1 /* $NetBSD: kern_descrip.c,v 1.217 2011/09/25 13:40:37 chs Exp $ */ 2 3 /*- 4 * Copyright (c) 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 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, 1991, 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 * @(#)kern_descrip.c 8.8 (Berkeley) 2/14/95 66 */ 67 68 /* 69 * File descriptor management. 70 */ 71 72 #include <sys/cdefs.h> 73 __KERNEL_RCSID(0, "$NetBSD: kern_descrip.c,v 1.217 2011/09/25 13:40:37 chs Exp $"); 74 75 #include <sys/param.h> 76 #include <sys/systm.h> 77 #include <sys/filedesc.h> 78 #include <sys/kernel.h> 79 #include <sys/proc.h> 80 #include <sys/file.h> 81 #include <sys/socket.h> 82 #include <sys/socketvar.h> 83 #include <sys/stat.h> 84 #include <sys/ioctl.h> 85 #include <sys/fcntl.h> 86 #include <sys/pool.h> 87 #include <sys/unistd.h> 88 #include <sys/resourcevar.h> 89 #include <sys/conf.h> 90 #include <sys/event.h> 91 #include <sys/kauth.h> 92 #include <sys/atomic.h> 93 #include <sys/syscallargs.h> 94 #include <sys/cpu.h> 95 #include <sys/kmem.h> 96 #include <sys/vnode.h> 97 #include <sys/sysctl.h> 98 #include <sys/ktrace.h> 99 100 /* 101 * A list (head) of open files, counter, and lock protecting them. 102 */ 103 struct filelist filehead __cacheline_aligned; 104 static u_int nfiles __cacheline_aligned; 105 kmutex_t filelist_lock __cacheline_aligned; 106 107 static pool_cache_t filedesc_cache __read_mostly; 108 static pool_cache_t file_cache __read_mostly; 109 static pool_cache_t fdfile_cache __read_mostly; 110 111 static int file_ctor(void *, void *, int); 112 static void file_dtor(void *, void *); 113 static int fdfile_ctor(void *, void *, int); 114 static void fdfile_dtor(void *, void *); 115 static int filedesc_ctor(void *, void *, int); 116 static void filedesc_dtor(void *, void *); 117 static int filedescopen(dev_t, int, int, lwp_t *); 118 119 static int sysctl_kern_file(SYSCTLFN_PROTO); 120 static int sysctl_kern_file2(SYSCTLFN_PROTO); 121 static void fill_file(struct kinfo_file *, const file_t *, const fdfile_t *, 122 int, pid_t); 123 124 const struct cdevsw filedesc_cdevsw = { 125 filedescopen, noclose, noread, nowrite, noioctl, 126 nostop, notty, nopoll, nommap, nokqfilter, D_OTHER | D_MPSAFE, 127 }; 128 129 /* For ease of reading. */ 130 __strong_alias(fd_putvnode,fd_putfile) 131 __strong_alias(fd_putsock,fd_putfile) 132 133 /* 134 * Initialize the descriptor system. 135 */ 136 void 137 fd_sys_init(void) 138 { 139 static struct sysctllog *clog; 140 141 mutex_init(&filelist_lock, MUTEX_DEFAULT, IPL_NONE); 142 143 file_cache = pool_cache_init(sizeof(file_t), coherency_unit, 0, 144 0, "file", NULL, IPL_NONE, file_ctor, file_dtor, NULL); 145 KASSERT(file_cache != NULL); 146 147 fdfile_cache = pool_cache_init(sizeof(fdfile_t), coherency_unit, 0, 148 PR_LARGECACHE, "fdfile", NULL, IPL_NONE, fdfile_ctor, fdfile_dtor, 149 NULL); 150 KASSERT(fdfile_cache != NULL); 151 152 filedesc_cache = pool_cache_init(sizeof(filedesc_t), coherency_unit, 153 0, 0, "filedesc", NULL, IPL_NONE, filedesc_ctor, filedesc_dtor, 154 NULL); 155 KASSERT(filedesc_cache != NULL); 156 157 sysctl_createv(&clog, 0, NULL, NULL, 158 CTLFLAG_PERMANENT, CTLTYPE_NODE, "kern", NULL, 159 NULL, 0, NULL, 0, CTL_KERN, CTL_EOL); 160 sysctl_createv(&clog, 0, NULL, NULL, 161 CTLFLAG_PERMANENT, 162 CTLTYPE_STRUCT, "file", 163 SYSCTL_DESCR("System open file table"), 164 sysctl_kern_file, 0, NULL, 0, 165 CTL_KERN, KERN_FILE, CTL_EOL); 166 sysctl_createv(&clog, 0, NULL, NULL, 167 CTLFLAG_PERMANENT, 168 CTLTYPE_STRUCT, "file2", 169 SYSCTL_DESCR("System open file table"), 170 sysctl_kern_file2, 0, NULL, 0, 171 CTL_KERN, KERN_FILE2, CTL_EOL); 172 } 173 174 static bool 175 fd_isused(filedesc_t *fdp, unsigned fd) 176 { 177 u_int off = fd >> NDENTRYSHIFT; 178 179 KASSERT(fd < fdp->fd_dt->dt_nfiles); 180 181 return (fdp->fd_lomap[off] & (1 << (fd & NDENTRYMASK))) != 0; 182 } 183 184 /* 185 * Verify that the bitmaps match the descriptor table. 186 */ 187 static inline void 188 fd_checkmaps(filedesc_t *fdp) 189 { 190 #ifdef DEBUG 191 fdtab_t *dt; 192 u_int fd; 193 194 dt = fdp->fd_dt; 195 if (fdp->fd_refcnt == -1) { 196 /* 197 * fd_free tears down the table without maintaining its bitmap. 198 */ 199 return; 200 } 201 for (fd = 0; fd < dt->dt_nfiles; fd++) { 202 if (fd < NDFDFILE) { 203 KASSERT(dt->dt_ff[fd] == 204 (fdfile_t *)fdp->fd_dfdfile[fd]); 205 } 206 if (dt->dt_ff[fd] == NULL) { 207 KASSERT(!fd_isused(fdp, fd)); 208 } else if (dt->dt_ff[fd]->ff_file != NULL) { 209 KASSERT(fd_isused(fdp, fd)); 210 } 211 } 212 #endif 213 } 214 215 static int 216 fd_next_zero(filedesc_t *fdp, uint32_t *bitmap, int want, u_int bits) 217 { 218 int i, off, maxoff; 219 uint32_t sub; 220 221 KASSERT(mutex_owned(&fdp->fd_lock)); 222 223 fd_checkmaps(fdp); 224 225 if (want > bits) 226 return -1; 227 228 off = want >> NDENTRYSHIFT; 229 i = want & NDENTRYMASK; 230 if (i) { 231 sub = bitmap[off] | ((u_int)~0 >> (NDENTRIES - i)); 232 if (sub != ~0) 233 goto found; 234 off++; 235 } 236 237 maxoff = NDLOSLOTS(bits); 238 while (off < maxoff) { 239 if ((sub = bitmap[off]) != ~0) 240 goto found; 241 off++; 242 } 243 244 return -1; 245 246 found: 247 return (off << NDENTRYSHIFT) + ffs(~sub) - 1; 248 } 249 250 static int 251 fd_last_set(filedesc_t *fd, int last) 252 { 253 int off, i; 254 fdfile_t **ff = fd->fd_dt->dt_ff; 255 uint32_t *bitmap = fd->fd_lomap; 256 257 KASSERT(mutex_owned(&fd->fd_lock)); 258 259 fd_checkmaps(fd); 260 261 off = (last - 1) >> NDENTRYSHIFT; 262 263 while (off >= 0 && !bitmap[off]) 264 off--; 265 266 if (off < 0) 267 return -1; 268 269 i = ((off + 1) << NDENTRYSHIFT) - 1; 270 if (i >= last) 271 i = last - 1; 272 273 /* XXX should use bitmap */ 274 while (i > 0 && (ff[i] == NULL || !ff[i]->ff_allocated)) 275 i--; 276 277 return i; 278 } 279 280 static inline void 281 fd_used(filedesc_t *fdp, unsigned fd) 282 { 283 u_int off = fd >> NDENTRYSHIFT; 284 fdfile_t *ff; 285 286 ff = fdp->fd_dt->dt_ff[fd]; 287 288 KASSERT(mutex_owned(&fdp->fd_lock)); 289 KASSERT((fdp->fd_lomap[off] & (1 << (fd & NDENTRYMASK))) == 0); 290 KASSERT(ff != NULL); 291 KASSERT(ff->ff_file == NULL); 292 KASSERT(!ff->ff_allocated); 293 294 ff->ff_allocated = true; 295 fdp->fd_lomap[off] |= 1 << (fd & NDENTRYMASK); 296 if (__predict_false(fdp->fd_lomap[off] == ~0)) { 297 KASSERT((fdp->fd_himap[off >> NDENTRYSHIFT] & 298 (1 << (off & NDENTRYMASK))) == 0); 299 fdp->fd_himap[off >> NDENTRYSHIFT] |= 1 << (off & NDENTRYMASK); 300 } 301 302 if ((int)fd > fdp->fd_lastfile) { 303 fdp->fd_lastfile = fd; 304 } 305 306 fd_checkmaps(fdp); 307 } 308 309 static inline void 310 fd_unused(filedesc_t *fdp, unsigned fd) 311 { 312 u_int off = fd >> NDENTRYSHIFT; 313 fdfile_t *ff; 314 315 ff = fdp->fd_dt->dt_ff[fd]; 316 317 /* 318 * Don't assert the lock is held here, as we may be copying 319 * the table during exec() and it is not needed there. 320 * procfs and sysctl are locked out by proc::p_reflock. 321 * 322 * KASSERT(mutex_owned(&fdp->fd_lock)); 323 */ 324 KASSERT(ff != NULL); 325 KASSERT(ff->ff_file == NULL); 326 KASSERT(ff->ff_allocated); 327 328 if (fd < fdp->fd_freefile) { 329 fdp->fd_freefile = fd; 330 } 331 332 if (fdp->fd_lomap[off] == ~0) { 333 KASSERT((fdp->fd_himap[off >> NDENTRYSHIFT] & 334 (1 << (off & NDENTRYMASK))) != 0); 335 fdp->fd_himap[off >> NDENTRYSHIFT] &= 336 ~(1 << (off & NDENTRYMASK)); 337 } 338 KASSERT((fdp->fd_lomap[off] & (1 << (fd & NDENTRYMASK))) != 0); 339 fdp->fd_lomap[off] &= ~(1 << (fd & NDENTRYMASK)); 340 ff->ff_allocated = false; 341 342 KASSERT(fd <= fdp->fd_lastfile); 343 if (fd == fdp->fd_lastfile) { 344 fdp->fd_lastfile = fd_last_set(fdp, fd); 345 } 346 fd_checkmaps(fdp); 347 } 348 349 /* 350 * Look up the file structure corresponding to a file descriptor 351 * and return the file, holding a reference on the descriptor. 352 */ 353 file_t * 354 fd_getfile(unsigned fd) 355 { 356 filedesc_t *fdp; 357 fdfile_t *ff; 358 file_t *fp; 359 fdtab_t *dt; 360 361 /* 362 * Look up the fdfile structure representing this descriptor. 363 * We are doing this unlocked. See fd_tryexpand(). 364 */ 365 fdp = curlwp->l_fd; 366 dt = fdp->fd_dt; 367 if (__predict_false(fd >= dt->dt_nfiles)) { 368 return NULL; 369 } 370 ff = dt->dt_ff[fd]; 371 KASSERT(fd >= NDFDFILE || ff == (fdfile_t *)fdp->fd_dfdfile[fd]); 372 if (__predict_false(ff == NULL)) { 373 return NULL; 374 } 375 376 /* Now get a reference to the descriptor. */ 377 if (fdp->fd_refcnt == 1) { 378 /* 379 * Single threaded: don't need to worry about concurrent 380 * access (other than earlier calls to kqueue, which may 381 * hold a reference to the descriptor). 382 */ 383 ff->ff_refcnt++; 384 } else { 385 /* 386 * Multi threaded: issue a memory barrier to ensure that we 387 * acquire the file pointer _after_ adding a reference. If 388 * no memory barrier, we could fetch a stale pointer. 389 */ 390 atomic_inc_uint(&ff->ff_refcnt); 391 #ifndef __HAVE_ATOMIC_AS_MEMBAR 392 membar_enter(); 393 #endif 394 } 395 396 /* 397 * If the file is not open or is being closed then put the 398 * reference back. 399 */ 400 fp = ff->ff_file; 401 if (__predict_true(fp != NULL)) { 402 return fp; 403 } 404 fd_putfile(fd); 405 return NULL; 406 } 407 408 /* 409 * Release a reference to a file descriptor acquired with fd_getfile(). 410 */ 411 void 412 fd_putfile(unsigned fd) 413 { 414 filedesc_t *fdp; 415 fdfile_t *ff; 416 u_int u, v; 417 418 fdp = curlwp->l_fd; 419 ff = fdp->fd_dt->dt_ff[fd]; 420 421 KASSERT(fd < fdp->fd_dt->dt_nfiles); 422 KASSERT(ff != NULL); 423 KASSERT((ff->ff_refcnt & FR_MASK) > 0); 424 KASSERT(fd >= NDFDFILE || ff == (fdfile_t *)fdp->fd_dfdfile[fd]); 425 426 if (fdp->fd_refcnt == 1) { 427 /* 428 * Single threaded: don't need to worry about concurrent 429 * access (other than earlier calls to kqueue, which may 430 * hold a reference to the descriptor). 431 */ 432 if (__predict_false((ff->ff_refcnt & FR_CLOSING) != 0)) { 433 fd_close(fd); 434 return; 435 } 436 ff->ff_refcnt--; 437 return; 438 } 439 440 /* 441 * Ensure that any use of the file is complete and globally 442 * visible before dropping the final reference. If no membar, 443 * the current CPU could still access memory associated with 444 * the file after it has been freed or recycled by another 445 * CPU. 446 */ 447 #ifndef __HAVE_ATOMIC_AS_MEMBAR 448 membar_exit(); 449 #endif 450 451 /* 452 * Be optimistic and start out with the assumption that no other 453 * threads are trying to close the descriptor. If the CAS fails, 454 * we lost a race and/or it's being closed. 455 */ 456 for (u = ff->ff_refcnt & FR_MASK;; u = v) { 457 v = atomic_cas_uint(&ff->ff_refcnt, u, u - 1); 458 if (__predict_true(u == v)) { 459 return; 460 } 461 if (__predict_false((v & FR_CLOSING) != 0)) { 462 break; 463 } 464 } 465 466 /* Another thread is waiting to close the file: join it. */ 467 (void)fd_close(fd); 468 } 469 470 /* 471 * Convenience wrapper around fd_getfile() that returns reference 472 * to a vnode. 473 */ 474 int 475 fd_getvnode(unsigned fd, file_t **fpp) 476 { 477 vnode_t *vp; 478 file_t *fp; 479 480 fp = fd_getfile(fd); 481 if (__predict_false(fp == NULL)) { 482 return EBADF; 483 } 484 if (__predict_false(fp->f_type != DTYPE_VNODE)) { 485 fd_putfile(fd); 486 return EINVAL; 487 } 488 vp = fp->f_data; 489 if (__predict_false(vp->v_type == VBAD)) { 490 /* XXX Is this case really necessary? */ 491 fd_putfile(fd); 492 return EBADF; 493 } 494 *fpp = fp; 495 return 0; 496 } 497 498 /* 499 * Convenience wrapper around fd_getfile() that returns reference 500 * to a socket. 501 */ 502 int 503 fd_getsock(unsigned fd, struct socket **sop) 504 { 505 file_t *fp; 506 507 fp = fd_getfile(fd); 508 if (__predict_false(fp == NULL)) { 509 return EBADF; 510 } 511 if (__predict_false(fp->f_type != DTYPE_SOCKET)) { 512 fd_putfile(fd); 513 return ENOTSOCK; 514 } 515 *sop = fp->f_data; 516 return 0; 517 } 518 519 /* 520 * Look up the file structure corresponding to a file descriptor 521 * and return it with a reference held on the file, not the 522 * descriptor. 523 * 524 * This is heavyweight and only used when accessing descriptors 525 * from a foreign process. The caller must ensure that `p' does 526 * not exit or fork across this call. 527 * 528 * To release the file (not descriptor) reference, use closef(). 529 */ 530 file_t * 531 fd_getfile2(proc_t *p, unsigned fd) 532 { 533 filedesc_t *fdp; 534 fdfile_t *ff; 535 file_t *fp; 536 fdtab_t *dt; 537 538 fdp = p->p_fd; 539 mutex_enter(&fdp->fd_lock); 540 dt = fdp->fd_dt; 541 if (fd >= dt->dt_nfiles) { 542 mutex_exit(&fdp->fd_lock); 543 return NULL; 544 } 545 if ((ff = dt->dt_ff[fd]) == NULL) { 546 mutex_exit(&fdp->fd_lock); 547 return NULL; 548 } 549 if ((fp = ff->ff_file) == NULL) { 550 mutex_exit(&fdp->fd_lock); 551 return NULL; 552 } 553 mutex_enter(&fp->f_lock); 554 fp->f_count++; 555 mutex_exit(&fp->f_lock); 556 mutex_exit(&fdp->fd_lock); 557 558 return fp; 559 } 560 561 /* 562 * Internal form of close. Must be called with a reference to the 563 * descriptor, and will drop the reference. When all descriptor 564 * references are dropped, releases the descriptor slot and a single 565 * reference to the file structure. 566 */ 567 int 568 fd_close(unsigned fd) 569 { 570 struct flock lf; 571 filedesc_t *fdp; 572 fdfile_t *ff; 573 file_t *fp; 574 proc_t *p; 575 lwp_t *l; 576 u_int refcnt; 577 578 l = curlwp; 579 p = l->l_proc; 580 fdp = l->l_fd; 581 ff = fdp->fd_dt->dt_ff[fd]; 582 583 KASSERT(fd >= NDFDFILE || ff == (fdfile_t *)fdp->fd_dfdfile[fd]); 584 585 mutex_enter(&fdp->fd_lock); 586 KASSERT((ff->ff_refcnt & FR_MASK) > 0); 587 if (__predict_false(ff->ff_file == NULL)) { 588 /* 589 * Another user of the file is already closing, and is 590 * waiting for other users of the file to drain. Release 591 * our reference, and wake up the closer. 592 */ 593 atomic_dec_uint(&ff->ff_refcnt); 594 cv_broadcast(&ff->ff_closing); 595 mutex_exit(&fdp->fd_lock); 596 597 /* 598 * An application error, so pretend that the descriptor 599 * was already closed. We can't safely wait for it to 600 * be closed without potentially deadlocking. 601 */ 602 return (EBADF); 603 } 604 KASSERT((ff->ff_refcnt & FR_CLOSING) == 0); 605 606 /* 607 * There may be multiple users of this file within the process. 608 * Notify existing and new users that the file is closing. This 609 * will prevent them from adding additional uses to this file 610 * while we are closing it. 611 */ 612 fp = ff->ff_file; 613 ff->ff_file = NULL; 614 ff->ff_exclose = false; 615 616 /* 617 * We expect the caller to hold a descriptor reference - drop it. 618 * The reference count may increase beyond zero at this point due 619 * to an erroneous descriptor reference by an application, but 620 * fd_getfile() will notice that the file is being closed and drop 621 * the reference again. 622 */ 623 if (fdp->fd_refcnt == 1) { 624 /* Single threaded. */ 625 refcnt = --(ff->ff_refcnt); 626 } else { 627 /* Multi threaded. */ 628 #ifndef __HAVE_ATOMIC_AS_MEMBAR 629 membar_producer(); 630 #endif 631 refcnt = atomic_dec_uint_nv(&ff->ff_refcnt); 632 } 633 if (__predict_false(refcnt != 0)) { 634 /* 635 * Wait for other references to drain. This is typically 636 * an application error - the descriptor is being closed 637 * while still in use. 638 * (Or just a threaded application trying to unblock its 639 * thread that sleeps in (say) accept()). 640 */ 641 atomic_or_uint(&ff->ff_refcnt, FR_CLOSING); 642 643 /* 644 * Remove any knotes attached to the file. A knote 645 * attached to the descriptor can hold references on it. 646 */ 647 mutex_exit(&fdp->fd_lock); 648 if (!SLIST_EMPTY(&ff->ff_knlist)) { 649 knote_fdclose(fd); 650 } 651 652 /* 653 * Since the file system code doesn't know which fd 654 * each request came from (think dup()), we have to 655 * ask it to return ERESTART for any long-term blocks. 656 * The re-entry through read/write/etc will detect the 657 * closed fd and return EBAFD. 658 * Blocked partial writes may return a short length. 659 */ 660 (*fp->f_ops->fo_restart)(fp); 661 mutex_enter(&fdp->fd_lock); 662 663 /* 664 * We need to see the count drop to zero at least once, 665 * in order to ensure that all pre-existing references 666 * have been drained. New references past this point are 667 * of no interest. 668 * XXX (dsl) this may need to call fo_restart() after a 669 * timeout to guarantee that all the system calls exit. 670 */ 671 while ((ff->ff_refcnt & FR_MASK) != 0) { 672 cv_wait(&ff->ff_closing, &fdp->fd_lock); 673 } 674 atomic_and_uint(&ff->ff_refcnt, ~FR_CLOSING); 675 } else { 676 /* If no references, there must be no knotes. */ 677 KASSERT(SLIST_EMPTY(&ff->ff_knlist)); 678 } 679 680 /* 681 * POSIX record locking dictates that any close releases ALL 682 * locks owned by this process. This is handled by setting 683 * a flag in the unlock to free ONLY locks obeying POSIX 684 * semantics, and not to free BSD-style file locks. 685 * If the descriptor was in a message, POSIX-style locks 686 * aren't passed with the descriptor. 687 */ 688 if (__predict_false((p->p_flag & PK_ADVLOCK) != 0 && 689 fp->f_type == DTYPE_VNODE)) { 690 lf.l_whence = SEEK_SET; 691 lf.l_start = 0; 692 lf.l_len = 0; 693 lf.l_type = F_UNLCK; 694 mutex_exit(&fdp->fd_lock); 695 (void)VOP_ADVLOCK(fp->f_data, p, F_UNLCK, &lf, F_POSIX); 696 mutex_enter(&fdp->fd_lock); 697 } 698 699 /* Free descriptor slot. */ 700 fd_unused(fdp, fd); 701 mutex_exit(&fdp->fd_lock); 702 703 /* Now drop reference to the file itself. */ 704 return closef(fp); 705 } 706 707 /* 708 * Duplicate a file descriptor. 709 */ 710 int 711 fd_dup(file_t *fp, int minfd, int *newp, bool exclose) 712 { 713 proc_t *p = curproc; 714 int error; 715 716 while ((error = fd_alloc(p, minfd, newp)) != 0) { 717 if (error != ENOSPC) { 718 return error; 719 } 720 fd_tryexpand(p); 721 } 722 723 curlwp->l_fd->fd_dt->dt_ff[*newp]->ff_exclose = exclose; 724 fd_affix(p, fp, *newp); 725 return 0; 726 } 727 728 /* 729 * dup2 operation. 730 */ 731 int 732 fd_dup2(file_t *fp, unsigned new, int flags) 733 { 734 filedesc_t *fdp = curlwp->l_fd; 735 fdfile_t *ff; 736 fdtab_t *dt; 737 738 if (flags & ~(O_CLOEXEC|O_NONBLOCK)) 739 return EINVAL; 740 /* 741 * Ensure there are enough slots in the descriptor table, 742 * and allocate an fdfile_t up front in case we need it. 743 */ 744 while (new >= fdp->fd_dt->dt_nfiles) { 745 fd_tryexpand(curproc); 746 } 747 ff = pool_cache_get(fdfile_cache, PR_WAITOK); 748 749 /* 750 * If there is already a file open, close it. If the file is 751 * half open, wait for it to be constructed before closing it. 752 * XXX Potential for deadlock here? 753 */ 754 mutex_enter(&fdp->fd_lock); 755 while (fd_isused(fdp, new)) { 756 mutex_exit(&fdp->fd_lock); 757 if (fd_getfile(new) != NULL) { 758 (void)fd_close(new); 759 } else { 760 /* 761 * Crummy, but unlikely to happen. 762 * Can occur if we interrupt another 763 * thread while it is opening a file. 764 */ 765 kpause("dup2", false, 1, NULL); 766 } 767 mutex_enter(&fdp->fd_lock); 768 } 769 dt = fdp->fd_dt; 770 if (dt->dt_ff[new] == NULL) { 771 KASSERT(new >= NDFDFILE); 772 dt->dt_ff[new] = ff; 773 ff = NULL; 774 } 775 fd_used(fdp, new); 776 mutex_exit(&fdp->fd_lock); 777 778 dt->dt_ff[new]->ff_exclose = (flags & O_CLOEXEC) != 0; 779 fp->f_flag |= flags & FNONBLOCK; 780 /* Slot is now allocated. Insert copy of the file. */ 781 fd_affix(curproc, fp, new); 782 if (ff != NULL) { 783 pool_cache_put(fdfile_cache, ff); 784 } 785 return 0; 786 } 787 788 /* 789 * Drop reference to a file structure. 790 */ 791 int 792 closef(file_t *fp) 793 { 794 struct flock lf; 795 int error; 796 797 /* 798 * Drop reference. If referenced elsewhere it's still open 799 * and we have nothing more to do. 800 */ 801 mutex_enter(&fp->f_lock); 802 KASSERT(fp->f_count > 0); 803 if (--fp->f_count > 0) { 804 mutex_exit(&fp->f_lock); 805 return 0; 806 } 807 KASSERT(fp->f_count == 0); 808 mutex_exit(&fp->f_lock); 809 810 /* We held the last reference - release locks, close and free. */ 811 if ((fp->f_flag & FHASLOCK) && fp->f_type == DTYPE_VNODE) { 812 lf.l_whence = SEEK_SET; 813 lf.l_start = 0; 814 lf.l_len = 0; 815 lf.l_type = F_UNLCK; 816 (void)VOP_ADVLOCK(fp->f_data, fp, F_UNLCK, &lf, F_FLOCK); 817 } 818 if (fp->f_ops != NULL) { 819 error = (*fp->f_ops->fo_close)(fp); 820 } else { 821 error = 0; 822 } 823 KASSERT(fp->f_count == 0); 824 KASSERT(fp->f_cred != NULL); 825 pool_cache_put(file_cache, fp); 826 827 return error; 828 } 829 830 /* 831 * Allocate a file descriptor for the process. 832 */ 833 int 834 fd_alloc(proc_t *p, int want, int *result) 835 { 836 filedesc_t *fdp = p->p_fd; 837 int i, lim, last, error; 838 u_int off, new; 839 fdtab_t *dt; 840 841 KASSERT(p == curproc || p == &proc0); 842 843 /* 844 * Search for a free descriptor starting at the higher 845 * of want or fd_freefile. 846 */ 847 mutex_enter(&fdp->fd_lock); 848 fd_checkmaps(fdp); 849 dt = fdp->fd_dt; 850 KASSERT(dt->dt_ff[0] == (fdfile_t *)fdp->fd_dfdfile[0]); 851 lim = min((int)p->p_rlimit[RLIMIT_NOFILE].rlim_cur, maxfiles); 852 last = min(dt->dt_nfiles, lim); 853 for (;;) { 854 if ((i = want) < fdp->fd_freefile) 855 i = fdp->fd_freefile; 856 off = i >> NDENTRYSHIFT; 857 new = fd_next_zero(fdp, fdp->fd_himap, off, 858 (last + NDENTRIES - 1) >> NDENTRYSHIFT); 859 if (new == -1) 860 break; 861 i = fd_next_zero(fdp, &fdp->fd_lomap[new], 862 new > off ? 0 : i & NDENTRYMASK, NDENTRIES); 863 if (i == -1) { 864 /* 865 * Free file descriptor in this block was 866 * below want, try again with higher want. 867 */ 868 want = (new + 1) << NDENTRYSHIFT; 869 continue; 870 } 871 i += (new << NDENTRYSHIFT); 872 if (i >= last) { 873 break; 874 } 875 if (dt->dt_ff[i] == NULL) { 876 KASSERT(i >= NDFDFILE); 877 dt->dt_ff[i] = pool_cache_get(fdfile_cache, PR_WAITOK); 878 } 879 KASSERT(dt->dt_ff[i]->ff_file == NULL); 880 fd_used(fdp, i); 881 if (want <= fdp->fd_freefile) { 882 fdp->fd_freefile = i; 883 } 884 *result = i; 885 KASSERT(i >= NDFDFILE || 886 dt->dt_ff[i] == (fdfile_t *)fdp->fd_dfdfile[i]); 887 fd_checkmaps(fdp); 888 mutex_exit(&fdp->fd_lock); 889 return 0; 890 } 891 892 /* No space in current array. Let the caller expand and retry. */ 893 error = (dt->dt_nfiles >= lim) ? EMFILE : ENOSPC; 894 mutex_exit(&fdp->fd_lock); 895 return error; 896 } 897 898 /* 899 * Allocate memory for a descriptor table. 900 */ 901 static fdtab_t * 902 fd_dtab_alloc(int n) 903 { 904 fdtab_t *dt; 905 size_t sz; 906 907 KASSERT(n > NDFILE); 908 909 sz = sizeof(*dt) + (n - NDFILE) * sizeof(dt->dt_ff[0]); 910 dt = kmem_alloc(sz, KM_SLEEP); 911 #ifdef DIAGNOSTIC 912 memset(dt, 0xff, sz); 913 #endif 914 dt->dt_nfiles = n; 915 dt->dt_link = NULL; 916 return dt; 917 } 918 919 /* 920 * Free a descriptor table, and all tables linked for deferred free. 921 */ 922 static void 923 fd_dtab_free(fdtab_t *dt) 924 { 925 fdtab_t *next; 926 size_t sz; 927 928 do { 929 next = dt->dt_link; 930 KASSERT(dt->dt_nfiles > NDFILE); 931 sz = sizeof(*dt) + 932 (dt->dt_nfiles - NDFILE) * sizeof(dt->dt_ff[0]); 933 #ifdef DIAGNOSTIC 934 memset(dt, 0xff, sz); 935 #endif 936 kmem_free(dt, sz); 937 dt = next; 938 } while (dt != NULL); 939 } 940 941 /* 942 * Allocate descriptor bitmap. 943 */ 944 static void 945 fd_map_alloc(int n, uint32_t **lo, uint32_t **hi) 946 { 947 uint8_t *ptr; 948 size_t szlo, szhi; 949 950 KASSERT(n > NDENTRIES); 951 952 szlo = NDLOSLOTS(n) * sizeof(uint32_t); 953 szhi = NDHISLOTS(n) * sizeof(uint32_t); 954 ptr = kmem_alloc(szlo + szhi, KM_SLEEP); 955 *lo = (uint32_t *)ptr; 956 *hi = (uint32_t *)(ptr + szlo); 957 } 958 959 /* 960 * Free descriptor bitmap. 961 */ 962 static void 963 fd_map_free(int n, uint32_t *lo, uint32_t *hi) 964 { 965 size_t szlo, szhi; 966 967 KASSERT(n > NDENTRIES); 968 969 szlo = NDLOSLOTS(n) * sizeof(uint32_t); 970 szhi = NDHISLOTS(n) * sizeof(uint32_t); 971 KASSERT(hi == (uint32_t *)((uint8_t *)lo + szlo)); 972 kmem_free(lo, szlo + szhi); 973 } 974 975 /* 976 * Expand a process' descriptor table. 977 */ 978 void 979 fd_tryexpand(proc_t *p) 980 { 981 filedesc_t *fdp; 982 int i, numfiles, oldnfiles; 983 fdtab_t *newdt, *dt; 984 uint32_t *newhimap, *newlomap; 985 986 KASSERT(p == curproc || p == &proc0); 987 988 fdp = p->p_fd; 989 newhimap = NULL; 990 newlomap = NULL; 991 oldnfiles = fdp->fd_dt->dt_nfiles; 992 993 if (oldnfiles < NDEXTENT) 994 numfiles = NDEXTENT; 995 else 996 numfiles = 2 * oldnfiles; 997 998 newdt = fd_dtab_alloc(numfiles); 999 if (NDHISLOTS(numfiles) > NDHISLOTS(oldnfiles)) { 1000 fd_map_alloc(numfiles, &newlomap, &newhimap); 1001 } 1002 1003 mutex_enter(&fdp->fd_lock); 1004 dt = fdp->fd_dt; 1005 KASSERT(dt->dt_ff[0] == (fdfile_t *)fdp->fd_dfdfile[0]); 1006 if (dt->dt_nfiles != oldnfiles) { 1007 /* fdp changed; caller must retry */ 1008 mutex_exit(&fdp->fd_lock); 1009 fd_dtab_free(newdt); 1010 if (NDHISLOTS(numfiles) > NDHISLOTS(oldnfiles)) { 1011 fd_map_free(numfiles, newlomap, newhimap); 1012 } 1013 return; 1014 } 1015 1016 /* Copy the existing descriptor table and zero the new portion. */ 1017 i = sizeof(fdfile_t *) * oldnfiles; 1018 memcpy(newdt->dt_ff, dt->dt_ff, i); 1019 memset((uint8_t *)newdt->dt_ff + i, 0, 1020 numfiles * sizeof(fdfile_t *) - i); 1021 1022 /* 1023 * Link old descriptor array into list to be discarded. We defer 1024 * freeing until the last reference to the descriptor table goes 1025 * away (usually process exit). This allows us to do lockless 1026 * lookups in fd_getfile(). 1027 */ 1028 if (oldnfiles > NDFILE) { 1029 if (fdp->fd_refcnt > 1) { 1030 newdt->dt_link = dt; 1031 } else { 1032 fd_dtab_free(dt); 1033 } 1034 } 1035 1036 if (NDHISLOTS(numfiles) > NDHISLOTS(oldnfiles)) { 1037 i = NDHISLOTS(oldnfiles) * sizeof(uint32_t); 1038 memcpy(newhimap, fdp->fd_himap, i); 1039 memset((uint8_t *)newhimap + i, 0, 1040 NDHISLOTS(numfiles) * sizeof(uint32_t) - i); 1041 1042 i = NDLOSLOTS(oldnfiles) * sizeof(uint32_t); 1043 memcpy(newlomap, fdp->fd_lomap, i); 1044 memset((uint8_t *)newlomap + i, 0, 1045 NDLOSLOTS(numfiles) * sizeof(uint32_t) - i); 1046 1047 if (NDHISLOTS(oldnfiles) > NDHISLOTS(NDFILE)) { 1048 fd_map_free(oldnfiles, fdp->fd_lomap, fdp->fd_himap); 1049 } 1050 fdp->fd_himap = newhimap; 1051 fdp->fd_lomap = newlomap; 1052 } 1053 1054 /* 1055 * All other modifications must become globally visible before 1056 * the change to fd_dt. See fd_getfile(). 1057 */ 1058 membar_producer(); 1059 fdp->fd_dt = newdt; 1060 KASSERT(newdt->dt_ff[0] == (fdfile_t *)fdp->fd_dfdfile[0]); 1061 fd_checkmaps(fdp); 1062 mutex_exit(&fdp->fd_lock); 1063 } 1064 1065 /* 1066 * Create a new open file structure and allocate a file descriptor 1067 * for the current process. 1068 */ 1069 int 1070 fd_allocfile(file_t **resultfp, int *resultfd) 1071 { 1072 proc_t *p = curproc; 1073 kauth_cred_t cred; 1074 file_t *fp; 1075 int error; 1076 1077 while ((error = fd_alloc(p, 0, resultfd)) != 0) { 1078 if (error != ENOSPC) { 1079 return error; 1080 } 1081 fd_tryexpand(p); 1082 } 1083 1084 fp = pool_cache_get(file_cache, PR_WAITOK); 1085 if (fp == NULL) { 1086 fd_abort(p, NULL, *resultfd); 1087 return ENFILE; 1088 } 1089 KASSERT(fp->f_count == 0); 1090 KASSERT(fp->f_msgcount == 0); 1091 KASSERT(fp->f_unpcount == 0); 1092 1093 /* Replace cached credentials if not what we need. */ 1094 cred = curlwp->l_cred; 1095 if (__predict_false(cred != fp->f_cred)) { 1096 kauth_cred_free(fp->f_cred); 1097 kauth_cred_hold(cred); 1098 fp->f_cred = cred; 1099 } 1100 1101 /* 1102 * Don't allow recycled files to be scanned. 1103 * See uipc_usrreq.c. 1104 */ 1105 if (__predict_false((fp->f_flag & FSCAN) != 0)) { 1106 mutex_enter(&fp->f_lock); 1107 atomic_and_uint(&fp->f_flag, ~FSCAN); 1108 mutex_exit(&fp->f_lock); 1109 } 1110 1111 fp->f_advice = 0; 1112 fp->f_offset = 0; 1113 *resultfp = fp; 1114 1115 return 0; 1116 } 1117 1118 /* 1119 * Successful creation of a new descriptor: make visible to the process. 1120 */ 1121 void 1122 fd_affix(proc_t *p, file_t *fp, unsigned fd) 1123 { 1124 fdfile_t *ff; 1125 filedesc_t *fdp; 1126 1127 KASSERT(p == curproc || p == &proc0); 1128 1129 /* Add a reference to the file structure. */ 1130 mutex_enter(&fp->f_lock); 1131 fp->f_count++; 1132 mutex_exit(&fp->f_lock); 1133 1134 /* 1135 * Insert the new file into the descriptor slot. 1136 * 1137 * The memory barriers provided by lock activity in this routine 1138 * ensure that any updates to the file structure become globally 1139 * visible before the file becomes visible to other LWPs in the 1140 * current process. 1141 */ 1142 fdp = p->p_fd; 1143 ff = fdp->fd_dt->dt_ff[fd]; 1144 1145 KASSERT(ff != NULL); 1146 KASSERT(ff->ff_file == NULL); 1147 KASSERT(ff->ff_allocated); 1148 KASSERT(fd_isused(fdp, fd)); 1149 KASSERT(fd >= NDFDFILE || ff == (fdfile_t *)fdp->fd_dfdfile[fd]); 1150 1151 /* No need to lock in order to make file initially visible. */ 1152 ff->ff_file = fp; 1153 } 1154 1155 /* 1156 * Abort creation of a new descriptor: free descriptor slot and file. 1157 */ 1158 void 1159 fd_abort(proc_t *p, file_t *fp, unsigned fd) 1160 { 1161 filedesc_t *fdp; 1162 fdfile_t *ff; 1163 1164 KASSERT(p == curproc || p == &proc0); 1165 1166 fdp = p->p_fd; 1167 ff = fdp->fd_dt->dt_ff[fd]; 1168 1169 KASSERT(fd >= NDFDFILE || ff == (fdfile_t *)fdp->fd_dfdfile[fd]); 1170 1171 mutex_enter(&fdp->fd_lock); 1172 KASSERT(fd_isused(fdp, fd)); 1173 fd_unused(fdp, fd); 1174 mutex_exit(&fdp->fd_lock); 1175 1176 if (fp != NULL) { 1177 KASSERT(fp->f_count == 0); 1178 KASSERT(fp->f_cred != NULL); 1179 pool_cache_put(file_cache, fp); 1180 } 1181 } 1182 1183 static int 1184 file_ctor(void *arg, void *obj, int flags) 1185 { 1186 file_t *fp = obj; 1187 1188 memset(fp, 0, sizeof(*fp)); 1189 1190 mutex_enter(&filelist_lock); 1191 if (__predict_false(nfiles >= maxfiles)) { 1192 mutex_exit(&filelist_lock); 1193 tablefull("file", "increase kern.maxfiles or MAXFILES"); 1194 return ENFILE; 1195 } 1196 nfiles++; 1197 LIST_INSERT_HEAD(&filehead, fp, f_list); 1198 mutex_init(&fp->f_lock, MUTEX_DEFAULT, IPL_NONE); 1199 fp->f_cred = curlwp->l_cred; 1200 kauth_cred_hold(fp->f_cred); 1201 mutex_exit(&filelist_lock); 1202 1203 return 0; 1204 } 1205 1206 static void 1207 file_dtor(void *arg, void *obj) 1208 { 1209 file_t *fp = obj; 1210 1211 mutex_enter(&filelist_lock); 1212 nfiles--; 1213 LIST_REMOVE(fp, f_list); 1214 mutex_exit(&filelist_lock); 1215 1216 kauth_cred_free(fp->f_cred); 1217 mutex_destroy(&fp->f_lock); 1218 } 1219 1220 static int 1221 fdfile_ctor(void *arg, void *obj, int flags) 1222 { 1223 fdfile_t *ff = obj; 1224 1225 memset(ff, 0, sizeof(*ff)); 1226 cv_init(&ff->ff_closing, "fdclose"); 1227 1228 return 0; 1229 } 1230 1231 static void 1232 fdfile_dtor(void *arg, void *obj) 1233 { 1234 fdfile_t *ff = obj; 1235 1236 cv_destroy(&ff->ff_closing); 1237 } 1238 1239 file_t * 1240 fgetdummy(void) 1241 { 1242 file_t *fp; 1243 1244 fp = kmem_zalloc(sizeof(*fp), KM_SLEEP); 1245 if (fp != NULL) { 1246 mutex_init(&fp->f_lock, MUTEX_DEFAULT, IPL_NONE); 1247 } 1248 return fp; 1249 } 1250 1251 void 1252 fputdummy(file_t *fp) 1253 { 1254 1255 mutex_destroy(&fp->f_lock); 1256 kmem_free(fp, sizeof(*fp)); 1257 } 1258 1259 /* 1260 * Create an initial filedesc structure. 1261 */ 1262 filedesc_t * 1263 fd_init(filedesc_t *fdp) 1264 { 1265 #ifdef DIAGNOSTIC 1266 unsigned fd; 1267 #endif 1268 1269 if (__predict_true(fdp == NULL)) { 1270 fdp = pool_cache_get(filedesc_cache, PR_WAITOK); 1271 } else { 1272 KASSERT(fdp == &filedesc0); 1273 filedesc_ctor(NULL, fdp, PR_WAITOK); 1274 } 1275 1276 #ifdef DIAGNOSTIC 1277 KASSERT(fdp->fd_lastfile == -1); 1278 KASSERT(fdp->fd_lastkqfile == -1); 1279 KASSERT(fdp->fd_knhash == NULL); 1280 KASSERT(fdp->fd_freefile == 0); 1281 KASSERT(fdp->fd_exclose == false); 1282 KASSERT(fdp->fd_dt == &fdp->fd_dtbuiltin); 1283 KASSERT(fdp->fd_dtbuiltin.dt_nfiles == NDFILE); 1284 for (fd = 0; fd < NDFDFILE; fd++) { 1285 KASSERT(fdp->fd_dtbuiltin.dt_ff[fd] == 1286 (fdfile_t *)fdp->fd_dfdfile[fd]); 1287 } 1288 for (fd = NDFDFILE; fd < NDFILE; fd++) { 1289 KASSERT(fdp->fd_dtbuiltin.dt_ff[fd] == NULL); 1290 } 1291 KASSERT(fdp->fd_himap == fdp->fd_dhimap); 1292 KASSERT(fdp->fd_lomap == fdp->fd_dlomap); 1293 #endif /* DIAGNOSTIC */ 1294 1295 fdp->fd_refcnt = 1; 1296 fd_checkmaps(fdp); 1297 1298 return fdp; 1299 } 1300 1301 /* 1302 * Initialize a file descriptor table. 1303 */ 1304 static int 1305 filedesc_ctor(void *arg, void *obj, int flag) 1306 { 1307 filedesc_t *fdp = obj; 1308 fdfile_t **ffp; 1309 int i; 1310 1311 memset(fdp, 0, sizeof(*fdp)); 1312 mutex_init(&fdp->fd_lock, MUTEX_DEFAULT, IPL_NONE); 1313 fdp->fd_lastfile = -1; 1314 fdp->fd_lastkqfile = -1; 1315 fdp->fd_dt = &fdp->fd_dtbuiltin; 1316 fdp->fd_dtbuiltin.dt_nfiles = NDFILE; 1317 fdp->fd_himap = fdp->fd_dhimap; 1318 fdp->fd_lomap = fdp->fd_dlomap; 1319 1320 CTASSERT(sizeof(fdp->fd_dfdfile[0]) >= sizeof(fdfile_t)); 1321 for (i = 0, ffp = fdp->fd_dt->dt_ff; i < NDFDFILE; i++, ffp++) { 1322 *ffp = (fdfile_t *)fdp->fd_dfdfile[i]; 1323 (void)fdfile_ctor(NULL, fdp->fd_dfdfile[i], PR_WAITOK); 1324 } 1325 1326 return 0; 1327 } 1328 1329 static void 1330 filedesc_dtor(void *arg, void *obj) 1331 { 1332 filedesc_t *fdp = obj; 1333 int i; 1334 1335 for (i = 0; i < NDFDFILE; i++) { 1336 fdfile_dtor(NULL, fdp->fd_dfdfile[i]); 1337 } 1338 1339 mutex_destroy(&fdp->fd_lock); 1340 } 1341 1342 /* 1343 * Make p share curproc's filedesc structure. 1344 */ 1345 void 1346 fd_share(struct proc *p) 1347 { 1348 filedesc_t *fdp; 1349 1350 fdp = curlwp->l_fd; 1351 p->p_fd = fdp; 1352 atomic_inc_uint(&fdp->fd_refcnt); 1353 } 1354 1355 /* 1356 * Acquire a hold on a filedesc structure. 1357 */ 1358 void 1359 fd_hold(lwp_t *l) 1360 { 1361 filedesc_t *fdp = l->l_fd; 1362 1363 atomic_inc_uint(&fdp->fd_refcnt); 1364 } 1365 1366 /* 1367 * Copy a filedesc structure. 1368 */ 1369 filedesc_t * 1370 fd_copy(void) 1371 { 1372 filedesc_t *newfdp, *fdp; 1373 fdfile_t *ff, **ffp, **nffp, *ff2; 1374 int i, j, numfiles, lastfile, newlast; 1375 file_t *fp; 1376 fdtab_t *newdt; 1377 1378 fdp = curproc->p_fd; 1379 newfdp = pool_cache_get(filedesc_cache, PR_WAITOK); 1380 newfdp->fd_refcnt = 1; 1381 1382 #ifdef DIAGNOSTIC 1383 KASSERT(newfdp->fd_lastfile == -1); 1384 KASSERT(newfdp->fd_lastkqfile == -1); 1385 KASSERT(newfdp->fd_knhash == NULL); 1386 KASSERT(newfdp->fd_freefile == 0); 1387 KASSERT(newfdp->fd_exclose == false); 1388 KASSERT(newfdp->fd_dt == &newfdp->fd_dtbuiltin); 1389 KASSERT(newfdp->fd_dtbuiltin.dt_nfiles == NDFILE); 1390 for (i = 0; i < NDFDFILE; i++) { 1391 KASSERT(newfdp->fd_dtbuiltin.dt_ff[i] == 1392 (fdfile_t *)&newfdp->fd_dfdfile[i]); 1393 } 1394 for (i = NDFDFILE; i < NDFILE; i++) { 1395 KASSERT(newfdp->fd_dtbuiltin.dt_ff[i] == NULL); 1396 } 1397 #endif /* DIAGNOSTIC */ 1398 1399 mutex_enter(&fdp->fd_lock); 1400 fd_checkmaps(fdp); 1401 numfiles = fdp->fd_dt->dt_nfiles; 1402 lastfile = fdp->fd_lastfile; 1403 1404 /* 1405 * If the number of open files fits in the internal arrays 1406 * of the open file structure, use them, otherwise allocate 1407 * additional memory for the number of descriptors currently 1408 * in use. 1409 */ 1410 if (lastfile < NDFILE) { 1411 i = NDFILE; 1412 newdt = newfdp->fd_dt; 1413 KASSERT(newfdp->fd_dt == &newfdp->fd_dtbuiltin); 1414 } else { 1415 /* 1416 * Compute the smallest multiple of NDEXTENT needed 1417 * for the file descriptors currently in use, 1418 * allowing the table to shrink. 1419 */ 1420 i = numfiles; 1421 while (i >= 2 * NDEXTENT && i > lastfile * 2) { 1422 i /= 2; 1423 } 1424 KASSERT(i > NDFILE); 1425 newdt = fd_dtab_alloc(i); 1426 newfdp->fd_dt = newdt; 1427 memcpy(newdt->dt_ff, newfdp->fd_dtbuiltin.dt_ff, 1428 NDFDFILE * sizeof(fdfile_t **)); 1429 memset(newdt->dt_ff + NDFDFILE, 0, 1430 (i - NDFDFILE) * sizeof(fdfile_t **)); 1431 } 1432 if (NDHISLOTS(i) <= NDHISLOTS(NDFILE)) { 1433 newfdp->fd_himap = newfdp->fd_dhimap; 1434 newfdp->fd_lomap = newfdp->fd_dlomap; 1435 } else { 1436 fd_map_alloc(i, &newfdp->fd_lomap, &newfdp->fd_himap); 1437 KASSERT(i >= NDENTRIES * NDENTRIES); 1438 memset(newfdp->fd_himap, 0, NDHISLOTS(i)*sizeof(uint32_t)); 1439 memset(newfdp->fd_lomap, 0, NDLOSLOTS(i)*sizeof(uint32_t)); 1440 } 1441 newfdp->fd_freefile = fdp->fd_freefile; 1442 newfdp->fd_exclose = fdp->fd_exclose; 1443 1444 ffp = fdp->fd_dt->dt_ff; 1445 nffp = newdt->dt_ff; 1446 newlast = -1; 1447 for (i = 0; i <= (int)lastfile; i++, ffp++, nffp++) { 1448 KASSERT(i >= NDFDFILE || 1449 *nffp == (fdfile_t *)newfdp->fd_dfdfile[i]); 1450 ff = *ffp; 1451 if (ff == NULL || (fp = ff->ff_file) == NULL) { 1452 /* Descriptor unused, or descriptor half open. */ 1453 KASSERT(!fd_isused(newfdp, i)); 1454 continue; 1455 } 1456 if (__predict_false(fp->f_type == DTYPE_KQUEUE)) { 1457 /* kqueue descriptors cannot be copied. */ 1458 if (i < newfdp->fd_freefile) { 1459 newfdp->fd_freefile = i; 1460 } 1461 continue; 1462 } 1463 /* It's active: add a reference to the file. */ 1464 mutex_enter(&fp->f_lock); 1465 fp->f_count++; 1466 mutex_exit(&fp->f_lock); 1467 1468 /* Allocate an fdfile_t to represent it. */ 1469 if (i >= NDFDFILE) { 1470 ff2 = pool_cache_get(fdfile_cache, PR_WAITOK); 1471 *nffp = ff2; 1472 } else { 1473 ff2 = newdt->dt_ff[i]; 1474 } 1475 ff2->ff_file = fp; 1476 ff2->ff_exclose = ff->ff_exclose; 1477 ff2->ff_allocated = true; 1478 1479 /* Fix up bitmaps. */ 1480 j = i >> NDENTRYSHIFT; 1481 KASSERT((newfdp->fd_lomap[j] & (1 << (i & NDENTRYMASK))) == 0); 1482 newfdp->fd_lomap[j] |= 1 << (i & NDENTRYMASK); 1483 if (__predict_false(newfdp->fd_lomap[j] == ~0)) { 1484 KASSERT((newfdp->fd_himap[j >> NDENTRYSHIFT] & 1485 (1 << (j & NDENTRYMASK))) == 0); 1486 newfdp->fd_himap[j >> NDENTRYSHIFT] |= 1487 1 << (j & NDENTRYMASK); 1488 } 1489 newlast = i; 1490 } 1491 KASSERT(newdt->dt_ff[0] == (fdfile_t *)newfdp->fd_dfdfile[0]); 1492 newfdp->fd_lastfile = newlast; 1493 fd_checkmaps(newfdp); 1494 mutex_exit(&fdp->fd_lock); 1495 1496 return newfdp; 1497 } 1498 1499 /* 1500 * Release a filedesc structure. 1501 */ 1502 void 1503 fd_free(void) 1504 { 1505 fdfile_t *ff; 1506 file_t *fp; 1507 int fd, nf; 1508 fdtab_t *dt; 1509 lwp_t * const l = curlwp; 1510 filedesc_t * const fdp = l->l_fd; 1511 const bool noadvlock = (l->l_proc->p_flag & PK_ADVLOCK) == 0; 1512 1513 KASSERT(fdp->fd_dt->dt_ff[0] == (fdfile_t *)fdp->fd_dfdfile[0]); 1514 KASSERT(fdp->fd_dtbuiltin.dt_nfiles == NDFILE); 1515 KASSERT(fdp->fd_dtbuiltin.dt_link == NULL); 1516 1517 #ifndef __HAVE_ATOMIC_AS_MEMBAR 1518 membar_exit(); 1519 #endif 1520 if (atomic_dec_uint_nv(&fdp->fd_refcnt) > 0) 1521 return; 1522 1523 /* 1524 * Close any files that the process holds open. 1525 */ 1526 dt = fdp->fd_dt; 1527 fd_checkmaps(fdp); 1528 #ifdef DEBUG 1529 fdp->fd_refcnt = -1; /* see fd_checkmaps */ 1530 #endif 1531 for (fd = 0, nf = dt->dt_nfiles; fd < nf; fd++) { 1532 ff = dt->dt_ff[fd]; 1533 KASSERT(fd >= NDFDFILE || 1534 ff == (fdfile_t *)fdp->fd_dfdfile[fd]); 1535 if (ff == NULL) 1536 continue; 1537 if ((fp = ff->ff_file) != NULL) { 1538 /* 1539 * Must use fd_close() here if there is 1540 * a reference from kqueue or we might have posix 1541 * advisory locks. 1542 */ 1543 if (__predict_true(ff->ff_refcnt == 0) && 1544 (noadvlock || fp->f_type != DTYPE_VNODE)) { 1545 ff->ff_file = NULL; 1546 ff->ff_exclose = false; 1547 ff->ff_allocated = false; 1548 closef(fp); 1549 } else { 1550 ff->ff_refcnt++; 1551 fd_close(fd); 1552 } 1553 } 1554 KASSERT(ff->ff_refcnt == 0); 1555 KASSERT(ff->ff_file == NULL); 1556 KASSERT(!ff->ff_exclose); 1557 KASSERT(!ff->ff_allocated); 1558 if (fd >= NDFDFILE) { 1559 pool_cache_put(fdfile_cache, ff); 1560 dt->dt_ff[fd] = NULL; 1561 } 1562 } 1563 1564 /* 1565 * Clean out the descriptor table for the next user and return 1566 * to the cache. 1567 */ 1568 if (__predict_false(dt != &fdp->fd_dtbuiltin)) { 1569 fd_dtab_free(fdp->fd_dt); 1570 /* Otherwise, done above. */ 1571 memset(&fdp->fd_dtbuiltin.dt_ff[NDFDFILE], 0, 1572 (NDFILE - NDFDFILE) * sizeof(fdp->fd_dtbuiltin.dt_ff[0])); 1573 fdp->fd_dt = &fdp->fd_dtbuiltin; 1574 } 1575 if (__predict_false(NDHISLOTS(nf) > NDHISLOTS(NDFILE))) { 1576 KASSERT(fdp->fd_himap != fdp->fd_dhimap); 1577 KASSERT(fdp->fd_lomap != fdp->fd_dlomap); 1578 fd_map_free(nf, fdp->fd_lomap, fdp->fd_himap); 1579 } 1580 if (__predict_false(fdp->fd_knhash != NULL)) { 1581 hashdone(fdp->fd_knhash, HASH_LIST, fdp->fd_knhashmask); 1582 fdp->fd_knhash = NULL; 1583 fdp->fd_knhashmask = 0; 1584 } else { 1585 KASSERT(fdp->fd_knhashmask == 0); 1586 } 1587 fdp->fd_dt = &fdp->fd_dtbuiltin; 1588 fdp->fd_lastkqfile = -1; 1589 fdp->fd_lastfile = -1; 1590 fdp->fd_freefile = 0; 1591 fdp->fd_exclose = false; 1592 memset(&fdp->fd_startzero, 0, sizeof(*fdp) - 1593 offsetof(filedesc_t, fd_startzero)); 1594 fdp->fd_himap = fdp->fd_dhimap; 1595 fdp->fd_lomap = fdp->fd_dlomap; 1596 KASSERT(fdp->fd_dtbuiltin.dt_nfiles == NDFILE); 1597 KASSERT(fdp->fd_dtbuiltin.dt_link == NULL); 1598 KASSERT(fdp->fd_dt == &fdp->fd_dtbuiltin); 1599 #ifdef DEBUG 1600 fdp->fd_refcnt = 0; /* see fd_checkmaps */ 1601 #endif 1602 fd_checkmaps(fdp); 1603 pool_cache_put(filedesc_cache, fdp); 1604 } 1605 1606 /* 1607 * File Descriptor pseudo-device driver (/dev/fd/). 1608 * 1609 * Opening minor device N dup()s the file (if any) connected to file 1610 * descriptor N belonging to the calling process. Note that this driver 1611 * consists of only the ``open()'' routine, because all subsequent 1612 * references to this file will be direct to the other driver. 1613 */ 1614 static int 1615 filedescopen(dev_t dev, int mode, int type, lwp_t *l) 1616 { 1617 1618 /* 1619 * XXX Kludge: set dupfd to contain the value of the 1620 * the file descriptor being sought for duplication. The error 1621 * return ensures that the vnode for this device will be released 1622 * by vn_open. Open will detect this special error and take the 1623 * actions in fd_dupopen below. Other callers of vn_open or VOP_OPEN 1624 * will simply report the error. 1625 */ 1626 l->l_dupfd = minor(dev); /* XXX */ 1627 return EDUPFD; 1628 } 1629 1630 /* 1631 * Duplicate the specified descriptor to a free descriptor. 1632 */ 1633 int 1634 fd_dupopen(int old, int *new, int mode, int error) 1635 { 1636 filedesc_t *fdp; 1637 fdfile_t *ff; 1638 file_t *fp; 1639 fdtab_t *dt; 1640 1641 if ((fp = fd_getfile(old)) == NULL) { 1642 return EBADF; 1643 } 1644 fdp = curlwp->l_fd; 1645 dt = fdp->fd_dt; 1646 ff = dt->dt_ff[old]; 1647 1648 /* 1649 * There are two cases of interest here. 1650 * 1651 * For EDUPFD simply dup (old) to file descriptor 1652 * (new) and return. 1653 * 1654 * For EMOVEFD steal away the file structure from (old) and 1655 * store it in (new). (old) is effectively closed by 1656 * this operation. 1657 * 1658 * Any other error code is just returned. 1659 */ 1660 switch (error) { 1661 case EDUPFD: 1662 /* 1663 * Check that the mode the file is being opened for is a 1664 * subset of the mode of the existing descriptor. 1665 */ 1666 if (((mode & (FREAD|FWRITE)) | fp->f_flag) != fp->f_flag) { 1667 error = EACCES; 1668 break; 1669 } 1670 1671 /* Copy it. */ 1672 error = fd_dup(fp, 0, new, ff->ff_exclose); 1673 break; 1674 1675 case EMOVEFD: 1676 /* Copy it. */ 1677 error = fd_dup(fp, 0, new, ff->ff_exclose); 1678 if (error != 0) { 1679 break; 1680 } 1681 1682 /* Steal away the file pointer from 'old'. */ 1683 (void)fd_close(old); 1684 return 0; 1685 } 1686 1687 fd_putfile(old); 1688 return error; 1689 } 1690 1691 /* 1692 * Close open files on exec. 1693 */ 1694 void 1695 fd_closeexec(void) 1696 { 1697 proc_t *p; 1698 filedesc_t *fdp; 1699 fdfile_t *ff; 1700 lwp_t *l; 1701 fdtab_t *dt; 1702 int fd; 1703 1704 l = curlwp; 1705 p = l->l_proc; 1706 fdp = p->p_fd; 1707 1708 if (fdp->fd_refcnt > 1) { 1709 fdp = fd_copy(); 1710 fd_free(); 1711 p->p_fd = fdp; 1712 l->l_fd = fdp; 1713 } 1714 if (!fdp->fd_exclose) { 1715 return; 1716 } 1717 fdp->fd_exclose = false; 1718 dt = fdp->fd_dt; 1719 1720 for (fd = 0; fd <= fdp->fd_lastfile; fd++) { 1721 if ((ff = dt->dt_ff[fd]) == NULL) { 1722 KASSERT(fd >= NDFDFILE); 1723 continue; 1724 } 1725 KASSERT(fd >= NDFDFILE || 1726 ff == (fdfile_t *)fdp->fd_dfdfile[fd]); 1727 if (ff->ff_file == NULL) 1728 continue; 1729 if (ff->ff_exclose) { 1730 /* 1731 * We need a reference to close the file. 1732 * No other threads can see the fdfile_t at 1733 * this point, so don't bother locking. 1734 */ 1735 KASSERT((ff->ff_refcnt & FR_CLOSING) == 0); 1736 ff->ff_refcnt++; 1737 fd_close(fd); 1738 } 1739 } 1740 } 1741 1742 /* 1743 * Sets descriptor owner. If the owner is a process, 'pgid' 1744 * is set to positive value, process ID. If the owner is process group, 1745 * 'pgid' is set to -pg_id. 1746 */ 1747 int 1748 fsetown(pid_t *pgid, u_long cmd, const void *data) 1749 { 1750 pid_t id = *(const pid_t *)data; 1751 int error; 1752 1753 switch (cmd) { 1754 case TIOCSPGRP: 1755 if (id < 0) 1756 return EINVAL; 1757 id = -id; 1758 break; 1759 default: 1760 break; 1761 } 1762 if (id > 0) { 1763 mutex_enter(proc_lock); 1764 error = proc_find(id) ? 0 : ESRCH; 1765 mutex_exit(proc_lock); 1766 } else if (id < 0) { 1767 error = pgid_in_session(curproc, -id); 1768 } else { 1769 error = 0; 1770 } 1771 if (!error) { 1772 *pgid = id; 1773 } 1774 return error; 1775 } 1776 1777 void 1778 fd_set_exclose(struct lwp *l, int fd, bool exclose) 1779 { 1780 filedesc_t *fdp = l->l_fd; 1781 fdfile_t *ff = fdp->fd_dt->dt_ff[fd]; 1782 1783 ff->ff_exclose = exclose; 1784 if (exclose) 1785 fdp->fd_exclose = true; 1786 } 1787 1788 /* 1789 * Return descriptor owner information. If the value is positive, 1790 * it's process ID. If it's negative, it's process group ID and 1791 * needs the sign removed before use. 1792 */ 1793 int 1794 fgetown(pid_t pgid, u_long cmd, void *data) 1795 { 1796 1797 switch (cmd) { 1798 case TIOCGPGRP: 1799 *(int *)data = -pgid; 1800 break; 1801 default: 1802 *(int *)data = pgid; 1803 break; 1804 } 1805 return 0; 1806 } 1807 1808 /* 1809 * Send signal to descriptor owner, either process or process group. 1810 */ 1811 void 1812 fownsignal(pid_t pgid, int signo, int code, int band, void *fdescdata) 1813 { 1814 ksiginfo_t ksi; 1815 1816 KASSERT(!cpu_intr_p()); 1817 1818 if (pgid == 0) { 1819 return; 1820 } 1821 1822 KSI_INIT(&ksi); 1823 ksi.ksi_signo = signo; 1824 ksi.ksi_code = code; 1825 ksi.ksi_band = band; 1826 1827 mutex_enter(proc_lock); 1828 if (pgid > 0) { 1829 struct proc *p1; 1830 1831 p1 = proc_find(pgid); 1832 if (p1 != NULL) { 1833 kpsignal(p1, &ksi, fdescdata); 1834 } 1835 } else { 1836 struct pgrp *pgrp; 1837 1838 KASSERT(pgid < 0); 1839 pgrp = pgrp_find(-pgid); 1840 if (pgrp != NULL) { 1841 kpgsignal(pgrp, &ksi, fdescdata, 0); 1842 } 1843 } 1844 mutex_exit(proc_lock); 1845 } 1846 1847 int 1848 fd_clone(file_t *fp, unsigned fd, int flag, const struct fileops *fops, 1849 void *data) 1850 { 1851 1852 fp->f_flag = flag; 1853 fp->f_type = DTYPE_MISC; 1854 fp->f_ops = fops; 1855 fp->f_data = data; 1856 curlwp->l_dupfd = fd; 1857 fd_affix(curproc, fp, fd); 1858 1859 return EMOVEFD; 1860 } 1861 1862 int 1863 fnullop_fcntl(file_t *fp, u_int cmd, void *data) 1864 { 1865 1866 if (cmd == F_SETFL) 1867 return 0; 1868 1869 return EOPNOTSUPP; 1870 } 1871 1872 int 1873 fnullop_poll(file_t *fp, int which) 1874 { 1875 1876 return 0; 1877 } 1878 1879 int 1880 fnullop_kqfilter(file_t *fp, struct knote *kn) 1881 { 1882 1883 return 0; 1884 } 1885 1886 void 1887 fnullop_restart(file_t *fp) 1888 { 1889 1890 } 1891 1892 int 1893 fbadop_read(file_t *fp, off_t *offset, struct uio *uio, 1894 kauth_cred_t cred, int flags) 1895 { 1896 1897 return EOPNOTSUPP; 1898 } 1899 1900 int 1901 fbadop_write(file_t *fp, off_t *offset, struct uio *uio, 1902 kauth_cred_t cred, int flags) 1903 { 1904 1905 return EOPNOTSUPP; 1906 } 1907 1908 int 1909 fbadop_ioctl(file_t *fp, u_long com, void *data) 1910 { 1911 1912 return EOPNOTSUPP; 1913 } 1914 1915 int 1916 fbadop_stat(file_t *fp, struct stat *sb) 1917 { 1918 1919 return EOPNOTSUPP; 1920 } 1921 1922 int 1923 fbadop_close(file_t *fp) 1924 { 1925 1926 return EOPNOTSUPP; 1927 } 1928 1929 /* 1930 * sysctl routines pertaining to file descriptors 1931 */ 1932 1933 /* Initialized in sysctl_init() for now... */ 1934 extern kmutex_t sysctl_file_marker_lock; 1935 static u_int sysctl_file_marker = 1; 1936 1937 /* 1938 * Expects to be called with proc_lock and sysctl_file_marker_lock locked. 1939 */ 1940 static void 1941 sysctl_file_marker_reset(void) 1942 { 1943 struct proc *p; 1944 1945 PROCLIST_FOREACH(p, &allproc) { 1946 struct filedesc *fd = p->p_fd; 1947 fdtab_t *dt; 1948 u_int i; 1949 1950 mutex_enter(&fd->fd_lock); 1951 dt = fd->fd_dt; 1952 for (i = 0; i < dt->dt_nfiles; i++) { 1953 struct file *fp; 1954 fdfile_t *ff; 1955 1956 if ((ff = dt->dt_ff[i]) == NULL) { 1957 continue; 1958 } 1959 if ((fp = ff->ff_file) == NULL) { 1960 continue; 1961 } 1962 fp->f_marker = 0; 1963 } 1964 mutex_exit(&fd->fd_lock); 1965 } 1966 } 1967 1968 /* 1969 * sysctl helper routine for kern.file pseudo-subtree. 1970 */ 1971 static int 1972 sysctl_kern_file(SYSCTLFN_ARGS) 1973 { 1974 int error; 1975 size_t buflen; 1976 struct file *fp, fbuf; 1977 char *start, *where; 1978 struct proc *p; 1979 1980 start = where = oldp; 1981 buflen = *oldlenp; 1982 1983 if (where == NULL) { 1984 /* 1985 * overestimate by 10 files 1986 */ 1987 *oldlenp = sizeof(filehead) + (nfiles + 10) * 1988 sizeof(struct file); 1989 return 0; 1990 } 1991 1992 /* 1993 * first sysctl_copyout filehead 1994 */ 1995 if (buflen < sizeof(filehead)) { 1996 *oldlenp = 0; 1997 return 0; 1998 } 1999 sysctl_unlock(); 2000 error = sysctl_copyout(l, &filehead, where, sizeof(filehead)); 2001 if (error) { 2002 sysctl_relock(); 2003 return error; 2004 } 2005 buflen -= sizeof(filehead); 2006 where += sizeof(filehead); 2007 2008 /* 2009 * followed by an array of file structures 2010 */ 2011 mutex_enter(&sysctl_file_marker_lock); 2012 mutex_enter(proc_lock); 2013 PROCLIST_FOREACH(p, &allproc) { 2014 struct filedesc *fd; 2015 fdtab_t *dt; 2016 u_int i; 2017 2018 if (p->p_stat == SIDL) { 2019 /* skip embryonic processes */ 2020 continue; 2021 } 2022 mutex_enter(p->p_lock); 2023 error = kauth_authorize_process(l->l_cred, 2024 KAUTH_PROCESS_CANSEE, p, 2025 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES), 2026 NULL, NULL); 2027 mutex_exit(p->p_lock); 2028 if (error != 0) { 2029 /* 2030 * Don't leak kauth retval if we're silently 2031 * skipping this entry. 2032 */ 2033 error = 0; 2034 continue; 2035 } 2036 2037 /* 2038 * Grab a hold on the process. 2039 */ 2040 if (!rw_tryenter(&p->p_reflock, RW_READER)) { 2041 continue; 2042 } 2043 mutex_exit(proc_lock); 2044 2045 fd = p->p_fd; 2046 mutex_enter(&fd->fd_lock); 2047 dt = fd->fd_dt; 2048 for (i = 0; i < dt->dt_nfiles; i++) { 2049 fdfile_t *ff; 2050 2051 if ((ff = dt->dt_ff[i]) == NULL) { 2052 continue; 2053 } 2054 if ((fp = ff->ff_file) == NULL) { 2055 continue; 2056 } 2057 2058 mutex_enter(&fp->f_lock); 2059 2060 if ((fp->f_count == 0) || 2061 (fp->f_marker == sysctl_file_marker)) { 2062 mutex_exit(&fp->f_lock); 2063 continue; 2064 } 2065 2066 /* Check that we have enough space. */ 2067 if (buflen < sizeof(struct file)) { 2068 *oldlenp = where - start; 2069 mutex_exit(&fp->f_lock); 2070 error = ENOMEM; 2071 break; 2072 } 2073 2074 memcpy(&fbuf, fp, sizeof(fbuf)); 2075 mutex_exit(&fp->f_lock); 2076 error = sysctl_copyout(l, &fbuf, where, sizeof(fbuf)); 2077 if (error) { 2078 break; 2079 } 2080 buflen -= sizeof(struct file); 2081 where += sizeof(struct file); 2082 2083 fp->f_marker = sysctl_file_marker; 2084 } 2085 mutex_exit(&fd->fd_lock); 2086 2087 /* 2088 * Release reference to process. 2089 */ 2090 mutex_enter(proc_lock); 2091 rw_exit(&p->p_reflock); 2092 2093 if (error) 2094 break; 2095 } 2096 2097 sysctl_file_marker++; 2098 /* Reset all markers if wrapped. */ 2099 if (sysctl_file_marker == 0) { 2100 sysctl_file_marker_reset(); 2101 sysctl_file_marker++; 2102 } 2103 2104 mutex_exit(proc_lock); 2105 mutex_exit(&sysctl_file_marker_lock); 2106 2107 *oldlenp = where - start; 2108 sysctl_relock(); 2109 return error; 2110 } 2111 2112 /* 2113 * sysctl helper function for kern.file2 2114 */ 2115 static int 2116 sysctl_kern_file2(SYSCTLFN_ARGS) 2117 { 2118 struct proc *p; 2119 struct file *fp; 2120 struct filedesc *fd; 2121 struct kinfo_file kf; 2122 char *dp; 2123 u_int i, op; 2124 size_t len, needed, elem_size, out_size; 2125 int error, arg, elem_count; 2126 fdfile_t *ff; 2127 fdtab_t *dt; 2128 2129 if (namelen == 1 && name[0] == CTL_QUERY) 2130 return sysctl_query(SYSCTLFN_CALL(rnode)); 2131 2132 if (namelen != 4) 2133 return EINVAL; 2134 2135 error = 0; 2136 dp = oldp; 2137 len = (oldp != NULL) ? *oldlenp : 0; 2138 op = name[0]; 2139 arg = name[1]; 2140 elem_size = name[2]; 2141 elem_count = name[3]; 2142 out_size = MIN(sizeof(kf), elem_size); 2143 needed = 0; 2144 2145 if (elem_size < 1 || elem_count < 0) 2146 return EINVAL; 2147 2148 switch (op) { 2149 case KERN_FILE_BYFILE: 2150 case KERN_FILE_BYPID: 2151 /* 2152 * We're traversing the process list in both cases; the BYFILE 2153 * case does additional work of keeping track of files already 2154 * looked at. 2155 */ 2156 2157 /* doesn't use arg so it must be zero */ 2158 if ((op == KERN_FILE_BYFILE) && (arg != 0)) 2159 return EINVAL; 2160 2161 if ((op == KERN_FILE_BYPID) && (arg < -1)) 2162 /* -1 means all processes */ 2163 return EINVAL; 2164 2165 sysctl_unlock(); 2166 if (op == KERN_FILE_BYFILE) 2167 mutex_enter(&sysctl_file_marker_lock); 2168 mutex_enter(proc_lock); 2169 PROCLIST_FOREACH(p, &allproc) { 2170 if (p->p_stat == SIDL) { 2171 /* skip embryonic processes */ 2172 continue; 2173 } 2174 if (arg > 0 && p->p_pid != arg) { 2175 /* pick only the one we want */ 2176 /* XXX want 0 to mean "kernel files" */ 2177 continue; 2178 } 2179 mutex_enter(p->p_lock); 2180 error = kauth_authorize_process(l->l_cred, 2181 KAUTH_PROCESS_CANSEE, p, 2182 KAUTH_ARG(KAUTH_REQ_PROCESS_CANSEE_OPENFILES), 2183 NULL, NULL); 2184 mutex_exit(p->p_lock); 2185 if (error != 0) { 2186 /* 2187 * Don't leak kauth retval if we're silently 2188 * skipping this entry. 2189 */ 2190 error = 0; 2191 continue; 2192 } 2193 2194 /* 2195 * Grab a hold on the process. 2196 */ 2197 if (!rw_tryenter(&p->p_reflock, RW_READER)) { 2198 continue; 2199 } 2200 mutex_exit(proc_lock); 2201 2202 fd = p->p_fd; 2203 mutex_enter(&fd->fd_lock); 2204 dt = fd->fd_dt; 2205 for (i = 0; i < dt->dt_nfiles; i++) { 2206 if ((ff = dt->dt_ff[i]) == NULL) { 2207 continue; 2208 } 2209 if ((fp = ff->ff_file) == NULL) { 2210 continue; 2211 } 2212 2213 if ((op == KERN_FILE_BYFILE) && 2214 (fp->f_marker == sysctl_file_marker)) { 2215 continue; 2216 } 2217 if (len >= elem_size && elem_count > 0) { 2218 mutex_enter(&fp->f_lock); 2219 fill_file(&kf, fp, ff, i, p->p_pid); 2220 mutex_exit(&fp->f_lock); 2221 mutex_exit(&fd->fd_lock); 2222 error = sysctl_copyout(l, 2223 &kf, dp, out_size); 2224 mutex_enter(&fd->fd_lock); 2225 if (error) 2226 break; 2227 dp += elem_size; 2228 len -= elem_size; 2229 } 2230 if (op == KERN_FILE_BYFILE) 2231 fp->f_marker = sysctl_file_marker; 2232 needed += elem_size; 2233 if (elem_count > 0 && elem_count != INT_MAX) 2234 elem_count--; 2235 } 2236 mutex_exit(&fd->fd_lock); 2237 2238 /* 2239 * Release reference to process. 2240 */ 2241 mutex_enter(proc_lock); 2242 rw_exit(&p->p_reflock); 2243 } 2244 if (op == KERN_FILE_BYFILE) { 2245 sysctl_file_marker++; 2246 2247 /* Reset all markers if wrapped. */ 2248 if (sysctl_file_marker == 0) { 2249 sysctl_file_marker_reset(); 2250 sysctl_file_marker++; 2251 } 2252 } 2253 mutex_exit(proc_lock); 2254 if (op == KERN_FILE_BYFILE) 2255 mutex_exit(&sysctl_file_marker_lock); 2256 sysctl_relock(); 2257 break; 2258 default: 2259 return EINVAL; 2260 } 2261 2262 if (oldp == NULL) 2263 needed += KERN_FILESLOP * elem_size; 2264 *oldlenp = needed; 2265 2266 return error; 2267 } 2268 2269 static void 2270 fill_file(struct kinfo_file *kp, const file_t *fp, const fdfile_t *ff, 2271 int i, pid_t pid) 2272 { 2273 2274 memset(kp, 0, sizeof(*kp)); 2275 2276 kp->ki_fileaddr = PTRTOUINT64(fp); 2277 kp->ki_flag = fp->f_flag; 2278 kp->ki_iflags = 0; 2279 kp->ki_ftype = fp->f_type; 2280 kp->ki_count = fp->f_count; 2281 kp->ki_msgcount = fp->f_msgcount; 2282 kp->ki_fucred = PTRTOUINT64(fp->f_cred); 2283 kp->ki_fuid = kauth_cred_geteuid(fp->f_cred); 2284 kp->ki_fgid = kauth_cred_getegid(fp->f_cred); 2285 kp->ki_fops = PTRTOUINT64(fp->f_ops); 2286 kp->ki_foffset = fp->f_offset; 2287 kp->ki_fdata = PTRTOUINT64(fp->f_data); 2288 2289 /* vnode information to glue this file to something */ 2290 if (fp->f_type == DTYPE_VNODE) { 2291 struct vnode *vp = (struct vnode *)fp->f_data; 2292 2293 kp->ki_vun = PTRTOUINT64(vp->v_un.vu_socket); 2294 kp->ki_vsize = vp->v_size; 2295 kp->ki_vtype = vp->v_type; 2296 kp->ki_vtag = vp->v_tag; 2297 kp->ki_vdata = PTRTOUINT64(vp->v_data); 2298 } 2299 2300 /* process information when retrieved via KERN_FILE_BYPID */ 2301 if (ff != NULL) { 2302 kp->ki_pid = pid; 2303 kp->ki_fd = i; 2304 kp->ki_ofileflags = ff->ff_exclose; 2305 kp->ki_usecount = ff->ff_refcnt; 2306 } 2307 } 2308