2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1989, 1993
5 * The Regents of the University of California. All rights reserved.
7 * This code is derived from software contributed
8 * to Berkeley by John Heidemann of the UCLA Ficus project.
10 * Source: * @(#)i405_init.c 2.10 92/04/27 UCLA Ficus project
12 * Redistribution and use in source and binary forms, with or without
13 * modification, are permitted provided that the following conditions
15 * 1. Redistributions of source code must retain the above copyright
16 * notice, this list of conditions and the following disclaimer.
17 * 2. Redistributions in binary form must reproduce the above copyright
18 * notice, this list of conditions and the following disclaimer in the
19 * documentation and/or other materials provided with the distribution.
20 * 3. Neither the name of the University nor the names of its contributors
21 * may be used to endorse or promote products derived from this software
22 * without specific prior written permission.
24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
37 #include <sys/cdefs.h>
38 #include <sys/param.h>
39 #include <sys/systm.h>
43 #include <sys/event.h>
44 #include <sys/filio.h>
45 #include <sys/kernel.h>
46 #include <sys/limits.h>
48 #include <sys/lockf.h>
49 #include <sys/malloc.h>
50 #include <sys/mount.h>
51 #include <sys/namei.h>
52 #include <sys/rwlock.h>
53 #include <sys/fcntl.h>
54 #include <sys/unistd.h>
55 #include <sys/vnode.h>
56 #include <sys/dirent.h>
59 #include <security/audit/audit.h>
62 #include <security/mac/mac_framework.h>
65 #include <vm/vm_object.h>
66 #include <vm/vm_extern.h>
68 #include <vm/vm_map.h>
69 #include <vm/vm_page.h>
70 #include <vm/vm_pager.h>
71 #include <vm/vnode_pager.h>
73 static int vop_nolookup(struct vop_lookup_args *);
74 static int vop_norename(struct vop_rename_args *);
75 static int vop_nostrategy(struct vop_strategy_args *);
76 static int dirent_exists(struct vnode *vp, const char *dirname,
79 static int vop_stdis_text(struct vop_is_text_args *ap);
80 static int vop_stdunset_text(struct vop_unset_text_args *ap);
81 static int vop_stdadd_writecount(struct vop_add_writecount_args *ap);
82 static int vop_stdcopy_file_range(struct vop_copy_file_range_args *ap);
83 static int vop_stdfdatasync(struct vop_fdatasync_args *ap);
84 static int vop_stdgetpages_async(struct vop_getpages_async_args *ap);
85 static int vop_stdread_pgcache(struct vop_read_pgcache_args *ap);
86 static int vop_stdstat(struct vop_stat_args *ap);
87 static int vop_stdvput_pair(struct vop_vput_pair_args *ap);
88 static int vop_stdgetlowvnode(struct vop_getlowvnode_args *ap);
91 * This vnode table stores what we want to do if the filesystem doesn't
92 * implement a particular VOP.
94 * If there is no specific entry here, we will return EOPNOTSUPP.
96 * Note that every filesystem has to implement either vop_access
97 * or vop_accessx; failing to do so will result in immediate crash
98 * due to stack overflow, as vop_stdaccess() calls vop_stdaccessx(),
99 * which calls vop_stdaccess() etc.
102 struct vop_vector default_vnodeops = {
104 .vop_bypass = VOP_EOPNOTSUPP,
106 .vop_access = vop_stdaccess,
107 .vop_accessx = vop_stdaccessx,
108 .vop_advise = vop_stdadvise,
109 .vop_advlock = vop_stdadvlock,
110 .vop_advlockasync = vop_stdadvlockasync,
111 .vop_advlockpurge = vop_stdadvlockpurge,
112 .vop_allocate = vop_stdallocate,
113 .vop_bmap = vop_stdbmap,
114 .vop_close = VOP_NULL,
115 .vop_fsync = VOP_NULL,
116 .vop_stat = vop_stdstat,
117 .vop_fdatasync = vop_stdfdatasync,
118 .vop_getlowvnode = vop_stdgetlowvnode,
119 .vop_getpages = vop_stdgetpages,
120 .vop_getpages_async = vop_stdgetpages_async,
121 .vop_getwritemount = vop_stdgetwritemount,
122 .vop_inactive = VOP_NULL,
123 .vop_need_inactive = vop_stdneed_inactive,
124 .vop_ioctl = vop_stdioctl,
125 .vop_kqfilter = vop_stdkqfilter,
126 .vop_islocked = vop_stdislocked,
127 .vop_lock1 = vop_stdlock,
128 .vop_lookup = vop_nolookup,
129 .vop_open = VOP_NULL,
130 .vop_pathconf = VOP_EINVAL,
131 .vop_poll = vop_nopoll,
132 .vop_putpages = vop_stdputpages,
133 .vop_readlink = VOP_EINVAL,
134 .vop_read_pgcache = vop_stdread_pgcache,
135 .vop_rename = vop_norename,
136 .vop_revoke = VOP_PANIC,
137 .vop_strategy = vop_nostrategy,
138 .vop_unlock = vop_stdunlock,
139 .vop_vptocnp = vop_stdvptocnp,
140 .vop_vptofh = vop_stdvptofh,
141 .vop_unp_bind = vop_stdunp_bind,
142 .vop_unp_connect = vop_stdunp_connect,
143 .vop_unp_detach = vop_stdunp_detach,
144 .vop_is_text = vop_stdis_text,
145 .vop_set_text = vop_stdset_text,
146 .vop_unset_text = vop_stdunset_text,
147 .vop_add_writecount = vop_stdadd_writecount,
148 .vop_copy_file_range = vop_stdcopy_file_range,
149 .vop_vput_pair = vop_stdvput_pair,
151 VFS_VOP_VECTOR_REGISTER(default_vnodeops);
154 * Series of placeholder functions for various error returns for
159 vop_eopnotsupp(struct vop_generic_args *ap)
162 printf("vop_notsupp[%s]\n", ap->a_desc->vdesc_name);
169 vop_ebadf(struct vop_generic_args *ap)
176 vop_enotty(struct vop_generic_args *ap)
183 vop_einval(struct vop_generic_args *ap)
190 vop_enoent(struct vop_generic_args *ap)
197 vop_eagain(struct vop_generic_args *ap)
204 vop_null(struct vop_generic_args *ap)
211 * Helper function to panic on some bad VOPs in some filesystems.
214 vop_panic(struct vop_generic_args *ap)
217 panic("filesystem goof: vop_panic[%s]", ap->a_desc->vdesc_name);
221 * vop_std<something> and vop_no<something> are default functions for use by
222 * filesystems that need the "default reasonable" implementation for a
223 * particular operation.
225 * The documentation for the operations they implement exists (if it exists)
226 * in the VOP_<SOMETHING>(9) manpage (all uppercase).
230 * Default vop for filesystems that do not support name lookup
233 vop_nolookup(struct vop_lookup_args *ap)
243 * Handle unlock and reference counting for arguments of vop_rename
244 * for filesystems that do not implement rename operation.
247 vop_norename(struct vop_rename_args *ap)
257 * Strategy routine for VFS devices that have none.
259 * BIO_ERROR and B_INVAL must be cleared prior to calling any strategy
260 * routine. Typically this is done for a BIO_READ strategy call.
261 * Typically B_INVAL is assumed to already be clear prior to a write
262 * and should not be cleared manually unless you just made the buffer
263 * invalid. BIO_ERROR should be cleared either way.
267 vop_nostrategy (struct vop_strategy_args *ap)
269 printf("No strategy for buffer at %p\n", ap->a_bp);
270 vn_printf(ap->a_vp, "vnode ");
271 ap->a_bp->b_ioflags |= BIO_ERROR;
272 ap->a_bp->b_error = EOPNOTSUPP;
278 * Check if a named file exists in a given directory vnode
280 * Returns 0 if the file exists, ENOENT if it doesn't, or errors returned by
281 * vn_dir_next_dirent().
284 dirent_exists(struct vnode *vp, const char *dirname, struct thread *td)
288 size_t dirbuflen, len;
293 ASSERT_VOP_LOCKED(vp, "vnode not locked");
294 KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp));
296 error = VOP_GETATTR(vp, &va, td->td_ucred);
300 dirbuflen = MAX(DEV_BSIZE, GENERIC_MAXDIRSIZ);
301 if (dirbuflen < va.va_blocksize)
302 dirbuflen = va.va_blocksize;
303 dirbuf = malloc(dirbuflen, M_TEMP, M_WAITOK);
310 error = vn_dir_next_dirent(vp, td, dirbuf, dirbuflen,
311 &dp, &len, &off, &eofflag);
318 if (dp->d_type != DT_WHT && dp->d_fileno != 0 &&
319 strcmp(dp->d_name, dirname) == 0)
326 free(dirbuf, M_TEMP);
331 vop_stdaccess(struct vop_access_args *ap)
334 KASSERT((ap->a_accmode & ~(VEXEC | VWRITE | VREAD | VADMIN |
335 VAPPEND)) == 0, ("invalid bit in accmode"));
337 return (VOP_ACCESSX(ap->a_vp, ap->a_accmode, ap->a_cred, ap->a_td));
341 vop_stdaccessx(struct vop_accessx_args *ap)
344 accmode_t accmode = ap->a_accmode;
346 error = vfs_unixify_accmode(&accmode);
353 return (VOP_ACCESS(ap->a_vp, accmode, ap->a_cred, ap->a_td));
357 * Advisory record locking support
360 vop_stdadvlock(struct vop_advlock_args *ap)
370 * Provide atomicity of open(O_CREAT | O_EXCL | O_EXLOCK) for
371 * local filesystems. See vn_open_cred() for reciprocal part.
374 if (mp != NULL && (mp->mnt_flag & MNT_LOCAL) != 0 &&
375 ap->a_op == F_SETLK && (ap->a_flags & F_FIRSTOPEN) == 0) {
377 while ((vp->v_iflag & VI_FOPENING) != 0)
378 msleep(vp, VI_MTX(vp), PLOCK, "lockfo", 0);
382 if (ap->a_fl->l_whence == SEEK_END) {
384 * The NFSv4 server must avoid doing a vn_lock() here, since it
385 * can deadlock the nfsd threads, due to a LOR. Fortunately
386 * the NFSv4 server always uses SEEK_SET and this code is
387 * only required for the SEEK_END case.
389 vn_lock(vp, LK_SHARED | LK_RETRY);
390 error = VOP_GETATTR(vp, &vattr, curthread->td_ucred);
397 return (lf_advlock(ap, &(vp->v_lockf), vattr.va_size));
401 vop_stdadvlockasync(struct vop_advlockasync_args *ap)
408 if (ap->a_fl->l_whence == SEEK_END) {
409 /* The size argument is only needed for SEEK_END. */
410 vn_lock(vp, LK_SHARED | LK_RETRY);
411 error = VOP_GETATTR(vp, &vattr, curthread->td_ucred);
418 return (lf_advlockasync(ap, &(vp->v_lockf), vattr.va_size));
422 vop_stdadvlockpurge(struct vop_advlockpurge_args *ap)
427 lf_purgelocks(vp, &vp->v_lockf);
434 * Standard implementation of POSIX pathconf, to get information about limits
436 * Override per filesystem for the case where the filesystem has smaller
440 vop_stdpathconf(struct vop_pathconf_args *ap)
443 switch (ap->a_name) {
445 *ap->a_retval = _POSIX_ASYNCHRONOUS_IO;
448 *ap->a_retval = PATH_MAX;
450 case _PC_ACL_EXTENDED:
452 case _PC_CAP_PRESENT:
453 case _PC_INF_PRESENT:
454 case _PC_MAC_PRESENT:
464 * Standard lock, unlock and islocked functions.
467 vop_stdlock(struct vop_lock1_args *ap)
469 struct vnode *vp = ap->a_vp;
473 return (lockmgr_lock_flags(vp->v_vnlock, ap->a_flags,
474 &ilk->lock_object, ap->a_file, ap->a_line));
479 vop_stdunlock(struct vop_unlock_args *ap)
481 struct vnode *vp = ap->a_vp;
483 return (lockmgr_unlock(vp->v_vnlock));
488 vop_stdislocked(struct vop_islocked_args *ap)
491 return (lockstatus(ap->a_vp->v_vnlock));
495 * Variants of the above set.
498 * - shared locking disablement is not supported
499 * - v_vnlock pointer is not honored
502 vop_lock(struct vop_lock1_args *ap)
504 struct vnode *vp = ap->a_vp;
505 int flags = ap->a_flags;
508 MPASS(vp->v_vnlock == &vp->v_lock);
510 if (__predict_false((flags & ~(LK_TYPE_MASK | LK_NODDLKTREAT | LK_RETRY)) != 0))
513 switch (flags & LK_TYPE_MASK) {
515 return (lockmgr_slock(&vp->v_lock, flags, ap->a_file, ap->a_line));
517 return (lockmgr_xlock(&vp->v_lock, flags, ap->a_file, ap->a_line));
521 return (lockmgr_lock_flags(&vp->v_lock, flags,
522 &ilk->lock_object, ap->a_file, ap->a_line));
526 vop_unlock(struct vop_unlock_args *ap)
528 struct vnode *vp = ap->a_vp;
530 MPASS(vp->v_vnlock == &vp->v_lock);
532 return (lockmgr_unlock(&vp->v_lock));
536 vop_islocked(struct vop_islocked_args *ap)
538 struct vnode *vp = ap->a_vp;
540 MPASS(vp->v_vnlock == &vp->v_lock);
542 return (lockstatus(&vp->v_lock));
546 * Return true for select/poll.
549 vop_nopoll(struct vop_poll_args *ap)
552 if (ap->a_events & ~POLLSTANDARD)
554 return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
558 * Implement poll for local filesystems that support it.
561 vop_stdpoll(struct vop_poll_args *ap)
563 if (ap->a_events & ~POLLSTANDARD)
564 return (vn_pollrecord(ap->a_vp, ap->a_td, ap->a_events));
565 return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
569 * Return our mount point, as we will take charge of the writes.
572 vop_stdgetwritemount(struct vop_getwritemount_args *ap)
578 * Note that having a reference does not prevent forced unmount from
579 * setting ->v_mount to NULL after the lock gets released. This is of
580 * no consequence for typical consumers (most notably vn_start_write)
581 * since in this case the vnode is VIRF_DOOMED. Unmount might have
582 * progressed far enough that its completion is only delayed by the
583 * reference obtained here. The consumer only needs to concern itself
587 mp = vfs_ref_from_vp(vp);
593 * If the file system doesn't implement VOP_BMAP, then return sensible defaults:
594 * - Return the vnode's bufobj instead of any underlying device's bufobj
595 * - Calculate the physical block number as if there were equal size
596 * consecutive blocks, but
597 * - Report no contiguous runs of blocks.
600 vop_stdbmap(struct vop_bmap_args *ap)
603 if (ap->a_bop != NULL)
604 *ap->a_bop = &ap->a_vp->v_bufobj;
605 if (ap->a_bnp != NULL)
606 *ap->a_bnp = ap->a_bn * btodb(ap->a_vp->v_mount->mnt_stat.f_iosize);
607 if (ap->a_runp != NULL)
609 if (ap->a_runb != NULL)
615 vop_stdfsync(struct vop_fsync_args *ap)
618 return (vn_fsync_buf(ap->a_vp, ap->a_waitfor));
622 vop_stdfdatasync(struct vop_fdatasync_args *ap)
625 return (VOP_FSYNC(ap->a_vp, MNT_WAIT, ap->a_td));
629 vop_stdfdatasync_buf(struct vop_fdatasync_args *ap)
632 return (vn_fsync_buf(ap->a_vp, MNT_WAIT));
635 /* XXX Needs good comment and more info in the manpage (VOP_GETPAGES(9)). */
637 vop_stdgetpages(struct vop_getpages_args *ap)
640 return vnode_pager_generic_getpages(ap->a_vp, ap->a_m,
641 ap->a_count, ap->a_rbehind, ap->a_rahead, NULL, NULL);
645 vop_stdgetpages_async(struct vop_getpages_async_args *ap)
649 error = VOP_GETPAGES(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind,
651 if (ap->a_iodone != NULL)
652 ap->a_iodone(ap->a_arg, ap->a_m, ap->a_count, error);
657 vop_stdkqfilter(struct vop_kqfilter_args *ap)
659 return vfs_kqfilter(ap);
662 /* XXX Needs good comment and more info in the manpage (VOP_PUTPAGES(9)). */
664 vop_stdputpages(struct vop_putpages_args *ap)
667 return vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count,
668 ap->a_sync, ap->a_rtvals);
672 vop_stdvptofh(struct vop_vptofh_args *ap)
678 vop_stdvptocnp(struct vop_vptocnp_args *ap)
680 struct vnode *const vp = ap->a_vp;
681 struct vnode **const dvp = ap->a_vpp;
682 char *buf = ap->a_buf;
683 size_t *buflen = ap->a_buflen;
686 int error = 0, covered = 0;
687 int eofflag, flags, locked;
688 size_t dirbuflen, len;
693 struct thread *const td = curthread;
694 struct ucred *const cred = td->td_ucred;
698 if (vp->v_type != VDIR)
701 error = VOP_GETATTR(vp, &va, cred);
706 locked = VOP_ISLOCKED(vp);
708 NDINIT_ATVP(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF, UIO_SYSSPACE,
711 error = vn_open_cred(&nd, &flags, 0, VN_OPEN_NOAUDIT, cred, NULL);
713 vn_lock(vp, locked | LK_RETRY);
716 NDFREE(&nd, NDF_ONLY_PNBUF);
718 mvp = *dvp = nd.ni_vp;
720 if (vp->v_mount != (*dvp)->v_mount &&
721 ((*dvp)->v_vflag & VV_ROOT) &&
722 ((*dvp)->v_mount->mnt_flag & MNT_UNION)) {
723 *dvp = (*dvp)->v_mount->mnt_vnodecovered;
726 vn_close(mvp, FREAD, cred, td);
728 vn_lock(*dvp, LK_SHARED | LK_RETRY);
732 fileno = va.va_fileid;
734 dirbuflen = MAX(DEV_BSIZE, GENERIC_MAXDIRSIZ);
735 if (dirbuflen < va.va_blocksize)
736 dirbuflen = va.va_blocksize;
737 dirbuf = malloc(dirbuflen, M_TEMP, M_WAITOK);
739 if ((*dvp)->v_type != VDIR) {
749 /* call VOP_READDIR of parent */
750 error = vn_dir_next_dirent(*dvp, td,
751 dirbuf, dirbuflen, &dp, &len, &off, &eofflag);
760 if ((dp->d_type != DT_WHT) &&
761 (dp->d_fileno == fileno)) {
764 vn_lock(mvp, LK_SHARED | LK_RETRY);
765 if (dirent_exists(mvp, dp->d_name, td) == 0) {
768 vn_lock(*dvp, LK_SHARED | LK_RETRY);
772 vn_lock(*dvp, LK_SHARED | LK_RETRY);
780 if (dp->d_namlen == 1 && dp->d_name[0] == '.') {
783 bcopy(dp->d_name, buf + i, dp->d_namlen);
791 free(dirbuf, M_TEMP);
801 vn_close(mvp, FREAD, cred, td);
803 vn_lock(vp, locked | LK_RETRY);
808 vop_stdallocate(struct vop_allocate_args *ap)
812 off_t maxfilesize = 0;
815 struct vattr vattr, *vap;
817 off_t fsize, len, cur, offset;
830 offset = *ap->a_offset;
832 error = VOP_GETATTR(vp, vap, ap->a_cred);
835 fsize = vap->va_size;
836 iosize = vap->va_blocksize;
838 iosize = BLKDEV_IOSIZE;
839 if (iosize > maxphys)
841 buf = malloc(iosize, M_TEMP, M_WAITOK);
845 * Check if the filesystem sets f_maxfilesize; if not use
846 * VOP_SETATTR to perform the check.
848 sfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK);
849 error = VFS_STATFS(vp->v_mount, sfs, td);
851 maxfilesize = sfs->f_maxfilesize;
856 if (offset > maxfilesize || len > maxfilesize ||
857 offset + len > maxfilesize) {
863 if (offset + len > vap->va_size) {
865 * Test offset + len against the filesystem's maxfilesize.
868 vap->va_size = offset + len;
869 error = VOP_SETATTR(vp, vap, ap->a_cred);
873 vap->va_size = fsize;
874 error = VOP_SETATTR(vp, vap, ap->a_cred);
881 * Read and write back anything below the nominal file
882 * size. There's currently no way outside the filesystem
883 * to know whether this area is sparse or not.
886 if ((offset % iosize) != 0)
887 cur -= (offset % iosize);
890 if (offset < fsize) {
893 auio.uio_iov = &aiov;
895 auio.uio_offset = offset;
896 auio.uio_resid = cur;
897 auio.uio_segflg = UIO_SYSSPACE;
898 auio.uio_rw = UIO_READ;
900 error = VOP_READ(vp, &auio, ap->a_ioflag, ap->a_cred);
903 if (auio.uio_resid > 0) {
904 bzero(buf + cur - auio.uio_resid,
913 auio.uio_iov = &aiov;
915 auio.uio_offset = offset;
916 auio.uio_resid = cur;
917 auio.uio_segflg = UIO_SYSSPACE;
918 auio.uio_rw = UIO_WRITE;
921 error = VOP_WRITE(vp, &auio, ap->a_ioflag, ap->a_cred);
935 *ap->a_offset = offset;
941 vop_stdadvise(struct vop_advise_args *ap)
945 daddr_t startn, endn;
946 off_t bstart, bend, start, end;
950 switch (ap->a_advice) {
951 case POSIX_FADV_WILLNEED:
953 * Do nothing for now. Filesystems should provide a
954 * custom method which starts an asynchronous read of
955 * the requested region.
959 case POSIX_FADV_DONTNEED:
961 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
962 if (VN_IS_DOOMED(vp)) {
968 * Round to block boundaries (and later possibly further to
969 * page boundaries). Applications cannot reasonably be aware
970 * of the boundaries, and the rounding must be to expand at
971 * both extremities to cover enough. It still doesn't cover
972 * read-ahead. For partial blocks, this gives unnecessary
973 * discarding of buffers but is efficient enough since the
974 * pages usually remain in VMIO for some time.
976 bsize = vp->v_bufobj.bo_bsize;
977 bstart = rounddown(ap->a_start, bsize);
978 bend = roundup(ap->a_end, bsize);
981 * Deactivate pages in the specified range from the backing VM
982 * object. Pages that are resident in the buffer cache will
983 * remain wired until their corresponding buffers are released
986 if (vp->v_object != NULL) {
987 start = trunc_page(bstart);
988 end = round_page(bend);
989 VM_OBJECT_RLOCK(vp->v_object);
990 vm_object_page_noreuse(vp->v_object, OFF_TO_IDX(start),
992 VM_OBJECT_RUNLOCK(vp->v_object);
997 startn = bstart / bsize;
999 error = bnoreuselist(&bo->bo_clean, bo, startn, endn);
1001 error = bnoreuselist(&bo->bo_dirty, bo, startn, endn);
1013 vop_stdunp_bind(struct vop_unp_bind_args *ap)
1016 ap->a_vp->v_unpcb = ap->a_unpcb;
1021 vop_stdunp_connect(struct vop_unp_connect_args *ap)
1024 *ap->a_unpcb = ap->a_vp->v_unpcb;
1029 vop_stdunp_detach(struct vop_unp_detach_args *ap)
1032 ap->a_vp->v_unpcb = NULL;
1037 vop_stdis_text(struct vop_is_text_args *ap)
1040 return (ap->a_vp->v_writecount < 0);
1044 vop_stdset_text(struct vop_set_text_args *ap)
1053 * Avoid the interlock if execs are already present.
1055 n = atomic_load_int(&vp->v_writecount);
1060 if (atomic_fcmpset_int(&vp->v_writecount, &n, n - 1)) {
1066 if (vp->v_writecount > 0) {
1070 * If requested by fs, keep a use reference to the
1071 * vnode until the last text reference is released.
1074 if (mp != NULL && (mp->mnt_kern_flag & MNTK_TEXT_REFS) != 0 &&
1075 vp->v_writecount == 0) {
1076 VNPASS((vp->v_iflag & VI_TEXT_REF) == 0, vp);
1077 vp->v_iflag |= VI_TEXT_REF;
1081 atomic_subtract_int(&vp->v_writecount, 1);
1089 vop_stdunset_text(struct vop_unset_text_args *ap)
1098 * Avoid the interlock if this is not the last exec.
1100 n = atomic_load_int(&vp->v_writecount);
1105 if (atomic_fcmpset_int(&vp->v_writecount, &n, n + 1)) {
1112 if (vp->v_writecount < 0) {
1113 if ((vp->v_iflag & VI_TEXT_REF) != 0 &&
1114 vp->v_writecount == -1) {
1116 vp->v_iflag &= ~VI_TEXT_REF;
1118 atomic_add_int(&vp->v_writecount, 1);
1130 vop_stdadd_writecount(struct vop_add_writecount_args *ap)
1137 VI_LOCK_FLAGS(vp, MTX_DUPOK);
1138 if (vp->v_writecount < 0) {
1141 VNASSERT(vp->v_writecount + ap->a_inc >= 0, vp,
1142 ("neg writecount increment %d", ap->a_inc));
1143 if (vp->v_writecount == 0) {
1145 if (mp != NULL && (mp->mnt_kern_flag & MNTK_NOMSYNC) == 0)
1148 vp->v_writecount += ap->a_inc;
1156 vop_stdneed_inactive(struct vop_need_inactive_args *ap)
1163 vop_stdioctl(struct vop_ioctl_args *ap)
1170 switch (ap->a_command) {
1174 error = vn_lock(vp, LK_SHARED);
1177 if (vp->v_type == VREG)
1178 error = VOP_GETATTR(vp, &va, ap->a_cred);
1183 if (*offp < 0 || *offp >= va.va_size)
1185 else if (ap->a_command == FIOSEEKHOLE)
1199 * used to fill the vfs function table to get reasonable default return values.
1202 vfs_stdroot(struct mount *mp, int flags, struct vnode **vpp)
1205 return (EOPNOTSUPP);
1209 vfs_stdstatfs(struct mount *mp, struct statfs *sbp)
1212 return (EOPNOTSUPP);
1216 vfs_stdquotactl(struct mount *mp, int cmds, uid_t uid, void *arg)
1219 return (EOPNOTSUPP);
1223 vfs_stdsync(struct mount *mp, int waitfor)
1225 struct vnode *vp, *mvp;
1227 int error, lockreq, allerror = 0;
1230 lockreq = LK_EXCLUSIVE | LK_INTERLOCK;
1231 if (waitfor != MNT_WAIT)
1232 lockreq |= LK_NOWAIT;
1234 * Force stale buffer cache information to be flushed.
1237 MNT_VNODE_FOREACH_ALL(vp, mp, mvp) {
1238 if (vp->v_bufobj.bo_dirty.bv_cnt == 0) {
1242 if ((error = vget(vp, lockreq)) != 0) {
1243 if (error == ENOENT) {
1244 MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
1249 error = VOP_FSYNC(vp, waitfor, td);
1258 vfs_stdnosync(struct mount *mp, int waitfor)
1265 vop_stdcopy_file_range(struct vop_copy_file_range_args *ap)
1269 error = vn_generic_copy_file_range(ap->a_invp, ap->a_inoffp,
1270 ap->a_outvp, ap->a_outoffp, ap->a_lenp, ap->a_flags, ap->a_incred,
1271 ap->a_outcred, ap->a_fsizetd);
1276 vfs_stdvget(struct mount *mp, ino_t ino, int flags, struct vnode **vpp)
1279 return (EOPNOTSUPP);
1283 vfs_stdfhtovp(struct mount *mp, struct fid *fhp, int flags, struct vnode **vpp)
1286 return (EOPNOTSUPP);
1290 vfs_stdinit(struct vfsconf *vfsp)
1297 vfs_stduninit(struct vfsconf *vfsp)
1304 vfs_stdextattrctl(struct mount *mp, int cmd, struct vnode *filename_vp,
1305 int attrnamespace, const char *attrname)
1308 if (filename_vp != NULL)
1309 VOP_UNLOCK(filename_vp);
1310 return (EOPNOTSUPP);
1314 vfs_stdsysctl(struct mount *mp, fsctlop_t op, struct sysctl_req *req)
1317 return (EOPNOTSUPP);
1320 static vop_bypass_t *
1321 bp_by_off(struct vop_vector *vop, struct vop_generic_args *a)
1324 return (*(vop_bypass_t **)((char *)vop + a->a_desc->vdesc_vop_offset));
1328 vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a)
1333 bp = bp_by_off(vop, a);
1336 prev_stops = sigdeferstop(SIGDEFERSTOP_SILENT);
1338 sigallowstop(prev_stops);
1343 vop_stdstat(struct vop_stat_args *a)
1355 error = vop_stat_helper_pre(a);
1362 * Initialize defaults for new and unusual fields, so that file
1363 * systems which don't support these fields don't need to know
1366 vap->va_birthtime.tv_sec = -1;
1367 vap->va_birthtime.tv_nsec = 0;
1368 vap->va_fsid = VNOVAL;
1369 vap->va_rdev = NODEV;
1371 error = VOP_GETATTR(vp, vap, a->a_active_cred);
1376 * Zero the spare stat fields
1378 bzero(sb, sizeof *sb);
1381 * Copy from vattr table
1383 if (vap->va_fsid != VNOVAL)
1384 sb->st_dev = vap->va_fsid;
1386 sb->st_dev = vp->v_mount->mnt_stat.f_fsid.val[0];
1387 sb->st_ino = vap->va_fileid;
1388 mode = vap->va_mode;
1389 switch (vap->va_type) {
1416 sb->st_nlink = vap->va_nlink;
1417 sb->st_uid = vap->va_uid;
1418 sb->st_gid = vap->va_gid;
1419 sb->st_rdev = vap->va_rdev;
1420 if (vap->va_size > OFF_MAX) {
1424 sb->st_size = vap->va_size;
1425 sb->st_atim.tv_sec = vap->va_atime.tv_sec;
1426 sb->st_atim.tv_nsec = vap->va_atime.tv_nsec;
1427 sb->st_mtim.tv_sec = vap->va_mtime.tv_sec;
1428 sb->st_mtim.tv_nsec = vap->va_mtime.tv_nsec;
1429 sb->st_ctim.tv_sec = vap->va_ctime.tv_sec;
1430 sb->st_ctim.tv_nsec = vap->va_ctime.tv_nsec;
1431 sb->st_birthtim.tv_sec = vap->va_birthtime.tv_sec;
1432 sb->st_birthtim.tv_nsec = vap->va_birthtime.tv_nsec;
1435 * According to www.opengroup.org, the meaning of st_blksize is
1436 * "a filesystem-specific preferred I/O block size for this
1437 * object. In some filesystem types, this may vary from file
1439 * Use minimum/default of PAGE_SIZE (e.g. for VCHR).
1442 sb->st_blksize = max(PAGE_SIZE, vap->va_blocksize);
1443 sb->st_flags = vap->va_flags;
1444 sb->st_blocks = vap->va_bytes / S_BLKSIZE;
1445 sb->st_gen = vap->va_gen;
1447 return (vop_stat_helper_post(a, error));
1451 vop_stdread_pgcache(struct vop_read_pgcache_args *ap __unused)
1453 return (EJUSTRETURN);
1457 vop_stdvput_pair(struct vop_vput_pair_args *ap)
1459 struct vnode *dvp, *vp, **vpp;
1464 if (vpp != NULL && ap->a_unlock_vp && (vp = *vpp) != NULL)
1470 vop_stdgetlowvnode(struct vop_getlowvnode_args *ap)
1473 *ap->a_vplp = ap->a_vp;