Now available: The Design and Implementation of the FreeBSD Operating System (Second Edition)
[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]
FreeBSD/Linux Kernel Cross Reference
sys/kern/vfs_subr.c
Version:
- FREEBSD - FREEBSD-13-STABLE - FREEBSD-13-0 - FREEBSD-12-STABLE - FREEBSD-12-0 - FREEBSD-11-STABLE - FREEBSD-11-0 - FREEBSD-10-STABLE - FREEBSD-10-0 - FREEBSD-9-STABLE - FREEBSD-9-0 - FREEBSD-8-STABLE - FREEBSD-8-0 - FREEBSD-7-STABLE - FREEBSD-7-0 - FREEBSD-6-STABLE - FREEBSD-6-0 - FREEBSD-5-STABLE - FREEBSD-5-0 - FREEBSD-4-STABLE - FREEBSD-3-STABLE - FREEBSD22 - l41 - OPENBSD - linux-2.6 - MK84 - PLAN9 - xnu-8792
SearchContext: - none - 3 - 10
SearchContext: - none - 3 - 10
1 /* $NetBSD: vfs_subr.c,v 1.496 2022/10/26 23:39:43 riastradh Exp $ */ 2 3 /*- 4 * Copyright (c) 1997, 1998, 2004, 2005, 2007, 2008, 2019, 2020 5 * The NetBSD Foundation, Inc. 6 * All rights reserved. 7 * 8 * This code is derived from software contributed to The NetBSD Foundation 9 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, 10 * NASA Ames Research Center, by Charles M. Hannum, by Andrew Doran, 11 * by Marshall Kirk McKusick and Greg Ganger at the University of Michigan. 12 * 13 * Redistribution and use in source and binary forms, with or without 14 * modification, are permitted provided that the following conditions 15 * are met: 16 * 1. Redistributions of source code must retain the above copyright 17 * notice, this list of conditions and the following disclaimer. 18 * 2. Redistributions in binary form must reproduce the above copyright 19 * notice, this list of conditions and the following disclaimer in the 20 * documentation and/or other materials provided with the distribution. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 23 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 24 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 25 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 26 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 27 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 28 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 29 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 30 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 31 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 32 * POSSIBILITY OF SUCH DAMAGE. 33 */ 34 35 /* 36 * Copyright (c) 1989, 1993 37 * The Regents of the University of California. All rights reserved. 38 * (c) UNIX System Laboratories, Inc. 39 * All or some portions of this file are derived from material licensed 40 * to the University of California by American Telephone and Telegraph 41 * Co. or Unix System Laboratories, Inc. and are reproduced herein with 42 * the permission of UNIX System Laboratories, Inc. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 3. Neither the name of the University nor the names of its contributors 53 * may be used to endorse or promote products derived from this software 54 * without specific prior written permission. 55 * 56 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 57 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 58 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 59 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 60 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 61 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 62 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 63 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 64 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 65 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 66 * SUCH DAMAGE. 67 * 68 * @(#)vfs_subr.c 8.13 (Berkeley) 4/18/94 69 */ 70 71 #include <sys/cdefs.h> 72 __KERNEL_RCSID(0, "$NetBSD: vfs_subr.c,v 1.496 2022/10/26 23:39:43 riastradh Exp $"); 73 74 #ifdef _KERNEL_OPT 75 #include "opt_ddb.h" 76 #include "opt_compat_netbsd.h" 77 #include "opt_compat_43.h" 78 #endif 79 80 #include <sys/param.h> 81 #include <sys/systm.h> 82 #include <sys/conf.h> 83 #include <sys/dirent.h> 84 #include <sys/filedesc.h> 85 #include <sys/kernel.h> 86 #include <sys/mount.h> 87 #include <sys/fstrans.h> 88 #include <sys/vnode_impl.h> 89 #include <sys/stat.h> 90 #include <sys/sysctl.h> 91 #include <sys/namei.h> 92 #include <sys/buf.h> 93 #include <sys/errno.h> 94 #include <sys/kmem.h> 95 #include <sys/syscallargs.h> 96 #include <sys/kauth.h> 97 #include <sys/module.h> 98 99 #include <miscfs/deadfs/deadfs.h> 100 #include <miscfs/genfs/genfs.h> 101 #include <miscfs/specfs/specdev.h> 102 103 #include <uvm/uvm_ddb.h> 104 105 const enum vtype iftovt_tab[16] = { 106 VNON, VFIFO, VCHR, VNON, VDIR, VNON, VBLK, VNON, 107 VREG, VNON, VLNK, VNON, VSOCK, VNON, VNON, VBAD, 108 }; 109 const int vttoif_tab[9] = { 110 0, S_IFREG, S_IFDIR, S_IFBLK, S_IFCHR, S_IFLNK, 111 S_IFSOCK, S_IFIFO, S_IFMT, 112 }; 113 114 /* 115 * Insq/Remq for the vnode usage lists. 116 */ 117 #define bufinsvn(bp, dp) LIST_INSERT_HEAD(dp, bp, b_vnbufs) 118 #define bufremvn(bp) { \ 119 LIST_REMOVE(bp, b_vnbufs); \ 120 (bp)->b_vnbufs.le_next = NOLIST; \ 121 } 122 123 int doforce = 1; /* 1 => permit forcible unmounting */ 124 125 /* 126 * Local declarations. 127 */ 128 129 static void vn_initialize_syncerd(void); 130 131 /* 132 * Initialize the vnode management data structures. 133 */ 134 void 135 vntblinit(void) 136 { 137 138 vn_initialize_syncerd(); 139 vfs_mount_sysinit(); 140 vfs_vnode_sysinit(); 141 } 142 143 /* 144 * Flush out and invalidate all buffers associated with a vnode. 145 * Called with the underlying vnode locked, which should prevent new dirty 146 * buffers from being queued. 147 */ 148 int 149 vinvalbuf(struct vnode *vp, int flags, kauth_cred_t cred, struct lwp *l, 150 bool catch_p, int slptimeo) 151 { 152 struct buf *bp, *nbp; 153 int error; 154 int flushflags = PGO_ALLPAGES | PGO_FREE | PGO_SYNCIO | 155 (flags & V_SAVE ? PGO_CLEANIT | PGO_RECLAIM : 0); 156 157 /* XXXUBC this doesn't look at flags or slp* */ 158 rw_enter(vp->v_uobj.vmobjlock, RW_WRITER); 159 error = VOP_PUTPAGES(vp, 0, 0, flushflags); 160 if (error) { 161 return error; 162 } 163 164 if (flags & V_SAVE) { 165 error = VOP_FSYNC(vp, cred, FSYNC_WAIT|FSYNC_RECLAIM, 0, 0); 166 if (error) 167 return (error); 168 KASSERT(LIST_EMPTY(&vp->v_dirtyblkhd)); 169 } 170 171 mutex_enter(&bufcache_lock); 172 restart: 173 for (bp = LIST_FIRST(&vp->v_dirtyblkhd); bp; bp = nbp) { 174 KASSERT(bp->b_vp == vp); 175 nbp = LIST_NEXT(bp, b_vnbufs); 176 error = bbusy(bp, catch_p, slptimeo, NULL); 177 if (error != 0) { 178 if (error == EPASSTHROUGH) 179 goto restart; 180 mutex_exit(&bufcache_lock); 181 return (error); 182 } 183 brelsel(bp, BC_INVAL | BC_VFLUSH); 184 } 185 186 for (bp = LIST_FIRST(&vp->v_cleanblkhd); bp; bp = nbp) { 187 KASSERT(bp->b_vp == vp); 188 nbp = LIST_NEXT(bp, b_vnbufs); 189 error = bbusy(bp, catch_p, slptimeo, NULL); 190 if (error != 0) { 191 if (error == EPASSTHROUGH) 192 goto restart; 193 mutex_exit(&bufcache_lock); 194 return (error); 195 } 196 /* 197 * XXX Since there are no node locks for NFS, I believe 198 * there is a slight chance that a delayed write will 199 * occur while sleeping just above, so check for it. 200 */ 201 if ((bp->b_oflags & BO_DELWRI) && (flags & V_SAVE)) { 202 #ifdef DEBUG 203 printf("buffer still DELWRI\n"); 204 #endif 205 bp->b_cflags |= BC_BUSY | BC_VFLUSH; 206 mutex_exit(&bufcache_lock); 207 VOP_BWRITE(bp->b_vp, bp); 208 mutex_enter(&bufcache_lock); 209 goto restart; 210 } 211 brelsel(bp, BC_INVAL | BC_VFLUSH); 212 } 213 214 #ifdef DIAGNOSTIC 215 if (!LIST_EMPTY(&vp->v_cleanblkhd) || !LIST_EMPTY(&vp->v_dirtyblkhd)) 216 panic("vinvalbuf: flush failed, vp %p", vp); 217 #endif 218 219 mutex_exit(&bufcache_lock); 220 221 return (0); 222 } 223 224 /* 225 * Destroy any in core blocks past the truncation length. 226 * Called with the underlying vnode locked, which should prevent new dirty 227 * buffers from being queued. 228 */ 229 int 230 vtruncbuf(struct vnode *vp, daddr_t lbn, bool catch_p, int slptimeo) 231 { 232 struct buf *bp, *nbp; 233 int error; 234 voff_t off; 235 236 off = round_page((voff_t)lbn << vp->v_mount->mnt_fs_bshift); 237 rw_enter(vp->v_uobj.vmobjlock, RW_WRITER); 238 error = VOP_PUTPAGES(vp, off, 0, PGO_FREE | PGO_SYNCIO); 239 if (error) { 240 return error; 241 } 242 243 mutex_enter(&bufcache_lock); 244 restart: 245 for (bp = LIST_FIRST(&vp->v_dirtyblkhd); bp; bp = nbp) { 246 KASSERT(bp->b_vp == vp); 247 nbp = LIST_NEXT(bp, b_vnbufs); 248 if (bp->b_lblkno < lbn) 249 continue; 250 error = bbusy(bp, catch_p, slptimeo, NULL); 251 if (error != 0) { 252 if (error == EPASSTHROUGH) 253 goto restart; 254 mutex_exit(&bufcache_lock); 255 return (error); 256 } 257 brelsel(bp, BC_INVAL | BC_VFLUSH); 258 } 259 260 for (bp = LIST_FIRST(&vp->v_cleanblkhd); bp; bp = nbp) { 261 KASSERT(bp->b_vp == vp); 262 nbp = LIST_NEXT(bp, b_vnbufs); 263 if (bp->b_lblkno < lbn) 264 continue; 265 error = bbusy(bp, catch_p, slptimeo, NULL); 266 if (error != 0) { 267 if (error == EPASSTHROUGH) 268 goto restart; 269 mutex_exit(&bufcache_lock); 270 return (error); 271 } 272 brelsel(bp, BC_INVAL | BC_VFLUSH); 273 } 274 mutex_exit(&bufcache_lock); 275 276 return (0); 277 } 278 279 /* 280 * Flush all dirty buffers from a vnode. 281 * Called with the underlying vnode locked, which should prevent new dirty 282 * buffers from being queued. 283 */ 284 int 285 vflushbuf(struct vnode *vp, int flags) 286 { 287 struct buf *bp, *nbp; 288 int error, pflags; 289 bool dirty, sync; 290 291 sync = (flags & FSYNC_WAIT) != 0; 292 pflags = PGO_CLEANIT | PGO_ALLPAGES | 293 (sync ? PGO_SYNCIO : 0) | 294 ((flags & FSYNC_LAZY) ? PGO_LAZY : 0); 295 rw_enter(vp->v_uobj.vmobjlock, RW_WRITER); 296 (void) VOP_PUTPAGES(vp, 0, 0, pflags); 297 298 loop: 299 mutex_enter(&bufcache_lock); 300 for (bp = LIST_FIRST(&vp->v_dirtyblkhd); bp; bp = nbp) { 301 KASSERT(bp->b_vp == vp); 302 nbp = LIST_NEXT(bp, b_vnbufs); 303 if ((bp->b_cflags & BC_BUSY)) 304 continue; 305 if ((bp->b_oflags & BO_DELWRI) == 0) 306 panic("vflushbuf: not dirty, bp %p", bp); 307 bp->b_cflags |= BC_BUSY | BC_VFLUSH; 308 mutex_exit(&bufcache_lock); 309 /* 310 * Wait for I/O associated with indirect blocks to complete, 311 * since there is no way to quickly wait for them below. 312 */ 313 if (bp->b_vp == vp || !sync) 314 (void) bawrite(bp); 315 else { 316 error = bwrite(bp); 317 if (error) 318 return error; 319 } 320 goto loop; 321 } 322 mutex_exit(&bufcache_lock); 323 324 if (!sync) 325 return 0; 326 327 mutex_enter(vp->v_interlock); 328 while (vp->v_numoutput != 0) 329 cv_wait(&vp->v_cv, vp->v_interlock); 330 dirty = !LIST_EMPTY(&vp->v_dirtyblkhd); 331 mutex_exit(vp->v_interlock); 332 333 if (dirty) { 334 vprint("vflushbuf: dirty", vp); 335 goto loop; 336 } 337 338 return 0; 339 } 340 341 /* 342 * Create a vnode for a block device. 343 * Used for root filesystem and swap areas. 344 * Also used for memory file system special devices. 345 */ 346 int 347 bdevvp(dev_t dev, vnode_t **vpp) 348 { 349 struct vattr va; 350 351 vattr_null(&va); 352 va.va_type = VBLK; 353 va.va_rdev = dev; 354 355 return vcache_new(dead_rootmount, NULL, &va, NOCRED, NULL, vpp); 356 } 357 358 /* 359 * Create a vnode for a character device. 360 * Used for kernfs and some console handling. 361 */ 362 int 363 cdevvp(dev_t dev, vnode_t **vpp) 364 { 365 struct vattr va; 366 367 vattr_null(&va); 368 va.va_type = VCHR; 369 va.va_rdev = dev; 370 371 return vcache_new(dead_rootmount, NULL, &va, NOCRED, NULL, vpp); 372 } 373 374 /* 375 * Associate a buffer with a vnode. There must already be a hold on 376 * the vnode. 377 */ 378 void 379 bgetvp(struct vnode *vp, struct buf *bp) 380 { 381 382 KASSERT(bp->b_vp == NULL); 383 KASSERT(bp->b_objlock == &buffer_lock); 384 KASSERT(mutex_owned(vp->v_interlock)); 385 KASSERT(mutex_owned(&bufcache_lock)); 386 KASSERT((bp->b_cflags & BC_BUSY) != 0); 387 KASSERT(!cv_has_waiters(&bp->b_done)); 388 389 vholdl(vp); 390 bp->b_vp = vp; 391 if (vp->v_type == VBLK || vp->v_type == VCHR) 392 bp->b_dev = vp->v_rdev; 393 else 394 bp->b_dev = NODEV; 395 396 /* 397 * Insert onto list for new vnode. 398 */ 399 bufinsvn(bp, &vp->v_cleanblkhd); 400 bp->b_objlock = vp->v_interlock; 401 } 402 403 /* 404 * Disassociate a buffer from a vnode. 405 */ 406 void 407 brelvp(struct buf *bp) 408 { 409 struct vnode *vp = bp->b_vp; 410 411 KASSERT(vp != NULL); 412 KASSERT(bp->b_objlock == vp->v_interlock); 413 KASSERT(mutex_owned(vp->v_interlock)); 414 KASSERT(mutex_owned(&bufcache_lock)); 415 KASSERT((bp->b_cflags & BC_BUSY) != 0); 416 KASSERT(!cv_has_waiters(&bp->b_done)); 417 418 /* 419 * Delete from old vnode list, if on one. 420 */ 421 if (LIST_NEXT(bp, b_vnbufs) != NOLIST) 422 bufremvn(bp); 423 424 if ((vp->v_iflag & (VI_ONWORKLST | VI_PAGES)) == VI_ONWORKLST && 425 LIST_FIRST(&vp->v_dirtyblkhd) == NULL) 426 vn_syncer_remove_from_worklist(vp); 427 428 bp->b_objlock = &buffer_lock; 429 bp->b_vp = NULL; 430 holdrelel(vp); 431 } 432 433 /* 434 * Reassign a buffer from one vnode list to another. 435 * The list reassignment must be within the same vnode. 436 * Used to assign file specific control information 437 * (indirect blocks) to the list to which they belong. 438 */ 439 void 440 reassignbuf(struct buf *bp, struct vnode *vp) 441 { 442 struct buflists *listheadp; 443 int delayx; 444 445 KASSERT(mutex_owned(&bufcache_lock)); 446 KASSERT(bp->b_objlock == vp->v_interlock); 447 KASSERT(mutex_owned(vp->v_interlock)); 448 KASSERT((bp->b_cflags & BC_BUSY) != 0); 449 450 /* 451 * Delete from old vnode list, if on one. 452 */ 453 if (LIST_NEXT(bp, b_vnbufs) != NOLIST) 454 bufremvn(bp); 455 456 /* 457 * If dirty, put on list of dirty buffers; 458 * otherwise insert onto list of clean buffers. 459 */ 460 if ((bp->b_oflags & BO_DELWRI) == 0) { 461 listheadp = &vp->v_cleanblkhd; 462 if ((vp->v_iflag & (VI_ONWORKLST | VI_PAGES)) == 463 VI_ONWORKLST && 464 LIST_FIRST(&vp->v_dirtyblkhd) == NULL) 465 vn_syncer_remove_from_worklist(vp); 466 } else { 467 listheadp = &vp->v_dirtyblkhd; 468 if ((vp->v_iflag & VI_ONWORKLST) == 0) { 469 switch (vp->v_type) { 470 case VDIR: 471 delayx = dirdelay; 472 break; 473 case VBLK: 474 if (spec_node_getmountedfs(vp) != NULL) { 475 delayx = metadelay; 476 break; 477 } 478 /* fall through */ 479 default: 480 delayx = filedelay; 481 break; 482 } 483 if (!vp->v_mount || 484 (vp->v_mount->mnt_flag & MNT_ASYNC) == 0) 485 vn_syncer_add_to_worklist(vp, delayx); 486 } 487 } 488 bufinsvn(bp, listheadp); 489 } 490 491 /* 492 * Lookup a vnode by device number and return it referenced. 493 */ 494 int 495 vfinddev(dev_t dev, enum vtype type, vnode_t **vpp) 496 { 497 498 return (spec_node_lookup_by_dev(type, dev, VDEAD_NOWAIT, vpp) == 0); 499 } 500 501 /* 502 * Revoke all the vnodes corresponding to the specified minor number 503 * range (endpoints inclusive) of the specified major. 504 */ 505 void 506 vdevgone(int maj, int minl, int minh, enum vtype type) 507 { 508 vnode_t *vp; 509 dev_t dev; 510 int mn; 511 512 for (mn = minl; mn <= minh; mn++) { 513 dev = makedev(maj, mn); 514 /* 515 * Notify anyone trying to get at this device that it 516 * has been detached, and then revoke it. 517 */ 518 switch (type) { 519 case VBLK: 520 bdev_detached(dev); 521 break; 522 case VCHR: 523 cdev_detached(dev); 524 break; 525 default: 526 panic("invalid specnode type: %d", type); 527 } 528 /* 529 * Passing 0 as flags, instead of VDEAD_NOWAIT, means 530 * spec_node_lookup_by_dev will wait for vnodes it 531 * finds concurrently being revoked before returning. 532 */ 533 while (spec_node_lookup_by_dev(type, dev, 0, &vp) == 0) { 534 VOP_REVOKE(vp, REVOKEALL); 535 vrele(vp); 536 } 537 } 538 } 539 540 /* 541 * The filesystem synchronizer mechanism - syncer. 542 * 543 * It is useful to delay writes of file data and filesystem metadata for 544 * a certain amount of time so that quickly created and deleted files need 545 * not waste disk bandwidth being created and removed. To implement this, 546 * vnodes are appended to a "workitem" queue. 547 * 548 * Most pending metadata should not wait for more than ten seconds. Thus, 549 * mounted on block devices are delayed only about a half the time that file 550 * data is delayed. Similarly, directory updates are more critical, so are 551 * only delayed about a third the time that file data is delayed. 552 * 553 * There are SYNCER_MAXDELAY queues that are processed in a round-robin 554 * manner at a rate of one each second (driven off the filesystem syner 555 * thread). The syncer_delayno variable indicates the next queue that is 556 * to be processed. Items that need to be processed soon are placed in 557 * this queue: 558 * 559 * syncer_workitem_pending[syncer_delayno] 560 * 561 * A delay of e.g. fifteen seconds is done by placing the request fifteen 562 * entries later in the queue: 563 * 564 * syncer_workitem_pending[(syncer_delayno + 15) & syncer_mask] 565 * 566 * Flag VI_ONWORKLST indicates that vnode is added into the queue. 567 */ 568 569 #define SYNCER_MAXDELAY 32 570 571 typedef TAILQ_HEAD(synclist, vnode_impl) synclist_t; 572 573 static void vn_syncer_add1(struct vnode *, int); 574 static void sysctl_vfs_syncfs_setup(struct sysctllog **); 575 576 /* 577 * Defines and variables for the syncer process. 578 */ 579 int syncer_maxdelay = SYNCER_MAXDELAY; /* maximum delay time */ 580 time_t syncdelay = 30; /* max time to delay syncing data */ 581 time_t filedelay = 30; /* time to delay syncing files */ 582 time_t dirdelay = 15; /* time to delay syncing directories */ 583 time_t metadelay = 10; /* time to delay syncing metadata */ 584 time_t lockdelay = 1; /* time to delay if locking fails */ 585 586 static kmutex_t syncer_data_lock; /* short term lock on data structs */ 587 588 static int syncer_delayno = 0; 589 static long syncer_last; 590 static synclist_t * syncer_workitem_pending; 591 592 static void 593 vn_initialize_syncerd(void) 594 { 595 int i; 596 597 syncer_last = SYNCER_MAXDELAY + 2; 598 599 sysctl_vfs_syncfs_setup(NULL); 600 601 syncer_workitem_pending = 602 kmem_alloc(syncer_last * sizeof (struct synclist), KM_SLEEP); 603 604 for (i = 0; i < syncer_last; i++) 605 TAILQ_INIT(&syncer_workitem_pending[i]); 606 607 mutex_init(&syncer_data_lock, MUTEX_DEFAULT, IPL_NONE); 608 } 609 610 /* 611 * Return delay factor appropriate for the given file system. For 612 * WAPBL we use the sync vnode to burst out metadata updates: sync 613 * those file systems more frequently. 614 */ 615 static inline int 616 sync_delay(struct mount *mp) 617 { 618 619 return mp->mnt_wapbl != NULL ? metadelay : syncdelay; 620 } 621 622 /* 623 * Compute the next slot index from delay. 624 */ 625 static inline int 626 sync_delay_slot(int delayx) 627 { 628 629 if (delayx > syncer_maxdelay - 2) 630 delayx = syncer_maxdelay - 2; 631 return (syncer_delayno + delayx) % syncer_last; 632 } 633 634 /* 635 * Add an item to the syncer work queue. 636 */ 637 static void 638 vn_syncer_add1(struct vnode *vp, int delayx) 639 { 640 synclist_t *slp; 641 vnode_impl_t *vip = VNODE_TO_VIMPL(vp); 642 643 KASSERT(mutex_owned(&syncer_data_lock)); 644 645 if (vp->v_iflag & VI_ONWORKLST) { 646 /* 647 * Remove in order to adjust the position of the vnode. 648 * Note: called from sched_sync(), which will not hold 649 * interlock, therefore we cannot modify v_iflag here. 650 */ 651 slp = &syncer_workitem_pending[vip->vi_synclist_slot]; 652 TAILQ_REMOVE(slp, vip, vi_synclist); 653 } else { 654 KASSERT(mutex_owned(vp->v_interlock)); 655 vp->v_iflag |= VI_ONWORKLST; 656 } 657 658 vip->vi_synclist_slot = sync_delay_slot(delayx); 659 660 slp = &syncer_workitem_pending[vip->vi_synclist_slot]; 661 TAILQ_INSERT_TAIL(slp, vip, vi_synclist); 662 } 663 664 void 665 vn_syncer_add_to_worklist(struct vnode *vp, int delayx) 666 { 667 668 KASSERT(mutex_owned(vp->v_interlock)); 669 670 mutex_enter(&syncer_data_lock); 671 vn_syncer_add1(vp, delayx); 672 mutex_exit(&syncer_data_lock); 673 } 674 675 /* 676 * Remove an item from the syncer work queue. 677 */ 678 void 679 vn_syncer_remove_from_worklist(struct vnode *vp) 680 { 681 synclist_t *slp; 682 vnode_impl_t *vip = VNODE_TO_VIMPL(vp); 683 684 KASSERT(mutex_owned(vp->v_interlock)); 685 686 if (vp->v_iflag & VI_ONWORKLST) { 687 mutex_enter(&syncer_data_lock); 688 vp->v_iflag &= ~VI_ONWORKLST; 689 slp = &syncer_workitem_pending[vip->vi_synclist_slot]; 690 TAILQ_REMOVE(slp, vip, vi_synclist); 691 mutex_exit(&syncer_data_lock); 692 } 693 } 694 695 /* 696 * Add this mount point to the syncer. 697 */ 698 void 699 vfs_syncer_add_to_worklist(struct mount *mp) 700 { 701 static int start, incr, next; 702 int vdelay; 703 704 KASSERT(mutex_owned(mp->mnt_updating)); 705 KASSERT((mp->mnt_iflag & IMNT_ONWORKLIST) == 0); 706 707 /* 708 * We attempt to scatter the mount points on the list 709 * so that they will go off at evenly distributed times 710 * even if all the filesystems are mounted at once. 711 */ 712 713 next += incr; 714 if (next == 0 || next > syncer_maxdelay) { 715 start /= 2; 716 incr /= 2; 717 if (start == 0) { 718 start = syncer_maxdelay / 2; 719 incr = syncer_maxdelay; 720 } 721 next = start; 722 } 723 mp->mnt_iflag |= IMNT_ONWORKLIST; 724 vdelay = sync_delay(mp); 725 mp->mnt_synclist_slot = vdelay > 0 ? next % vdelay : 0; 726 } 727 728 /* 729 * Remove the mount point from the syncer. 730 */ 731 void 732 vfs_syncer_remove_from_worklist(struct mount *mp) 733 { 734 735 KASSERT(mutex_owned(mp->mnt_updating)); 736 KASSERT((mp->mnt_iflag & IMNT_ONWORKLIST) != 0); 737 738 mp->mnt_iflag &= ~IMNT_ONWORKLIST; 739 } 740 741 /* 742 * Try lazy sync, return true on success. 743 */ 744 static bool 745 lazy_sync_vnode(struct vnode *vp) 746 { 747 bool synced; 748 749 KASSERT(mutex_owned(&syncer_data_lock)); 750 751 synced = false; 752 if (vcache_tryvget(vp) == 0) { 753 mutex_exit(&syncer_data_lock); 754 if (vn_lock(vp, LK_EXCLUSIVE | LK_NOWAIT) == 0) { 755 synced = true; 756 (void) VOP_FSYNC(vp, curlwp->l_cred, 757 FSYNC_LAZY, 0, 0); 758 vput(vp); 759 } else 760 vrele(vp); 761 mutex_enter(&syncer_data_lock); 762 } 763 return synced; 764 } 765 766 /* 767 * System filesystem synchronizer daemon. 768 */ 769 void 770 sched_sync(void *arg) 771 { 772 mount_iterator_t *iter; 773 synclist_t *slp; 774 struct vnode_impl *vi; 775 struct vnode *vp; 776 struct mount *mp; 777 time_t starttime; 778 bool synced; 779 780 for (;;) { 781 starttime = time_second; 782 783 /* 784 * Sync mounts whose dirty time has expired. 785 */ 786 mountlist_iterator_init(&iter); 787 while ((mp = mountlist_iterator_trynext(iter)) != NULL) { 788 if ((mp->mnt_iflag & IMNT_ONWORKLIST) == 0 || 789 mp->mnt_synclist_slot != syncer_delayno) { 790 continue; 791 } 792 mp->mnt_synclist_slot = sync_delay_slot(sync_delay(mp)); 793 VFS_SYNC(mp, MNT_LAZY, curlwp->l_cred); 794 } 795 mountlist_iterator_destroy(iter); 796 797 mutex_enter(&syncer_data_lock); 798 799 /* 800 * Push files whose dirty time has expired. 801 */ 802 slp = &syncer_workitem_pending[syncer_delayno]; 803 syncer_delayno += 1; 804 if (syncer_delayno >= syncer_last) 805 syncer_delayno = 0; 806 807 while ((vi = TAILQ_FIRST(slp)) != NULL) { 808 vp = VIMPL_TO_VNODE(vi); 809 synced = lazy_sync_vnode(vp); 810 811 /* 812 * XXX The vnode may have been recycled, in which 813 * case it may have a new identity. 814 */ 815 vi = TAILQ_FIRST(slp); 816 if (vi != NULL && VIMPL_TO_VNODE(vi) == vp) { 817 /* 818 * Put us back on the worklist. The worklist 819 * routine will remove us from our current 820 * position and then add us back in at a later 821 * position. 822 * 823 * Try again sooner rather than later if 824 * we were unable to lock the vnode. Lock 825 * failure should not prevent us from doing 826 * the sync "soon". 827 * 828 * If we locked it yet arrive here, it's 829 * likely that lazy sync is in progress and 830 * so the vnode still has dirty metadata. 831 * syncdelay is mainly to get this vnode out 832 * of the way so we do not consider it again 833 * "soon" in this loop, so the delay time is 834 * not critical as long as it is not "soon". 835 * While write-back strategy is the file 836 * system's domain, we expect write-back to 837 * occur no later than syncdelay seconds 838 * into the future. 839 */ 840 vn_syncer_add1(vp, 841 synced ? syncdelay : lockdelay); 842 } 843 } 844 845 /* 846 * If it has taken us less than a second to process the 847 * current work, then wait. Otherwise start right over 848 * again. We can still lose time if any single round 849 * takes more than two seconds, but it does not really 850 * matter as we are just trying to generally pace the 851 * filesystem activity. 852 */ 853 if (time_second == starttime) { 854 kpause("syncer", false, hz, &syncer_data_lock); 855 } 856 mutex_exit(&syncer_data_lock); 857 } 858 } 859 860 static void 861 sysctl_vfs_syncfs_setup(struct sysctllog **clog) 862 { 863 const struct sysctlnode *rnode, *cnode; 864 865 sysctl_createv(clog, 0, NULL, &rnode, 866 CTLFLAG_PERMANENT, 867 CTLTYPE_NODE, "sync", 868 SYSCTL_DESCR("syncer options"), 869 NULL, 0, NULL, 0, 870 CTL_VFS, CTL_CREATE, CTL_EOL); 871 872 sysctl_createv(clog, 0, &rnode, &cnode, 873 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 874 CTLTYPE_QUAD, "delay", 875 SYSCTL_DESCR("max time to delay syncing data"), 876 NULL, 0, &syncdelay, 0, 877 CTL_CREATE, CTL_EOL); 878 879 sysctl_createv(clog, 0, &rnode, &cnode, 880 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 881 CTLTYPE_QUAD, "filedelay", 882 SYSCTL_DESCR("time to delay syncing files"), 883 NULL, 0, &filedelay, 0, 884 CTL_CREATE, CTL_EOL); 885 886 sysctl_createv(clog, 0, &rnode, &cnode, 887 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 888 CTLTYPE_QUAD, "dirdelay", 889 SYSCTL_DESCR("time to delay syncing directories"), 890 NULL, 0, &dirdelay, 0, 891 CTL_CREATE, CTL_EOL); 892 893 sysctl_createv(clog, 0, &rnode, &cnode, 894 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 895 CTLTYPE_QUAD, "metadelay", 896 SYSCTL_DESCR("time to delay syncing metadata"), 897 NULL, 0, &metadelay, 0, 898 CTL_CREATE, CTL_EOL); 899 } 900 901 /* 902 * sysctl helper routine to return list of supported fstypes 903 */ 904 int 905 sysctl_vfs_generic_fstypes(SYSCTLFN_ARGS) 906 { 907 char bf[sizeof(((struct statvfs *)NULL)->f_fstypename)]; 908 char *where = oldp; 909 struct vfsops *v; 910 size_t needed, left, slen; 911 int error, first; 912 913 if (newp != NULL) 914 return (EPERM); 915 if (namelen != 0) 916 return (EINVAL); 917 918 first = 1; 919 error = 0; 920 needed = 0; 921 left = *oldlenp; 922 923 sysctl_unlock(); 924 mutex_enter(&vfs_list_lock); 925 LIST_FOREACH(v, &vfs_list, vfs_list) { 926 if (where == NULL) 927 needed += strlen(v->vfs_name) + 1; 928 else { 929 memset(bf, 0, sizeof(bf)); 930 if (first) { 931 strncpy(bf, v->vfs_name, sizeof(bf)); 932 first = 0; 933 } else { 934 bf[0] = ' '; 935 strncpy(bf + 1, v->vfs_name, sizeof(bf) - 1); 936 } 937 bf[sizeof(bf)-1] = '\0'; 938 slen = strlen(bf); 939 if (left < slen + 1) 940 break; 941 v->vfs_refcount++; 942 mutex_exit(&vfs_list_lock); 943 /* +1 to copy out the trailing NUL byte */ 944 error = copyout(bf, where, slen + 1); 945 mutex_enter(&vfs_list_lock); 946 v->vfs_refcount--; 947 if (error) 948 break; 949 where += slen; 950 needed += slen; 951 left -= slen; 952 } 953 } 954 mutex_exit(&vfs_list_lock); 955 sysctl_relock(); 956 *oldlenp = needed; 957 return (error); 958 } 959 960 int kinfo_vdebug = 1; 961 int kinfo_vgetfailed; 962 963 #define KINFO_VNODESLOP 10 964 965 /* 966 * Dump vnode list (via sysctl). 967 * Copyout address of vnode followed by vnode. 968 */ 969 int 970 sysctl_kern_vnode(SYSCTLFN_ARGS) 971 { 972 char *where = oldp; 973 size_t *sizep = oldlenp; 974 struct mount *mp; 975 vnode_t *vp, vbuf; 976 mount_iterator_t *iter; 977 struct vnode_iterator *marker; 978 char *bp = where; 979 char *ewhere; 980 int error; 981 982 if (namelen != 0) 983 return (EOPNOTSUPP); 984 if (newp != NULL) 985 return (EPERM); 986 987 #define VPTRSZ sizeof(vnode_t *) 988 #define VNODESZ sizeof(vnode_t) 989 if (where == NULL) { 990 *sizep = (numvnodes + KINFO_VNODESLOP) * (VPTRSZ + VNODESZ); 991 return (0); 992 } 993 ewhere = where + *sizep; 994 995 sysctl_unlock(); 996 mountlist_iterator_init(&iter); 997 while ((mp = mountlist_iterator_next(iter)) != NULL) { 998 vfs_vnode_iterator_init(mp, &marker); 999 while ((vp = vfs_vnode_iterator_next(marker, NULL, NULL))) { 1000 if (bp + VPTRSZ + VNODESZ > ewhere) { 1001 vrele(vp); 1002 vfs_vnode_iterator_destroy(marker); 1003 mountlist_iterator_destroy(iter); 1004 sysctl_relock(); 1005 *sizep = bp - where; 1006 return (ENOMEM); 1007 } 1008 memcpy(&vbuf, vp, VNODESZ); 1009 if ((error = copyout(&vp, bp, VPTRSZ)) || 1010 (error = copyout(&vbuf, bp + VPTRSZ, VNODESZ))) { 1011 vrele(vp); 1012 vfs_vnode_iterator_destroy(marker); 1013 mountlist_iterator_destroy(iter); 1014 sysctl_relock(); 1015 return (error); 1016 } 1017 vrele(vp); 1018 bp += VPTRSZ + VNODESZ; 1019 } 1020 vfs_vnode_iterator_destroy(marker); 1021 } 1022 mountlist_iterator_destroy(iter); 1023 sysctl_relock(); 1024 1025 *sizep = bp - where; 1026 return (0); 1027 } 1028 1029 /* 1030 * Set vnode attributes to VNOVAL 1031 */ 1032 void 1033 vattr_null(struct vattr *vap) 1034 { 1035 1036 memset(vap, 0, sizeof(*vap)); 1037 1038 vap->va_type = VNON; 1039 1040 /* 1041 * Assign individually so that it is safe even if size and 1042 * sign of each member are varied. 1043 */ 1044 vap->va_mode = VNOVAL; 1045 vap->va_nlink = VNOVAL; 1046 vap->va_uid = VNOVAL; 1047 vap->va_gid = VNOVAL; 1048 vap->va_fsid = VNOVAL; 1049 vap->va_fileid = VNOVAL; 1050 vap->va_size = VNOVAL; 1051 vap->va_blocksize = VNOVAL; 1052 vap->va_atime.tv_sec = 1053 vap->va_mtime.tv_sec = 1054 vap->va_ctime.tv_sec = 1055 vap->va_birthtime.tv_sec = VNOVAL; 1056 vap->va_atime.tv_nsec = 1057 vap->va_mtime.tv_nsec = 1058 vap->va_ctime.tv_nsec = 1059 vap->va_birthtime.tv_nsec = VNOVAL; 1060 vap->va_gen = VNOVAL; 1061 vap->va_flags = VNOVAL; 1062 vap->va_rdev = VNOVAL; 1063 vap->va_bytes = VNOVAL; 1064 } 1065 1066 /* 1067 * Vnode state to string. 1068 */ 1069 const char * 1070 vstate_name(enum vnode_state state) 1071 { 1072 1073 switch (state) { 1074 case VS_ACTIVE: 1075 return "ACTIVE"; 1076 case VS_MARKER: 1077 return "MARKER"; 1078 case VS_LOADING: 1079 return "LOADING"; 1080 case VS_LOADED: 1081 return "LOADED"; 1082 case VS_BLOCKED: 1083 return "BLOCKED"; 1084 case VS_RECLAIMING: 1085 return "RECLAIMING"; 1086 case VS_RECLAIMED: 1087 return "RECLAIMED"; 1088 default: 1089 return "ILLEGAL"; 1090 } 1091 } 1092 1093 /* 1094 * Print a description of a vnode (common part). 1095 */ 1096 static void 1097 vprint_common(struct vnode *vp, const char *prefix, 1098 void (*pr)(const char *, ...) __printflike(1, 2)) 1099 { 1100 int n; 1101 char bf[96]; 1102 const uint8_t *cp; 1103 vnode_impl_t *vip; 1104 const char * const vnode_tags[] = { VNODE_TAGS }; 1105 const char * const vnode_types[] = { VNODE_TYPES }; 1106 const char vnode_flagbits[] = VNODE_FLAGBITS; 1107 1108 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof(arr[0])) 1109 #define ARRAY_PRINT(idx, arr) \ 1110 ((unsigned int)(idx) < ARRAY_SIZE(arr) ? (arr)[(idx)] : "UNKNOWN") 1111 1112 vip = VNODE_TO_VIMPL(vp); 1113 1114 snprintb(bf, sizeof(bf), 1115 vnode_flagbits, vp->v_iflag | vp->v_vflag | vp->v_uflag); 1116 1117 (*pr)("vnode %p flags %s\n", vp, bf); 1118 (*pr)("%stag %s(%d) type %s(%d) mount %p typedata %p\n", prefix, 1119 ARRAY_PRINT(vp->v_tag, vnode_tags), vp->v_tag, 1120 ARRAY_PRINT(vp->v_type, vnode_types), vp->v_type, 1121 vp->v_mount, vp->v_mountedhere); 1122 (*pr)("%susecount %d writecount %d holdcount %d\n", prefix, 1123 vrefcnt(vp), vp->v_writecount, vp->v_holdcnt); 1124 (*pr)("%ssize %" PRIx64 " writesize %" PRIx64 " numoutput %d\n", 1125 prefix, vp->v_size, vp->v_writesize, vp->v_numoutput); 1126 (*pr)("%sdata %p lock %p\n", prefix, vp->v_data, &vip->vi_lock); 1127 1128 (*pr)("%sstate %s key(%p %zd)", prefix, vstate_name(vip->vi_state), 1129 vip->vi_key.vk_mount, vip->vi_key.vk_key_len); 1130 n = vip->vi_key.vk_key_len; 1131 cp = vip->vi_key.vk_key; 1132 while (n-- > 0) 1133 (*pr)(" %02x", *cp++); 1134 (*pr)("\n"); 1135 (*pr)("%slrulisthd %p\n", prefix, vip->vi_lrulisthd); 1136 1137 #undef ARRAY_PRINT 1138 #undef ARRAY_SIZE 1139 } 1140 1141 /* 1142 * Print out a description of a vnode. 1143 */ 1144 void 1145 vprint(const char *label, struct vnode *vp) 1146 { 1147 1148 if (label != NULL) 1149 printf("%s: ", label); 1150 vprint_common(vp, "\t", printf); 1151 if (vp->v_data != NULL) { 1152 printf("\t"); 1153 VOP_PRINT(vp); 1154 } 1155 } 1156 1157 /* 1158 * Given a file system name, look up the vfsops for that 1159 * file system, or return NULL if file system isn't present 1160 * in the kernel. 1161 */ 1162 struct vfsops * 1163 vfs_getopsbyname(const char *name) 1164 { 1165 struct vfsops *v; 1166 1167 mutex_enter(&vfs_list_lock); 1168 LIST_FOREACH(v, &vfs_list, vfs_list) { 1169 if (strcmp(v->vfs_name, name) == 0) 1170 break; 1171 } 1172 if (v != NULL) 1173 v->vfs_refcount++; 1174 mutex_exit(&vfs_list_lock); 1175 1176 return (v); 1177 } 1178 1179 void 1180 copy_statvfs_info(struct statvfs *sbp, const struct mount *mp) 1181 { 1182 const struct statvfs *mbp; 1183 1184 if (sbp == (mbp = &mp->mnt_stat)) 1185 return; 1186 1187 (void)memcpy(&sbp->f_fsidx, &mbp->f_fsidx, sizeof(sbp->f_fsidx)); 1188 sbp->f_fsid = mbp->f_fsid; 1189 sbp->f_owner = mbp->f_owner; 1190 sbp->f_flag = mbp->f_flag; 1191 sbp->f_syncwrites = mbp->f_syncwrites; 1192 sbp->f_asyncwrites = mbp->f_asyncwrites; 1193 sbp->f_syncreads = mbp->f_syncreads; 1194 sbp->f_asyncreads = mbp->f_asyncreads; 1195 (void)memcpy(sbp->f_spare, mbp->f_spare, sizeof(mbp->f_spare)); 1196 (void)memcpy(sbp->f_fstypename, mbp->f_fstypename, 1197 sizeof(sbp->f_fstypename)); 1198 (void)memcpy(sbp->f_mntonname, mbp->f_mntonname, 1199 sizeof(sbp->f_mntonname)); 1200 (void)memcpy(sbp->f_mntfromname, mp->mnt_stat.f_mntfromname, 1201 sizeof(sbp->f_mntfromname)); 1202 (void)memcpy(sbp->f_mntfromlabel, mp->mnt_stat.f_mntfromlabel, 1203 sizeof(sbp->f_mntfromlabel)); 1204 sbp->f_namemax = mbp->f_namemax; 1205 } 1206 1207 int 1208 set_statvfs_info(const char *onp, int ukon, const char *fromp, int ukfrom, 1209 const char *vfsname, struct mount *mp, struct lwp *l) 1210 { 1211 int error; 1212 size_t size; 1213 struct statvfs *sfs = &mp->mnt_stat; 1214 int (*fun)(const void *, void *, size_t, size_t *); 1215 1216 (void)strlcpy(mp->mnt_stat.f_fstypename, vfsname, 1217 sizeof(mp->mnt_stat.f_fstypename)); 1218 1219 if (onp) { 1220 struct cwdinfo *cwdi = l->l_proc->p_cwdi; 1221 fun = (ukon == UIO_SYSSPACE) ? copystr : copyinstr; 1222 if (cwdi->cwdi_rdir != NULL) { 1223 size_t len; 1224 char *bp; 1225 char *path = PNBUF_GET(); 1226 1227 bp = path + MAXPATHLEN; 1228 *--bp = '\0'; 1229 rw_enter(&cwdi->cwdi_lock, RW_READER); 1230 error = getcwd_common(cwdi->cwdi_rdir, rootvnode, &bp, 1231 path, MAXPATHLEN / 2, 0, l); 1232 rw_exit(&cwdi->cwdi_lock); 1233 if (error) { 1234 PNBUF_PUT(path); 1235 return error; 1236 } 1237 1238 len = strlen(bp); 1239 if (len > sizeof(sfs->f_mntonname) - 1) 1240 len = sizeof(sfs->f_mntonname) - 1; 1241 (void)strncpy(sfs->f_mntonname, bp, len); 1242 PNBUF_PUT(path); 1243 1244 if (len < sizeof(sfs->f_mntonname) - 1) { 1245 error = (*fun)(onp, &sfs->f_mntonname[len], 1246 sizeof(sfs->f_mntonname) - len - 1, &size); 1247 if (error) 1248 return error; 1249 size += len; 1250 } else { 1251 size = len; 1252 } 1253 } else { 1254 error = (*fun)(onp, &sfs->f_mntonname, 1255 sizeof(sfs->f_mntonname) - 1, &size); 1256 if (error) 1257 return error; 1258 } 1259 (void)memset(sfs->f_mntonname + size, 0, 1260 sizeof(sfs->f_mntonname) - size); 1261 } 1262 1263 if (fromp) { 1264 fun = (ukfrom == UIO_SYSSPACE) ? copystr : copyinstr; 1265 error = (*fun)(fromp, sfs->f_mntfromname, 1266 sizeof(sfs->f_mntfromname) - 1, &size); 1267 if (error) 1268 return error; 1269 (void)memset(sfs->f_mntfromname + size, 0, 1270 sizeof(sfs->f_mntfromname) - size); 1271 } 1272 return 0; 1273 } 1274 1275 /* 1276 * Knob to control the precision of file timestamps: 1277 * 1278 * 0 = seconds only; nanoseconds zeroed. 1279 * 1 = seconds and nanoseconds, accurate within 1/HZ. 1280 * 2 = seconds and nanoseconds, truncated to microseconds. 1281 * >=3 = seconds and nanoseconds, maximum precision. 1282 */ 1283 enum { TSP_SEC, TSP_HZ, TSP_USEC, TSP_NSEC }; 1284 1285 int vfs_timestamp_precision __read_mostly = TSP_NSEC; 1286 1287 void 1288 vfs_timestamp(struct timespec *tsp) 1289 { 1290 struct timeval tv; 1291 1292 switch (vfs_timestamp_precision) { 1293 case TSP_SEC: 1294 tsp->tv_sec = time_second; 1295 tsp->tv_nsec = 0; 1296 break; 1297 case TSP_HZ: 1298 getnanotime(tsp); 1299 break; 1300 case TSP_USEC: 1301 microtime(&tv); 1302 TIMEVAL_TO_TIMESPEC(&tv, tsp); 1303 break; 1304 case TSP_NSEC: 1305 default: 1306 nanotime(tsp); 1307 break; 1308 } 1309 } 1310 1311 /* 1312 * The purpose of this routine is to remove granularity from accmode_t, 1313 * reducing it into standard unix access bits - VEXEC, VREAD, VWRITE, 1314 * VADMIN and VAPPEND. 1315 * 1316 * If it returns 0, the caller is supposed to continue with the usual 1317 * access checks using 'accmode' as modified by this routine. If it 1318 * returns nonzero value, the caller is supposed to return that value 1319 * as errno. 1320 * 1321 * Note that after this routine runs, accmode may be zero. 1322 */ 1323 int 1324 vfs_unixify_accmode(accmode_t *accmode) 1325 { 1326 /* 1327 * There is no way to specify explicit "deny" rule using 1328 * file mode or POSIX.1e ACLs. 1329 */ 1330 if (*accmode & VEXPLICIT_DENY) { 1331 *accmode = 0; 1332 return (0); 1333 } 1334 1335 /* 1336 * None of these can be translated into usual access bits. 1337 * Also, the common case for NFSv4 ACLs is to not contain 1338 * either of these bits. Caller should check for VWRITE 1339 * on the containing directory instead. 1340 */ 1341 if (*accmode & (VDELETE_CHILD | VDELETE)) 1342 return (EPERM); 1343 1344 if (*accmode & VADMIN_PERMS) { 1345 *accmode &= ~VADMIN_PERMS; 1346 *accmode |= VADMIN; 1347 } 1348 1349 /* 1350 * There is no way to deny VREAD_ATTRIBUTES, VREAD_ACL 1351 * or VSYNCHRONIZE using file mode or POSIX.1e ACL. 1352 */ 1353 *accmode &= ~(VSTAT_PERMS | VSYNCHRONIZE); 1354 1355 return (0); 1356 } 1357 1358 time_t rootfstime; /* recorded root fs time, if known */ 1359 void 1360 setrootfstime(time_t t) 1361 { 1362 rootfstime = t; 1363 } 1364 1365 static const uint8_t vttodt_tab[ ] = { 1366 [VNON] = DT_UNKNOWN, 1367 [VREG] = DT_REG, 1368 [VDIR] = DT_DIR, 1369 [VBLK] = DT_BLK, 1370 [VCHR] = DT_CHR, 1371 [VLNK] = DT_LNK, 1372 [VSOCK] = DT_SOCK, 1373 [VFIFO] = DT_FIFO, 1374 [VBAD] = DT_UNKNOWN 1375 }; 1376 1377 uint8_t 1378 vtype2dt(enum vtype vt) 1379 { 1380 1381 CTASSERT(VBAD == __arraycount(vttodt_tab) - 1); 1382 return vttodt_tab[vt]; 1383 } 1384 1385 int 1386 VFS_MOUNT(struct mount *mp, const char *a, void *b, size_t *c) 1387 { 1388 int mpsafe = mp->mnt_iflag & IMNT_MPSAFE; 1389 int error; 1390 1391 /* 1392 * Note: The first time through, the vfs_mount function may set 1393 * IMNT_MPSAFE, so we have to cache it on entry in order to 1394 * avoid leaking a kernel lock. 1395 * 1396 * XXX Maybe the MPSAFE bit should be set in struct vfsops and 1397 * not in struct mount. 1398 */ 1399 if (mpsafe) { 1400 KERNEL_LOCK(1, NULL); 1401 } 1402 error = (*(mp->mnt_op->vfs_mount))(mp, a, b, c); 1403 if (mpsafe) { 1404 KERNEL_UNLOCK_ONE(NULL); 1405 } 1406 1407 return error; 1408 } 1409 1410 int 1411 VFS_START(struct mount *mp, int a) 1412 { 1413 int error; 1414 1415 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1416 KERNEL_LOCK(1, NULL); 1417 } 1418 error = (*(mp->mnt_op->vfs_start))(mp, a); 1419 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1420 KERNEL_UNLOCK_ONE(NULL); 1421 } 1422 1423 return error; 1424 } 1425 1426 int 1427 VFS_UNMOUNT(struct mount *mp, int a) 1428 { 1429 int error; 1430 1431 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1432 KERNEL_LOCK(1, NULL); 1433 } 1434 error = (*(mp->mnt_op->vfs_unmount))(mp, a); 1435 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1436 KERNEL_UNLOCK_ONE(NULL); 1437 } 1438 1439 return error; 1440 } 1441 1442 int 1443 VFS_ROOT(struct mount *mp, int lktype, struct vnode **a) 1444 { 1445 int error; 1446 1447 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1448 KERNEL_LOCK(1, NULL); 1449 } 1450 error = (*(mp->mnt_op->vfs_root))(mp, lktype, a); 1451 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1452 KERNEL_UNLOCK_ONE(NULL); 1453 } 1454 1455 return error; 1456 } 1457 1458 int 1459 VFS_QUOTACTL(struct mount *mp, struct quotactl_args *args) 1460 { 1461 int error; 1462 1463 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1464 KERNEL_LOCK(1, NULL); 1465 } 1466 error = (*(mp->mnt_op->vfs_quotactl))(mp, args); 1467 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1468 KERNEL_UNLOCK_ONE(NULL); 1469 } 1470 1471 return error; 1472 } 1473 1474 int 1475 VFS_STATVFS(struct mount *mp, struct statvfs *a) 1476 { 1477 int error; 1478 1479 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1480 KERNEL_LOCK(1, NULL); 1481 } 1482 error = (*(mp->mnt_op->vfs_statvfs))(mp, a); 1483 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1484 KERNEL_UNLOCK_ONE(NULL); 1485 } 1486 1487 return error; 1488 } 1489 1490 int 1491 VFS_SYNC(struct mount *mp, int a, struct kauth_cred *b) 1492 { 1493 int error; 1494 1495 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1496 KERNEL_LOCK(1, NULL); 1497 } 1498 error = (*(mp->mnt_op->vfs_sync))(mp, a, b); 1499 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1500 KERNEL_UNLOCK_ONE(NULL); 1501 } 1502 1503 return error; 1504 } 1505 1506 int 1507 VFS_FHTOVP(struct mount *mp, struct fid *a, int b, struct vnode **c) 1508 { 1509 int error; 1510 1511 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1512 KERNEL_LOCK(1, NULL); 1513 } 1514 error = (*(mp->mnt_op->vfs_fhtovp))(mp, a, b, c); 1515 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1516 KERNEL_UNLOCK_ONE(NULL); 1517 } 1518 1519 return error; 1520 } 1521 1522 int 1523 VFS_VPTOFH(struct vnode *vp, struct fid *a, size_t *b) 1524 { 1525 int error; 1526 1527 if ((vp->v_vflag & VV_MPSAFE) == 0) { 1528 KERNEL_LOCK(1, NULL); 1529 } 1530 error = (*(vp->v_mount->mnt_op->vfs_vptofh))(vp, a, b); 1531 if ((vp->v_vflag & VV_MPSAFE) == 0) { 1532 KERNEL_UNLOCK_ONE(NULL); 1533 } 1534 1535 return error; 1536 } 1537 1538 int 1539 VFS_SNAPSHOT(struct mount *mp, struct vnode *a, struct timespec *b) 1540 { 1541 int error; 1542 1543 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1544 KERNEL_LOCK(1, NULL); 1545 } 1546 error = (*(mp->mnt_op->vfs_snapshot))(mp, a, b); 1547 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1548 KERNEL_UNLOCK_ONE(NULL); 1549 } 1550 1551 return error; 1552 } 1553 1554 int 1555 VFS_EXTATTRCTL(struct mount *mp, int a, struct vnode *b, int c, const char *d) 1556 { 1557 int error; 1558 1559 KERNEL_LOCK(1, NULL); /* XXXSMP check ffs */ 1560 error = (*(mp->mnt_op->vfs_extattrctl))(mp, a, b, c, d); 1561 KERNEL_UNLOCK_ONE(NULL); /* XXX */ 1562 1563 return error; 1564 } 1565 1566 int 1567 VFS_SUSPENDCTL(struct mount *mp, int a) 1568 { 1569 int error; 1570 1571 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1572 KERNEL_LOCK(1, NULL); 1573 } 1574 error = (*(mp->mnt_op->vfs_suspendctl))(mp, a); 1575 if ((mp->mnt_iflag & IMNT_MPSAFE) == 0) { 1576 KERNEL_UNLOCK_ONE(NULL); 1577 } 1578 1579 return error; 1580 } 1581 1582 #if defined(DDB) || defined(DEBUGPRINT) 1583 static const char buf_flagbits[] = BUF_FLAGBITS; 1584 1585 void 1586 vfs_buf_print(struct buf *bp, int full, void (*pr)(const char *, ...)) 1587 { 1588 char bf[1024]; 1589 1590 (*pr)(" vp %p lblkno 0x%"PRIx64" blkno 0x%"PRIx64" rawblkno 0x%" 1591 PRIx64 " dev 0x%x\n", 1592 bp->b_vp, bp->b_lblkno, bp->b_blkno, bp->b_rawblkno, bp->b_dev); 1593 1594 snprintb(bf, sizeof(bf), 1595 buf_flagbits, bp->b_flags | bp->b_oflags | bp->b_cflags); 1596 (*pr)(" error %d flags %s\n", bp->b_error, bf); 1597 1598 (*pr)(" bufsize 0x%lx bcount 0x%lx resid 0x%lx\n", 1599 bp->b_bufsize, bp->b_bcount, bp->b_resid); 1600 (*pr)(" data %p saveaddr %p\n", 1601 bp->b_data, bp->b_saveaddr); 1602 (*pr)(" iodone %p objlock %p\n", bp->b_iodone, bp->b_objlock); 1603 } 1604 1605 void 1606 vfs_vnode_print(struct vnode *vp, int full, void (*pr)(const char *, ...)) 1607 { 1608 1609 uvm_object_printit(&vp->v_uobj, full, pr); 1610 (*pr)("\n"); 1611 vprint_common(vp, "", pr); 1612 if (full) { 1613 struct buf *bp; 1614 1615 (*pr)("clean bufs:\n"); 1616 LIST_FOREACH(bp, &vp->v_cleanblkhd, b_vnbufs) { 1617 (*pr)(" bp %p\n", bp); 1618 vfs_buf_print(bp, full, pr); 1619 } 1620 1621 (*pr)("dirty bufs:\n"); 1622 LIST_FOREACH(bp, &vp->v_dirtyblkhd, b_vnbufs) { 1623 (*pr)(" bp %p\n", bp); 1624 vfs_buf_print(bp, full, pr); 1625 } 1626 } 1627 } 1628 1629 void 1630 vfs_vnode_lock_print(void *vlock, int full, void (*pr)(const char *, ...)) 1631 { 1632 struct mount *mp; 1633 vnode_impl_t *vip; 1634 1635 for (mp = _mountlist_next(NULL); mp; mp = _mountlist_next(mp)) { 1636 TAILQ_FOREACH(vip, &mp->mnt_vnodelist, vi_mntvnodes) { 1637 if (&vip->vi_lock == vlock || 1638 VIMPL_TO_VNODE(vip)->v_interlock == vlock) 1639 vfs_vnode_print(VIMPL_TO_VNODE(vip), full, pr); 1640 } 1641 } 1642 } 1643 1644 void 1645 vfs_mount_print_all(int full, void (*pr)(const char *, ...)) 1646 { 1647 struct mount *mp; 1648 for (mp = _mountlist_next(NULL); mp; mp = _mountlist_next(mp)) 1649 vfs_mount_print(mp, full, pr); 1650 } 1651 1652 void 1653 vfs_mount_print(struct mount *mp, int full, void (*pr)(const char *, ...)) 1654 { 1655 char sbuf[256]; 1656 1657 (*pr)("vnodecovered = %p data = %p\n", 1658 mp->mnt_vnodecovered, mp->mnt_data); 1659 1660 (*pr)("fs_bshift %d dev_bshift = %d\n", 1661 mp->mnt_fs_bshift, mp->mnt_dev_bshift); 1662 1663 snprintb(sbuf, sizeof(sbuf), __MNT_FLAG_BITS, mp->mnt_flag); 1664 (*pr)("flag = %s\n", sbuf); 1665 1666 snprintb(sbuf, sizeof(sbuf), __IMNT_FLAG_BITS, mp->mnt_iflag); 1667 (*pr)("iflag = %s\n", sbuf); 1668 1669 (*pr)("refcnt = %d updating @ %p\n", mp->mnt_refcnt, mp->mnt_updating); 1670 1671 (*pr)("statvfs cache:\n"); 1672 (*pr)("\tbsize = %lu\n", mp->mnt_stat.f_bsize); 1673 (*pr)("\tfrsize = %lu\n", mp->mnt_stat.f_frsize); 1674 (*pr)("\tiosize = %lu\n", mp->mnt_stat.f_iosize); 1675 1676 (*pr)("\tblocks = %"PRIu64"\n", mp->mnt_stat.f_blocks); 1677 (*pr)("\tbfree = %"PRIu64"\n", mp->mnt_stat.f_bfree); 1678 (*pr)("\tbavail = %"PRIu64"\n", mp->mnt_stat.f_bavail); 1679 (*pr)("\tbresvd = %"PRIu64"\n", mp->mnt_stat.f_bresvd); 1680 1681 (*pr)("\tfiles = %"PRIu64"\n", mp->mnt_stat.f_files); 1682 (*pr)("\tffree = %"PRIu64"\n", mp->mnt_stat.f_ffree); 1683 (*pr)("\tfavail = %"PRIu64"\n", mp->mnt_stat.f_favail); 1684 (*pr)("\tfresvd = %"PRIu64"\n", mp->mnt_stat.f_fresvd); 1685 1686 (*pr)("\tf_fsidx = { 0x%"PRIx32", 0x%"PRIx32" }\n", 1687 mp->mnt_stat.f_fsidx.__fsid_val[0], 1688 mp->mnt_stat.f_fsidx.__fsid_val[1]); 1689 1690 (*pr)("\towner = %"PRIu32"\n", mp->mnt_stat.f_owner); 1691 (*pr)("\tnamemax = %lu\n", mp->mnt_stat.f_namemax); 1692 1693 snprintb(sbuf, sizeof(sbuf), __MNT_FLAG_BITS, mp->mnt_stat.f_flag); 1694 1695 (*pr)("\tflag = %s\n", sbuf); 1696 (*pr)("\tsyncwrites = %" PRIu64 "\n", mp->mnt_stat.f_syncwrites); 1697 (*pr)("\tasyncwrites = %" PRIu64 "\n", mp->mnt_stat.f_asyncwrites); 1698 (*pr)("\tsyncreads = %" PRIu64 "\n", mp->mnt_stat.f_syncreads); 1699 (*pr)("\tasyncreads = %" PRIu64 "\n", mp->mnt_stat.f_asyncreads); 1700 (*pr)("\tfstypename = %s\n", mp->mnt_stat.f_fstypename); 1701 (*pr)("\tmntonname = %s\n", mp->mnt_stat.f_mntonname); 1702 (*pr)("\tmntfromname = %s\n", mp->mnt_stat.f_mntfromname); 1703 1704 { 1705 int cnt = 0; 1706 vnode_t *vp; 1707 vnode_impl_t *vip; 1708 (*pr)("locked vnodes ="); 1709 TAILQ_FOREACH(vip, &mp->mnt_vnodelist, vi_mntvnodes) { 1710 vp = VIMPL_TO_VNODE(vip); 1711 if (VOP_ISLOCKED(vp)) { 1712 if ((++cnt % 6) == 0) { 1713 (*pr)(" %p,\n\t", vp); 1714 } else { 1715 (*pr)(" %p,", vp); 1716 } 1717 } 1718 } 1719 (*pr)("\n"); 1720 } 1721 1722 if (full) { 1723 int cnt = 0; 1724 vnode_t *vp; 1725 vnode_impl_t *vip; 1726 (*pr)("all vnodes ="); 1727 TAILQ_FOREACH(vip, &mp->mnt_vnodelist, vi_mntvnodes) { 1728 vp = VIMPL_TO_VNODE(vip); 1729 if (!TAILQ_NEXT(vip, vi_mntvnodes)) { 1730 (*pr)(" %p", vp); 1731 } else if ((++cnt % 6) == 0) { 1732 (*pr)(" %p,\n\t", vp); 1733 } else { 1734 (*pr)(" %p,", vp); 1735 } 1736 } 1737 (*pr)("\n"); 1738 } 1739 } 1740 1741 /* 1742 * List all of the locked vnodes in the system. 1743 */ 1744 void printlockedvnodes(void); 1745 1746 void 1747 printlockedvnodes(void) 1748 { 1749 struct mount *mp; 1750 vnode_t *vp; 1751 vnode_impl_t *vip; 1752 1753 printf("Locked vnodes\n"); 1754 for (mp = _mountlist_next(NULL); mp; mp = _mountlist_next(mp)) { 1755 TAILQ_FOREACH(vip, &mp->mnt_vnodelist, vi_mntvnodes) { 1756 vp = VIMPL_TO_VNODE(vip); 1757 if (VOP_ISLOCKED(vp)) 1758 vprint(NULL, vp); 1759 } 1760 } 1761 } 1762 1763 #endif /* DDB || DEBUGPRINT */
Cache object: 891e41b4329c8c499273e180ede10c79
[ source navigation ] [ diff markup ] [ identifier search ] [ freetext search ] [ file search ] [ list types ] [ track identifier ]