1 /*- 2 * Copyright (c) 1982, 1986, 1989, 1993 3 * The Regents of the University of California. All rights reserved. 4 * (c) UNIX System Laboratories, Inc. 5 * All or some portions of this file are derived from material licensed 6 * to the University of California by American Telephone and Telegraph 7 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 8 * the permission of UNIX System Laboratories, Inc. 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 * 4. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 * 34 * @(#)sys_generic.c 8.5 (Berkeley) 1/21/94 35 */ 36 37 #include <sys/cdefs.h> 38 __FBSDID("$FreeBSD$"); 39 40 #include "opt_compat.h" 41 #include "opt_ktrace.h" 42 43 #include <sys/param.h> 44 #include <sys/systm.h> 45 #include <sys/sysproto.h> 46 #include <sys/filedesc.h> 47 #include <sys/filio.h> 48 #include <sys/fcntl.h> 49 #include <sys/file.h> 50 #include <sys/proc.h> 51 #include <sys/signalvar.h> 52 #include <sys/socketvar.h> 53 #include <sys/uio.h> 54 #include <sys/kernel.h> 55 #include <sys/ktr.h> 56 #include <sys/limits.h> 57 #include <sys/malloc.h> 58 #include <sys/poll.h> 59 #include <sys/resourcevar.h> 60 #include <sys/selinfo.h> 61 #include <sys/sleepqueue.h> 62 #include <sys/syscallsubr.h> 63 #include <sys/sysctl.h> 64 #include <sys/sysent.h> 65 #include <sys/vnode.h> 66 #include <sys/bio.h> 67 #include <sys/buf.h> 68 #include <sys/condvar.h> 69 #ifdef KTRACE 70 #include <sys/ktrace.h> 71 #endif 72 73 #include <security/audit/audit.h> 74 75 static MALLOC_DEFINE(M_IOCTLOPS, "ioctlops", "ioctl data buffer"); 76 static MALLOC_DEFINE(M_SELECT, "select", "select() buffer"); 77 MALLOC_DEFINE(M_IOV, "iov", "large iov's"); 78 79 static int pollout(struct pollfd *, struct pollfd *, u_int); 80 static int pollscan(struct thread *, struct pollfd *, u_int); 81 static int pollrescan(struct thread *); 82 static int selscan(struct thread *, fd_mask **, fd_mask **, int); 83 static int selrescan(struct thread *, fd_mask **, fd_mask **); 84 static void selfdalloc(struct thread *, void *); 85 static void selfdfree(struct seltd *, struct selfd *); 86 static int dofileread(struct thread *, int, struct file *, struct uio *, 87 off_t, int); 88 static int dofilewrite(struct thread *, int, struct file *, struct uio *, 89 off_t, int); 90 static void doselwakeup(struct selinfo *, int); 91 static void seltdinit(struct thread *); 92 static int seltdwait(struct thread *, int); 93 static void seltdclear(struct thread *); 94 95 /* 96 * One seltd per-thread allocated on demand as needed. 97 * 98 * t - protected by st_mtx 99 * k - Only accessed by curthread or read-only 100 */ 101 struct seltd { 102 STAILQ_HEAD(, selfd) st_selq; /* (k) List of selfds. */ 103 struct selfd *st_free1; /* (k) free fd for read set. */ 104 struct selfd *st_free2; /* (k) free fd for write set. */ 105 struct mtx st_mtx; /* Protects struct seltd */ 106 struct cv st_wait; /* (t) Wait channel. */ 107 int st_flags; /* (t) SELTD_ flags. */ 108 }; 109 110 #define SELTD_PENDING 0x0001 /* We have pending events. */ 111 #define SELTD_RESCAN 0x0002 /* Doing a rescan. */ 112 113 /* 114 * One selfd allocated per-thread per-file-descriptor. 115 * f - protected by sf_mtx 116 */ 117 struct selfd { 118 STAILQ_ENTRY(selfd) sf_link; /* (k) fds owned by this td. */ 119 TAILQ_ENTRY(selfd) sf_threads; /* (f) fds on this selinfo. */ 120 struct selinfo *sf_si; /* (f) selinfo when linked. */ 121 struct mtx *sf_mtx; /* Pointer to selinfo mtx. */ 122 struct seltd *sf_td; /* (k) owning seltd. */ 123 void *sf_cookie; /* (k) fd or pollfd. */ 124 }; 125 126 static uma_zone_t selfd_zone; 127 static struct mtx_pool *mtxpool_select; 128 129 #ifndef _SYS_SYSPROTO_H_ 130 struct read_args { 131 int fd; 132 void *buf; 133 size_t nbyte; 134 }; 135 #endif 136 int 137 read(td, uap) 138 struct thread *td; 139 struct read_args *uap; 140 { 141 struct uio auio; 142 struct iovec aiov; 143 int error; 144 145 if (uap->nbyte > INT_MAX) 146 return (EINVAL); 147 aiov.iov_base = uap->buf; 148 aiov.iov_len = uap->nbyte; 149 auio.uio_iov = &aiov; 150 auio.uio_iovcnt = 1; 151 auio.uio_resid = uap->nbyte; 152 auio.uio_segflg = UIO_USERSPACE; 153 error = kern_readv(td, uap->fd, &auio); 154 return(error); 155 } 156 157 /* 158 * Positioned read system call 159 */ 160 #ifndef _SYS_SYSPROTO_H_ 161 struct pread_args { 162 int fd; 163 void *buf; 164 size_t nbyte; 165 int pad; 166 off_t offset; 167 }; 168 #endif 169 int 170 pread(td, uap) 171 struct thread *td; 172 struct pread_args *uap; 173 { 174 struct uio auio; 175 struct iovec aiov; 176 int error; 177 178 if (uap->nbyte > INT_MAX) 179 return (EINVAL); 180 aiov.iov_base = uap->buf; 181 aiov.iov_len = uap->nbyte; 182 auio.uio_iov = &aiov; 183 auio.uio_iovcnt = 1; 184 auio.uio_resid = uap->nbyte; 185 auio.uio_segflg = UIO_USERSPACE; 186 error = kern_preadv(td, uap->fd, &auio, uap->offset); 187 return(error); 188 } 189 190 int 191 freebsd6_pread(td, uap) 192 struct thread *td; 193 struct freebsd6_pread_args *uap; 194 { 195 struct pread_args oargs; 196 197 oargs.fd = uap->fd; 198 oargs.buf = uap->buf; 199 oargs.nbyte = uap->nbyte; 200 oargs.offset = uap->offset; 201 return (pread(td, &oargs)); 202 } 203 204 /* 205 * Scatter read system call. 206 */ 207 #ifndef _SYS_SYSPROTO_H_ 208 struct readv_args { 209 int fd; 210 struct iovec *iovp; 211 u_int iovcnt; 212 }; 213 #endif 214 int 215 readv(struct thread *td, struct readv_args *uap) 216 { 217 struct uio *auio; 218 int error; 219 220 error = copyinuio(uap->iovp, uap->iovcnt, &auio); 221 if (error) 222 return (error); 223 error = kern_readv(td, uap->fd, auio); 224 free(auio, M_IOV); 225 return (error); 226 } 227 228 int 229 kern_readv(struct thread *td, int fd, struct uio *auio) 230 { 231 struct file *fp; 232 int error; 233 234 error = fget_read(td, fd, &fp); 235 if (error) 236 return (error); 237 error = dofileread(td, fd, fp, auio, (off_t)-1, 0); 238 fdrop(fp, td); 239 return (error); 240 } 241 242 /* 243 * Scatter positioned read system call. 244 */ 245 #ifndef _SYS_SYSPROTO_H_ 246 struct preadv_args { 247 int fd; 248 struct iovec *iovp; 249 u_int iovcnt; 250 off_t offset; 251 }; 252 #endif 253 int 254 preadv(struct thread *td, struct preadv_args *uap) 255 { 256 struct uio *auio; 257 int error; 258 259 error = copyinuio(uap->iovp, uap->iovcnt, &auio); 260 if (error) 261 return (error); 262 error = kern_preadv(td, uap->fd, auio, uap->offset); 263 free(auio, M_IOV); 264 return (error); 265 } 266 267 int 268 kern_preadv(td, fd, auio, offset) 269 struct thread *td; 270 int fd; 271 struct uio *auio; 272 off_t offset; 273 { 274 struct file *fp; 275 int error; 276 277 error = fget_read(td, fd, &fp); 278 if (error) 279 return (error); 280 if (!(fp->f_ops->fo_flags & DFLAG_SEEKABLE)) 281 error = ESPIPE; 282 else if (offset < 0 && fp->f_vnode->v_type != VCHR) 283 error = EINVAL; 284 else 285 error = dofileread(td, fd, fp, auio, offset, FOF_OFFSET); 286 fdrop(fp, td); 287 return (error); 288 } 289 290 /* 291 * Common code for readv and preadv that reads data in 292 * from a file using the passed in uio, offset, and flags. 293 */ 294 static int 295 dofileread(td, fd, fp, auio, offset, flags) 296 struct thread *td; 297 int fd; 298 struct file *fp; 299 struct uio *auio; 300 off_t offset; 301 int flags; 302 { 303 ssize_t cnt; 304 int error; 305 #ifdef KTRACE 306 struct uio *ktruio = NULL; 307 #endif 308 309 /* Finish zero length reads right here */ 310 if (auio->uio_resid == 0) { 311 td->td_retval[0] = 0; 312 return(0); 313 } 314 auio->uio_rw = UIO_READ; 315 auio->uio_offset = offset; 316 auio->uio_td = td; 317 #ifdef KTRACE 318 if (KTRPOINT(td, KTR_GENIO)) 319 ktruio = cloneuio(auio); 320 #endif 321 cnt = auio->uio_resid; 322 if ((error = fo_read(fp, auio, td->td_ucred, flags, td))) { 323 if (auio->uio_resid != cnt && (error == ERESTART || 324 error == EINTR || error == EWOULDBLOCK)) 325 error = 0; 326 } 327 cnt -= auio->uio_resid; 328 #ifdef KTRACE 329 if (ktruio != NULL) { 330 ktruio->uio_resid = cnt; 331 ktrgenio(fd, UIO_READ, ktruio, error); 332 } 333 #endif 334 td->td_retval[0] = cnt; 335 return (error); 336 } 337 338 #ifndef _SYS_SYSPROTO_H_ 339 struct write_args { 340 int fd; 341 const void *buf; 342 size_t nbyte; 343 }; 344 #endif 345 int 346 write(td, uap) 347 struct thread *td; 348 struct write_args *uap; 349 { 350 struct uio auio; 351 struct iovec aiov; 352 int error; 353 354 if (uap->nbyte > INT_MAX) 355 return (EINVAL); 356 aiov.iov_base = (void *)(uintptr_t)uap->buf; 357 aiov.iov_len = uap->nbyte; 358 auio.uio_iov = &aiov; 359 auio.uio_iovcnt = 1; 360 auio.uio_resid = uap->nbyte; 361 auio.uio_segflg = UIO_USERSPACE; 362 error = kern_writev(td, uap->fd, &auio); 363 return(error); 364 } 365 366 /* 367 * Positioned write system call. 368 */ 369 #ifndef _SYS_SYSPROTO_H_ 370 struct pwrite_args { 371 int fd; 372 const void *buf; 373 size_t nbyte; 374 int pad; 375 off_t offset; 376 }; 377 #endif 378 int 379 pwrite(td, uap) 380 struct thread *td; 381 struct pwrite_args *uap; 382 { 383 struct uio auio; 384 struct iovec aiov; 385 int error; 386 387 if (uap->nbyte > INT_MAX) 388 return (EINVAL); 389 aiov.iov_base = (void *)(uintptr_t)uap->buf; 390 aiov.iov_len = uap->nbyte; 391 auio.uio_iov = &aiov; 392 auio.uio_iovcnt = 1; 393 auio.uio_resid = uap->nbyte; 394 auio.uio_segflg = UIO_USERSPACE; 395 error = kern_pwritev(td, uap->fd, &auio, uap->offset); 396 return(error); 397 } 398 399 int 400 freebsd6_pwrite(td, uap) 401 struct thread *td; 402 struct freebsd6_pwrite_args *uap; 403 { 404 struct pwrite_args oargs; 405 406 oargs.fd = uap->fd; 407 oargs.buf = uap->buf; 408 oargs.nbyte = uap->nbyte; 409 oargs.offset = uap->offset; 410 return (pwrite(td, &oargs)); 411 } 412 413 /* 414 * Gather write system call. 415 */ 416 #ifndef _SYS_SYSPROTO_H_ 417 struct writev_args { 418 int fd; 419 struct iovec *iovp; 420 u_int iovcnt; 421 }; 422 #endif 423 int 424 writev(struct thread *td, struct writev_args *uap) 425 { 426 struct uio *auio; 427 int error; 428 429 error = copyinuio(uap->iovp, uap->iovcnt, &auio); 430 if (error) 431 return (error); 432 error = kern_writev(td, uap->fd, auio); 433 free(auio, M_IOV); 434 return (error); 435 } 436 437 int 438 kern_writev(struct thread *td, int fd, struct uio *auio) 439 { 440 struct file *fp; 441 int error; 442 443 error = fget_write(td, fd, &fp); 444 if (error) 445 return (error); 446 error = dofilewrite(td, fd, fp, auio, (off_t)-1, 0); 447 fdrop(fp, td); 448 return (error); 449 } 450 451 /* 452 * Gather positioned write system call. 453 */ 454 #ifndef _SYS_SYSPROTO_H_ 455 struct pwritev_args { 456 int fd; 457 struct iovec *iovp; 458 u_int iovcnt; 459 off_t offset; 460 }; 461 #endif 462 int 463 pwritev(struct thread *td, struct pwritev_args *uap) 464 { 465 struct uio *auio; 466 int error; 467 468 error = copyinuio(uap->iovp, uap->iovcnt, &auio); 469 if (error) 470 return (error); 471 error = kern_pwritev(td, uap->fd, auio, uap->offset); 472 free(auio, M_IOV); 473 return (error); 474 } 475 476 int 477 kern_pwritev(td, fd, auio, offset) 478 struct thread *td; 479 struct uio *auio; 480 int fd; 481 off_t offset; 482 { 483 struct file *fp; 484 int error; 485 486 error = fget_write(td, fd, &fp); 487 if (error) 488 return (error); 489 if (!(fp->f_ops->fo_flags & DFLAG_SEEKABLE)) 490 error = ESPIPE; 491 else if (offset < 0 && fp->f_vnode->v_type != VCHR) 492 error = EINVAL; 493 else 494 error = dofilewrite(td, fd, fp, auio, offset, FOF_OFFSET); 495 fdrop(fp, td); 496 return (error); 497 } 498 499 /* 500 * Common code for writev and pwritev that writes data to 501 * a file using the passed in uio, offset, and flags. 502 */ 503 static int 504 dofilewrite(td, fd, fp, auio, offset, flags) 505 struct thread *td; 506 int fd; 507 struct file *fp; 508 struct uio *auio; 509 off_t offset; 510 int flags; 511 { 512 ssize_t cnt; 513 int error; 514 #ifdef KTRACE 515 struct uio *ktruio = NULL; 516 #endif 517 518 auio->uio_rw = UIO_WRITE; 519 auio->uio_td = td; 520 auio->uio_offset = offset; 521 #ifdef KTRACE 522 if (KTRPOINT(td, KTR_GENIO)) 523 ktruio = cloneuio(auio); 524 #endif 525 cnt = auio->uio_resid; 526 if (fp->f_type == DTYPE_VNODE) 527 bwillwrite(); 528 if ((error = fo_write(fp, auio, td->td_ucred, flags, td))) { 529 if (auio->uio_resid != cnt && (error == ERESTART || 530 error == EINTR || error == EWOULDBLOCK)) 531 error = 0; 532 /* Socket layer is responsible for issuing SIGPIPE. */ 533 if (fp->f_type != DTYPE_SOCKET && error == EPIPE) { 534 PROC_LOCK(td->td_proc); 535 psignal(td->td_proc, SIGPIPE); 536 PROC_UNLOCK(td->td_proc); 537 } 538 } 539 cnt -= auio->uio_resid; 540 #ifdef KTRACE 541 if (ktruio != NULL) { 542 ktruio->uio_resid = cnt; 543 ktrgenio(fd, UIO_WRITE, ktruio, error); 544 } 545 #endif 546 td->td_retval[0] = cnt; 547 return (error); 548 } 549 550 /* 551 * Truncate a file given a file descriptor. 552 * 553 * Can't use fget_write() here, since must return EINVAL and not EBADF if the 554 * descriptor isn't writable. 555 */ 556 int 557 kern_ftruncate(td, fd, length) 558 struct thread *td; 559 int fd; 560 off_t length; 561 { 562 struct file *fp; 563 int error; 564 565 AUDIT_ARG_FD(fd); 566 if (length < 0) 567 return (EINVAL); 568 error = fget(td, fd, &fp); 569 if (error) 570 return (error); 571 AUDIT_ARG_FILE(td->td_proc, fp); 572 if (!(fp->f_flag & FWRITE)) { 573 fdrop(fp, td); 574 return (EINVAL); 575 } 576 error = fo_truncate(fp, length, td->td_ucred, td); 577 fdrop(fp, td); 578 return (error); 579 } 580 581 #ifndef _SYS_SYSPROTO_H_ 582 struct ftruncate_args { 583 int fd; 584 int pad; 585 off_t length; 586 }; 587 #endif 588 int 589 ftruncate(td, uap) 590 struct thread *td; 591 struct ftruncate_args *uap; 592 { 593 594 return (kern_ftruncate(td, uap->fd, uap->length)); 595 } 596 597 #if defined(COMPAT_43) 598 #ifndef _SYS_SYSPROTO_H_ 599 struct oftruncate_args { 600 int fd; 601 long length; 602 }; 603 #endif 604 int 605 oftruncate(td, uap) 606 struct thread *td; 607 struct oftruncate_args *uap; 608 { 609 610 return (kern_ftruncate(td, uap->fd, uap->length)); 611 } 612 #endif /* COMPAT_43 */ 613 614 #ifndef _SYS_SYSPROTO_H_ 615 struct ioctl_args { 616 int fd; 617 u_long com; 618 caddr_t data; 619 }; 620 #endif 621 /* ARGSUSED */ 622 int 623 ioctl(struct thread *td, struct ioctl_args *uap) 624 { 625 u_long com; 626 int arg, error; 627 u_int size; 628 caddr_t data; 629 630 if (uap->com > 0xffffffff) { 631 printf( 632 "WARNING pid %d (%s): ioctl sign-extension ioctl %lx\n", 633 td->td_proc->p_pid, td->td_name, uap->com); 634 uap->com &= 0xffffffff; 635 } 636 com = uap->com; 637 638 /* 639 * Interpret high order word to find amount of data to be 640 * copied to/from the user's address space. 641 */ 642 size = IOCPARM_LEN(com); 643 if ((size > IOCPARM_MAX) || 644 ((com & (IOC_VOID | IOC_IN | IOC_OUT)) == 0) || 645 #if defined(COMPAT_FREEBSD5) || defined(COMPAT_FREEBSD4) || defined(COMPAT_43) 646 ((com & IOC_OUT) && size == 0) || 647 #else 648 ((com & (IOC_IN | IOC_OUT)) && size == 0) || 649 #endif 650 ((com & IOC_VOID) && size > 0 && size != sizeof(int))) 651 return (ENOTTY); 652 653 if (size > 0) { 654 if (com & IOC_VOID) { 655 /* Integer argument. */ 656 arg = (intptr_t)uap->data; 657 data = (void *)&arg; 658 size = 0; 659 } else 660 data = malloc((u_long)size, M_IOCTLOPS, M_WAITOK); 661 } else 662 data = (void *)&uap->data; 663 if (com & IOC_IN) { 664 error = copyin(uap->data, data, (u_int)size); 665 if (error) { 666 if (size > 0) 667 free(data, M_IOCTLOPS); 668 return (error); 669 } 670 } else if (com & IOC_OUT) { 671 /* 672 * Zero the buffer so the user always 673 * gets back something deterministic. 674 */ 675 bzero(data, size); 676 } 677 678 error = kern_ioctl(td, uap->fd, com, data); 679 680 if (error == 0 && (com & IOC_OUT)) 681 error = copyout(data, uap->data, (u_int)size); 682 683 if (size > 0) 684 free(data, M_IOCTLOPS); 685 return (error); 686 } 687 688 int 689 kern_ioctl(struct thread *td, int fd, u_long com, caddr_t data) 690 { 691 struct file *fp; 692 struct filedesc *fdp; 693 int error; 694 int tmp; 695 696 AUDIT_ARG_FD(fd); 697 AUDIT_ARG_CMD(com); 698 if ((error = fget(td, fd, &fp)) != 0) 699 return (error); 700 if ((fp->f_flag & (FREAD | FWRITE)) == 0) { 701 fdrop(fp, td); 702 return (EBADF); 703 } 704 fdp = td->td_proc->p_fd; 705 switch (com) { 706 case FIONCLEX: 707 FILEDESC_XLOCK(fdp); 708 fdp->fd_ofileflags[fd] &= ~UF_EXCLOSE; 709 FILEDESC_XUNLOCK(fdp); 710 goto out; 711 case FIOCLEX: 712 FILEDESC_XLOCK(fdp); 713 fdp->fd_ofileflags[fd] |= UF_EXCLOSE; 714 FILEDESC_XUNLOCK(fdp); 715 goto out; 716 case FIONBIO: 717 if ((tmp = *(int *)data)) 718 atomic_set_int(&fp->f_flag, FNONBLOCK); 719 else 720 atomic_clear_int(&fp->f_flag, FNONBLOCK); 721 data = (void *)&tmp; 722 break; 723 case FIOASYNC: 724 if ((tmp = *(int *)data)) 725 atomic_set_int(&fp->f_flag, FASYNC); 726 else 727 atomic_clear_int(&fp->f_flag, FASYNC); 728 data = (void *)&tmp; 729 break; 730 } 731 732 error = fo_ioctl(fp, com, data, td->td_ucred, td); 733 out: 734 fdrop(fp, td); 735 return (error); 736 } 737 738 int 739 poll_no_poll(int events) 740 { 741 /* 742 * Return true for read/write. If the user asked for something 743 * special, return POLLNVAL, so that clients have a way of 744 * determining reliably whether or not the extended 745 * functionality is present without hard-coding knowledge 746 * of specific filesystem implementations. 747 */ 748 if (events & ~POLLSTANDARD) 749 return (POLLNVAL); 750 751 return (events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM)); 752 } 753 754 int 755 pselect(struct thread *td, struct pselect_args *uap) 756 { 757 struct timespec ts; 758 struct timeval tv, *tvp; 759 sigset_t set, *uset; 760 int error; 761 762 if (uap->ts != NULL) { 763 error = copyin(uap->ts, &ts, sizeof(ts)); 764 if (error != 0) 765 return (error); 766 TIMESPEC_TO_TIMEVAL(&tv, &ts); 767 tvp = &tv; 768 } else 769 tvp = NULL; 770 if (uap->sm != NULL) { 771 error = copyin(uap->sm, &set, sizeof(set)); 772 if (error != 0) 773 return (error); 774 uset = &set; 775 } else 776 uset = NULL; 777 return (kern_pselect(td, uap->nd, uap->in, uap->ou, uap->ex, tvp, 778 uset, NFDBITS)); 779 } 780 781 int 782 kern_pselect(struct thread *td, int nd, fd_set *in, fd_set *ou, fd_set *ex, 783 struct timeval *tvp, sigset_t *uset, int abi_nfdbits) 784 { 785 int error; 786 787 if (uset != NULL) { 788 error = kern_sigprocmask(td, SIG_SETMASK, uset, 789 &td->td_oldsigmask, 0); 790 if (error != 0) 791 return (error); 792 td->td_pflags |= TDP_OLDMASK; 793 /* 794 * Make sure that ast() is called on return to 795 * usermode and TDP_OLDMASK is cleared, restoring old 796 * sigmask. 797 */ 798 thread_lock(td); 799 td->td_flags |= TDF_ASTPENDING; 800 thread_unlock(td); 801 } 802 error = kern_select(td, nd, in, ou, ex, tvp, abi_nfdbits); 803 return (error); 804 } 805 806 #ifndef _SYS_SYSPROTO_H_ 807 struct select_args { 808 int nd; 809 fd_set *in, *ou, *ex; 810 struct timeval *tv; 811 }; 812 #endif 813 int 814 select(struct thread *td, struct select_args *uap) 815 { 816 struct timeval tv, *tvp; 817 int error; 818 819 if (uap->tv != NULL) { 820 error = copyin(uap->tv, &tv, sizeof(tv)); 821 if (error) 822 return (error); 823 tvp = &tv; 824 } else 825 tvp = NULL; 826 827 return (kern_select(td, uap->nd, uap->in, uap->ou, uap->ex, tvp, 828 NFDBITS)); 829 } 830 831 int 832 kern_select(struct thread *td, int nd, fd_set *fd_in, fd_set *fd_ou, 833 fd_set *fd_ex, struct timeval *tvp, int abi_nfdbits) 834 { 835 struct filedesc *fdp; 836 /* 837 * The magic 2048 here is chosen to be just enough for FD_SETSIZE 838 * infds with the new FD_SETSIZE of 1024, and more than enough for 839 * FD_SETSIZE infds, outfds and exceptfds with the old FD_SETSIZE 840 * of 256. 841 */ 842 fd_mask s_selbits[howmany(2048, NFDBITS)]; 843 fd_mask *ibits[3], *obits[3], *selbits, *sbp; 844 struct timeval atv, rtv, ttv; 845 int error, timo; 846 u_int nbufbytes, ncpbytes, ncpubytes, nfdbits; 847 848 if (nd < 0) 849 return (EINVAL); 850 fdp = td->td_proc->p_fd; 851 if (nd > fdp->fd_lastfile + 1) 852 nd = fdp->fd_lastfile + 1; 853 854 /* 855 * Allocate just enough bits for the non-null fd_sets. Use the 856 * preallocated auto buffer if possible. 857 */ 858 nfdbits = roundup(nd, NFDBITS); 859 ncpbytes = nfdbits / NBBY; 860 ncpubytes = roundup(nd, abi_nfdbits) / NBBY; 861 nbufbytes = 0; 862 if (fd_in != NULL) 863 nbufbytes += 2 * ncpbytes; 864 if (fd_ou != NULL) 865 nbufbytes += 2 * ncpbytes; 866 if (fd_ex != NULL) 867 nbufbytes += 2 * ncpbytes; 868 if (nbufbytes <= sizeof s_selbits) 869 selbits = &s_selbits[0]; 870 else 871 selbits = malloc(nbufbytes, M_SELECT, M_WAITOK); 872 873 /* 874 * Assign pointers into the bit buffers and fetch the input bits. 875 * Put the output buffers together so that they can be bzeroed 876 * together. 877 */ 878 sbp = selbits; 879 #define getbits(name, x) \ 880 do { \ 881 if (name == NULL) \ 882 ibits[x] = NULL; \ 883 else { \ 884 ibits[x] = sbp + nbufbytes / 2 / sizeof *sbp; \ 885 obits[x] = sbp; \ 886 sbp += ncpbytes / sizeof *sbp; \ 887 error = copyin(name, ibits[x], ncpubytes); \ 888 if (error != 0) \ 889 goto done; \ 890 bzero((char *)ibits[x] + ncpubytes, \ 891 ncpbytes - ncpubytes); \ 892 } \ 893 } while (0) 894 getbits(fd_in, 0); 895 getbits(fd_ou, 1); 896 getbits(fd_ex, 2); 897 #undef getbits 898 if (nbufbytes != 0) 899 bzero(selbits, nbufbytes / 2); 900 901 if (tvp != NULL) { 902 atv = *tvp; 903 if (itimerfix(&atv)) { 904 error = EINVAL; 905 goto done; 906 } 907 getmicrouptime(&rtv); 908 timevaladd(&atv, &rtv); 909 } else { 910 atv.tv_sec = 0; 911 atv.tv_usec = 0; 912 } 913 timo = 0; 914 seltdinit(td); 915 /* Iterate until the timeout expires or descriptors become ready. */ 916 for (;;) { 917 error = selscan(td, ibits, obits, nd); 918 if (error || td->td_retval[0] != 0) 919 break; 920 if (atv.tv_sec || atv.tv_usec) { 921 getmicrouptime(&rtv); 922 if (timevalcmp(&rtv, &atv, >=)) 923 break; 924 ttv = atv; 925 timevalsub(&ttv, &rtv); 926 timo = ttv.tv_sec > 24 * 60 * 60 ? 927 24 * 60 * 60 * hz : tvtohz(&ttv); 928 } 929 error = seltdwait(td, timo); 930 if (error) 931 break; 932 error = selrescan(td, ibits, obits); 933 if (error || td->td_retval[0] != 0) 934 break; 935 } 936 seltdclear(td); 937 938 done: 939 /* select is not restarted after signals... */ 940 if (error == ERESTART) 941 error = EINTR; 942 if (error == EWOULDBLOCK) 943 error = 0; 944 #define putbits(name, x) \ 945 if (name && (error2 = copyout(obits[x], name, ncpubytes))) \ 946 error = error2; 947 if (error == 0) { 948 int error2; 949 950 putbits(fd_in, 0); 951 putbits(fd_ou, 1); 952 putbits(fd_ex, 2); 953 #undef putbits 954 } 955 if (selbits != &s_selbits[0]) 956 free(selbits, M_SELECT); 957 958 return (error); 959 } 960 /* 961 * Convert a select bit set to poll flags. 962 * 963 * The backend always returns POLLHUP/POLLERR if appropriate and we 964 * return this as a set bit in any set. 965 */ 966 static int select_flags[3] = { 967 POLLRDNORM | POLLHUP | POLLERR, 968 POLLWRNORM | POLLHUP | POLLERR, 969 POLLRDBAND | POLLHUP | POLLERR 970 }; 971 972 /* 973 * Compute the fo_poll flags required for a fd given by the index and 974 * bit position in the fd_mask array. 975 */ 976 static __inline int 977 selflags(fd_mask **ibits, int idx, fd_mask bit) 978 { 979 int flags; 980 int msk; 981 982 flags = 0; 983 for (msk = 0; msk < 3; msk++) { 984 if (ibits[msk] == NULL) 985 continue; 986 if ((ibits[msk][idx] & bit) == 0) 987 continue; 988 flags |= select_flags[msk]; 989 } 990 return (flags); 991 } 992 993 /* 994 * Set the appropriate output bits given a mask of fired events and the 995 * input bits originally requested. 996 */ 997 static __inline int 998 selsetbits(fd_mask **ibits, fd_mask **obits, int idx, fd_mask bit, int events) 999 { 1000 int msk; 1001 int n; 1002 1003 n = 0; 1004 for (msk = 0; msk < 3; msk++) { 1005 if ((events & select_flags[msk]) == 0) 1006 continue; 1007 if (ibits[msk] == NULL) 1008 continue; 1009 if ((ibits[msk][idx] & bit) == 0) 1010 continue; 1011 /* 1012 * XXX Check for a duplicate set. This can occur because a 1013 * socket calls selrecord() twice for each poll() call 1014 * resulting in two selfds per real fd. selrescan() will 1015 * call selsetbits twice as a result. 1016 */ 1017 if ((obits[msk][idx] & bit) != 0) 1018 continue; 1019 obits[msk][idx] |= bit; 1020 n++; 1021 } 1022 1023 return (n); 1024 } 1025 1026 /* 1027 * Traverse the list of fds attached to this thread's seltd and check for 1028 * completion. 1029 */ 1030 static int 1031 selrescan(struct thread *td, fd_mask **ibits, fd_mask **obits) 1032 { 1033 struct filedesc *fdp; 1034 struct selinfo *si; 1035 struct seltd *stp; 1036 struct selfd *sfp; 1037 struct selfd *sfn; 1038 struct file *fp; 1039 fd_mask bit; 1040 int fd, ev, n, idx; 1041 1042 fdp = td->td_proc->p_fd; 1043 stp = td->td_sel; 1044 n = 0; 1045 STAILQ_FOREACH_SAFE(sfp, &stp->st_selq, sf_link, sfn) { 1046 fd = (int)(uintptr_t)sfp->sf_cookie; 1047 si = sfp->sf_si; 1048 selfdfree(stp, sfp); 1049 /* If the selinfo wasn't cleared the event didn't fire. */ 1050 if (si != NULL) 1051 continue; 1052 if ((fp = fget_unlocked(fdp, fd)) == NULL) 1053 return (EBADF); 1054 idx = fd / NFDBITS; 1055 bit = (fd_mask)1 << (fd % NFDBITS); 1056 ev = fo_poll(fp, selflags(ibits, idx, bit), td->td_ucred, td); 1057 fdrop(fp, td); 1058 if (ev != 0) 1059 n += selsetbits(ibits, obits, idx, bit, ev); 1060 } 1061 stp->st_flags = 0; 1062 td->td_retval[0] = n; 1063 return (0); 1064 } 1065 1066 /* 1067 * Perform the initial filedescriptor scan and register ourselves with 1068 * each selinfo. 1069 */ 1070 static int 1071 selscan(td, ibits, obits, nfd) 1072 struct thread *td; 1073 fd_mask **ibits, **obits; 1074 int nfd; 1075 { 1076 struct filedesc *fdp; 1077 struct file *fp; 1078 fd_mask bit; 1079 int ev, flags, end, fd; 1080 int n, idx; 1081 1082 fdp = td->td_proc->p_fd; 1083 n = 0; 1084 for (idx = 0, fd = 0; fd < nfd; idx++) { 1085 end = imin(fd + NFDBITS, nfd); 1086 for (bit = 1; fd < end; bit <<= 1, fd++) { 1087 /* Compute the list of events we're interested in. */ 1088 flags = selflags(ibits, idx, bit); 1089 if (flags == 0) 1090 continue; 1091 if ((fp = fget_unlocked(fdp, fd)) == NULL) 1092 return (EBADF); 1093 selfdalloc(td, (void *)(uintptr_t)fd); 1094 ev = fo_poll(fp, flags, td->td_ucred, td); 1095 fdrop(fp, td); 1096 if (ev != 0) 1097 n += selsetbits(ibits, obits, idx, bit, ev); 1098 } 1099 } 1100 1101 td->td_retval[0] = n; 1102 return (0); 1103 } 1104 1105 #ifndef _SYS_SYSPROTO_H_ 1106 struct poll_args { 1107 struct pollfd *fds; 1108 u_int nfds; 1109 int timeout; 1110 }; 1111 #endif 1112 int 1113 poll(td, uap) 1114 struct thread *td; 1115 struct poll_args *uap; 1116 { 1117 struct pollfd *bits; 1118 struct pollfd smallbits[32]; 1119 struct timeval atv, rtv, ttv; 1120 int error = 0, timo; 1121 u_int nfds; 1122 size_t ni; 1123 1124 nfds = uap->nfds; 1125 if (nfds > maxfilesperproc && nfds > FD_SETSIZE) 1126 return (EINVAL); 1127 ni = nfds * sizeof(struct pollfd); 1128 if (ni > sizeof(smallbits)) 1129 bits = malloc(ni, M_TEMP, M_WAITOK); 1130 else 1131 bits = smallbits; 1132 error = copyin(uap->fds, bits, ni); 1133 if (error) 1134 goto done; 1135 if (uap->timeout != INFTIM) { 1136 atv.tv_sec = uap->timeout / 1000; 1137 atv.tv_usec = (uap->timeout % 1000) * 1000; 1138 if (itimerfix(&atv)) { 1139 error = EINVAL; 1140 goto done; 1141 } 1142 getmicrouptime(&rtv); 1143 timevaladd(&atv, &rtv); 1144 } else { 1145 atv.tv_sec = 0; 1146 atv.tv_usec = 0; 1147 } 1148 timo = 0; 1149 seltdinit(td); 1150 /* Iterate until the timeout expires or descriptors become ready. */ 1151 for (;;) { 1152 error = pollscan(td, bits, nfds); 1153 if (error || td->td_retval[0] != 0) 1154 break; 1155 if (atv.tv_sec || atv.tv_usec) { 1156 getmicrouptime(&rtv); 1157 if (timevalcmp(&rtv, &atv, >=)) 1158 break; 1159 ttv = atv; 1160 timevalsub(&ttv, &rtv); 1161 timo = ttv.tv_sec > 24 * 60 * 60 ? 1162 24 * 60 * 60 * hz : tvtohz(&ttv); 1163 } 1164 error = seltdwait(td, timo); 1165 if (error) 1166 break; 1167 error = pollrescan(td); 1168 if (error || td->td_retval[0] != 0) 1169 break; 1170 } 1171 seltdclear(td); 1172 1173 done: 1174 /* poll is not restarted after signals... */ 1175 if (error == ERESTART) 1176 error = EINTR; 1177 if (error == EWOULDBLOCK) 1178 error = 0; 1179 if (error == 0) { 1180 error = pollout(bits, uap->fds, nfds); 1181 if (error) 1182 goto out; 1183 } 1184 out: 1185 if (ni > sizeof(smallbits)) 1186 free(bits, M_TEMP); 1187 return (error); 1188 } 1189 1190 static int 1191 pollrescan(struct thread *td) 1192 { 1193 struct seltd *stp; 1194 struct selfd *sfp; 1195 struct selfd *sfn; 1196 struct selinfo *si; 1197 struct filedesc *fdp; 1198 struct file *fp; 1199 struct pollfd *fd; 1200 int n; 1201 1202 n = 0; 1203 fdp = td->td_proc->p_fd; 1204 stp = td->td_sel; 1205 FILEDESC_SLOCK(fdp); 1206 STAILQ_FOREACH_SAFE(sfp, &stp->st_selq, sf_link, sfn) { 1207 fd = (struct pollfd *)sfp->sf_cookie; 1208 si = sfp->sf_si; 1209 selfdfree(stp, sfp); 1210 /* If the selinfo wasn't cleared the event didn't fire. */ 1211 if (si != NULL) 1212 continue; 1213 fp = fdp->fd_ofiles[fd->fd]; 1214 if (fp == NULL) { 1215 fd->revents = POLLNVAL; 1216 n++; 1217 continue; 1218 } 1219 /* 1220 * Note: backend also returns POLLHUP and 1221 * POLLERR if appropriate. 1222 */ 1223 fd->revents = fo_poll(fp, fd->events, td->td_ucred, td); 1224 if (fd->revents != 0) 1225 n++; 1226 } 1227 FILEDESC_SUNLOCK(fdp); 1228 stp->st_flags = 0; 1229 td->td_retval[0] = n; 1230 return (0); 1231 } 1232 1233 1234 static int 1235 pollout(fds, ufds, nfd) 1236 struct pollfd *fds; 1237 struct pollfd *ufds; 1238 u_int nfd; 1239 { 1240 int error = 0; 1241 u_int i = 0; 1242 1243 for (i = 0; i < nfd; i++) { 1244 error = copyout(&fds->revents, &ufds->revents, 1245 sizeof(ufds->revents)); 1246 if (error) 1247 return (error); 1248 fds++; 1249 ufds++; 1250 } 1251 return (0); 1252 } 1253 1254 static int 1255 pollscan(td, fds, nfd) 1256 struct thread *td; 1257 struct pollfd *fds; 1258 u_int nfd; 1259 { 1260 struct filedesc *fdp = td->td_proc->p_fd; 1261 int i; 1262 struct file *fp; 1263 int n = 0; 1264 1265 FILEDESC_SLOCK(fdp); 1266 for (i = 0; i < nfd; i++, fds++) { 1267 if (fds->fd >= fdp->fd_nfiles) { 1268 fds->revents = POLLNVAL; 1269 n++; 1270 } else if (fds->fd < 0) { 1271 fds->revents = 0; 1272 } else { 1273 fp = fdp->fd_ofiles[fds->fd]; 1274 if (fp == NULL) { 1275 fds->revents = POLLNVAL; 1276 n++; 1277 } else { 1278 /* 1279 * Note: backend also returns POLLHUP and 1280 * POLLERR if appropriate. 1281 */ 1282 selfdalloc(td, fds); 1283 fds->revents = fo_poll(fp, fds->events, 1284 td->td_ucred, td); 1285 /* 1286 * POSIX requires POLLOUT to be never 1287 * set simultaneously with POLLHUP. 1288 */ 1289 if ((fds->revents & POLLHUP) != 0) 1290 fds->revents &= ~POLLOUT; 1291 1292 if (fds->revents != 0) 1293 n++; 1294 } 1295 } 1296 } 1297 FILEDESC_SUNLOCK(fdp); 1298 td->td_retval[0] = n; 1299 return (0); 1300 } 1301 1302 /* 1303 * OpenBSD poll system call. 1304 * 1305 * XXX this isn't quite a true representation.. OpenBSD uses select ops. 1306 */ 1307 #ifndef _SYS_SYSPROTO_H_ 1308 struct openbsd_poll_args { 1309 struct pollfd *fds; 1310 u_int nfds; 1311 int timeout; 1312 }; 1313 #endif 1314 int 1315 openbsd_poll(td, uap) 1316 register struct thread *td; 1317 register struct openbsd_poll_args *uap; 1318 { 1319 return (poll(td, (struct poll_args *)uap)); 1320 } 1321 1322 /* 1323 * XXX This was created specifically to support netncp and netsmb. This 1324 * allows the caller to specify a socket to wait for events on. It returns 1325 * 0 if any events matched and an error otherwise. There is no way to 1326 * determine which events fired. 1327 */ 1328 int 1329 selsocket(struct socket *so, int events, struct timeval *tvp, struct thread *td) 1330 { 1331 struct timeval atv, rtv, ttv; 1332 int error, timo; 1333 1334 if (tvp != NULL) { 1335 atv = *tvp; 1336 if (itimerfix(&atv)) 1337 return (EINVAL); 1338 getmicrouptime(&rtv); 1339 timevaladd(&atv, &rtv); 1340 } else { 1341 atv.tv_sec = 0; 1342 atv.tv_usec = 0; 1343 } 1344 1345 timo = 0; 1346 seltdinit(td); 1347 /* 1348 * Iterate until the timeout expires or the socket becomes ready. 1349 */ 1350 for (;;) { 1351 selfdalloc(td, NULL); 1352 error = sopoll(so, events, NULL, td); 1353 /* error here is actually the ready events. */ 1354 if (error) 1355 return (0); 1356 if (atv.tv_sec || atv.tv_usec) { 1357 getmicrouptime(&rtv); 1358 if (timevalcmp(&rtv, &atv, >=)) { 1359 seltdclear(td); 1360 return (EWOULDBLOCK); 1361 } 1362 ttv = atv; 1363 timevalsub(&ttv, &rtv); 1364 timo = ttv.tv_sec > 24 * 60 * 60 ? 1365 24 * 60 * 60 * hz : tvtohz(&ttv); 1366 } 1367 error = seltdwait(td, timo); 1368 seltdclear(td); 1369 if (error) 1370 break; 1371 } 1372 /* XXX Duplicates ncp/smb behavior. */ 1373 if (error == ERESTART) 1374 error = 0; 1375 return (error); 1376 } 1377 1378 /* 1379 * Preallocate two selfds associated with 'cookie'. Some fo_poll routines 1380 * have two select sets, one for read and another for write. 1381 */ 1382 static void 1383 selfdalloc(struct thread *td, void *cookie) 1384 { 1385 struct seltd *stp; 1386 1387 stp = td->td_sel; 1388 if (stp->st_free1 == NULL) 1389 stp->st_free1 = uma_zalloc(selfd_zone, M_WAITOK|M_ZERO); 1390 stp->st_free1->sf_td = stp; 1391 stp->st_free1->sf_cookie = cookie; 1392 if (stp->st_free2 == NULL) 1393 stp->st_free2 = uma_zalloc(selfd_zone, M_WAITOK|M_ZERO); 1394 stp->st_free2->sf_td = stp; 1395 stp->st_free2->sf_cookie = cookie; 1396 } 1397 1398 static void 1399 selfdfree(struct seltd *stp, struct selfd *sfp) 1400 { 1401 STAILQ_REMOVE(&stp->st_selq, sfp, selfd, sf_link); 1402 mtx_lock(sfp->sf_mtx); 1403 if (sfp->sf_si) 1404 TAILQ_REMOVE(&sfp->sf_si->si_tdlist, sfp, sf_threads); 1405 mtx_unlock(sfp->sf_mtx); 1406 uma_zfree(selfd_zone, sfp); 1407 } 1408 1409 /* 1410 * Record a select request. 1411 */ 1412 void 1413 selrecord(selector, sip) 1414 struct thread *selector; 1415 struct selinfo *sip; 1416 { 1417 struct selfd *sfp; 1418 struct seltd *stp; 1419 struct mtx *mtxp; 1420 1421 stp = selector->td_sel; 1422 /* 1423 * Don't record when doing a rescan. 1424 */ 1425 if (stp->st_flags & SELTD_RESCAN) 1426 return; 1427 /* 1428 * Grab one of the preallocated descriptors. 1429 */ 1430 sfp = NULL; 1431 if ((sfp = stp->st_free1) != NULL) 1432 stp->st_free1 = NULL; 1433 else if ((sfp = stp->st_free2) != NULL) 1434 stp->st_free2 = NULL; 1435 else 1436 panic("selrecord: No free selfd on selq"); 1437 mtxp = sip->si_mtx; 1438 if (mtxp == NULL) 1439 mtxp = mtx_pool_find(mtxpool_select, sip); 1440 /* 1441 * Initialize the sfp and queue it in the thread. 1442 */ 1443 sfp->sf_si = sip; 1444 sfp->sf_mtx = mtxp; 1445 STAILQ_INSERT_TAIL(&stp->st_selq, sfp, sf_link); 1446 /* 1447 * Now that we've locked the sip, check for initialization. 1448 */ 1449 mtx_lock(mtxp); 1450 if (sip->si_mtx == NULL) { 1451 sip->si_mtx = mtxp; 1452 TAILQ_INIT(&sip->si_tdlist); 1453 } 1454 /* 1455 * Add this thread to the list of selfds listening on this selinfo. 1456 */ 1457 TAILQ_INSERT_TAIL(&sip->si_tdlist, sfp, sf_threads); 1458 mtx_unlock(sip->si_mtx); 1459 } 1460 1461 /* Wake up a selecting thread. */ 1462 void 1463 selwakeup(sip) 1464 struct selinfo *sip; 1465 { 1466 doselwakeup(sip, -1); 1467 } 1468 1469 /* Wake up a selecting thread, and set its priority. */ 1470 void 1471 selwakeuppri(sip, pri) 1472 struct selinfo *sip; 1473 int pri; 1474 { 1475 doselwakeup(sip, pri); 1476 } 1477 1478 /* 1479 * Do a wakeup when a selectable event occurs. 1480 */ 1481 static void 1482 doselwakeup(sip, pri) 1483 struct selinfo *sip; 1484 int pri; 1485 { 1486 struct selfd *sfp; 1487 struct selfd *sfn; 1488 struct seltd *stp; 1489 1490 /* If it's not initialized there can't be any waiters. */ 1491 if (sip->si_mtx == NULL) 1492 return; 1493 /* 1494 * Locking the selinfo locks all selfds associated with it. 1495 */ 1496 mtx_lock(sip->si_mtx); 1497 TAILQ_FOREACH_SAFE(sfp, &sip->si_tdlist, sf_threads, sfn) { 1498 /* 1499 * Once we remove this sfp from the list and clear the 1500 * sf_si seltdclear will know to ignore this si. 1501 */ 1502 TAILQ_REMOVE(&sip->si_tdlist, sfp, sf_threads); 1503 sfp->sf_si = NULL; 1504 stp = sfp->sf_td; 1505 mtx_lock(&stp->st_mtx); 1506 stp->st_flags |= SELTD_PENDING; 1507 cv_broadcastpri(&stp->st_wait, pri); 1508 mtx_unlock(&stp->st_mtx); 1509 } 1510 mtx_unlock(sip->si_mtx); 1511 } 1512 1513 static void 1514 seltdinit(struct thread *td) 1515 { 1516 struct seltd *stp; 1517 1518 if ((stp = td->td_sel) != NULL) 1519 goto out; 1520 td->td_sel = stp = malloc(sizeof(*stp), M_SELECT, M_WAITOK|M_ZERO); 1521 mtx_init(&stp->st_mtx, "sellck", NULL, MTX_DEF); 1522 cv_init(&stp->st_wait, "select"); 1523 out: 1524 stp->st_flags = 0; 1525 STAILQ_INIT(&stp->st_selq); 1526 } 1527 1528 static int 1529 seltdwait(struct thread *td, int timo) 1530 { 1531 struct seltd *stp; 1532 int error; 1533 1534 stp = td->td_sel; 1535 /* 1536 * An event of interest may occur while we do not hold the seltd 1537 * locked so check the pending flag before we sleep. 1538 */ 1539 mtx_lock(&stp->st_mtx); 1540 /* 1541 * Any further calls to selrecord will be a rescan. 1542 */ 1543 stp->st_flags |= SELTD_RESCAN; 1544 if (stp->st_flags & SELTD_PENDING) { 1545 mtx_unlock(&stp->st_mtx); 1546 return (0); 1547 } 1548 if (timo > 0) 1549 error = cv_timedwait_sig(&stp->st_wait, &stp->st_mtx, timo); 1550 else 1551 error = cv_wait_sig(&stp->st_wait, &stp->st_mtx); 1552 mtx_unlock(&stp->st_mtx); 1553 1554 return (error); 1555 } 1556 1557 void 1558 seltdfini(struct thread *td) 1559 { 1560 struct seltd *stp; 1561 1562 stp = td->td_sel; 1563 if (stp == NULL) 1564 return; 1565 if (stp->st_free1) 1566 uma_zfree(selfd_zone, stp->st_free1); 1567 if (stp->st_free2) 1568 uma_zfree(selfd_zone, stp->st_free2); 1569 td->td_sel = NULL; 1570 free(stp, M_SELECT); 1571 } 1572 1573 /* 1574 * Remove the references to the thread from all of the objects we were 1575 * polling. 1576 */ 1577 static void 1578 seltdclear(struct thread *td) 1579 { 1580 struct seltd *stp; 1581 struct selfd *sfp; 1582 struct selfd *sfn; 1583 1584 stp = td->td_sel; 1585 STAILQ_FOREACH_SAFE(sfp, &stp->st_selq, sf_link, sfn) 1586 selfdfree(stp, sfp); 1587 stp->st_flags = 0; 1588 } 1589 1590 static void selectinit(void *); 1591 SYSINIT(select, SI_SUB_SYSCALLS, SI_ORDER_ANY, selectinit, NULL); 1592 static void 1593 selectinit(void *dummy __unused) 1594 { 1595 1596 selfd_zone = uma_zcreate("selfd", sizeof(struct selfd), NULL, NULL, 1597 NULL, NULL, UMA_ALIGN_PTR, 0); 1598 mtxpool_select = mtx_pool_create("select mtxpool", 128, MTX_DEF); 1599 } 1600