sys_select.c revision 1.3.4.3
1 /* $NetBSD: sys_select.c,v 1.3.4.3 2008/06/02 13:24:12 mjf Exp $ */ 2 3 /*- 4 * Copyright (c) 2007, 2008 The NetBSD Foundation, Inc. 5 * All rights reserved. 6 * 7 * This code is derived from software contributed to The NetBSD Foundation 8 * 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) 1982, 1986, 1989, 1993 34 * The Regents of the University of California. All rights reserved. 35 * (c) UNIX System Laboratories, Inc. 36 * All or some portions of this file are derived from material licensed 37 * to the University of California by American Telephone and Telegraph 38 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 39 * the permission of UNIX System Laboratories, Inc. 40 * 41 * Redistribution and use in source and binary forms, with or without 42 * modification, are permitted provided that the following conditions 43 * are met: 44 * 1. Redistributions of source code must retain the above copyright 45 * notice, this list of conditions and the following disclaimer. 46 * 2. Redistributions in binary form must reproduce the above copyright 47 * notice, this list of conditions and the following disclaimer in the 48 * documentation and/or other materials provided with the distribution. 49 * 3. Neither the name of the University nor the names of its contributors 50 * may be used to endorse or promote products derived from this software 51 * without specific prior written permission. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 63 * SUCH DAMAGE. 64 * 65 * @(#)sys_generic.c 8.9 (Berkeley) 2/14/95 66 */ 67 68 /* 69 * System calls relating to files. 70 */ 71 72 #include <sys/cdefs.h> 73 __KERNEL_RCSID(0, "$NetBSD: sys_select.c,v 1.3.4.3 2008/06/02 13:24:12 mjf Exp $"); 74 75 #include <sys/param.h> 76 #include <sys/systm.h> 77 #include <sys/filedesc.h> 78 #include <sys/ioctl.h> 79 #include <sys/file.h> 80 #include <sys/proc.h> 81 #include <sys/socketvar.h> 82 #include <sys/signalvar.h> 83 #include <sys/uio.h> 84 #include <sys/kernel.h> 85 #include <sys/stat.h> 86 #include <sys/poll.h> 87 #include <sys/vnode.h> 88 #include <sys/mount.h> 89 #include <sys/syscallargs.h> 90 #include <sys/cpu.h> 91 #include <sys/atomic.h> 92 #include <sys/socketvar.h> 93 #include <sys/sleepq.h> 94 95 /* Flags for lwp::l_selflag. */ 96 #define SEL_RESET 0 /* awoken, interrupted, or not yet polling */ 97 #define SEL_SCANNING 1 /* polling descriptors */ 98 #define SEL_BLOCKING 2 /* about to block on select_cv */ 99 100 /* Per-CPU state for select()/poll(). */ 101 #if MAXCPUS > 32 102 #error adjust this code 103 #endif 104 typedef struct selcpu { 105 kmutex_t sc_lock; 106 sleepq_t sc_sleepq; 107 int sc_ncoll; 108 uint32_t sc_mask; 109 } selcpu_t; 110 111 static int selscan(lwp_t *, fd_mask *, fd_mask *, int, register_t *); 112 static int pollscan(lwp_t *, struct pollfd *, int, register_t *); 113 static void selclear(void); 114 115 static syncobj_t select_sobj = { 116 SOBJ_SLEEPQ_FIFO, 117 sleepq_unsleep, 118 sleepq_changepri, 119 sleepq_lendpri, 120 syncobj_noowner, 121 }; 122 123 /* 124 * Select system call. 125 */ 126 int 127 sys_pselect(struct lwp *l, const struct sys_pselect_args *uap, register_t *retval) 128 { 129 /* { 130 syscallarg(int) nd; 131 syscallarg(fd_set *) in; 132 syscallarg(fd_set *) ou; 133 syscallarg(fd_set *) ex; 134 syscallarg(const struct timespec *) ts; 135 syscallarg(sigset_t *) mask; 136 } */ 137 struct timespec ats; 138 struct timeval atv, *tv = NULL; 139 sigset_t amask, *mask = NULL; 140 int error; 141 142 if (SCARG(uap, ts)) { 143 error = copyin(SCARG(uap, ts), &ats, sizeof(ats)); 144 if (error) 145 return error; 146 atv.tv_sec = ats.tv_sec; 147 atv.tv_usec = ats.tv_nsec / 1000; 148 tv = &atv; 149 } 150 if (SCARG(uap, mask) != NULL) { 151 error = copyin(SCARG(uap, mask), &amask, sizeof(amask)); 152 if (error) 153 return error; 154 mask = &amask; 155 } 156 157 return selcommon(l, retval, SCARG(uap, nd), SCARG(uap, in), 158 SCARG(uap, ou), SCARG(uap, ex), tv, mask); 159 } 160 161 int 162 inittimeleft(struct timeval *tv, struct timeval *sleeptv) 163 { 164 if (itimerfix(tv)) 165 return -1; 166 getmicrouptime(sleeptv); 167 return 0; 168 } 169 170 int 171 gettimeleft(struct timeval *tv, struct timeval *sleeptv) 172 { 173 /* 174 * We have to recalculate the timeout on every retry. 175 */ 176 struct timeval slepttv; 177 /* 178 * reduce tv by elapsed time 179 * based on monotonic time scale 180 */ 181 getmicrouptime(&slepttv); 182 timeradd(tv, sleeptv, tv); 183 timersub(tv, &slepttv, tv); 184 *sleeptv = slepttv; 185 return tvtohz(tv); 186 } 187 188 int 189 sys_select(struct lwp *l, const struct sys_select_args *uap, register_t *retval) 190 { 191 /* { 192 syscallarg(int) nd; 193 syscallarg(fd_set *) in; 194 syscallarg(fd_set *) ou; 195 syscallarg(fd_set *) ex; 196 syscallarg(struct timeval *) tv; 197 } */ 198 struct timeval atv, *tv = NULL; 199 int error; 200 201 if (SCARG(uap, tv)) { 202 error = copyin(SCARG(uap, tv), (void *)&atv, 203 sizeof(atv)); 204 if (error) 205 return error; 206 tv = &atv; 207 } 208 209 return selcommon(l, retval, SCARG(uap, nd), SCARG(uap, in), 210 SCARG(uap, ou), SCARG(uap, ex), tv, NULL); 211 } 212 213 int 214 selcommon(lwp_t *l, register_t *retval, int nd, fd_set *u_in, 215 fd_set *u_ou, fd_set *u_ex, struct timeval *tv, sigset_t *mask) 216 { 217 char smallbits[howmany(FD_SETSIZE, NFDBITS) * 218 sizeof(fd_mask) * 6]; 219 proc_t * const p = l->l_proc; 220 char *bits; 221 int ncoll, error, timo; 222 size_t ni; 223 sigset_t oldmask; 224 struct timeval sleeptv; 225 selcpu_t *sc; 226 227 error = 0; 228 if (nd < 0) 229 return (EINVAL); 230 if (nd > p->p_fd->fd_nfiles) { 231 /* forgiving; slightly wrong */ 232 nd = p->p_fd->fd_nfiles; 233 } 234 ni = howmany(nd, NFDBITS) * sizeof(fd_mask); 235 if (ni * 6 > sizeof(smallbits)) 236 bits = kmem_alloc(ni * 6, KM_SLEEP); 237 else 238 bits = smallbits; 239 240 #define getbits(name, x) \ 241 if (u_ ## name) { \ 242 error = copyin(u_ ## name, bits + ni * x, ni); \ 243 if (error) \ 244 goto done; \ 245 } else \ 246 memset(bits + ni * x, 0, ni); 247 getbits(in, 0); 248 getbits(ou, 1); 249 getbits(ex, 2); 250 #undef getbits 251 252 timo = 0; 253 if (tv && inittimeleft(tv, &sleeptv) == -1) { 254 error = EINVAL; 255 goto done; 256 } 257 258 if (mask) { 259 sigminusset(&sigcantmask, mask); 260 mutex_enter(p->p_lock); 261 oldmask = l->l_sigmask; 262 l->l_sigmask = *mask; 263 mutex_exit(p->p_lock); 264 } else 265 oldmask = l->l_sigmask; /* XXXgcc */ 266 267 sc = curcpu()->ci_data.cpu_selcpu; 268 l->l_selcpu = sc; 269 SLIST_INIT(&l->l_selwait); 270 for (;;) { 271 /* 272 * No need to lock. If this is overwritten by another 273 * value while scanning, we will retry below. We only 274 * need to see exact state from the descriptors that 275 * we are about to poll, and lock activity resulting 276 * from fo_poll is enough to provide an up to date value 277 * for new polling activity. 278 */ 279 l->l_selflag = SEL_SCANNING; 280 ncoll = sc->sc_ncoll; 281 282 error = selscan(l, (fd_mask *)(bits + ni * 0), 283 (fd_mask *)(bits + ni * 3), nd, retval); 284 285 if (error || *retval) 286 break; 287 if (tv && (timo = gettimeleft(tv, &sleeptv)) <= 0) 288 break; 289 mutex_spin_enter(&sc->sc_lock); 290 if (l->l_selflag != SEL_SCANNING || sc->sc_ncoll != ncoll) { 291 mutex_spin_exit(&sc->sc_lock); 292 continue; 293 } 294 l->l_selflag = SEL_BLOCKING; 295 l->l_kpriority = true; 296 sleepq_enter(&sc->sc_sleepq, l, &sc->sc_lock); 297 sleepq_enqueue(&sc->sc_sleepq, sc, "select", &select_sobj); 298 error = sleepq_block(timo, true); 299 if (error != 0) 300 break; 301 } 302 selclear(); 303 304 if (mask) { 305 mutex_enter(p->p_lock); 306 l->l_sigmask = oldmask; 307 mutex_exit(p->p_lock); 308 } 309 310 done: 311 /* select is not restarted after signals... */ 312 if (error == ERESTART) 313 error = EINTR; 314 if (error == EWOULDBLOCK) 315 error = 0; 316 if (error == 0 && u_in != NULL) 317 error = copyout(bits + ni * 3, u_in, ni); 318 if (error == 0 && u_ou != NULL) 319 error = copyout(bits + ni * 4, u_ou, ni); 320 if (error == 0 && u_ex != NULL) 321 error = copyout(bits + ni * 5, u_ex, ni); 322 if (bits != smallbits) 323 kmem_free(bits, ni * 6); 324 return (error); 325 } 326 327 int 328 selscan(lwp_t *l, fd_mask *ibitp, fd_mask *obitp, int nfd, 329 register_t *retval) 330 { 331 static const int flag[3] = { POLLRDNORM | POLLHUP | POLLERR, 332 POLLWRNORM | POLLHUP | POLLERR, 333 POLLRDBAND }; 334 int msk, i, j, fd, n; 335 fd_mask ibits, obits; 336 file_t *fp; 337 338 n = 0; 339 for (msk = 0; msk < 3; msk++) { 340 for (i = 0; i < nfd; i += NFDBITS) { 341 ibits = *ibitp++; 342 obits = 0; 343 while ((j = ffs(ibits)) && (fd = i + --j) < nfd) { 344 ibits &= ~(1 << j); 345 if ((fp = fd_getfile(fd)) == NULL) 346 return (EBADF); 347 if ((*fp->f_ops->fo_poll)(fp, flag[msk])) { 348 obits |= (1 << j); 349 n++; 350 } 351 fd_putfile(fd); 352 } 353 *obitp++ = obits; 354 } 355 } 356 *retval = n; 357 return (0); 358 } 359 360 /* 361 * Poll system call. 362 */ 363 int 364 sys_poll(struct lwp *l, const struct sys_poll_args *uap, register_t *retval) 365 { 366 /* { 367 syscallarg(struct pollfd *) fds; 368 syscallarg(u_int) nfds; 369 syscallarg(int) timeout; 370 } */ 371 struct timeval atv, *tv = NULL; 372 373 if (SCARG(uap, timeout) != INFTIM) { 374 atv.tv_sec = SCARG(uap, timeout) / 1000; 375 atv.tv_usec = (SCARG(uap, timeout) % 1000) * 1000; 376 tv = &atv; 377 } 378 379 return pollcommon(l, retval, SCARG(uap, fds), SCARG(uap, nfds), 380 tv, NULL); 381 } 382 383 /* 384 * Poll system call. 385 */ 386 int 387 sys_pollts(struct lwp *l, const struct sys_pollts_args *uap, register_t *retval) 388 { 389 /* { 390 syscallarg(struct pollfd *) fds; 391 syscallarg(u_int) nfds; 392 syscallarg(const struct timespec *) ts; 393 syscallarg(const sigset_t *) mask; 394 } */ 395 struct timespec ats; 396 struct timeval atv, *tv = NULL; 397 sigset_t amask, *mask = NULL; 398 int error; 399 400 if (SCARG(uap, ts)) { 401 error = copyin(SCARG(uap, ts), &ats, sizeof(ats)); 402 if (error) 403 return error; 404 atv.tv_sec = ats.tv_sec; 405 atv.tv_usec = ats.tv_nsec / 1000; 406 tv = &atv; 407 } 408 if (SCARG(uap, mask)) { 409 error = copyin(SCARG(uap, mask), &amask, sizeof(amask)); 410 if (error) 411 return error; 412 mask = &amask; 413 } 414 415 return pollcommon(l, retval, SCARG(uap, fds), SCARG(uap, nfds), 416 tv, mask); 417 } 418 419 int 420 pollcommon(lwp_t *l, register_t *retval, 421 struct pollfd *u_fds, u_int nfds, 422 struct timeval *tv, sigset_t *mask) 423 { 424 char smallbits[32 * sizeof(struct pollfd)]; 425 proc_t * const p = l->l_proc; 426 void * bits; 427 sigset_t oldmask; 428 int ncoll, error, timo; 429 size_t ni; 430 struct timeval sleeptv; 431 selcpu_t *sc; 432 433 if (nfds > p->p_fd->fd_nfiles) { 434 /* forgiving; slightly wrong */ 435 nfds = p->p_fd->fd_nfiles; 436 } 437 ni = nfds * sizeof(struct pollfd); 438 if (ni > sizeof(smallbits)) 439 bits = kmem_alloc(ni, KM_SLEEP); 440 else 441 bits = smallbits; 442 443 error = copyin(u_fds, bits, ni); 444 if (error) 445 goto done; 446 447 timo = 0; 448 if (tv && inittimeleft(tv, &sleeptv) == -1) { 449 error = EINVAL; 450 goto done; 451 } 452 453 if (mask) { 454 sigminusset(&sigcantmask, mask); 455 mutex_enter(p->p_lock); 456 oldmask = l->l_sigmask; 457 l->l_sigmask = *mask; 458 mutex_exit(p->p_lock); 459 } else 460 oldmask = l->l_sigmask; /* XXXgcc */ 461 462 sc = curcpu()->ci_data.cpu_selcpu; 463 l->l_selcpu = sc; 464 SLIST_INIT(&l->l_selwait); 465 for (;;) { 466 /* 467 * No need to lock. If this is overwritten by another 468 * value while scanning, we will retry below. We only 469 * need to see exact state from the descriptors that 470 * we are about to poll, and lock activity resulting 471 * from fo_poll is enough to provide an up to date value 472 * for new polling activity. 473 */ 474 ncoll = sc->sc_ncoll; 475 l->l_selflag = SEL_SCANNING; 476 477 error = pollscan(l, (struct pollfd *)bits, nfds, retval); 478 479 if (error || *retval) 480 break; 481 if (tv && (timo = gettimeleft(tv, &sleeptv)) <= 0) 482 break; 483 mutex_spin_enter(&sc->sc_lock); 484 if (l->l_selflag != SEL_SCANNING || sc->sc_ncoll != ncoll) { 485 mutex_spin_exit(&sc->sc_lock); 486 continue; 487 } 488 l->l_selflag = SEL_BLOCKING; 489 l->l_kpriority = true; 490 sleepq_enter(&sc->sc_sleepq, l, &sc->sc_lock); 491 sleepq_enqueue(&sc->sc_sleepq, sc, "select", &select_sobj); 492 error = sleepq_block(timo, true); 493 if (error != 0) 494 break; 495 } 496 selclear(); 497 498 if (mask) { 499 mutex_enter(p->p_lock); 500 l->l_sigmask = oldmask; 501 mutex_exit(p->p_lock); 502 } 503 done: 504 /* poll is not restarted after signals... */ 505 if (error == ERESTART) 506 error = EINTR; 507 if (error == EWOULDBLOCK) 508 error = 0; 509 if (error == 0) 510 error = copyout(bits, u_fds, ni); 511 if (bits != smallbits) 512 kmem_free(bits, ni); 513 return (error); 514 } 515 516 int 517 pollscan(lwp_t *l, struct pollfd *fds, int nfd, register_t *retval) 518 { 519 int i, n; 520 file_t *fp; 521 522 n = 0; 523 for (i = 0; i < nfd; i++, fds++) { 524 if (fds->fd < 0) { 525 fds->revents = 0; 526 } else if ((fp = fd_getfile(fds->fd)) == NULL) { 527 fds->revents = POLLNVAL; 528 n++; 529 } else { 530 fds->revents = (*fp->f_ops->fo_poll)(fp, 531 fds->events | POLLERR | POLLHUP); 532 if (fds->revents != 0) 533 n++; 534 fd_putfile(fds->fd); 535 } 536 } 537 *retval = n; 538 return (0); 539 } 540 541 /*ARGSUSED*/ 542 int 543 seltrue(dev_t dev, int events, lwp_t *l) 544 { 545 546 return (events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM)); 547 } 548 549 /* 550 * Record a select request. Concurrency issues: 551 * 552 * The caller holds the same lock across calls to selrecord() and 553 * selnotify(), so we don't need to consider a concurrent wakeup 554 * while in this routine. 555 * 556 * The only activity we need to guard against is selclear(), called by 557 * another thread that is exiting selcommon() or pollcommon(). 558 * `sel_lwp' can only become non-NULL while the caller's lock is held, 559 * so it cannot become non-NULL due to a change made by another thread 560 * while we are in this routine. It can only become _NULL_ due to a 561 * call to selclear(). 562 * 563 * If it is non-NULL and != selector there is the potential for 564 * selclear() to be called by another thread. If either of those 565 * conditions are true, we're not interested in touching the `named 566 * waiter' part of the selinfo record because we need to record a 567 * collision. Hence there is no need for additional locking in this 568 * routine. 569 */ 570 void 571 selrecord(lwp_t *selector, struct selinfo *sip) 572 { 573 selcpu_t *sc; 574 lwp_t *other; 575 576 KASSERT(selector == curlwp); 577 578 sc = selector->l_selcpu; 579 other = sip->sel_lwp; 580 581 if (other == selector) { 582 /* `selector' has already claimed it. */ 583 KASSERT(sip->sel_cpu = sc); 584 } else if (other == NULL) { 585 /* 586 * First named waiter, although there may be unnamed 587 * waiters (collisions). Issue a memory barrier to 588 * ensure that we access sel_lwp (above) before other 589 * fields - this guards against a call to selclear(). 590 */ 591 membar_enter(); 592 sip->sel_lwp = selector; 593 SLIST_INSERT_HEAD(&selector->l_selwait, sip, sel_chain); 594 /* Replace selinfo's lock with our chosen CPU's lock. */ 595 sip->sel_cpu = sc; 596 } else { 597 /* Multiple waiters: record a collision. */ 598 sip->sel_collision |= sc->sc_mask; 599 KASSERT(sip->sel_cpu != NULL); 600 } 601 } 602 603 /* 604 * Do a wakeup when a selectable event occurs. Concurrency issues: 605 * 606 * As per selrecord(), the caller's object lock is held. If there 607 * is a named waiter, we must acquire the associated selcpu's lock 608 * in order to synchronize with selclear() and pollers going to sleep 609 * in selcommon() and/or pollcommon(). 610 * 611 * sip->sel_cpu cannot change at this point, as it is only changed 612 * in selrecord(), and concurrent calls to selrecord() are locked 613 * out by the caller. 614 */ 615 void 616 selnotify(struct selinfo *sip, int events, long knhint) 617 { 618 selcpu_t *sc; 619 uint32_t mask; 620 int index, oflag, swapin; 621 lwp_t *l; 622 623 KNOTE(&sip->sel_klist, knhint); 624 625 if (sip->sel_lwp != NULL) { 626 /* One named LWP is waiting. */ 627 swapin = 0; 628 sc = sip->sel_cpu; 629 mutex_spin_enter(&sc->sc_lock); 630 /* Still there? */ 631 if (sip->sel_lwp != NULL) { 632 l = sip->sel_lwp; 633 /* 634 * If thread is sleeping, wake it up. If it's not 635 * yet asleep, it will notice the change in state 636 * and will re-poll the descriptors. 637 */ 638 oflag = l->l_selflag; 639 l->l_selflag = SEL_RESET; 640 if (oflag == SEL_BLOCKING && 641 l->l_mutex == &sc->sc_lock) { 642 KASSERT(l->l_wchan == sc); 643 swapin = sleepq_unsleep(l, false); 644 } 645 } 646 mutex_spin_exit(&sc->sc_lock); 647 if (swapin) 648 uvm_kick_scheduler(); 649 } 650 651 if ((mask = sip->sel_collision) != 0) { 652 /* 653 * There was a collision (multiple waiters): we must 654 * inform all potentially interested waiters. 655 */ 656 sip->sel_collision = 0; 657 do { 658 index = ffs(mask) - 1; 659 mask &= ~(1 << index); 660 sc = cpu_lookup_byindex(index)->ci_data.cpu_selcpu; 661 mutex_spin_enter(&sc->sc_lock); 662 sc->sc_ncoll++; 663 sleepq_wake(&sc->sc_sleepq, sc, (u_int)-1, 664 &sc->sc_lock); 665 } while (__predict_false(mask != 0)); 666 } 667 } 668 669 /* 670 * Remove an LWP from all objects that it is waiting for. Concurrency 671 * issues: 672 * 673 * The object owner's (e.g. device driver) lock is not held here. Calls 674 * can be made to selrecord() and we do not synchronize against those 675 * directly using locks. However, we use `sel_lwp' to lock out changes. 676 * Before clearing it we must use memory barriers to ensure that we can 677 * safely traverse the list of selinfo records. 678 */ 679 static void 680 selclear(void) 681 { 682 struct selinfo *sip, *next; 683 selcpu_t *sc; 684 lwp_t *l; 685 686 l = curlwp; 687 sc = l->l_selcpu; 688 689 mutex_spin_enter(&sc->sc_lock); 690 for (sip = SLIST_FIRST(&l->l_selwait); sip != NULL; sip = next) { 691 KASSERT(sip->sel_lwp == l); 692 KASSERT(sip->sel_cpu == l->l_selcpu); 693 /* 694 * Read link to next selinfo record, if any. 695 * It's no longer safe to touch `sip' after clearing 696 * `sel_lwp', so ensure that the read of `sel_chain' 697 * completes before the clearing of sel_lwp becomes 698 * globally visible. 699 */ 700 next = SLIST_NEXT(sip, sel_chain); 701 membar_exit(); 702 /* Release the record for another named waiter to use. */ 703 sip->sel_lwp = NULL; 704 } 705 mutex_spin_exit(&sc->sc_lock); 706 } 707 708 /* 709 * Initialize the select/poll system calls. Called once for each 710 * CPU in the system, as they are attached. 711 */ 712 void 713 selsysinit(struct cpu_info *ci) 714 { 715 selcpu_t *sc; 716 717 sc = kmem_alloc(roundup2(sizeof(selcpu_t), coherency_unit) + 718 coherency_unit, KM_SLEEP); 719 sc = (void *)roundup2((uintptr_t)sc, coherency_unit); 720 mutex_init(&sc->sc_lock, MUTEX_DEFAULT, IPL_SCHED); 721 sleepq_init(&sc->sc_sleepq); 722 sc->sc_ncoll = 0; 723 sc->sc_mask = (1 << cpu_index(ci)); 724 ci->ci_data.cpu_selcpu = sc; 725 } 726 727 /* 728 * Initialize a selinfo record. 729 */ 730 void 731 selinit(struct selinfo *sip) 732 { 733 734 memset(sip, 0, sizeof(*sip)); 735 } 736 737 /* 738 * Destroy a selinfo record. The owning object must not gain new 739 * references while this is in progress: all activity on the record 740 * must be stopped. 741 * 742 * Concurrency issues: we only need guard against a call to selclear() 743 * by a thread exiting selcommon() and/or pollcommon(). The caller has 744 * prevented further references being made to the selinfo record via 745 * selrecord(), and it won't call selwakeup() again. 746 */ 747 void 748 seldestroy(struct selinfo *sip) 749 { 750 selcpu_t *sc; 751 lwp_t *l; 752 753 if (sip->sel_lwp == NULL) 754 return; 755 756 /* 757 * Lock out selclear(). The selcpu pointer can't change while 758 * we are here since it is only ever changed in selrecord(), 759 * and that will not be entered again for this record because 760 * it is dying. 761 */ 762 KASSERT(sip->sel_cpu != NULL); 763 sc = sip->sel_cpu; 764 mutex_spin_enter(&sc->sc_lock); 765 if ((l = sip->sel_lwp) != NULL) { 766 /* 767 * This should rarely happen, so although SLIST_REMOVE() 768 * is slow, using it here is not a problem. 769 */ 770 KASSERT(l->l_selcpu == sc); 771 SLIST_REMOVE(&l->l_selwait, sip, selinfo, sel_chain); 772 sip->sel_lwp = NULL; 773 } 774 mutex_spin_exit(&sc->sc_lock); 775 } 776 777 int 778 pollsock(struct socket *so, const struct timeval *tvp, int events) 779 { 780 int ncoll, error, timo; 781 struct timeval sleeptv, tv; 782 selcpu_t *sc; 783 lwp_t *l; 784 785 timo = 0; 786 if (tvp != NULL) { 787 tv = *tvp; 788 if (inittimeleft(&tv, &sleeptv) == -1) 789 return EINVAL; 790 } 791 792 l = curlwp; 793 sc = l->l_cpu->ci_data.cpu_selcpu; 794 l->l_selcpu = sc; 795 SLIST_INIT(&l->l_selwait); 796 error = 0; 797 for (;;) { 798 /* 799 * No need to lock. If this is overwritten by another 800 * value while scanning, we will retry below. We only 801 * need to see exact state from the descriptors that 802 * we are about to poll, and lock activity resulting 803 * from fo_poll is enough to provide an up to date value 804 * for new polling activity. 805 */ 806 ncoll = sc->sc_ncoll; 807 l->l_selflag = SEL_SCANNING; 808 if (sopoll(so, events) != 0) 809 break; 810 if (tvp && (timo = gettimeleft(&tv, &sleeptv)) <= 0) 811 break; 812 mutex_spin_enter(&sc->sc_lock); 813 if (l->l_selflag != SEL_SCANNING || sc->sc_ncoll != ncoll) { 814 mutex_spin_exit(&sc->sc_lock); 815 continue; 816 } 817 l->l_selflag = SEL_BLOCKING; 818 sleepq_enter(&sc->sc_sleepq, l, &sc->sc_lock); 819 sleepq_enqueue(&sc->sc_sleepq, sc, "pollsock", &select_sobj); 820 error = sleepq_block(timo, true); 821 if (error != 0) 822 break; 823 } 824 selclear(); 825 /* poll is not restarted after signals... */ 826 if (error == ERESTART) 827 error = EINTR; 828 if (error == EWOULDBLOCK) 829 error = 0; 830 return (error); 831 } 832
Indexes created Sun Sep 21 20:09:37 GMT 2025