uipc_socket.c revision 1.235.2.3
1 /* $NetBSD: uipc_socket.c,v 1.235.2.3 2015/09/22 12:06:07 skrll Exp $ */ 2 3 /*- 4 * Copyright (c) 2002, 2007, 2008, 2009 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * by Jason R. Thorpe of Wasabi Systems, Inc, and by Andrew Doran. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 29 * POSSIBILITY OF SUCH DAMAGE. 30 */ 31 32 /* 33 * Copyright (c) 2004 The FreeBSD Foundation 34 * Copyright (c) 2004 Robert Watson 35 * Copyright (c) 1982, 1986, 1988, 1990, 1993 36 * The Regents of the University of California. All rights reserved. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. Neither the name of the University nor the names of its contributors 47 * may be used to endorse or promote products derived from this software 48 * without specific prior written permission. 49 * 50 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 51 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 52 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 53 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 54 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 55 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 56 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 57 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 58 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 59 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 60 * SUCH DAMAGE. 61 * 62 * @(#)uipc_socket.c 8.6 (Berkeley) 5/2/95 63 */ 64 65 /* 66 * Socket operation routines. 67 * 68 * These routines are called by the routines in sys_socket.c or from a 69 * system process, and implement the semantics of socket operations by 70 * switching out to the protocol specific routines. 71 */ 72 73 #include <sys/cdefs.h> 74 __KERNEL_RCSID(0, "$NetBSD: uipc_socket.c,v 1.235.2.3 2015/09/22 12:06:07 skrll Exp $"); 75 76 #ifdef _KERNEL_OPT 77 #include "opt_compat_netbsd.h" 78 #include "opt_sock_counters.h" 79 #include "opt_sosend_loan.h" 80 #include "opt_mbuftrace.h" 81 #include "opt_somaxkva.h" 82 #include "opt_multiprocessor.h" /* XXX */ 83 #endif 84 85 #include <sys/param.h> 86 #include <sys/systm.h> 87 #include <sys/proc.h> 88 #include <sys/file.h> 89 #include <sys/filedesc.h> 90 #include <sys/kmem.h> 91 #include <sys/mbuf.h> 92 #include <sys/domain.h> 93 #include <sys/kernel.h> 94 #include <sys/protosw.h> 95 #include <sys/socket.h> 96 #include <sys/socketvar.h> 97 #include <sys/signalvar.h> 98 #include <sys/resourcevar.h> 99 #include <sys/uidinfo.h> 100 #include <sys/event.h> 101 #include <sys/poll.h> 102 #include <sys/kauth.h> 103 #include <sys/mutex.h> 104 #include <sys/condvar.h> 105 #include <sys/kthread.h> 106 107 #ifdef COMPAT_50 108 #include <compat/sys/time.h> 109 #include <compat/sys/socket.h> 110 #endif 111 112 #include <uvm/uvm_extern.h> 113 #include <uvm/uvm_loan.h> 114 #include <uvm/uvm_page.h> 115 116 MALLOC_DEFINE(M_SONAME, "soname", "socket name"); 117 118 extern const struct fileops socketops; 119 120 extern int somaxconn; /* patchable (XXX sysctl) */ 121 int somaxconn = SOMAXCONN; 122 kmutex_t *softnet_lock; 123 124 #ifdef SOSEND_COUNTERS 125 #include <sys/device.h> 126 127 static struct evcnt sosend_loan_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 128 NULL, "sosend", "loan big"); 129 static struct evcnt sosend_copy_big = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 130 NULL, "sosend", "copy big"); 131 static struct evcnt sosend_copy_small = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 132 NULL, "sosend", "copy small"); 133 static struct evcnt sosend_kvalimit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 134 NULL, "sosend", "kva limit"); 135 136 #define SOSEND_COUNTER_INCR(ev) (ev)->ev_count++ 137 138 EVCNT_ATTACH_STATIC(sosend_loan_big); 139 EVCNT_ATTACH_STATIC(sosend_copy_big); 140 EVCNT_ATTACH_STATIC(sosend_copy_small); 141 EVCNT_ATTACH_STATIC(sosend_kvalimit); 142 #else 143 144 #define SOSEND_COUNTER_INCR(ev) /* nothing */ 145 146 #endif /* SOSEND_COUNTERS */ 147 148 #if defined(SOSEND_NO_LOAN) || defined(MULTIPROCESSOR) 149 int sock_loan_thresh = -1; 150 #else 151 int sock_loan_thresh = 4096; 152 #endif 153 154 static kmutex_t so_pendfree_lock; 155 static struct mbuf *so_pendfree = NULL; 156 157 #ifndef SOMAXKVA 158 #define SOMAXKVA (16 * 1024 * 1024) 159 #endif 160 int somaxkva = SOMAXKVA; 161 static int socurkva; 162 static kcondvar_t socurkva_cv; 163 164 static kauth_listener_t socket_listener; 165 166 #define SOCK_LOAN_CHUNK 65536 167 168 static void sopendfree_thread(void *); 169 static kcondvar_t pendfree_thread_cv; 170 static lwp_t *sopendfree_lwp; 171 172 static void sysctl_kern_socket_setup(void); 173 static struct sysctllog *socket_sysctllog; 174 175 static vsize_t 176 sokvareserve(struct socket *so, vsize_t len) 177 { 178 int error; 179 180 mutex_enter(&so_pendfree_lock); 181 while (socurkva + len > somaxkva) { 182 SOSEND_COUNTER_INCR(&sosend_kvalimit); 183 error = cv_wait_sig(&socurkva_cv, &so_pendfree_lock); 184 if (error) { 185 len = 0; 186 break; 187 } 188 } 189 socurkva += len; 190 mutex_exit(&so_pendfree_lock); 191 return len; 192 } 193 194 static void 195 sokvaunreserve(vsize_t len) 196 { 197 198 mutex_enter(&so_pendfree_lock); 199 socurkva -= len; 200 cv_broadcast(&socurkva_cv); 201 mutex_exit(&so_pendfree_lock); 202 } 203 204 /* 205 * sokvaalloc: allocate kva for loan. 206 */ 207 208 vaddr_t 209 sokvaalloc(vaddr_t sva, vsize_t len, struct socket *so) 210 { 211 vaddr_t lva; 212 213 /* 214 * reserve kva. 215 */ 216 217 if (sokvareserve(so, len) == 0) 218 return 0; 219 220 /* 221 * allocate kva. 222 */ 223 224 lva = uvm_km_alloc(kernel_map, len, atop(sva) & uvmexp.colormask, 225 UVM_KMF_COLORMATCH | UVM_KMF_VAONLY | UVM_KMF_WAITVA); 226 if (lva == 0) { 227 sokvaunreserve(len); 228 return (0); 229 } 230 231 return lva; 232 } 233 234 /* 235 * sokvafree: free kva for loan. 236 */ 237 238 void 239 sokvafree(vaddr_t sva, vsize_t len) 240 { 241 242 /* 243 * free kva. 244 */ 245 246 uvm_km_free(kernel_map, sva, len, UVM_KMF_VAONLY); 247 248 /* 249 * unreserve kva. 250 */ 251 252 sokvaunreserve(len); 253 } 254 255 static void 256 sodoloanfree(struct vm_page **pgs, void *buf, size_t size) 257 { 258 vaddr_t sva, eva; 259 vsize_t len; 260 int npgs; 261 262 KASSERT(pgs != NULL); 263 264 eva = round_page((vaddr_t) buf + size); 265 sva = trunc_page((vaddr_t) buf); 266 len = eva - sva; 267 npgs = len >> PAGE_SHIFT; 268 269 pmap_kremove(sva, len); 270 pmap_update(pmap_kernel()); 271 uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE); 272 sokvafree(sva, len); 273 } 274 275 /* 276 * sopendfree_thread: free mbufs on "pendfree" list. 277 * unlock and relock so_pendfree_lock when freeing mbufs. 278 */ 279 280 static void 281 sopendfree_thread(void *v) 282 { 283 struct mbuf *m, *next; 284 size_t rv; 285 286 mutex_enter(&so_pendfree_lock); 287 288 for (;;) { 289 rv = 0; 290 while (so_pendfree != NULL) { 291 m = so_pendfree; 292 so_pendfree = NULL; 293 mutex_exit(&so_pendfree_lock); 294 295 for (; m != NULL; m = next) { 296 next = m->m_next; 297 KASSERT((~m->m_flags & (M_EXT|M_EXT_PAGES)) == 0); 298 KASSERT(m->m_ext.ext_refcnt == 0); 299 300 rv += m->m_ext.ext_size; 301 sodoloanfree(m->m_ext.ext_pgs, m->m_ext.ext_buf, 302 m->m_ext.ext_size); 303 pool_cache_put(mb_cache, m); 304 } 305 306 mutex_enter(&so_pendfree_lock); 307 } 308 if (rv) 309 cv_broadcast(&socurkva_cv); 310 cv_wait(&pendfree_thread_cv, &so_pendfree_lock); 311 } 312 panic("sopendfree_thread"); 313 /* NOTREACHED */ 314 } 315 316 void 317 soloanfree(struct mbuf *m, void *buf, size_t size, void *arg) 318 { 319 320 KASSERT(m != NULL); 321 322 /* 323 * postpone freeing mbuf. 324 * 325 * we can't do it in interrupt context 326 * because we need to put kva back to kernel_map. 327 */ 328 329 mutex_enter(&so_pendfree_lock); 330 m->m_next = so_pendfree; 331 so_pendfree = m; 332 cv_signal(&pendfree_thread_cv); 333 mutex_exit(&so_pendfree_lock); 334 } 335 336 static long 337 sosend_loan(struct socket *so, struct uio *uio, struct mbuf *m, long space) 338 { 339 struct iovec *iov = uio->uio_iov; 340 vaddr_t sva, eva; 341 vsize_t len; 342 vaddr_t lva; 343 int npgs, error; 344 vaddr_t va; 345 int i; 346 347 if (VMSPACE_IS_KERNEL_P(uio->uio_vmspace)) 348 return (0); 349 350 if (iov->iov_len < (size_t) space) 351 space = iov->iov_len; 352 if (space > SOCK_LOAN_CHUNK) 353 space = SOCK_LOAN_CHUNK; 354 355 eva = round_page((vaddr_t) iov->iov_base + space); 356 sva = trunc_page((vaddr_t) iov->iov_base); 357 len = eva - sva; 358 npgs = len >> PAGE_SHIFT; 359 360 KASSERT(npgs <= M_EXT_MAXPAGES); 361 362 lva = sokvaalloc(sva, len, so); 363 if (lva == 0) 364 return 0; 365 366 error = uvm_loan(&uio->uio_vmspace->vm_map, sva, len, 367 m->m_ext.ext_pgs, UVM_LOAN_TOPAGE); 368 if (error) { 369 sokvafree(lva, len); 370 return (0); 371 } 372 373 for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE) 374 pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]), 375 VM_PROT_READ, 0); 376 pmap_update(pmap_kernel()); 377 378 lva += (vaddr_t) iov->iov_base & PAGE_MASK; 379 380 MEXTADD(m, (void *) lva, space, M_MBUF, soloanfree, so); 381 m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP; 382 383 uio->uio_resid -= space; 384 /* uio_offset not updated, not set/used for write(2) */ 385 uio->uio_iov->iov_base = (char *)uio->uio_iov->iov_base + space; 386 uio->uio_iov->iov_len -= space; 387 if (uio->uio_iov->iov_len == 0) { 388 uio->uio_iov++; 389 uio->uio_iovcnt--; 390 } 391 392 return (space); 393 } 394 395 struct mbuf * 396 getsombuf(struct socket *so, int type) 397 { 398 struct mbuf *m; 399 400 m = m_get(M_WAIT, type); 401 MCLAIM(m, so->so_mowner); 402 return m; 403 } 404 405 static int 406 socket_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie, 407 void *arg0, void *arg1, void *arg2, void *arg3) 408 { 409 int result; 410 enum kauth_network_req req; 411 412 result = KAUTH_RESULT_DEFER; 413 req = (enum kauth_network_req)arg0; 414 415 if ((action != KAUTH_NETWORK_SOCKET) && 416 (action != KAUTH_NETWORK_BIND)) 417 return result; 418 419 switch (req) { 420 case KAUTH_REQ_NETWORK_BIND_PORT: 421 result = KAUTH_RESULT_ALLOW; 422 break; 423 424 case KAUTH_REQ_NETWORK_SOCKET_DROP: { 425 /* Normal users can only drop their own connections. */ 426 struct socket *so = (struct socket *)arg1; 427 428 if (so->so_cred && proc_uidmatch(cred, so->so_cred) == 0) 429 result = KAUTH_RESULT_ALLOW; 430 431 break; 432 } 433 434 case KAUTH_REQ_NETWORK_SOCKET_OPEN: 435 /* We allow "raw" routing/bluetooth sockets to anyone. */ 436 if ((u_long)arg1 == PF_ROUTE || (u_long)arg1 == PF_OROUTE 437 || (u_long)arg1 == PF_BLUETOOTH) { 438 result = KAUTH_RESULT_ALLOW; 439 } else { 440 /* Privileged, let secmodel handle this. */ 441 if ((u_long)arg2 == SOCK_RAW) 442 break; 443 } 444 445 result = KAUTH_RESULT_ALLOW; 446 447 break; 448 449 case KAUTH_REQ_NETWORK_SOCKET_CANSEE: 450 result = KAUTH_RESULT_ALLOW; 451 452 break; 453 454 default: 455 break; 456 } 457 458 return result; 459 } 460 461 void 462 soinit(void) 463 { 464 465 sysctl_kern_socket_setup(); 466 467 mutex_init(&so_pendfree_lock, MUTEX_DEFAULT, IPL_VM); 468 softnet_lock = mutex_obj_alloc(MUTEX_DEFAULT, IPL_NONE); 469 cv_init(&socurkva_cv, "sokva"); 470 cv_init(&pendfree_thread_cv, "sopendfr"); 471 soinit2(); 472 473 /* Set the initial adjusted socket buffer size. */ 474 if (sb_max_set(sb_max)) 475 panic("bad initial sb_max value: %lu", sb_max); 476 477 socket_listener = kauth_listen_scope(KAUTH_SCOPE_NETWORK, 478 socket_listener_cb, NULL); 479 } 480 481 void 482 soinit1(void) 483 { 484 int error = kthread_create(PRI_NONE, KTHREAD_MPSAFE, NULL, 485 sopendfree_thread, NULL, &sopendfree_lwp, "sopendfree"); 486 if (error) 487 panic("soinit1 %d", error); 488 } 489 490 /* 491 * socreate: create a new socket of the specified type and the protocol. 492 * 493 * => Caller may specify another socket for lock sharing (must not be held). 494 * => Returns the new socket without lock held. 495 */ 496 int 497 socreate(int dom, struct socket **aso, int type, int proto, struct lwp *l, 498 struct socket *lockso) 499 { 500 const struct protosw *prp; 501 struct socket *so; 502 uid_t uid; 503 int error; 504 kmutex_t *lock; 505 506 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 507 KAUTH_REQ_NETWORK_SOCKET_OPEN, KAUTH_ARG(dom), KAUTH_ARG(type), 508 KAUTH_ARG(proto)); 509 if (error != 0) 510 return error; 511 512 if (proto) 513 prp = pffindproto(dom, proto, type); 514 else 515 prp = pffindtype(dom, type); 516 if (prp == NULL) { 517 /* no support for domain */ 518 if (pffinddomain(dom) == 0) 519 return EAFNOSUPPORT; 520 /* no support for socket type */ 521 if (proto == 0 && type != 0) 522 return EPROTOTYPE; 523 return EPROTONOSUPPORT; 524 } 525 if (prp->pr_usrreqs == NULL) 526 return EPROTONOSUPPORT; 527 if (prp->pr_type != type) 528 return EPROTOTYPE; 529 530 so = soget(true); 531 so->so_type = type; 532 so->so_proto = prp; 533 so->so_send = sosend; 534 so->so_receive = soreceive; 535 #ifdef MBUFTRACE 536 so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner; 537 so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner; 538 so->so_mowner = &prp->pr_domain->dom_mowner; 539 #endif 540 uid = kauth_cred_geteuid(l->l_cred); 541 so->so_uidinfo = uid_find(uid); 542 so->so_cpid = l->l_proc->p_pid; 543 544 /* 545 * Lock assigned and taken during PCB attach, unless we share 546 * the lock with another socket, e.g. socketpair(2) case. 547 */ 548 if (lockso) { 549 lock = lockso->so_lock; 550 so->so_lock = lock; 551 mutex_obj_hold(lock); 552 mutex_enter(lock); 553 } 554 555 /* Attach the PCB (returns with the socket lock held). */ 556 error = (*prp->pr_usrreqs->pr_attach)(so, proto); 557 KASSERT(solocked(so)); 558 559 if (error) { 560 KASSERT(so->so_pcb == NULL); 561 so->so_state |= SS_NOFDREF; 562 sofree(so); 563 return error; 564 } 565 so->so_cred = kauth_cred_dup(l->l_cred); 566 sounlock(so); 567 568 *aso = so; 569 return 0; 570 } 571 572 /* 573 * fsocreate: create a socket and a file descriptor associated with it. 574 * 575 * => On success, write file descriptor to fdout and return zero. 576 * => On failure, return non-zero; *fdout will be undefined. 577 */ 578 int 579 fsocreate(int domain, struct socket **sop, int type, int proto, int *fdout) 580 { 581 lwp_t *l = curlwp; 582 int error, fd, flags; 583 struct socket *so; 584 struct file *fp; 585 586 if ((error = fd_allocfile(&fp, &fd)) != 0) { 587 return error; 588 } 589 flags = type & SOCK_FLAGS_MASK; 590 fd_set_exclose(l, fd, (flags & SOCK_CLOEXEC) != 0); 591 fp->f_flag = FREAD|FWRITE|((flags & SOCK_NONBLOCK) ? FNONBLOCK : 0)| 592 ((flags & SOCK_NOSIGPIPE) ? FNOSIGPIPE : 0); 593 fp->f_type = DTYPE_SOCKET; 594 fp->f_ops = &socketops; 595 596 type &= ~SOCK_FLAGS_MASK; 597 error = socreate(domain, &so, type, proto, l, NULL); 598 if (error) { 599 fd_abort(curproc, fp, fd); 600 return error; 601 } 602 if (flags & SOCK_NONBLOCK) { 603 so->so_state |= SS_NBIO; 604 } 605 fp->f_socket = so; 606 fd_affix(curproc, fp, fd); 607 608 if (sop != NULL) { 609 *sop = so; 610 } 611 *fdout = fd; 612 return error; 613 } 614 615 int 616 sofamily(const struct socket *so) 617 { 618 const struct protosw *pr; 619 const struct domain *dom; 620 621 if ((pr = so->so_proto) == NULL) 622 return AF_UNSPEC; 623 if ((dom = pr->pr_domain) == NULL) 624 return AF_UNSPEC; 625 return dom->dom_family; 626 } 627 628 int 629 sobind(struct socket *so, struct sockaddr *nam, struct lwp *l) 630 { 631 int error; 632 633 solock(so); 634 if (nam->sa_family != so->so_proto->pr_domain->dom_family) { 635 sounlock(so); 636 return EAFNOSUPPORT; 637 } 638 error = (*so->so_proto->pr_usrreqs->pr_bind)(so, nam, l); 639 sounlock(so); 640 return error; 641 } 642 643 int 644 solisten(struct socket *so, int backlog, struct lwp *l) 645 { 646 int error; 647 648 solock(so); 649 if ((so->so_state & (SS_ISCONNECTED | SS_ISCONNECTING | 650 SS_ISDISCONNECTING)) != 0) { 651 sounlock(so); 652 return EINVAL; 653 } 654 error = (*so->so_proto->pr_usrreqs->pr_listen)(so, l); 655 if (error != 0) { 656 sounlock(so); 657 return error; 658 } 659 if (TAILQ_EMPTY(&so->so_q)) 660 so->so_options |= SO_ACCEPTCONN; 661 if (backlog < 0) 662 backlog = 0; 663 so->so_qlimit = min(backlog, somaxconn); 664 sounlock(so); 665 return 0; 666 } 667 668 void 669 sofree(struct socket *so) 670 { 671 u_int refs; 672 673 KASSERT(solocked(so)); 674 675 if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) { 676 sounlock(so); 677 return; 678 } 679 if (so->so_head) { 680 /* 681 * We must not decommission a socket that's on the accept(2) 682 * queue. If we do, then accept(2) may hang after select(2) 683 * indicated that the listening socket was ready. 684 */ 685 if (!soqremque(so, 0)) { 686 sounlock(so); 687 return; 688 } 689 } 690 if (so->so_rcv.sb_hiwat) 691 (void)chgsbsize(so->so_uidinfo, &so->so_rcv.sb_hiwat, 0, 692 RLIM_INFINITY); 693 if (so->so_snd.sb_hiwat) 694 (void)chgsbsize(so->so_uidinfo, &so->so_snd.sb_hiwat, 0, 695 RLIM_INFINITY); 696 sbrelease(&so->so_snd, so); 697 KASSERT(!cv_has_waiters(&so->so_cv)); 698 KASSERT(!cv_has_waiters(&so->so_rcv.sb_cv)); 699 KASSERT(!cv_has_waiters(&so->so_snd.sb_cv)); 700 sorflush(so); 701 refs = so->so_aborting; /* XXX */ 702 /* Remove acccept filter if one is present. */ 703 if (so->so_accf != NULL) 704 (void)accept_filt_clear(so); 705 sounlock(so); 706 if (refs == 0) /* XXX */ 707 soput(so); 708 } 709 710 /* 711 * soclose: close a socket on last file table reference removal. 712 * Initiate disconnect if connected. Free socket when disconnect complete. 713 */ 714 int 715 soclose(struct socket *so) 716 { 717 struct socket *so2; 718 int error = 0; 719 720 solock(so); 721 if (so->so_options & SO_ACCEPTCONN) { 722 for (;;) { 723 if ((so2 = TAILQ_FIRST(&so->so_q0)) != 0) { 724 KASSERT(solocked2(so, so2)); 725 (void) soqremque(so2, 0); 726 /* soabort drops the lock. */ 727 (void) soabort(so2); 728 solock(so); 729 continue; 730 } 731 if ((so2 = TAILQ_FIRST(&so->so_q)) != 0) { 732 KASSERT(solocked2(so, so2)); 733 (void) soqremque(so2, 1); 734 /* soabort drops the lock. */ 735 (void) soabort(so2); 736 solock(so); 737 continue; 738 } 739 break; 740 } 741 } 742 if (so->so_pcb == NULL) 743 goto discard; 744 if (so->so_state & SS_ISCONNECTED) { 745 if ((so->so_state & SS_ISDISCONNECTING) == 0) { 746 error = sodisconnect(so); 747 if (error) 748 goto drop; 749 } 750 if (so->so_options & SO_LINGER) { 751 if ((so->so_state & (SS_ISDISCONNECTING|SS_NBIO)) == 752 (SS_ISDISCONNECTING|SS_NBIO)) 753 goto drop; 754 while (so->so_state & SS_ISCONNECTED) { 755 error = sowait(so, true, so->so_linger * hz); 756 if (error) 757 break; 758 } 759 } 760 } 761 drop: 762 if (so->so_pcb) { 763 KASSERT(solocked(so)); 764 (*so->so_proto->pr_usrreqs->pr_detach)(so); 765 } 766 discard: 767 KASSERT((so->so_state & SS_NOFDREF) == 0); 768 kauth_cred_free(so->so_cred); 769 so->so_state |= SS_NOFDREF; 770 sofree(so); 771 return error; 772 } 773 774 /* 775 * Must be called with the socket locked.. Will return with it unlocked. 776 */ 777 int 778 soabort(struct socket *so) 779 { 780 u_int refs; 781 int error; 782 783 KASSERT(solocked(so)); 784 KASSERT(so->so_head == NULL); 785 786 so->so_aborting++; /* XXX */ 787 error = (*so->so_proto->pr_usrreqs->pr_abort)(so); 788 refs = --so->so_aborting; /* XXX */ 789 if (error || (refs == 0)) { 790 sofree(so); 791 } else { 792 sounlock(so); 793 } 794 return error; 795 } 796 797 int 798 soaccept(struct socket *so, struct sockaddr *nam) 799 { 800 int error; 801 802 KASSERT(solocked(so)); 803 KASSERT((so->so_state & SS_NOFDREF) != 0); 804 805 so->so_state &= ~SS_NOFDREF; 806 if ((so->so_state & SS_ISDISCONNECTED) == 0 || 807 (so->so_proto->pr_flags & PR_ABRTACPTDIS) == 0) 808 error = (*so->so_proto->pr_usrreqs->pr_accept)(so, nam); 809 else 810 error = ECONNABORTED; 811 812 return error; 813 } 814 815 int 816 soconnect(struct socket *so, struct sockaddr *nam, struct lwp *l) 817 { 818 int error; 819 820 KASSERT(solocked(so)); 821 822 if (so->so_options & SO_ACCEPTCONN) 823 return EOPNOTSUPP; 824 /* 825 * If protocol is connection-based, can only connect once. 826 * Otherwise, if connected, try to disconnect first. 827 * This allows user to disconnect by connecting to, e.g., 828 * a null address. 829 */ 830 if (so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING) && 831 ((so->so_proto->pr_flags & PR_CONNREQUIRED) || 832 (error = sodisconnect(so)))) { 833 error = EISCONN; 834 } else { 835 if (nam->sa_family != so->so_proto->pr_domain->dom_family) { 836 return EAFNOSUPPORT; 837 } 838 error = (*so->so_proto->pr_usrreqs->pr_connect)(so, nam, l); 839 } 840 841 return error; 842 } 843 844 int 845 soconnect2(struct socket *so1, struct socket *so2) 846 { 847 KASSERT(solocked2(so1, so2)); 848 849 return (*so1->so_proto->pr_usrreqs->pr_connect2)(so1, so2); 850 } 851 852 int 853 sodisconnect(struct socket *so) 854 { 855 int error; 856 857 KASSERT(solocked(so)); 858 859 if ((so->so_state & SS_ISCONNECTED) == 0) { 860 error = ENOTCONN; 861 } else if (so->so_state & SS_ISDISCONNECTING) { 862 error = EALREADY; 863 } else { 864 error = (*so->so_proto->pr_usrreqs->pr_disconnect)(so); 865 } 866 return (error); 867 } 868 869 #define SBLOCKWAIT(f) (((f) & MSG_DONTWAIT) ? M_NOWAIT : M_WAITOK) 870 /* 871 * Send on a socket. 872 * If send must go all at once and message is larger than 873 * send buffering, then hard error. 874 * Lock against other senders. 875 * If must go all at once and not enough room now, then 876 * inform user that this would block and do nothing. 877 * Otherwise, if nonblocking, send as much as possible. 878 * The data to be sent is described by "uio" if nonzero, 879 * otherwise by the mbuf chain "top" (which must be null 880 * if uio is not). Data provided in mbuf chain must be small 881 * enough to send all at once. 882 * 883 * Returns nonzero on error, timeout or signal; callers 884 * must check for short counts if EINTR/ERESTART are returned. 885 * Data and control buffers are freed on return. 886 */ 887 int 888 sosend(struct socket *so, struct sockaddr *addr, struct uio *uio, 889 struct mbuf *top, struct mbuf *control, int flags, struct lwp *l) 890 { 891 struct mbuf **mp, *m; 892 long space, len, resid, clen, mlen; 893 int error, s, dontroute, atomic; 894 short wakeup_state = 0; 895 896 clen = 0; 897 898 /* 899 * solock() provides atomicity of access. splsoftnet() prevents 900 * protocol processing soft interrupts from interrupting us and 901 * blocking (expensive). 902 */ 903 s = splsoftnet(); 904 solock(so); 905 atomic = sosendallatonce(so) || top; 906 if (uio) 907 resid = uio->uio_resid; 908 else 909 resid = top->m_pkthdr.len; 910 /* 911 * In theory resid should be unsigned. 912 * However, space must be signed, as it might be less than 0 913 * if we over-committed, and we must use a signed comparison 914 * of space and resid. On the other hand, a negative resid 915 * causes us to loop sending 0-length segments to the protocol. 916 */ 917 if (resid < 0) { 918 error = EINVAL; 919 goto out; 920 } 921 dontroute = 922 (flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 && 923 (so->so_proto->pr_flags & PR_ATOMIC); 924 l->l_ru.ru_msgsnd++; 925 if (control) 926 clen = control->m_len; 927 restart: 928 if ((error = sblock(&so->so_snd, SBLOCKWAIT(flags))) != 0) 929 goto out; 930 do { 931 if (so->so_state & SS_CANTSENDMORE) { 932 error = EPIPE; 933 goto release; 934 } 935 if (so->so_error) { 936 error = so->so_error; 937 so->so_error = 0; 938 goto release; 939 } 940 if ((so->so_state & SS_ISCONNECTED) == 0) { 941 if (so->so_proto->pr_flags & PR_CONNREQUIRED) { 942 if (resid || clen == 0) { 943 error = ENOTCONN; 944 goto release; 945 } 946 } else if (addr == NULL) { 947 error = EDESTADDRREQ; 948 goto release; 949 } 950 } 951 space = sbspace(&so->so_snd); 952 if (flags & MSG_OOB) 953 space += 1024; 954 if ((atomic && resid > so->so_snd.sb_hiwat) || 955 clen > so->so_snd.sb_hiwat) { 956 error = EMSGSIZE; 957 goto release; 958 } 959 if (space < resid + clen && 960 (atomic || space < so->so_snd.sb_lowat || space < clen)) { 961 if ((so->so_state & SS_NBIO) || (flags & MSG_NBIO)) { 962 error = EWOULDBLOCK; 963 goto release; 964 } 965 sbunlock(&so->so_snd); 966 if (wakeup_state & SS_RESTARTSYS) { 967 error = ERESTART; 968 goto out; 969 } 970 error = sbwait(&so->so_snd); 971 if (error) 972 goto out; 973 wakeup_state = so->so_state; 974 goto restart; 975 } 976 wakeup_state = 0; 977 mp = ⊤ 978 space -= clen; 979 do { 980 if (uio == NULL) { 981 /* 982 * Data is prepackaged in "top". 983 */ 984 resid = 0; 985 if (flags & MSG_EOR) 986 top->m_flags |= M_EOR; 987 } else do { 988 sounlock(so); 989 splx(s); 990 if (top == NULL) { 991 m = m_gethdr(M_WAIT, MT_DATA); 992 mlen = MHLEN; 993 m->m_pkthdr.len = 0; 994 m->m_pkthdr.rcvif = NULL; 995 } else { 996 m = m_get(M_WAIT, MT_DATA); 997 mlen = MLEN; 998 } 999 MCLAIM(m, so->so_snd.sb_mowner); 1000 if (sock_loan_thresh >= 0 && 1001 uio->uio_iov->iov_len >= sock_loan_thresh && 1002 space >= sock_loan_thresh && 1003 (len = sosend_loan(so, uio, m, 1004 space)) != 0) { 1005 SOSEND_COUNTER_INCR(&sosend_loan_big); 1006 space -= len; 1007 goto have_data; 1008 } 1009 if (resid >= MINCLSIZE && space >= MCLBYTES) { 1010 SOSEND_COUNTER_INCR(&sosend_copy_big); 1011 m_clget(m, M_DONTWAIT); 1012 if ((m->m_flags & M_EXT) == 0) 1013 goto nopages; 1014 mlen = MCLBYTES; 1015 if (atomic && top == 0) { 1016 len = lmin(MCLBYTES - max_hdr, 1017 resid); 1018 m->m_data += max_hdr; 1019 } else 1020 len = lmin(MCLBYTES, resid); 1021 space -= len; 1022 } else { 1023 nopages: 1024 SOSEND_COUNTER_INCR(&sosend_copy_small); 1025 len = lmin(lmin(mlen, resid), space); 1026 space -= len; 1027 /* 1028 * For datagram protocols, leave room 1029 * for protocol headers in first mbuf. 1030 */ 1031 if (atomic && top == 0 && len < mlen) 1032 MH_ALIGN(m, len); 1033 } 1034 error = uiomove(mtod(m, void *), (int)len, uio); 1035 have_data: 1036 resid = uio->uio_resid; 1037 m->m_len = len; 1038 *mp = m; 1039 top->m_pkthdr.len += len; 1040 s = splsoftnet(); 1041 solock(so); 1042 if (error != 0) 1043 goto release; 1044 mp = &m->m_next; 1045 if (resid <= 0) { 1046 if (flags & MSG_EOR) 1047 top->m_flags |= M_EOR; 1048 break; 1049 } 1050 } while (space > 0 && atomic); 1051 1052 if (so->so_state & SS_CANTSENDMORE) { 1053 error = EPIPE; 1054 goto release; 1055 } 1056 if (dontroute) 1057 so->so_options |= SO_DONTROUTE; 1058 if (resid > 0) 1059 so->so_state |= SS_MORETOCOME; 1060 if (flags & MSG_OOB) { 1061 error = (*so->so_proto->pr_usrreqs->pr_sendoob)(so, 1062 top, control); 1063 } else { 1064 error = (*so->so_proto->pr_usrreqs->pr_send)(so, 1065 top, addr, control, l); 1066 } 1067 if (dontroute) 1068 so->so_options &= ~SO_DONTROUTE; 1069 if (resid > 0) 1070 so->so_state &= ~SS_MORETOCOME; 1071 clen = 0; 1072 control = NULL; 1073 top = NULL; 1074 mp = ⊤ 1075 if (error != 0) 1076 goto release; 1077 } while (resid && space > 0); 1078 } while (resid); 1079 1080 release: 1081 sbunlock(&so->so_snd); 1082 out: 1083 sounlock(so); 1084 splx(s); 1085 if (top) 1086 m_freem(top); 1087 if (control) 1088 m_freem(control); 1089 return (error); 1090 } 1091 1092 /* 1093 * Following replacement or removal of the first mbuf on the first 1094 * mbuf chain of a socket buffer, push necessary state changes back 1095 * into the socket buffer so that other consumers see the values 1096 * consistently. 'nextrecord' is the callers locally stored value of 1097 * the original value of sb->sb_mb->m_nextpkt which must be restored 1098 * when the lead mbuf changes. NOTE: 'nextrecord' may be NULL. 1099 */ 1100 static void 1101 sbsync(struct sockbuf *sb, struct mbuf *nextrecord) 1102 { 1103 1104 KASSERT(solocked(sb->sb_so)); 1105 1106 /* 1107 * First, update for the new value of nextrecord. If necessary, 1108 * make it the first record. 1109 */ 1110 if (sb->sb_mb != NULL) 1111 sb->sb_mb->m_nextpkt = nextrecord; 1112 else 1113 sb->sb_mb = nextrecord; 1114 1115 /* 1116 * Now update any dependent socket buffer fields to reflect 1117 * the new state. This is an inline of SB_EMPTY_FIXUP, with 1118 * the addition of a second clause that takes care of the 1119 * case where sb_mb has been updated, but remains the last 1120 * record. 1121 */ 1122 if (sb->sb_mb == NULL) { 1123 sb->sb_mbtail = NULL; 1124 sb->sb_lastrecord = NULL; 1125 } else if (sb->sb_mb->m_nextpkt == NULL) 1126 sb->sb_lastrecord = sb->sb_mb; 1127 } 1128 1129 /* 1130 * Implement receive operations on a socket. 1131 * We depend on the way that records are added to the sockbuf 1132 * by sbappend*. In particular, each record (mbufs linked through m_next) 1133 * must begin with an address if the protocol so specifies, 1134 * followed by an optional mbuf or mbufs containing ancillary data, 1135 * and then zero or more mbufs of data. 1136 * In order to avoid blocking network interrupts for the entire time here, 1137 * we splx() while doing the actual copy to user space. 1138 * Although the sockbuf is locked, new data may still be appended, 1139 * and thus we must maintain consistency of the sockbuf during that time. 1140 * 1141 * The caller may receive the data as a single mbuf chain by supplying 1142 * an mbuf **mp0 for use in returning the chain. The uio is then used 1143 * only for the count in uio_resid. 1144 */ 1145 int 1146 soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio, 1147 struct mbuf **mp0, struct mbuf **controlp, int *flagsp) 1148 { 1149 struct lwp *l = curlwp; 1150 struct mbuf *m, **mp, *mt; 1151 size_t len, offset, moff, orig_resid; 1152 int atomic, flags, error, s, type; 1153 const struct protosw *pr; 1154 struct mbuf *nextrecord; 1155 int mbuf_removed = 0; 1156 const struct domain *dom; 1157 short wakeup_state = 0; 1158 1159 pr = so->so_proto; 1160 atomic = pr->pr_flags & PR_ATOMIC; 1161 dom = pr->pr_domain; 1162 mp = mp0; 1163 type = 0; 1164 orig_resid = uio->uio_resid; 1165 1166 if (paddr != NULL) 1167 *paddr = NULL; 1168 if (controlp != NULL) 1169 *controlp = NULL; 1170 if (flagsp != NULL) 1171 flags = *flagsp &~ MSG_EOR; 1172 else 1173 flags = 0; 1174 1175 if (flags & MSG_OOB) { 1176 m = m_get(M_WAIT, MT_DATA); 1177 solock(so); 1178 error = (*pr->pr_usrreqs->pr_recvoob)(so, m, flags & MSG_PEEK); 1179 sounlock(so); 1180 if (error) 1181 goto bad; 1182 do { 1183 error = uiomove(mtod(m, void *), 1184 MIN(uio->uio_resid, m->m_len), uio); 1185 m = m_free(m); 1186 } while (uio->uio_resid > 0 && error == 0 && m); 1187 bad: 1188 if (m != NULL) 1189 m_freem(m); 1190 return error; 1191 } 1192 if (mp != NULL) 1193 *mp = NULL; 1194 1195 /* 1196 * solock() provides atomicity of access. splsoftnet() prevents 1197 * protocol processing soft interrupts from interrupting us and 1198 * blocking (expensive). 1199 */ 1200 s = splsoftnet(); 1201 solock(so); 1202 restart: 1203 if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) { 1204 sounlock(so); 1205 splx(s); 1206 return error; 1207 } 1208 1209 m = so->so_rcv.sb_mb; 1210 /* 1211 * If we have less data than requested, block awaiting more 1212 * (subject to any timeout) if: 1213 * 1. the current count is less than the low water mark, 1214 * 2. MSG_WAITALL is set, and it is possible to do the entire 1215 * receive operation at once if we block (resid <= hiwat), or 1216 * 3. MSG_DONTWAIT is not set. 1217 * If MSG_WAITALL is set but resid is larger than the receive buffer, 1218 * we have to do the receive in sections, and thus risk returning 1219 * a short count if a timeout or signal occurs after we start. 1220 */ 1221 if (m == NULL || 1222 ((flags & MSG_DONTWAIT) == 0 && 1223 so->so_rcv.sb_cc < uio->uio_resid && 1224 (so->so_rcv.sb_cc < so->so_rcv.sb_lowat || 1225 ((flags & MSG_WAITALL) && 1226 uio->uio_resid <= so->so_rcv.sb_hiwat)) && 1227 m->m_nextpkt == NULL && !atomic)) { 1228 #ifdef DIAGNOSTIC 1229 if (m == NULL && so->so_rcv.sb_cc) 1230 panic("receive 1"); 1231 #endif 1232 if (so->so_error) { 1233 if (m != NULL) 1234 goto dontblock; 1235 error = so->so_error; 1236 if ((flags & MSG_PEEK) == 0) 1237 so->so_error = 0; 1238 goto release; 1239 } 1240 if (so->so_state & SS_CANTRCVMORE) { 1241 if (m != NULL) 1242 goto dontblock; 1243 else 1244 goto release; 1245 } 1246 for (; m != NULL; m = m->m_next) 1247 if (m->m_type == MT_OOBDATA || (m->m_flags & M_EOR)) { 1248 m = so->so_rcv.sb_mb; 1249 goto dontblock; 1250 } 1251 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) == 0 && 1252 (so->so_proto->pr_flags & PR_CONNREQUIRED)) { 1253 error = ENOTCONN; 1254 goto release; 1255 } 1256 if (uio->uio_resid == 0) 1257 goto release; 1258 if ((so->so_state & SS_NBIO) || 1259 (flags & (MSG_DONTWAIT|MSG_NBIO))) { 1260 error = EWOULDBLOCK; 1261 goto release; 1262 } 1263 SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 1"); 1264 SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 1"); 1265 sbunlock(&so->so_rcv); 1266 if (wakeup_state & SS_RESTARTSYS) 1267 error = ERESTART; 1268 else 1269 error = sbwait(&so->so_rcv); 1270 if (error != 0) { 1271 sounlock(so); 1272 splx(s); 1273 return error; 1274 } 1275 wakeup_state = so->so_state; 1276 goto restart; 1277 } 1278 dontblock: 1279 /* 1280 * On entry here, m points to the first record of the socket buffer. 1281 * From this point onward, we maintain 'nextrecord' as a cache of the 1282 * pointer to the next record in the socket buffer. We must keep the 1283 * various socket buffer pointers and local stack versions of the 1284 * pointers in sync, pushing out modifications before dropping the 1285 * socket lock, and re-reading them when picking it up. 1286 * 1287 * Otherwise, we will race with the network stack appending new data 1288 * or records onto the socket buffer by using inconsistent/stale 1289 * versions of the field, possibly resulting in socket buffer 1290 * corruption. 1291 * 1292 * By holding the high-level sblock(), we prevent simultaneous 1293 * readers from pulling off the front of the socket buffer. 1294 */ 1295 if (l != NULL) 1296 l->l_ru.ru_msgrcv++; 1297 KASSERT(m == so->so_rcv.sb_mb); 1298 SBLASTRECORDCHK(&so->so_rcv, "soreceive 1"); 1299 SBLASTMBUFCHK(&so->so_rcv, "soreceive 1"); 1300 nextrecord = m->m_nextpkt; 1301 if (pr->pr_flags & PR_ADDR) { 1302 #ifdef DIAGNOSTIC 1303 if (m->m_type != MT_SONAME) 1304 panic("receive 1a"); 1305 #endif 1306 orig_resid = 0; 1307 if (flags & MSG_PEEK) { 1308 if (paddr) 1309 *paddr = m_copy(m, 0, m->m_len); 1310 m = m->m_next; 1311 } else { 1312 sbfree(&so->so_rcv, m); 1313 mbuf_removed = 1; 1314 if (paddr != NULL) { 1315 *paddr = m; 1316 so->so_rcv.sb_mb = m->m_next; 1317 m->m_next = NULL; 1318 m = so->so_rcv.sb_mb; 1319 } else { 1320 MFREE(m, so->so_rcv.sb_mb); 1321 m = so->so_rcv.sb_mb; 1322 } 1323 sbsync(&so->so_rcv, nextrecord); 1324 } 1325 } 1326 1327 /* 1328 * Process one or more MT_CONTROL mbufs present before any data mbufs 1329 * in the first mbuf chain on the socket buffer. If MSG_PEEK, we 1330 * just copy the data; if !MSG_PEEK, we call into the protocol to 1331 * perform externalization (or freeing if controlp == NULL). 1332 */ 1333 if (__predict_false(m != NULL && m->m_type == MT_CONTROL)) { 1334 struct mbuf *cm = NULL, *cmn; 1335 struct mbuf **cme = &cm; 1336 1337 do { 1338 if (flags & MSG_PEEK) { 1339 if (controlp != NULL) { 1340 *controlp = m_copy(m, 0, m->m_len); 1341 controlp = &(*controlp)->m_next; 1342 } 1343 m = m->m_next; 1344 } else { 1345 sbfree(&so->so_rcv, m); 1346 so->so_rcv.sb_mb = m->m_next; 1347 m->m_next = NULL; 1348 *cme = m; 1349 cme = &(*cme)->m_next; 1350 m = so->so_rcv.sb_mb; 1351 } 1352 } while (m != NULL && m->m_type == MT_CONTROL); 1353 if ((flags & MSG_PEEK) == 0) 1354 sbsync(&so->so_rcv, nextrecord); 1355 for (; cm != NULL; cm = cmn) { 1356 cmn = cm->m_next; 1357 cm->m_next = NULL; 1358 type = mtod(cm, struct cmsghdr *)->cmsg_type; 1359 if (controlp != NULL) { 1360 if (dom->dom_externalize != NULL && 1361 type == SCM_RIGHTS) { 1362 sounlock(so); 1363 splx(s); 1364 error = (*dom->dom_externalize)(cm, l, 1365 (flags & MSG_CMSG_CLOEXEC) ? 1366 O_CLOEXEC : 0); 1367 s = splsoftnet(); 1368 solock(so); 1369 } 1370 *controlp = cm; 1371 while (*controlp != NULL) 1372 controlp = &(*controlp)->m_next; 1373 } else { 1374 /* 1375 * Dispose of any SCM_RIGHTS message that went 1376 * through the read path rather than recv. 1377 */ 1378 if (dom->dom_dispose != NULL && 1379 type == SCM_RIGHTS) { 1380 sounlock(so); 1381 (*dom->dom_dispose)(cm); 1382 solock(so); 1383 } 1384 m_freem(cm); 1385 } 1386 } 1387 if (m != NULL) 1388 nextrecord = so->so_rcv.sb_mb->m_nextpkt; 1389 else 1390 nextrecord = so->so_rcv.sb_mb; 1391 orig_resid = 0; 1392 } 1393 1394 /* If m is non-NULL, we have some data to read. */ 1395 if (__predict_true(m != NULL)) { 1396 type = m->m_type; 1397 if (type == MT_OOBDATA) 1398 flags |= MSG_OOB; 1399 } 1400 SBLASTRECORDCHK(&so->so_rcv, "soreceive 2"); 1401 SBLASTMBUFCHK(&so->so_rcv, "soreceive 2"); 1402 1403 moff = 0; 1404 offset = 0; 1405 while (m != NULL && uio->uio_resid > 0 && error == 0) { 1406 if (m->m_type == MT_OOBDATA) { 1407 if (type != MT_OOBDATA) 1408 break; 1409 } else if (type == MT_OOBDATA) 1410 break; 1411 #ifdef DIAGNOSTIC 1412 else if (m->m_type != MT_DATA && m->m_type != MT_HEADER) 1413 panic("receive 3"); 1414 #endif 1415 so->so_state &= ~SS_RCVATMARK; 1416 wakeup_state = 0; 1417 len = uio->uio_resid; 1418 if (so->so_oobmark && len > so->so_oobmark - offset) 1419 len = so->so_oobmark - offset; 1420 if (len > m->m_len - moff) 1421 len = m->m_len - moff; 1422 /* 1423 * If mp is set, just pass back the mbufs. 1424 * Otherwise copy them out via the uio, then free. 1425 * Sockbuf must be consistent here (points to current mbuf, 1426 * it points to next record) when we drop priority; 1427 * we must note any additions to the sockbuf when we 1428 * block interrupts again. 1429 */ 1430 if (mp == NULL) { 1431 SBLASTRECORDCHK(&so->so_rcv, "soreceive uiomove"); 1432 SBLASTMBUFCHK(&so->so_rcv, "soreceive uiomove"); 1433 sounlock(so); 1434 splx(s); 1435 error = uiomove(mtod(m, char *) + moff, len, uio); 1436 s = splsoftnet(); 1437 solock(so); 1438 if (error != 0) { 1439 /* 1440 * If any part of the record has been removed 1441 * (such as the MT_SONAME mbuf, which will 1442 * happen when PR_ADDR, and thus also 1443 * PR_ATOMIC, is set), then drop the entire 1444 * record to maintain the atomicity of the 1445 * receive operation. 1446 * 1447 * This avoids a later panic("receive 1a") 1448 * when compiled with DIAGNOSTIC. 1449 */ 1450 if (m && mbuf_removed && atomic) 1451 (void) sbdroprecord(&so->so_rcv); 1452 1453 goto release; 1454 } 1455 } else 1456 uio->uio_resid -= len; 1457 if (len == m->m_len - moff) { 1458 if (m->m_flags & M_EOR) 1459 flags |= MSG_EOR; 1460 if (flags & MSG_PEEK) { 1461 m = m->m_next; 1462 moff = 0; 1463 } else { 1464 nextrecord = m->m_nextpkt; 1465 sbfree(&so->so_rcv, m); 1466 if (mp) { 1467 *mp = m; 1468 mp = &m->m_next; 1469 so->so_rcv.sb_mb = m = m->m_next; 1470 *mp = NULL; 1471 } else { 1472 MFREE(m, so->so_rcv.sb_mb); 1473 m = so->so_rcv.sb_mb; 1474 } 1475 /* 1476 * If m != NULL, we also know that 1477 * so->so_rcv.sb_mb != NULL. 1478 */ 1479 KASSERT(so->so_rcv.sb_mb == m); 1480 if (m) { 1481 m->m_nextpkt = nextrecord; 1482 if (nextrecord == NULL) 1483 so->so_rcv.sb_lastrecord = m; 1484 } else { 1485 so->so_rcv.sb_mb = nextrecord; 1486 SB_EMPTY_FIXUP(&so->so_rcv); 1487 } 1488 SBLASTRECORDCHK(&so->so_rcv, "soreceive 3"); 1489 SBLASTMBUFCHK(&so->so_rcv, "soreceive 3"); 1490 } 1491 } else if (flags & MSG_PEEK) 1492 moff += len; 1493 else { 1494 if (mp != NULL) { 1495 mt = m_copym(m, 0, len, M_NOWAIT); 1496 if (__predict_false(mt == NULL)) { 1497 sounlock(so); 1498 mt = m_copym(m, 0, len, M_WAIT); 1499 solock(so); 1500 } 1501 *mp = mt; 1502 } 1503 m->m_data += len; 1504 m->m_len -= len; 1505 so->so_rcv.sb_cc -= len; 1506 } 1507 if (so->so_oobmark) { 1508 if ((flags & MSG_PEEK) == 0) { 1509 so->so_oobmark -= len; 1510 if (so->so_oobmark == 0) { 1511 so->so_state |= SS_RCVATMARK; 1512 break; 1513 } 1514 } else { 1515 offset += len; 1516 if (offset == so->so_oobmark) 1517 break; 1518 } 1519 } 1520 if (flags & MSG_EOR) 1521 break; 1522 /* 1523 * If the MSG_WAITALL flag is set (for non-atomic socket), 1524 * we must not quit until "uio->uio_resid == 0" or an error 1525 * termination. If a signal/timeout occurs, return 1526 * with a short count but without error. 1527 * Keep sockbuf locked against other readers. 1528 */ 1529 while (flags & MSG_WAITALL && m == NULL && uio->uio_resid > 0 && 1530 !sosendallatonce(so) && !nextrecord) { 1531 if (so->so_error || so->so_state & SS_CANTRCVMORE) 1532 break; 1533 /* 1534 * If we are peeking and the socket receive buffer is 1535 * full, stop since we can't get more data to peek at. 1536 */ 1537 if ((flags & MSG_PEEK) && sbspace(&so->so_rcv) <= 0) 1538 break; 1539 /* 1540 * If we've drained the socket buffer, tell the 1541 * protocol in case it needs to do something to 1542 * get it filled again. 1543 */ 1544 if ((pr->pr_flags & PR_WANTRCVD) && so->so_pcb) 1545 (*pr->pr_usrreqs->pr_rcvd)(so, flags, l); 1546 SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2"); 1547 SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2"); 1548 if (wakeup_state & SS_RESTARTSYS) 1549 error = ERESTART; 1550 else 1551 error = sbwait(&so->so_rcv); 1552 if (error != 0) { 1553 sbunlock(&so->so_rcv); 1554 sounlock(so); 1555 splx(s); 1556 return 0; 1557 } 1558 if ((m = so->so_rcv.sb_mb) != NULL) 1559 nextrecord = m->m_nextpkt; 1560 wakeup_state = so->so_state; 1561 } 1562 } 1563 1564 if (m && atomic) { 1565 flags |= MSG_TRUNC; 1566 if ((flags & MSG_PEEK) == 0) 1567 (void) sbdroprecord(&so->so_rcv); 1568 } 1569 if ((flags & MSG_PEEK) == 0) { 1570 if (m == NULL) { 1571 /* 1572 * First part is an inline SB_EMPTY_FIXUP(). Second 1573 * part makes sure sb_lastrecord is up-to-date if 1574 * there is still data in the socket buffer. 1575 */ 1576 so->so_rcv.sb_mb = nextrecord; 1577 if (so->so_rcv.sb_mb == NULL) { 1578 so->so_rcv.sb_mbtail = NULL; 1579 so->so_rcv.sb_lastrecord = NULL; 1580 } else if (nextrecord->m_nextpkt == NULL) 1581 so->so_rcv.sb_lastrecord = nextrecord; 1582 } 1583 SBLASTRECORDCHK(&so->so_rcv, "soreceive 4"); 1584 SBLASTMBUFCHK(&so->so_rcv, "soreceive 4"); 1585 if (pr->pr_flags & PR_WANTRCVD && so->so_pcb) 1586 (*pr->pr_usrreqs->pr_rcvd)(so, flags, l); 1587 } 1588 if (orig_resid == uio->uio_resid && orig_resid && 1589 (flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) { 1590 sbunlock(&so->so_rcv); 1591 goto restart; 1592 } 1593 1594 if (flagsp != NULL) 1595 *flagsp |= flags; 1596 release: 1597 sbunlock(&so->so_rcv); 1598 sounlock(so); 1599 splx(s); 1600 return error; 1601 } 1602 1603 int 1604 soshutdown(struct socket *so, int how) 1605 { 1606 const struct protosw *pr; 1607 int error; 1608 1609 KASSERT(solocked(so)); 1610 1611 pr = so->so_proto; 1612 if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR)) 1613 return (EINVAL); 1614 1615 if (how == SHUT_RD || how == SHUT_RDWR) { 1616 sorflush(so); 1617 error = 0; 1618 } 1619 if (how == SHUT_WR || how == SHUT_RDWR) 1620 error = (*pr->pr_usrreqs->pr_shutdown)(so); 1621 1622 return error; 1623 } 1624 1625 void 1626 sorestart(struct socket *so) 1627 { 1628 /* 1629 * An application has called close() on an fd on which another 1630 * of its threads has called a socket system call. 1631 * Mark this and wake everyone up, and code that would block again 1632 * instead returns ERESTART. 1633 * On system call re-entry the fd is validated and EBADF returned. 1634 * Any other fd will block again on the 2nd syscall. 1635 */ 1636 solock(so); 1637 so->so_state |= SS_RESTARTSYS; 1638 cv_broadcast(&so->so_cv); 1639 cv_broadcast(&so->so_snd.sb_cv); 1640 cv_broadcast(&so->so_rcv.sb_cv); 1641 sounlock(so); 1642 } 1643 1644 void 1645 sorflush(struct socket *so) 1646 { 1647 struct sockbuf *sb, asb; 1648 const struct protosw *pr; 1649 1650 KASSERT(solocked(so)); 1651 1652 sb = &so->so_rcv; 1653 pr = so->so_proto; 1654 socantrcvmore(so); 1655 sb->sb_flags |= SB_NOINTR; 1656 (void )sblock(sb, M_WAITOK); 1657 sbunlock(sb); 1658 asb = *sb; 1659 /* 1660 * Clear most of the sockbuf structure, but leave some of the 1661 * fields valid. 1662 */ 1663 memset(&sb->sb_startzero, 0, 1664 sizeof(*sb) - offsetof(struct sockbuf, sb_startzero)); 1665 if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) { 1666 sounlock(so); 1667 (*pr->pr_domain->dom_dispose)(asb.sb_mb); 1668 solock(so); 1669 } 1670 sbrelease(&asb, so); 1671 } 1672 1673 /* 1674 * internal set SOL_SOCKET options 1675 */ 1676 static int 1677 sosetopt1(struct socket *so, const struct sockopt *sopt) 1678 { 1679 int error = EINVAL, opt; 1680 int optval = 0; /* XXX: gcc */ 1681 struct linger l; 1682 struct timeval tv; 1683 1684 switch ((opt = sopt->sopt_name)) { 1685 1686 case SO_ACCEPTFILTER: 1687 error = accept_filt_setopt(so, sopt); 1688 KASSERT(solocked(so)); 1689 break; 1690 1691 case SO_LINGER: 1692 error = sockopt_get(sopt, &l, sizeof(l)); 1693 solock(so); 1694 if (error) 1695 break; 1696 if (l.l_linger < 0 || l.l_linger > USHRT_MAX || 1697 l.l_linger > (INT_MAX / hz)) { 1698 error = EDOM; 1699 break; 1700 } 1701 so->so_linger = l.l_linger; 1702 if (l.l_onoff) 1703 so->so_options |= SO_LINGER; 1704 else 1705 so->so_options &= ~SO_LINGER; 1706 break; 1707 1708 case SO_DEBUG: 1709 case SO_KEEPALIVE: 1710 case SO_DONTROUTE: 1711 case SO_USELOOPBACK: 1712 case SO_BROADCAST: 1713 case SO_REUSEADDR: 1714 case SO_REUSEPORT: 1715 case SO_OOBINLINE: 1716 case SO_TIMESTAMP: 1717 case SO_NOSIGPIPE: 1718 #ifdef SO_OTIMESTAMP 1719 case SO_OTIMESTAMP: 1720 #endif 1721 error = sockopt_getint(sopt, &optval); 1722 solock(so); 1723 if (error) 1724 break; 1725 if (optval) 1726 so->so_options |= opt; 1727 else 1728 so->so_options &= ~opt; 1729 break; 1730 1731 case SO_SNDBUF: 1732 case SO_RCVBUF: 1733 case SO_SNDLOWAT: 1734 case SO_RCVLOWAT: 1735 error = sockopt_getint(sopt, &optval); 1736 solock(so); 1737 if (error) 1738 break; 1739 1740 /* 1741 * Values < 1 make no sense for any of these 1742 * options, so disallow them. 1743 */ 1744 if (optval < 1) { 1745 error = EINVAL; 1746 break; 1747 } 1748 1749 switch (opt) { 1750 case SO_SNDBUF: 1751 if (sbreserve(&so->so_snd, (u_long)optval, so) == 0) { 1752 error = ENOBUFS; 1753 break; 1754 } 1755 so->so_snd.sb_flags &= ~SB_AUTOSIZE; 1756 break; 1757 1758 case SO_RCVBUF: 1759 if (sbreserve(&so->so_rcv, (u_long)optval, so) == 0) { 1760 error = ENOBUFS; 1761 break; 1762 } 1763 so->so_rcv.sb_flags &= ~SB_AUTOSIZE; 1764 break; 1765 1766 /* 1767 * Make sure the low-water is never greater than 1768 * the high-water. 1769 */ 1770 case SO_SNDLOWAT: 1771 if (optval > so->so_snd.sb_hiwat) 1772 optval = so->so_snd.sb_hiwat; 1773 1774 so->so_snd.sb_lowat = optval; 1775 break; 1776 1777 case SO_RCVLOWAT: 1778 if (optval > so->so_rcv.sb_hiwat) 1779 optval = so->so_rcv.sb_hiwat; 1780 1781 so->so_rcv.sb_lowat = optval; 1782 break; 1783 } 1784 break; 1785 1786 #ifdef COMPAT_50 1787 case SO_OSNDTIMEO: 1788 case SO_ORCVTIMEO: { 1789 struct timeval50 otv; 1790 error = sockopt_get(sopt, &otv, sizeof(otv)); 1791 if (error) { 1792 solock(so); 1793 break; 1794 } 1795 timeval50_to_timeval(&otv, &tv); 1796 opt = opt == SO_OSNDTIMEO ? SO_SNDTIMEO : SO_RCVTIMEO; 1797 error = 0; 1798 /*FALLTHROUGH*/ 1799 } 1800 #endif /* COMPAT_50 */ 1801 1802 case SO_SNDTIMEO: 1803 case SO_RCVTIMEO: 1804 if (error) 1805 error = sockopt_get(sopt, &tv, sizeof(tv)); 1806 solock(so); 1807 if (error) 1808 break; 1809 1810 if (tv.tv_sec > (INT_MAX - tv.tv_usec / tick) / hz) { 1811 error = EDOM; 1812 break; 1813 } 1814 1815 optval = tv.tv_sec * hz + tv.tv_usec / tick; 1816 if (optval == 0 && tv.tv_usec != 0) 1817 optval = 1; 1818 1819 switch (opt) { 1820 case SO_SNDTIMEO: 1821 so->so_snd.sb_timeo = optval; 1822 break; 1823 case SO_RCVTIMEO: 1824 so->so_rcv.sb_timeo = optval; 1825 break; 1826 } 1827 break; 1828 1829 default: 1830 solock(so); 1831 error = ENOPROTOOPT; 1832 break; 1833 } 1834 KASSERT(solocked(so)); 1835 return error; 1836 } 1837 1838 int 1839 sosetopt(struct socket *so, struct sockopt *sopt) 1840 { 1841 int error, prerr; 1842 1843 if (sopt->sopt_level == SOL_SOCKET) { 1844 error = sosetopt1(so, sopt); 1845 KASSERT(solocked(so)); 1846 } else { 1847 error = ENOPROTOOPT; 1848 solock(so); 1849 } 1850 1851 if ((error == 0 || error == ENOPROTOOPT) && 1852 so->so_proto != NULL && so->so_proto->pr_ctloutput != NULL) { 1853 /* give the protocol stack a shot */ 1854 prerr = (*so->so_proto->pr_ctloutput)(PRCO_SETOPT, so, sopt); 1855 if (prerr == 0) 1856 error = 0; 1857 else if (prerr != ENOPROTOOPT) 1858 error = prerr; 1859 } 1860 sounlock(so); 1861 return error; 1862 } 1863 1864 /* 1865 * so_setsockopt() is a wrapper providing a sockopt structure for sosetopt() 1866 */ 1867 int 1868 so_setsockopt(struct lwp *l, struct socket *so, int level, int name, 1869 const void *val, size_t valsize) 1870 { 1871 struct sockopt sopt; 1872 int error; 1873 1874 KASSERT(valsize == 0 || val != NULL); 1875 1876 sockopt_init(&sopt, level, name, valsize); 1877 sockopt_set(&sopt, val, valsize); 1878 1879 error = sosetopt(so, &sopt); 1880 1881 sockopt_destroy(&sopt); 1882 1883 return error; 1884 } 1885 1886 /* 1887 * internal get SOL_SOCKET options 1888 */ 1889 static int 1890 sogetopt1(struct socket *so, struct sockopt *sopt) 1891 { 1892 int error, optval, opt; 1893 struct linger l; 1894 struct timeval tv; 1895 1896 switch ((opt = sopt->sopt_name)) { 1897 1898 case SO_ACCEPTFILTER: 1899 error = accept_filt_getopt(so, sopt); 1900 break; 1901 1902 case SO_LINGER: 1903 l.l_onoff = (so->so_options & SO_LINGER) ? 1 : 0; 1904 l.l_linger = so->so_linger; 1905 1906 error = sockopt_set(sopt, &l, sizeof(l)); 1907 break; 1908 1909 case SO_USELOOPBACK: 1910 case SO_DONTROUTE: 1911 case SO_DEBUG: 1912 case SO_KEEPALIVE: 1913 case SO_REUSEADDR: 1914 case SO_REUSEPORT: 1915 case SO_BROADCAST: 1916 case SO_OOBINLINE: 1917 case SO_TIMESTAMP: 1918 case SO_NOSIGPIPE: 1919 #ifdef SO_OTIMESTAMP 1920 case SO_OTIMESTAMP: 1921 #endif 1922 case SO_ACCEPTCONN: 1923 error = sockopt_setint(sopt, (so->so_options & opt) ? 1 : 0); 1924 break; 1925 1926 case SO_TYPE: 1927 error = sockopt_setint(sopt, so->so_type); 1928 break; 1929 1930 case SO_ERROR: 1931 error = sockopt_setint(sopt, so->so_error); 1932 so->so_error = 0; 1933 break; 1934 1935 case SO_SNDBUF: 1936 error = sockopt_setint(sopt, so->so_snd.sb_hiwat); 1937 break; 1938 1939 case SO_RCVBUF: 1940 error = sockopt_setint(sopt, so->so_rcv.sb_hiwat); 1941 break; 1942 1943 case SO_SNDLOWAT: 1944 error = sockopt_setint(sopt, so->so_snd.sb_lowat); 1945 break; 1946 1947 case SO_RCVLOWAT: 1948 error = sockopt_setint(sopt, so->so_rcv.sb_lowat); 1949 break; 1950 1951 #ifdef COMPAT_50 1952 case SO_OSNDTIMEO: 1953 case SO_ORCVTIMEO: { 1954 struct timeval50 otv; 1955 1956 optval = (opt == SO_OSNDTIMEO ? 1957 so->so_snd.sb_timeo : so->so_rcv.sb_timeo); 1958 1959 otv.tv_sec = optval / hz; 1960 otv.tv_usec = (optval % hz) * tick; 1961 1962 error = sockopt_set(sopt, &otv, sizeof(otv)); 1963 break; 1964 } 1965 #endif /* COMPAT_50 */ 1966 1967 case SO_SNDTIMEO: 1968 case SO_RCVTIMEO: 1969 optval = (opt == SO_SNDTIMEO ? 1970 so->so_snd.sb_timeo : so->so_rcv.sb_timeo); 1971 1972 tv.tv_sec = optval / hz; 1973 tv.tv_usec = (optval % hz) * tick; 1974 1975 error = sockopt_set(sopt, &tv, sizeof(tv)); 1976 break; 1977 1978 case SO_OVERFLOWED: 1979 error = sockopt_setint(sopt, so->so_rcv.sb_overflowed); 1980 break; 1981 1982 default: 1983 error = ENOPROTOOPT; 1984 break; 1985 } 1986 1987 return (error); 1988 } 1989 1990 int 1991 sogetopt(struct socket *so, struct sockopt *sopt) 1992 { 1993 int error; 1994 1995 solock(so); 1996 if (sopt->sopt_level != SOL_SOCKET) { 1997 if (so->so_proto && so->so_proto->pr_ctloutput) { 1998 error = ((*so->so_proto->pr_ctloutput) 1999 (PRCO_GETOPT, so, sopt)); 2000 } else 2001 error = (ENOPROTOOPT); 2002 } else { 2003 error = sogetopt1(so, sopt); 2004 } 2005 sounlock(so); 2006 return (error); 2007 } 2008 2009 /* 2010 * alloc sockopt data buffer buffer 2011 * - will be released at destroy 2012 */ 2013 static int 2014 sockopt_alloc(struct sockopt *sopt, size_t len, km_flag_t kmflag) 2015 { 2016 2017 KASSERT(sopt->sopt_size == 0); 2018 2019 if (len > sizeof(sopt->sopt_buf)) { 2020 sopt->sopt_data = kmem_zalloc(len, kmflag); 2021 if (sopt->sopt_data == NULL) 2022 return ENOMEM; 2023 } else 2024 sopt->sopt_data = sopt->sopt_buf; 2025 2026 sopt->sopt_size = len; 2027 return 0; 2028 } 2029 2030 /* 2031 * initialise sockopt storage 2032 * - MAY sleep during allocation 2033 */ 2034 void 2035 sockopt_init(struct sockopt *sopt, int level, int name, size_t size) 2036 { 2037 2038 memset(sopt, 0, sizeof(*sopt)); 2039 2040 sopt->sopt_level = level; 2041 sopt->sopt_name = name; 2042 (void)sockopt_alloc(sopt, size, KM_SLEEP); 2043 } 2044 2045 /* 2046 * destroy sockopt storage 2047 * - will release any held memory references 2048 */ 2049 void 2050 sockopt_destroy(struct sockopt *sopt) 2051 { 2052 2053 if (sopt->sopt_data != sopt->sopt_buf) 2054 kmem_free(sopt->sopt_data, sopt->sopt_size); 2055 2056 memset(sopt, 0, sizeof(*sopt)); 2057 } 2058 2059 /* 2060 * set sockopt value 2061 * - value is copied into sockopt 2062 * - memory is allocated when necessary, will not sleep 2063 */ 2064 int 2065 sockopt_set(struct sockopt *sopt, const void *buf, size_t len) 2066 { 2067 int error; 2068 2069 if (sopt->sopt_size == 0) { 2070 error = sockopt_alloc(sopt, len, KM_NOSLEEP); 2071 if (error) 2072 return error; 2073 } 2074 2075 KASSERT(sopt->sopt_size == len); 2076 memcpy(sopt->sopt_data, buf, len); 2077 return 0; 2078 } 2079 2080 /* 2081 * common case of set sockopt integer value 2082 */ 2083 int 2084 sockopt_setint(struct sockopt *sopt, int val) 2085 { 2086 2087 return sockopt_set(sopt, &val, sizeof(int)); 2088 } 2089 2090 /* 2091 * get sockopt value 2092 * - correct size must be given 2093 */ 2094 int 2095 sockopt_get(const struct sockopt *sopt, void *buf, size_t len) 2096 { 2097 2098 if (sopt->sopt_size != len) 2099 return EINVAL; 2100 2101 memcpy(buf, sopt->sopt_data, len); 2102 return 0; 2103 } 2104 2105 /* 2106 * common case of get sockopt integer value 2107 */ 2108 int 2109 sockopt_getint(const struct sockopt *sopt, int *valp) 2110 { 2111 2112 return sockopt_get(sopt, valp, sizeof(int)); 2113 } 2114 2115 /* 2116 * set sockopt value from mbuf 2117 * - ONLY for legacy code 2118 * - mbuf is released by sockopt 2119 * - will not sleep 2120 */ 2121 int 2122 sockopt_setmbuf(struct sockopt *sopt, struct mbuf *m) 2123 { 2124 size_t len; 2125 int error; 2126 2127 len = m_length(m); 2128 2129 if (sopt->sopt_size == 0) { 2130 error = sockopt_alloc(sopt, len, KM_NOSLEEP); 2131 if (error) 2132 return error; 2133 } 2134 2135 KASSERT(sopt->sopt_size == len); 2136 m_copydata(m, 0, len, sopt->sopt_data); 2137 m_freem(m); 2138 2139 return 0; 2140 } 2141 2142 /* 2143 * get sockopt value into mbuf 2144 * - ONLY for legacy code 2145 * - mbuf to be released by the caller 2146 * - will not sleep 2147 */ 2148 struct mbuf * 2149 sockopt_getmbuf(const struct sockopt *sopt) 2150 { 2151 struct mbuf *m; 2152 2153 if (sopt->sopt_size > MCLBYTES) 2154 return NULL; 2155 2156 m = m_get(M_DONTWAIT, MT_SOOPTS); 2157 if (m == NULL) 2158 return NULL; 2159 2160 if (sopt->sopt_size > MLEN) { 2161 MCLGET(m, M_DONTWAIT); 2162 if ((m->m_flags & M_EXT) == 0) { 2163 m_free(m); 2164 return NULL; 2165 } 2166 } 2167 2168 memcpy(mtod(m, void *), sopt->sopt_data, sopt->sopt_size); 2169 m->m_len = sopt->sopt_size; 2170 2171 return m; 2172 } 2173 2174 void 2175 sohasoutofband(struct socket *so) 2176 { 2177 2178 fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so); 2179 selnotify(&so->so_rcv.sb_sel, POLLPRI | POLLRDBAND, NOTE_SUBMIT); 2180 } 2181 2182 static void 2183 filt_sordetach(struct knote *kn) 2184 { 2185 struct socket *so; 2186 2187 so = ((file_t *)kn->kn_obj)->f_socket; 2188 solock(so); 2189 SLIST_REMOVE(&so->so_rcv.sb_sel.sel_klist, kn, knote, kn_selnext); 2190 if (SLIST_EMPTY(&so->so_rcv.sb_sel.sel_klist)) 2191 so->so_rcv.sb_flags &= ~SB_KNOTE; 2192 sounlock(so); 2193 } 2194 2195 /*ARGSUSED*/ 2196 static int 2197 filt_soread(struct knote *kn, long hint) 2198 { 2199 struct socket *so; 2200 int rv; 2201 2202 so = ((file_t *)kn->kn_obj)->f_socket; 2203 if (hint != NOTE_SUBMIT) 2204 solock(so); 2205 kn->kn_data = so->so_rcv.sb_cc; 2206 if (so->so_state & SS_CANTRCVMORE) { 2207 kn->kn_flags |= EV_EOF; 2208 kn->kn_fflags = so->so_error; 2209 rv = 1; 2210 } else if (so->so_error) /* temporary udp error */ 2211 rv = 1; 2212 else if (kn->kn_sfflags & NOTE_LOWAT) 2213 rv = (kn->kn_data >= kn->kn_sdata); 2214 else 2215 rv = (kn->kn_data >= so->so_rcv.sb_lowat); 2216 if (hint != NOTE_SUBMIT) 2217 sounlock(so); 2218 return rv; 2219 } 2220 2221 static void 2222 filt_sowdetach(struct knote *kn) 2223 { 2224 struct socket *so; 2225 2226 so = ((file_t *)kn->kn_obj)->f_socket; 2227 solock(so); 2228 SLIST_REMOVE(&so->so_snd.sb_sel.sel_klist, kn, knote, kn_selnext); 2229 if (SLIST_EMPTY(&so->so_snd.sb_sel.sel_klist)) 2230 so->so_snd.sb_flags &= ~SB_KNOTE; 2231 sounlock(so); 2232 } 2233 2234 /*ARGSUSED*/ 2235 static int 2236 filt_sowrite(struct knote *kn, long hint) 2237 { 2238 struct socket *so; 2239 int rv; 2240 2241 so = ((file_t *)kn->kn_obj)->f_socket; 2242 if (hint != NOTE_SUBMIT) 2243 solock(so); 2244 kn->kn_data = sbspace(&so->so_snd); 2245 if (so->so_state & SS_CANTSENDMORE) { 2246 kn->kn_flags |= EV_EOF; 2247 kn->kn_fflags = so->so_error; 2248 rv = 1; 2249 } else if (so->so_error) /* temporary udp error */ 2250 rv = 1; 2251 else if (((so->so_state & SS_ISCONNECTED) == 0) && 2252 (so->so_proto->pr_flags & PR_CONNREQUIRED)) 2253 rv = 0; 2254 else if (kn->kn_sfflags & NOTE_LOWAT) 2255 rv = (kn->kn_data >= kn->kn_sdata); 2256 else 2257 rv = (kn->kn_data >= so->so_snd.sb_lowat); 2258 if (hint != NOTE_SUBMIT) 2259 sounlock(so); 2260 return rv; 2261 } 2262 2263 /*ARGSUSED*/ 2264 static int 2265 filt_solisten(struct knote *kn, long hint) 2266 { 2267 struct socket *so; 2268 int rv; 2269 2270 so = ((file_t *)kn->kn_obj)->f_socket; 2271 2272 /* 2273 * Set kn_data to number of incoming connections, not 2274 * counting partial (incomplete) connections. 2275 */ 2276 if (hint != NOTE_SUBMIT) 2277 solock(so); 2278 kn->kn_data = so->so_qlen; 2279 rv = (kn->kn_data > 0); 2280 if (hint != NOTE_SUBMIT) 2281 sounlock(so); 2282 return rv; 2283 } 2284 2285 static const struct filterops solisten_filtops = 2286 { 1, NULL, filt_sordetach, filt_solisten }; 2287 static const struct filterops soread_filtops = 2288 { 1, NULL, filt_sordetach, filt_soread }; 2289 static const struct filterops sowrite_filtops = 2290 { 1, NULL, filt_sowdetach, filt_sowrite }; 2291 2292 int 2293 soo_kqfilter(struct file *fp, struct knote *kn) 2294 { 2295 struct socket *so; 2296 struct sockbuf *sb; 2297 2298 so = ((file_t *)kn->kn_obj)->f_socket; 2299 solock(so); 2300 switch (kn->kn_filter) { 2301 case EVFILT_READ: 2302 if (so->so_options & SO_ACCEPTCONN) 2303 kn->kn_fop = &solisten_filtops; 2304 else 2305 kn->kn_fop = &soread_filtops; 2306 sb = &so->so_rcv; 2307 break; 2308 case EVFILT_WRITE: 2309 kn->kn_fop = &sowrite_filtops; 2310 sb = &so->so_snd; 2311 break; 2312 default: 2313 sounlock(so); 2314 return (EINVAL); 2315 } 2316 SLIST_INSERT_HEAD(&sb->sb_sel.sel_klist, kn, kn_selnext); 2317 sb->sb_flags |= SB_KNOTE; 2318 sounlock(so); 2319 return (0); 2320 } 2321 2322 static int 2323 sodopoll(struct socket *so, int events) 2324 { 2325 int revents; 2326 2327 revents = 0; 2328 2329 if (events & (POLLIN | POLLRDNORM)) 2330 if (soreadable(so)) 2331 revents |= events & (POLLIN | POLLRDNORM); 2332 2333 if (events & (POLLOUT | POLLWRNORM)) 2334 if (sowritable(so)) 2335 revents |= events & (POLLOUT | POLLWRNORM); 2336 2337 if (events & (POLLPRI | POLLRDBAND)) 2338 if (so->so_oobmark || (so->so_state & SS_RCVATMARK)) 2339 revents |= events & (POLLPRI | POLLRDBAND); 2340 2341 return revents; 2342 } 2343 2344 int 2345 sopoll(struct socket *so, int events) 2346 { 2347 int revents = 0; 2348 2349 #ifndef DIAGNOSTIC 2350 /* 2351 * Do a quick, unlocked check in expectation that the socket 2352 * will be ready for I/O. Don't do this check if DIAGNOSTIC, 2353 * as the solocked() assertions will fail. 2354 */ 2355 if ((revents = sodopoll(so, events)) != 0) 2356 return revents; 2357 #endif 2358 2359 solock(so); 2360 if ((revents = sodopoll(so, events)) == 0) { 2361 if (events & (POLLIN | POLLPRI | POLLRDNORM | POLLRDBAND)) { 2362 selrecord(curlwp, &so->so_rcv.sb_sel); 2363 so->so_rcv.sb_flags |= SB_NOTIFY; 2364 } 2365 2366 if (events & (POLLOUT | POLLWRNORM)) { 2367 selrecord(curlwp, &so->so_snd.sb_sel); 2368 so->so_snd.sb_flags |= SB_NOTIFY; 2369 } 2370 } 2371 sounlock(so); 2372 2373 return revents; 2374 } 2375 2376 2377 #include <sys/sysctl.h> 2378 2379 static int sysctl_kern_somaxkva(SYSCTLFN_PROTO); 2380 static int sysctl_kern_sbmax(SYSCTLFN_PROTO); 2381 2382 /* 2383 * sysctl helper routine for kern.somaxkva. ensures that the given 2384 * value is not too small. 2385 * (XXX should we maybe make sure it's not too large as well?) 2386 */ 2387 static int 2388 sysctl_kern_somaxkva(SYSCTLFN_ARGS) 2389 { 2390 int error, new_somaxkva; 2391 struct sysctlnode node; 2392 2393 new_somaxkva = somaxkva; 2394 node = *rnode; 2395 node.sysctl_data = &new_somaxkva; 2396 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 2397 if (error || newp == NULL) 2398 return (error); 2399 2400 if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */ 2401 return (EINVAL); 2402 2403 mutex_enter(&so_pendfree_lock); 2404 somaxkva = new_somaxkva; 2405 cv_broadcast(&socurkva_cv); 2406 mutex_exit(&so_pendfree_lock); 2407 2408 return (error); 2409 } 2410 2411 /* 2412 * sysctl helper routine for kern.sbmax. Basically just ensures that 2413 * any new value is not too small. 2414 */ 2415 static int 2416 sysctl_kern_sbmax(SYSCTLFN_ARGS) 2417 { 2418 int error, new_sbmax; 2419 struct sysctlnode node; 2420 2421 new_sbmax = sb_max; 2422 node = *rnode; 2423 node.sysctl_data = &new_sbmax; 2424 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 2425 if (error || newp == NULL) 2426 return (error); 2427 2428 KERNEL_LOCK(1, NULL); 2429 error = sb_max_set(new_sbmax); 2430 KERNEL_UNLOCK_ONE(NULL); 2431 2432 return (error); 2433 } 2434 2435 static void 2436 sysctl_kern_socket_setup(void) 2437 { 2438 2439 KASSERT(socket_sysctllog == NULL); 2440 2441 sysctl_createv(&socket_sysctllog, 0, NULL, NULL, 2442 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2443 CTLTYPE_INT, "somaxkva", 2444 SYSCTL_DESCR("Maximum amount of kernel memory to be " 2445 "used for socket buffers"), 2446 sysctl_kern_somaxkva, 0, NULL, 0, 2447 CTL_KERN, KERN_SOMAXKVA, CTL_EOL); 2448 2449 sysctl_createv(&socket_sysctllog, 0, NULL, NULL, 2450 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2451 CTLTYPE_INT, "sbmax", 2452 SYSCTL_DESCR("Maximum socket buffer size"), 2453 sysctl_kern_sbmax, 0, NULL, 0, 2454 CTL_KERN, KERN_SBMAX, CTL_EOL); 2455 } 2456
Indexes created Sat Sep 20 22:09:52 GMT 2025