfo_t soo_fill_kinfo; 97 static fo_aio_queue_t soo_aio_queue; 98 99 static void soo_aio_cancel(struct kaiocb *job); 100 101 struct fileops socketops = { 102 .fo_read = soo_read, 103 .fo_write = soo_write, 104 .fo_truncate = invfo_truncate, 105 .fo_ioctl = soo_ioctl, 106 .fo_poll = soo_poll, 107 .fo_kqfilter = soo_kqfilter, 108 .fo_stat = soo_stat, 109 .fo_close = soo_close, 110 .fo_chmod = invfo_chmod, 111 .fo_chown = invfo_chown, 112 .fo_sendfile = invfo_sendfile, 113 .fo_fill_kinfo = soo_fill_kinfo, 114 .fo_aio_queue = soo_aio_queue, 115 .fo_flags = DFLAG_PASSABLE 116 }; 117 118 static int 119 soo_read(struct file *fp, struct uio *uio, struct ucred *active_cred, 120 int flags, struct thread *td) 121 { 122 struct socket *so = fp->f_data; 123 int error; 124 125 #ifdef MAC 126 error = mac_socket_check_receive(active_cred, so); 127 if (error) 128 return (error); 129 #endif 130 error = soreceive(so, 0, uio, 0, 0, 0); 131 return (error); 132 } 133 134 static int 135 soo_write(struct file *fp, struct uio *uio, struct ucred *active_cred, 136 int flags, struct thread *td) 137 { 138 struct socket *so = fp->f_data; 139 int error; 140 141 #ifdef MAC 142 error = mac_socket_check_send(active_cred, so); 143 if (error) 144 return (error); 145 #endif 146 error = sosend(so, 0, uio, 0, 0, 0, uio->uio_td); 147 if (error == EPIPE && (so->so_options & SO_NOSIGPIPE) == 0) { 148 PROC_LOCK(uio->uio_td->td_proc); 149 tdsignal(uio->uio_td, SIGPIPE); 150 PROC_UNLOCK(uio->uio_td->td_proc); 151 } 152 return (error); 153 } 154 155 static int 156 soo_ioctl(struct file *fp, u_long cmd, void *data, struct ucred *active_cred, 157 struct thread *td) 158 { 159 struct socket *so = fp->f_data; 160 int error = 0; 161 162 switch (cmd) { 163 case FIONBIO: 164 SOCK_LOCK(so); 165 if (*(int *)data) 166 so->so_state |= SS_NBIO; 167 else 168 so->so_state &= ~SS_NBIO; 169 SOCK_UNLOCK(so); 170 break; 171 172 case FIOASYNC: 173 /* 174 * XXXRW: This code separately acquires SOCK_LOCK(so) and 175 * SOCKBUF_LOCK(&so->so_rcv) even though they are the same 176 * mutex to avoid introducing the assumption that they are 177 * the same. 178 */ 179 if (*(int *)data) { 180 SOCK_LOCK(so); 181 so->so_state |= SS_ASYNC; 182 SOCK_UNLOCK(so); 183 SOCKBUF_LOCK(&so->so_rcv); 184 so->so_rcv.sb_flags |= SB_ASYNC; 185 SOCKBUF_UNLOCK(&so->so_rcv); 186 SOCKBUF_LOCK(&so->so_snd); 187 so->so_snd.sb_flags |= SB_ASYNC; 188 SOCKBUF_UNLOCK(&so->so_snd); 189 } else { 190 SOCK_LOCK(so); 191 so->so_state &= ~SS_ASYNC; 192 SOCK_UNLOCK(so); 193 SOCKBUF_LOCK(&so->so_rcv); 194 so->so_rcv.sb_flags &= ~SB_ASYNC; 195 SOCKBUF_UNLOCK(&so->so_rcv); 196 SOCKBUF_LOCK(&so->so_snd); 197 so->so_snd.sb_flags &= ~SB_ASYNC; 198 SOCKBUF_UNLOCK(&so->so_snd); 199 } 200 break; 201 202 case FIONREAD: 203 /* Unlocked read. */ 204 *(int *)data = sbavail(&so->so_rcv); 205 break; 206 207 case FIONWRITE: 208 /* Unlocked read. */ 209 *(int *)data = sbavail(&so->so_snd); 210 break; 211 212 case FIONSPACE: 213 /* Unlocked read. */ 214 if ((so->so_snd.sb_hiwat < sbused(&so->so_snd)) || 215 (so->so_snd.sb_mbmax < so->so_snd.sb_mbcnt)) 216 *(int *)data = 0; 217 else 218 *(int *)data = sbspace(&so->so_snd); 219 break; 220 221 case FIOSETOWN: 222 error = fsetown(*(int *)data, &so->so_sigio); 223 break; 224 225 case FIOGETOWN: 226 *(int *)data = fgetown(&so->so_sigio); 227 break; 228 229 case SIOCSPGRP: 230 error = fsetown(-(*(int *)data), &so->so_sigio); 231 break; 232 233 case SIOCGPGRP: 234 *(int *)data = -fgetown(&so->so_sigio); 235 break; 236 237 case SIOCATMARK: 238 /* Unlocked read. */ 239 *(int *)data = (so->so_rcv.sb_state & SBS_RCVATMARK) != 0; 240 break; 241 default: 242 /* 243 * Interface/routing/protocol specific ioctls: interface and 244 * routing ioctls should have a different entry since a 245 * socket is unnecessary. 246 */ 247 if (IOCGROUP(cmd) == 'i') 248 error = ifioctl(so, cmd, data, td); 249 else if (IOCGROUP(cmd) == 'r') { 250 CURVNET_SET(so->so_vnet); 251 error = rtioctl_fib(cmd, data, so->so_fibnum); 252 CURVNET_RESTORE(); 253 } else { 254 CURVNET_SET(so->so_vnet); 255 error = ((*so->so_proto->pr_usrreqs->pru_control) 256 (so, cmd, data, 0, td)); 257 CURVNET_RESTORE(); 258 } 259 break; 260 } 261 return (error); 262 } 263 264 static int 265 soo_poll(struct file *fp, int events, struct ucred *active_cred, 266 struct thread *td) 267 { 268 struct socket *so = fp->f_data; 269 #ifdef MAC 270 int error; 271 272 error = mac_socket_check_poll(active_cred, so); 273 if (error) 274 return (error); 275 #endif 276 return (sopoll(so, events, fp->f_cred, td)); 277 } 278 279 static int 280 soo_stat(struct file *fp, struct stat *ub, struct ucred *active_cred, 281 struct thread *td) 282 { 283 struct socket *so = fp->f_data; 284 struct sockbuf *sb; 285 #ifdef MAC 286 int error; 287 #endif 288 289 bzero((caddr_t)ub, sizeof (*ub)); 290 ub->st_mode = S_IFSOCK; 291 #ifdef MAC 292 error = mac_socket_check_stat(active_cred, so); 293 if (error) 294 return (error); 295 #endif 296 /* 297 * If SBS_CANTRCVMORE is set, but there's still data left in the 298 * receive buffer, the socket is still readable. 299 */ 300 sb = &so->so_rcv; 301 SOCKBUF_LOCK(sb); 302 if ((sb->sb_state & SBS_CANTRCVMORE) == 0 || sbavail(sb)) 303 ub->st_mode |= S_IRUSR | S_IRGRP | S_IROTH; 304 ub->st_size = sbavail(sb) - sb->sb_ctl; 305 SOCKBUF_UNLOCK(sb); 306 307 sb = &so->so_snd; 308 SOCKBUF_LOCK(sb); 309 if ((sb->sb_state & SBS_CANTSENDMORE) == 0) 310 ub->st_mode |= S_IWUSR | S_IWGRP | S_IWOTH; 311 SOCKBUF_UNLOCK(sb); 312 ub->st_uid = so->so_cred->cr_uid; 313 ub->st_gid = so->so_cred->cr_gid; 314 return (*so->so_proto->pr_usrreqs->pru_sense)(so, ub); 315 } 316 317 /* 318 * API socket close on file pointer. We call soclose() to close the socket 319 * (including initiating closing protocols). soclose() will sorele() the 320 * file reference but the actual socket will not go away until the socket's 321 * ref count hits 0. 322 */ 323 static int 324 soo_close(struct file *fp, struct thread *td) 325 { 326 int error = 0; 327 struct socket *so; 328 329 so = fp->f_data; 330 fp->f_ops = &badfileops; 331 fp->f_data = NULL; 332 333 if (so) 334 error = soclose(so); 335 return (error); 336 } 337 338 static int 339 soo_fill_kinfo(struct file *fp, struct kinfo_file *kif, struct filedesc *fdp) 340 { 341 struct sockaddr *sa; 342 struct inpcb *inpcb; 343 struct unpcb *unpcb; 344 struct socket *so; 345 int error; 346 347 kif->kf_type = KF_TYPE_SOCKET; 348 so = fp->f_data; 349 kif->kf_sock_domain = so->so_proto->pr_domain->dom_family; 350 kif->kf_sock_type = so->so_type; 351 kif->kf_sock_protocol = so->so_proto->pr_protocol; 352 kif->kf_un.kf_sock.kf_sock_pcb = (uintptr_t)so->so_pcb; 353 switch (kif->kf_sock_domain) { 354 case AF_INET: 355 case AF_INET6: 356 if (kif->kf_sock_protocol == IPPROTO_TCP) { 357 if (so->so_pcb != NULL) { 358 inpcb = (struct inpcb *)(so->so_pcb); 359 kif->kf_un.kf_sock.kf_sock_inpcb = 360 (uintptr_t)inpcb->inp_ppcb; 361 } 362 } 363 break; 364 case AF_UNIX: 365 if (so->so_pcb != NULL) { 366 unpcb = (struct unpcb *)(so->so_pcb); 367 if (unpcb->unp_conn) { 368 kif->kf_un.kf_sock.kf_sock_unpconn = 369 (uintptr_t)unpcb->unp_conn; 370 kif->kf_un.kf_sock.kf_sock_rcv_sb_state = 371 so->so_rcv.sb_state; 372 kif->kf_un.kf_sock.kf_sock_snd_sb_state = 373 so->so_snd.sb_state; 374 } 375 } 376 break; 377 } 378 error = so->so_proto->pr_usrreqs->pru_sockaddr(so, &sa); 379 if (error == 0 && sa->sa_len <= sizeof(kif->kf_sa_local)) { 380 bcopy(sa, &kif->kf_sa_local, sa->sa_len); 381 free(sa, M_SONAME); 382 } 383 error = so->so_proto->pr_usrreqs->pru_peeraddr(so, &sa); 384 if (error == 0 && sa->sa_len <= sizeof(kif->kf_sa_peer)) { 385 bcopy(sa, &kif->kf_sa_peer, sa->sa_len); 386 free(sa, M_SONAME); 387 } 388 strncpy(kif->kf_path, so->so_proto->pr_domain->dom_name, 389 sizeof(kif->kf_path)); 390 return (0); 391 } 392 393 /* 394 * Use the 'backend3' field in AIO jobs to store the amount of data 395 * completed by the AIO job so far. 396 */ 397 #define aio_done backend3 398 399 static STAILQ_HEAD(, task) soaio_jobs; 400 static struct mtx soaio_jobs_lock; 401 static struct task soaio_kproc_task; 402 static int soaio_starting, soaio_idle, soaio_queued; 403 static struct unrhdr *soaio_kproc_unr; 404 405 static int soaio_max_procs = MAX_AIO_PROCS; 406 SYSCTL_INT(_kern_ipc_aio, OID_AUTO, max_procs, CTLFLAG_RW, &soaio_max_procs, 0, 407 "Maximum number of kernel processes to use for async socket IO"); 408 409 static int soaio_num_procs; 410 SYSCTL_INT(_kern_ipc_aio, OID_AUTO, num_procs, CTLFLAG_RD, &soaio_num_procs, 0, 411 "Number of active kernel processes for async socket IO"); 412 413 static int soaio_target_procs = TARGET_AIO_PROCS; 414 SYSCTL_INT(_kern_ipc_aio, OID_AUTO, target_procs, CTLFLAG_RD, 415 &soaio_target_procs, 0, 416 "Preferred number of ready kernel processes for async socket IO"); 417 418 static int soaio_lifetime; 419 SYSCTL_INT(_kern_ipc_aio, OID_AUTO, lifetime, CTLFLAG_RW, &soaio_lifetime, 0, 420 "Maximum lifetime for idle aiod"); 421 422 static void 423 soaio_kproc_loop(void *arg) 424 { 425 struct proc *p; 426 struct vmspace *myvm; 427 struct task *task; 428 int error, id, pending; 429 430 id = (intptr_t)arg; 431 432 /* 433 * Grab an extra reference on the daemon's vmspace so that it 434 * doesn't get freed by jobs that switch to a different 435 * vmspace. 436 */ 437 p = curproc; 438 myvm = vmspace_acquire_ref(p); 439 440 mtx_lock(&soaio_jobs_lock); 441 MPASS(soaio_starting > 0); 442 soaio_starting--; 443 for (;;) { 444 while (!STAILQ_EMPTY(&soaio_jobs)) { 445 task = STAILQ_FIRST(&soaio_jobs); 446 STAILQ_REMOVE_HEAD(&soaio_jobs, ta_link); 447 soaio_queued--; 448 pending = task->ta_pending; 449 task->ta_pending = 0; 450 mtx_unlock(&soaio_jobs_lock); 451 452 task->ta_func(task->ta_context, pending); 453 454 mtx_lock(&soaio_jobs_lock); 455 } 456 MPASS(soaio_queued == 0); 457 458 if (p->p_vmspace != myvm) { 459 mtx_unlock(&soaio_jobs_lock); 460 vmspace_switch_aio(myvm); 461 mtx_lock(&soaio_jobs_lock); 462 continue; 463 } 464 465 soaio_idle++; 466 error = mtx_sleep(&soaio_idle, &soaio_jobs_lock, 0, "-", 467 soaio_lifetime); 468 soaio_idle--; 469 if (error == EWOULDBLOCK && STAILQ_EMPTY(&soaio_jobs) && 470 soaio_num_procs > soaio_target_procs) 471 break; 472 } 473 soaio_num_procs--; 474 mtx_unlock(&soaio_jobs_lock); 475 free_unr(soaio_kproc_unr, id); 476 kproc_exit(0); 477 } 478 479 static void 480 soaio_kproc_create(void *context, int pending) 481 { 482 struct proc *p; 483 int error, id; 484 485 mtx_lock(&soaio_jobs_lock); 486 for (;;) { 487 if (soaio_num_procs < soaio_target_procs) { 488 /* Must create */ 489 } else if (soaio_num_procs >= soaio_max_procs) { 490 /* 491 * Hit the limit on kernel processes, don't 492 * create another one. 493 */ 494 break; 495 } else if (soaio_queued <= soaio_idle + soaio_starting) { 496 /* 497 * No more AIO jobs waiting for a process to be 498 * created, so stop. 499 */ 500 break; 501 } 502 soaio_starting++; 503 mtx_unlock(&soaio_jobs_lock); 504 505 id = alloc_unr(soaio_kproc_unr); 506 error = kproc_create(soaio_kproc_loop, (void *)(intptr_t)id, 507 &p, 0, 0, "soaiod%d", id); 508 if (error != 0) { 509 free_unr(soaio_kproc_unr, id); 510 mtx_lock(&soaio_jobs_lock); 511 soaio_starting--; 512 break; 513 } 514 515 mtx_lock(&soaio_jobs_lock); 516 soaio_num_procs++; 517 } 518 mtx_unlock(&soaio_jobs_lock); 519 } 520 521 void 522 soaio_enqueue(struct task *task) 523 { 524 525 mtx_lock(&soaio_jobs_lock); 526 MPASS(task->ta_pending == 0); 527 task->ta_pending++; 528 STAILQ_INSERT_TAIL(&soaio_jobs, task, ta_link); 529 soaio_queued++; 530 if (soaio_queued <= soaio_idle) 531 wakeup_one(&soaio_idle); 532 else if (soaio_num_procs < soaio_max_procs) 533 taskqueue_enqueue(taskqueue_thread, &soaio_kproc_task); 534 mtx_unlock(&soaio_jobs_lock); 535 } 536 537 static void 538 soaio_init(void) 539 { 540 541 soaio_lifetime = AIOD_LIFETIME_DEFAULT; 542 STAILQ_INIT(&soaio_jobs); 543 mtx_init(&soaio_jobs_lock, "soaio jobs", NULL, MTX_DEF); 544 soaio_kproc_unr = new_unrhdr(1, INT_MAX, NULL); 545 TASK_INIT(&soaio_kproc_task, 0, soaio_kproc_create, NULL); 546 if (soaio_target_procs > 0) 547 taskqueue_enqueue(taskqueue_thread, &soaio_kproc_task); 548 } 549 SYSINIT(soaio, SI_SUB_VFS, SI_ORDER_ANY, soaio_init, NULL); 550 551 static __inline int 552 soaio_ready(struct socket *so, struct sockbuf *sb) 553 { 554 return (sb == &so->so_rcv ? soreadable(so) : sowriteable(so)); 555 } 556 557 static void 558 soaio_process_job(struct socket *so, struct sockbuf *sb, struct kaiocb *job) 559 { 560 struct ucred *td_savedcred; 561 struct thread *td; 562 struct file *fp; 563 struct uio uio; 564 struct iovec iov; 565 size_t cnt, done; 566 long ru_before; 567 int error, flags; 568 569 SOCKBUF_UNLOCK(sb); 570 aio_switch_vmspace(job); 571 td = curthread; 572 fp = job->fd_file; 573 retry: 574 td_savedcred = td->td_ucred; 575 td->td_ucred = job->cred; 576 577 done = job->aio_done; 578 cnt = job->uaiocb.aio_nbytes - done; 579 iov.iov_base = (void *)((uintptr_t)job->uaiocb.aio_buf + done); 580 iov.iov_len = cnt; 581 uio.uio_iov = &iov; 582 uio.uio_iovcnt = 1; 583 uio.uio_offset = 0; 584 uio.uio_resid = cnt; 585 uio.uio_segflg = UIO_USERSPACE; 586 uio.uio_td = td; 587 flags = MSG_NBIO; 588 589 /* 590 * For resource usage accounting, only count a completed request 591 * as a single message to avoid counting multiple calls to 592 * sosend/soreceive on a blocking socket. 593 */ 594 595 if (sb == &so->so_rcv) { 596 uio.uio_rw = UIO_READ; 597 ru_before = td->td_ru.ru_msgrcv; 598 #ifdef MAC 599 error = mac_socket_check_receive(fp->f_cred, so); 600 if (error == 0) 601 602 #endif 603 error = soreceive(so, NULL, &uio, NULL, NULL, &flags); 604 if (td->td_ru.ru_msgrcv != ru_before) 605 job->msgrcv = 1; 606 } else { 607 uio.uio_rw = UIO_WRITE; 608 ru_before = td->td_ru.ru_msgsnd; 609 #ifdef MAC 610 error = mac_socket_check_send(fp->f_cred, so); 611 if (error == 0) 612 #endif 613 error = sosend(so, NULL, &uio, NULL, NULL, flags, td); 614 if (td->td_ru.ru_msgsnd != ru_before) 615 job->msgsnd = 1; 616 if (error == EPIPE && (so->so_options & SO_NOSIGPIPE) == 0) { 617 PROC_LOCK(job->userproc); 618 kern_psignal(job->userproc, SIGPIPE); 619 PROC_UNLOCK(job->userproc); 620 } 621 } 622 623 done += cnt - uio.uio_resid; 624 job->aio_done = done; 625 td->td_ucred = td_savedcred; 626 627 if (error == EWOULDBLOCK) { 628 /* 629 * The request was either partially completed or not 630 * completed at all due to racing with a read() or 631 * write() on the socket. If the socket is 632 * non-blocking, return with any partial completion. 633 * If the socket is blocking or if no progress has 634 * been made, requeue this request at the head of the 635 * queue to try again when the socket is ready. 636 */ 637 MPASS(done != job->uaiocb.aio_nbytes); 638 SOCKBUF_LOCK(sb); 639 if (done == 0 || !(so->so_state & SS_NBIO)) { 640 empty_results++; 641 if (soaio_ready(so, sb)) { 642 empty_retries++; 643 SOCKBUF_UNLOCK(sb); 644 goto retry; 645 } 646 647 if (!aio_set_cancel_function(job, soo_aio_cancel)) { 648 SOCKBUF_UNLOCK(sb); 649 if (done != 0) 650 aio_complete(job, done, 0); 651 else 652 aio_cancel(job); 653 SOCKBUF_LOCK(sb); 654 } else { 655 TAILQ_INSERT_HEAD(&sb->sb_aiojobq, job, list); 656 } 657 return; 658 } 659 SOCKBUF_UNLOCK(sb); 660 } 661 if (done != 0 && (error == ERESTART || error == EINTR || 662 error == EWOULDBLOCK)) 663 error = 0; 664 if (error) 665 aio_complete(job, -1, error); 666 else 667 aio_complete(job, done, 0); 668 SOCKBUF_LOCK(sb); 669 } 670 671 static void 672 soaio_process_sb(struct socket *so, struct sockbuf *sb) 673 { 674 struct kaiocb *job; 675 676 SOCKBUF_LOCK(sb); 677 while (!TAILQ_EMPTY(&sb->sb_aiojobq) && soaio_ready(so, sb)) { 678 job = TAILQ_FIRST(&sb->sb_aiojobq); 679 TAILQ_REMOVE(&sb->sb_aiojobq, job, list); 680 if (!aio_clear_cancel_function(job)) 681 continue; 682 683 soaio_process_job(so, sb, job); 684 } 685 686 /* 687 * If there are still pending requests, the socket must not be 688 * ready so set SB_AIO to request a wakeup when the socket 689 * becomes ready. 690 */ 691 if (!TAILQ_EMPTY(&sb->sb_aiojobq)) 692 sb->sb_flags |= SB_AIO; 693 sb->sb_flags &= ~SB_AIO_RUNNING; 694 SOCKBUF_UNLOCK(sb); 695 696 ACCEPT_LOCK(); 697 SOCK_LOCK(so); 698 sorele(so); 699 } 700 701 void 702 soaio_rcv(void *context, int pending) 703 { 704 struct socket *so; 705 706 so = context; 707 soaio_process_sb(so, &so->so_rcv); 708 } 709 710 void 711 soaio_snd(void *context, int pending) 712 { 713 struct socket *so; 714 715 so = context; 716 soaio_process_sb(so, &so->so_snd); 717 } 718 719 void 720 sowakeup_aio(struct socket *so, struct sockbuf *sb) 721 { 722 723 SOCKBUF_LOCK_ASSERT(sb); 724 sb->sb_flags &= ~SB_AIO; 725 if (sb->sb_flags & SB_AIO_RUNNING) 726 return; 727 sb->sb_flags |= SB_AIO_RUNNING; 728 if (sb == &so->so_snd) 729 SOCK_LOCK(so); 730 soref(so); 731 if (sb == &so->so_snd) 732 SOCK_UNLOCK(so); 733 soaio_enqueue(&sb->sb_aiotask); 734 } 735 736 static void 737 soo_aio_cancel(struct kaiocb *job) 738 { 739 struct socket *so; 740 struct sockbuf *sb; 741 long done; 742 int opcode; 743 744 so = job->fd_file->f_data; 745 opcode = job->uaiocb.aio_lio_opcode; 746 if (opcode == LIO_READ) 747 sb = &so->so_rcv; 748 else { 749 MPASS(opcode == LIO_WRITE); 750 sb = &so->so_snd; 751 } 752 753 SOCKBUF_LOCK(sb); 754 if (!aio_cancel_cleared(job)) 755 TAILQ_REMOVE(&sb->sb_aiojobq, job, list); 756 if (TAILQ_EMPTY(&sb->sb_aiojobq)) 757 sb->sb_flags &= ~SB_AIO; 758 SOCKBUF_UNLOCK(sb); 759 760 done = job->aio_done; 761 if (done != 0) 762 aio_complete(job, done, 0); 763 else 764 aio_cancel(job); 765 } 766 767 static int 768 soo_aio_queue(struct file *fp, struct kaiocb *job) 769 { 770 struct socket *so; 771 struct sockbuf *sb; 772 int error; 773 774 so = fp->f_data; 775 error = (*so->so_proto->pr_usrreqs->pru_aio_queue)(so, job); 776 if (error == 0) 777 return (0); 778 779 switch (job->uaiocb.aio_lio_opcode) { 780 case LIO_READ: 781 sb = &so->so_rcv; 782 break; 783 case LIO_WRITE: 784 sb = &so->so_snd; 785 break; 786 default: 787 return (EINVAL); 788 } 789 790 SOCKBUF_LOCK(sb); 791 if (!aio_set_cancel_function(job, soo_aio_cancel)) 792 panic("new job was cancelled"); 793 TAILQ_INSERT_TAIL(&sb->sb_aiojobq, job, list); 794 if (!(sb->sb_flags & SB_AIO_RUNNING)) { 795 if (soaio_ready(so, sb)) 796 sowakeup_aio(so, sb); 797 else 798 sb->sb_flags |= SB_AIO; 799 } 800 SOCKBUF_UNLOCK(sb); 801 return (0); 802 } 803