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 __FBSDID("$FreeBSD$");
40 #include <sys/param.h>
41 #include <sys/systm.h>
45 #include <sys/event.h>
46 #include <sys/filio.h>
47 #include <sys/kernel.h>
48 #include <sys/limits.h>
50 #include <sys/lockf.h>
51 #include <sys/malloc.h>
52 #include <sys/mount.h>
53 #include <sys/namei.h>
54 #include <sys/rwlock.h>
55 #include <sys/fcntl.h>
56 #include <sys/unistd.h>
57 #include <sys/vnode.h>
58 #include <sys/dirent.h>
61 #include <security/audit/audit.h>
64 #include <security/mac/mac_framework.h>
67 #include <vm/vm_object.h>
68 #include <vm/vm_extern.h>
70 #include <vm/vm_map.h>
71 #include <vm/vm_page.h>
72 #include <vm/vm_pager.h>
73 #include <vm/vnode_pager.h>
75 static int vop_nolookup(struct vop_lookup_args *);
76 static int vop_norename(struct vop_rename_args *);
77 static int vop_nostrategy(struct vop_strategy_args *);
78 static int get_next_dirent(struct vnode *vp, struct dirent **dpp,
79 char *dirbuf, int dirbuflen, off_t *off,
80 char **cpos, int *len, int *eofflag,
82 static int dirent_exists(struct vnode *vp, const char *dirname,
85 #define DIRENT_MINSIZE (sizeof(struct dirent) - (MAXNAMLEN+1) + 4)
87 static int vop_stdis_text(struct vop_is_text_args *ap);
88 static int vop_stdunset_text(struct vop_unset_text_args *ap);
89 static int vop_stdadd_writecount(struct vop_add_writecount_args *ap);
90 static int vop_stdcopy_file_range(struct vop_copy_file_range_args *ap);
91 static int vop_stdfdatasync(struct vop_fdatasync_args *ap);
92 static int vop_stdgetpages_async(struct vop_getpages_async_args *ap);
93 static int vop_stdread_pgcache(struct vop_read_pgcache_args *ap);
94 static int vop_stdstat(struct vop_stat_args *ap);
95 static int vop_stdvput_pair(struct vop_vput_pair_args *ap);
98 * This vnode table stores what we want to do if the filesystem doesn't
99 * implement a particular VOP.
101 * If there is no specific entry here, we will return EOPNOTSUPP.
103 * Note that every filesystem has to implement either vop_access
104 * or vop_accessx; failing to do so will result in immediate crash
105 * due to stack overflow, as vop_stdaccess() calls vop_stdaccessx(),
106 * which calls vop_stdaccess() etc.
109 struct vop_vector default_vnodeops = {
111 .vop_bypass = VOP_EOPNOTSUPP,
113 .vop_access = vop_stdaccess,
114 .vop_accessx = vop_stdaccessx,
115 .vop_advise = vop_stdadvise,
116 .vop_advlock = vop_stdadvlock,
117 .vop_advlockasync = vop_stdadvlockasync,
118 .vop_advlockpurge = vop_stdadvlockpurge,
119 .vop_allocate = vop_stdallocate,
120 .vop_bmap = vop_stdbmap,
121 .vop_close = VOP_NULL,
122 .vop_fsync = VOP_NULL,
123 .vop_stat = vop_stdstat,
124 .vop_fdatasync = vop_stdfdatasync,
125 .vop_getpages = vop_stdgetpages,
126 .vop_getpages_async = vop_stdgetpages_async,
127 .vop_getwritemount = vop_stdgetwritemount,
128 .vop_inactive = VOP_NULL,
129 .vop_need_inactive = vop_stdneed_inactive,
130 .vop_ioctl = vop_stdioctl,
131 .vop_kqfilter = vop_stdkqfilter,
132 .vop_islocked = vop_stdislocked,
133 .vop_lock1 = vop_stdlock,
134 .vop_lookup = vop_nolookup,
135 .vop_open = VOP_NULL,
136 .vop_pathconf = VOP_EINVAL,
137 .vop_poll = vop_nopoll,
138 .vop_putpages = vop_stdputpages,
139 .vop_readlink = VOP_EINVAL,
140 .vop_read_pgcache = vop_stdread_pgcache,
141 .vop_rename = vop_norename,
142 .vop_revoke = VOP_PANIC,
143 .vop_strategy = vop_nostrategy,
144 .vop_unlock = vop_stdunlock,
145 .vop_vptocnp = vop_stdvptocnp,
146 .vop_vptofh = vop_stdvptofh,
147 .vop_unp_bind = vop_stdunp_bind,
148 .vop_unp_connect = vop_stdunp_connect,
149 .vop_unp_detach = vop_stdunp_detach,
150 .vop_is_text = vop_stdis_text,
151 .vop_set_text = vop_stdset_text,
152 .vop_unset_text = vop_stdunset_text,
153 .vop_add_writecount = vop_stdadd_writecount,
154 .vop_copy_file_range = vop_stdcopy_file_range,
155 .vop_vput_pair = vop_stdvput_pair,
157 VFS_VOP_VECTOR_REGISTER(default_vnodeops);
160 * Series of placeholder functions for various error returns for
165 vop_eopnotsupp(struct vop_generic_args *ap)
168 printf("vop_notsupp[%s]\n", ap->a_desc->vdesc_name);
175 vop_ebadf(struct vop_generic_args *ap)
182 vop_enotty(struct vop_generic_args *ap)
189 vop_einval(struct vop_generic_args *ap)
196 vop_enoent(struct vop_generic_args *ap)
203 vop_eagain(struct vop_generic_args *ap)
210 vop_null(struct vop_generic_args *ap)
217 * Helper function to panic on some bad VOPs in some filesystems.
220 vop_panic(struct vop_generic_args *ap)
223 panic("filesystem goof: vop_panic[%s]", ap->a_desc->vdesc_name);
227 * vop_std<something> and vop_no<something> are default functions for use by
228 * filesystems that need the "default reasonable" implementation for a
229 * particular operation.
231 * The documentation for the operations they implement exists (if it exists)
232 * in the VOP_<SOMETHING>(9) manpage (all uppercase).
236 * Default vop for filesystems that do not support name lookup
240 struct vop_lookup_args /* {
242 struct vnode **a_vpp;
243 struct componentname *a_cnp;
254 * Handle unlock and reference counting for arguments of vop_rename
255 * for filesystems that do not implement rename operation.
258 vop_norename(struct vop_rename_args *ap)
268 * Strategy routine for VFS devices that have none.
270 * BIO_ERROR and B_INVAL must be cleared prior to calling any strategy
271 * routine. Typically this is done for a BIO_READ strategy call.
272 * Typically B_INVAL is assumed to already be clear prior to a write
273 * and should not be cleared manually unless you just made the buffer
274 * invalid. BIO_ERROR should be cleared either way.
278 vop_nostrategy (struct vop_strategy_args *ap)
280 printf("No strategy for buffer at %p\n", ap->a_bp);
281 vn_printf(ap->a_vp, "vnode ");
282 ap->a_bp->b_ioflags |= BIO_ERROR;
283 ap->a_bp->b_error = EOPNOTSUPP;
289 get_next_dirent(struct vnode *vp, struct dirent **dpp, char *dirbuf,
290 int dirbuflen, off_t *off, char **cpos, int *len,
291 int *eofflag, struct thread *td)
298 KASSERT(VOP_ISLOCKED(vp), ("vp %p is not locked", vp));
299 KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp));
302 iov.iov_base = dirbuf;
303 iov.iov_len = dirbuflen;
307 uio.uio_offset = *off;
308 uio.uio_resid = dirbuflen;
309 uio.uio_segflg = UIO_SYSSPACE;
310 uio.uio_rw = UIO_READ;
316 error = mac_vnode_check_readdir(td->td_ucred, vp);
319 error = VOP_READDIR(vp, &uio, td->td_ucred, eofflag,
324 *off = uio.uio_offset;
327 *len = (dirbuflen - uio.uio_resid);
333 dp = (struct dirent *)(*cpos);
334 reclen = dp->d_reclen;
337 /* check for malformed directory.. */
338 if (reclen < DIRENT_MINSIZE)
348 * Check if a named file exists in a given directory vnode.
351 dirent_exists(struct vnode *vp, const char *dirname, struct thread *td)
354 int error, eofflag, dirbuflen, len, found;
359 KASSERT(VOP_ISLOCKED(vp), ("vp %p is not locked", vp));
360 KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp));
364 error = VOP_GETATTR(vp, &va, td->td_ucred);
368 dirbuflen = DEV_BSIZE;
369 if (dirbuflen < va.va_blocksize)
370 dirbuflen = va.va_blocksize;
371 dirbuf = (char *)malloc(dirbuflen, M_TEMP, M_WAITOK);
376 error = get_next_dirent(vp, &dp, dirbuf, dirbuflen, &off,
377 &cpos, &len, &eofflag, td);
381 if (dp->d_type != DT_WHT && dp->d_fileno != 0 &&
382 strcmp(dp->d_name, dirname) == 0) {
386 } while (len > 0 || !eofflag);
389 free(dirbuf, M_TEMP);
394 vop_stdaccess(struct vop_access_args *ap)
397 KASSERT((ap->a_accmode & ~(VEXEC | VWRITE | VREAD | VADMIN |
398 VAPPEND)) == 0, ("invalid bit in accmode"));
400 return (VOP_ACCESSX(ap->a_vp, ap->a_accmode, ap->a_cred, ap->a_td));
404 vop_stdaccessx(struct vop_accessx_args *ap)
407 accmode_t accmode = ap->a_accmode;
409 error = vfs_unixify_accmode(&accmode);
416 return (VOP_ACCESS(ap->a_vp, accmode, ap->a_cred, ap->a_td));
420 * Advisory record locking support
423 vop_stdadvlock(struct vop_advlock_args *ap)
433 * Provide atomicity of open(O_CREAT | O_EXCL | O_EXLOCK) for
434 * local filesystems. See vn_open_cred() for reciprocal part.
437 if (mp != NULL && (mp->mnt_flag & MNT_LOCAL) != 0 &&
438 ap->a_op == F_SETLK && (ap->a_flags & F_FIRSTOPEN) == 0) {
440 while ((vp->v_iflag & VI_FOPENING) != 0)
441 msleep(vp, VI_MTX(vp), PLOCK, "lockfo", 0);
445 if (ap->a_fl->l_whence == SEEK_END) {
447 * The NFSv4 server must avoid doing a vn_lock() here, since it
448 * can deadlock the nfsd threads, due to a LOR. Fortunately
449 * the NFSv4 server always uses SEEK_SET and this code is
450 * only required for the SEEK_END case.
452 vn_lock(vp, LK_SHARED | LK_RETRY);
453 error = VOP_GETATTR(vp, &vattr, curthread->td_ucred);
460 return (lf_advlock(ap, &(vp->v_lockf), vattr.va_size));
464 vop_stdadvlockasync(struct vop_advlockasync_args *ap)
471 if (ap->a_fl->l_whence == SEEK_END) {
472 /* The size argument is only needed for SEEK_END. */
473 vn_lock(vp, LK_SHARED | LK_RETRY);
474 error = VOP_GETATTR(vp, &vattr, curthread->td_ucred);
481 return (lf_advlockasync(ap, &(vp->v_lockf), vattr.va_size));
485 vop_stdadvlockpurge(struct vop_advlockpurge_args *ap)
490 lf_purgelocks(vp, &vp->v_lockf);
497 * Standard implementation of POSIX pathconf, to get information about limits
499 * Override per filesystem for the case where the filesystem has smaller
504 struct vop_pathconf_args /* {
511 switch (ap->a_name) {
513 *ap->a_retval = _POSIX_ASYNCHRONOUS_IO;
516 *ap->a_retval = PATH_MAX;
518 case _PC_ACL_EXTENDED:
520 case _PC_CAP_PRESENT:
521 case _PC_INF_PRESENT:
522 case _PC_MAC_PRESENT:
532 * Standard lock, unlock and islocked functions.
536 struct vop_lock1_args /* {
543 struct vnode *vp = ap->a_vp;
547 return (lockmgr_lock_flags(vp->v_vnlock, ap->a_flags,
548 &ilk->lock_object, ap->a_file, ap->a_line));
554 struct vop_unlock_args /* {
558 struct vnode *vp = ap->a_vp;
560 return (lockmgr_unlock(vp->v_vnlock));
566 struct vop_islocked_args /* {
571 return (lockstatus(ap->a_vp->v_vnlock));
575 * Variants of the above set.
578 * - shared locking disablement is not supported
579 * - v_vnlock pointer is not honored
583 struct vop_lock1_args /* {
590 struct vnode *vp = ap->a_vp;
591 int flags = ap->a_flags;
594 MPASS(vp->v_vnlock == &vp->v_lock);
596 if (__predict_false((flags & ~(LK_TYPE_MASK | LK_NODDLKTREAT | LK_RETRY)) != 0))
599 switch (flags & LK_TYPE_MASK) {
601 return (lockmgr_slock(&vp->v_lock, flags, ap->a_file, ap->a_line));
603 return (lockmgr_xlock(&vp->v_lock, flags, ap->a_file, ap->a_line));
607 return (lockmgr_lock_flags(&vp->v_lock, flags,
608 &ilk->lock_object, ap->a_file, ap->a_line));
613 struct vop_unlock_args /* {
617 struct vnode *vp = ap->a_vp;
619 MPASS(vp->v_vnlock == &vp->v_lock);
621 return (lockmgr_unlock(&vp->v_lock));
626 struct vop_islocked_args /* {
630 struct vnode *vp = ap->a_vp;
632 MPASS(vp->v_vnlock == &vp->v_lock);
634 return (lockstatus(&vp->v_lock));
638 * Return true for select/poll.
642 struct vop_poll_args /* {
645 struct ucred *a_cred;
650 if (ap->a_events & ~POLLSTANDARD)
652 return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
656 * Implement poll for local filesystems that support it.
660 struct vop_poll_args /* {
663 struct ucred *a_cred;
667 if (ap->a_events & ~POLLSTANDARD)
668 return (vn_pollrecord(ap->a_vp, ap->a_td, ap->a_events));
669 return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
673 * Return our mount point, as we will take charge of the writes.
676 vop_stdgetwritemount(ap)
677 struct vop_getwritemount_args /* {
679 struct mount **a_mpp;
686 * Note that having a reference does not prevent forced unmount from
687 * setting ->v_mount to NULL after the lock gets released. This is of
688 * no consequence for typical consumers (most notably vn_start_write)
689 * since in this case the vnode is VIRF_DOOMED. Unmount might have
690 * progressed far enough that its completion is only delayed by the
691 * reference obtained here. The consumer only needs to concern itself
695 mp = vfs_ref_from_vp(vp);
701 * If the file system doesn't implement VOP_BMAP, then return sensible defaults:
702 * - Return the vnode's bufobj instead of any underlying device's bufobj
703 * - Calculate the physical block number as if there were equal size
704 * consecutive blocks, but
705 * - Report no contiguous runs of blocks.
709 struct vop_bmap_args /* {
712 struct bufobj **a_bop;
719 if (ap->a_bop != NULL)
720 *ap->a_bop = &ap->a_vp->v_bufobj;
721 if (ap->a_bnp != NULL)
722 *ap->a_bnp = ap->a_bn * btodb(ap->a_vp->v_mount->mnt_stat.f_iosize);
723 if (ap->a_runp != NULL)
725 if (ap->a_runb != NULL)
732 struct vop_fsync_args /* {
739 return (vn_fsync_buf(ap->a_vp, ap->a_waitfor));
743 vop_stdfdatasync(struct vop_fdatasync_args *ap)
746 return (VOP_FSYNC(ap->a_vp, MNT_WAIT, ap->a_td));
750 vop_stdfdatasync_buf(struct vop_fdatasync_args *ap)
753 return (vn_fsync_buf(ap->a_vp, MNT_WAIT));
756 /* XXX Needs good comment and more info in the manpage (VOP_GETPAGES(9)). */
759 struct vop_getpages_args /* {
768 return vnode_pager_generic_getpages(ap->a_vp, ap->a_m,
769 ap->a_count, ap->a_rbehind, ap->a_rahead, NULL, NULL);
773 vop_stdgetpages_async(struct vop_getpages_async_args *ap)
777 error = VOP_GETPAGES(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind,
779 if (ap->a_iodone != NULL)
780 ap->a_iodone(ap->a_arg, ap->a_m, ap->a_count, error);
785 vop_stdkqfilter(struct vop_kqfilter_args *ap)
787 return vfs_kqfilter(ap);
790 /* XXX Needs good comment and more info in the manpage (VOP_PUTPAGES(9)). */
793 struct vop_putpages_args /* {
802 return vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count,
803 ap->a_sync, ap->a_rtvals);
807 vop_stdvptofh(struct vop_vptofh_args *ap)
813 vop_stdvptocnp(struct vop_vptocnp_args *ap)
815 struct vnode *vp = ap->a_vp;
816 struct vnode **dvp = ap->a_vpp;
818 char *buf = ap->a_buf;
819 size_t *buflen = ap->a_buflen;
821 int i, error, eofflag, dirbuflen, flags, locked, len, covered;
836 if (vp->v_type != VDIR)
839 error = VOP_GETATTR(vp, &va, cred);
844 locked = VOP_ISLOCKED(vp);
846 NDINIT_ATVP(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF, UIO_SYSSPACE,
849 error = vn_open_cred(&nd, &flags, 0, VN_OPEN_NOAUDIT, cred, NULL);
851 vn_lock(vp, locked | LK_RETRY);
854 NDFREE(&nd, NDF_ONLY_PNBUF);
856 mvp = *dvp = nd.ni_vp;
858 if (vp->v_mount != (*dvp)->v_mount &&
859 ((*dvp)->v_vflag & VV_ROOT) &&
860 ((*dvp)->v_mount->mnt_flag & MNT_UNION)) {
861 *dvp = (*dvp)->v_mount->mnt_vnodecovered;
864 vn_close(mvp, FREAD, cred, td);
866 vn_lock(*dvp, LK_SHARED | LK_RETRY);
870 fileno = va.va_fileid;
872 dirbuflen = DEV_BSIZE;
873 if (dirbuflen < va.va_blocksize)
874 dirbuflen = va.va_blocksize;
875 dirbuf = (char *)malloc(dirbuflen, M_TEMP, M_WAITOK);
877 if ((*dvp)->v_type != VDIR) {
885 /* call VOP_READDIR of parent */
886 error = get_next_dirent(*dvp, &dp, dirbuf, dirbuflen, &off,
887 &cpos, &len, &eofflag, td);
891 if ((dp->d_type != DT_WHT) &&
892 (dp->d_fileno == fileno)) {
895 vn_lock(mvp, LK_SHARED | LK_RETRY);
896 if (dirent_exists(mvp, dp->d_name, td)) {
899 vn_lock(*dvp, LK_SHARED | LK_RETRY);
903 vn_lock(*dvp, LK_SHARED | LK_RETRY);
911 if (dp->d_namlen == 1 && dp->d_name[0] == '.') {
914 bcopy(dp->d_name, buf + i, dp->d_namlen);
919 } while (len > 0 || !eofflag);
923 free(dirbuf, M_TEMP);
933 vn_close(mvp, FREAD, cred, td);
935 vn_lock(vp, locked | LK_RETRY);
940 vop_stdallocate(struct vop_allocate_args *ap)
944 off_t maxfilesize = 0;
947 struct vattr vattr, *vap;
949 off_t fsize, len, cur, offset;
962 offset = *ap->a_offset;
964 error = VOP_GETATTR(vp, vap, td->td_ucred);
967 fsize = vap->va_size;
968 iosize = vap->va_blocksize;
970 iosize = BLKDEV_IOSIZE;
971 if (iosize > maxphys)
973 buf = malloc(iosize, M_TEMP, M_WAITOK);
977 * Check if the filesystem sets f_maxfilesize; if not use
978 * VOP_SETATTR to perform the check.
980 sfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK);
981 error = VFS_STATFS(vp->v_mount, sfs, td);
983 maxfilesize = sfs->f_maxfilesize;
988 if (offset > maxfilesize || len > maxfilesize ||
989 offset + len > maxfilesize) {
995 if (offset + len > vap->va_size) {
997 * Test offset + len against the filesystem's maxfilesize.
1000 vap->va_size = offset + len;
1001 error = VOP_SETATTR(vp, vap, td->td_ucred);
1005 vap->va_size = fsize;
1006 error = VOP_SETATTR(vp, vap, td->td_ucred);
1013 * Read and write back anything below the nominal file
1014 * size. There's currently no way outside the filesystem
1015 * to know whether this area is sparse or not.
1018 if ((offset % iosize) != 0)
1019 cur -= (offset % iosize);
1022 if (offset < fsize) {
1023 aiov.iov_base = buf;
1025 auio.uio_iov = &aiov;
1026 auio.uio_iovcnt = 1;
1027 auio.uio_offset = offset;
1028 auio.uio_resid = cur;
1029 auio.uio_segflg = UIO_SYSSPACE;
1030 auio.uio_rw = UIO_READ;
1032 error = VOP_READ(vp, &auio, 0, td->td_ucred);
1035 if (auio.uio_resid > 0) {
1036 bzero(buf + cur - auio.uio_resid,
1043 aiov.iov_base = buf;
1045 auio.uio_iov = &aiov;
1046 auio.uio_iovcnt = 1;
1047 auio.uio_offset = offset;
1048 auio.uio_resid = cur;
1049 auio.uio_segflg = UIO_SYSSPACE;
1050 auio.uio_rw = UIO_WRITE;
1053 error = VOP_WRITE(vp, &auio, 0, td->td_ucred);
1067 *ap->a_offset = offset;
1073 vop_stdadvise(struct vop_advise_args *ap)
1077 daddr_t startn, endn;
1078 off_t bstart, bend, start, end;
1082 switch (ap->a_advice) {
1083 case POSIX_FADV_WILLNEED:
1085 * Do nothing for now. Filesystems should provide a
1086 * custom method which starts an asynchronous read of
1087 * the requested region.
1091 case POSIX_FADV_DONTNEED:
1093 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
1094 if (VN_IS_DOOMED(vp)) {
1100 * Round to block boundaries (and later possibly further to
1101 * page boundaries). Applications cannot reasonably be aware
1102 * of the boundaries, and the rounding must be to expand at
1103 * both extremities to cover enough. It still doesn't cover
1104 * read-ahead. For partial blocks, this gives unnecessary
1105 * discarding of buffers but is efficient enough since the
1106 * pages usually remain in VMIO for some time.
1108 bsize = vp->v_bufobj.bo_bsize;
1109 bstart = rounddown(ap->a_start, bsize);
1110 bend = roundup(ap->a_end, bsize);
1113 * Deactivate pages in the specified range from the backing VM
1114 * object. Pages that are resident in the buffer cache will
1115 * remain wired until their corresponding buffers are released
1118 if (vp->v_object != NULL) {
1119 start = trunc_page(bstart);
1120 end = round_page(bend);
1121 VM_OBJECT_RLOCK(vp->v_object);
1122 vm_object_page_noreuse(vp->v_object, OFF_TO_IDX(start),
1124 VM_OBJECT_RUNLOCK(vp->v_object);
1129 startn = bstart / bsize;
1130 endn = bend / bsize;
1131 error = bnoreuselist(&bo->bo_clean, bo, startn, endn);
1133 error = bnoreuselist(&bo->bo_dirty, bo, startn, endn);
1145 vop_stdunp_bind(struct vop_unp_bind_args *ap)
1148 ap->a_vp->v_unpcb = ap->a_unpcb;
1153 vop_stdunp_connect(struct vop_unp_connect_args *ap)
1156 *ap->a_unpcb = ap->a_vp->v_unpcb;
1161 vop_stdunp_detach(struct vop_unp_detach_args *ap)
1164 ap->a_vp->v_unpcb = NULL;
1169 vop_stdis_text(struct vop_is_text_args *ap)
1172 return (ap->a_vp->v_writecount < 0);
1176 vop_stdset_text(struct vop_set_text_args *ap)
1184 if (vp->v_writecount > 0) {
1188 * If requested by fs, keep a use reference to the
1189 * vnode until the last text reference is released.
1192 if (mp != NULL && (mp->mnt_kern_flag & MNTK_TEXT_REFS) != 0 &&
1193 vp->v_writecount == 0) {
1194 VNPASS((vp->v_iflag & VI_TEXT_REF) == 0, vp);
1195 vp->v_iflag |= VI_TEXT_REF;
1207 vop_stdunset_text(struct vop_unset_text_args *ap)
1216 if (vp->v_writecount < 0) {
1217 if ((vp->v_iflag & VI_TEXT_REF) != 0 &&
1218 vp->v_writecount == -1) {
1220 vp->v_iflag &= ~VI_TEXT_REF;
1234 vop_stdadd_writecount(struct vop_add_writecount_args *ap)
1241 VI_LOCK_FLAGS(vp, MTX_DUPOK);
1242 if (vp->v_writecount < 0) {
1245 VNASSERT(vp->v_writecount + ap->a_inc >= 0, vp,
1246 ("neg writecount increment %d", ap->a_inc));
1247 if (vp->v_writecount == 0) {
1249 if (mp != NULL && (mp->mnt_kern_flag & MNTK_NOMSYNC) == 0)
1252 vp->v_writecount += ap->a_inc;
1260 vop_stdneed_inactive(struct vop_need_inactive_args *ap)
1267 vop_stdioctl(struct vop_ioctl_args *ap)
1274 switch (ap->a_command) {
1278 error = vn_lock(vp, LK_SHARED);
1281 if (vp->v_type == VREG)
1282 error = VOP_GETATTR(vp, &va, ap->a_cred);
1287 if (*offp < 0 || *offp >= va.va_size)
1289 else if (ap->a_command == FIOSEEKHOLE)
1303 * used to fill the vfs function table to get reasonable default return values.
1306 vfs_stdroot (mp, flags, vpp)
1312 return (EOPNOTSUPP);
1316 vfs_stdstatfs (mp, sbp)
1321 return (EOPNOTSUPP);
1325 vfs_stdquotactl (mp, cmds, uid, arg)
1332 return (EOPNOTSUPP);
1336 vfs_stdsync(mp, waitfor)
1340 struct vnode *vp, *mvp;
1342 int error, lockreq, allerror = 0;
1345 lockreq = LK_EXCLUSIVE | LK_INTERLOCK;
1346 if (waitfor != MNT_WAIT)
1347 lockreq |= LK_NOWAIT;
1349 * Force stale buffer cache information to be flushed.
1352 MNT_VNODE_FOREACH_ALL(vp, mp, mvp) {
1353 if (vp->v_bufobj.bo_dirty.bv_cnt == 0) {
1357 if ((error = vget(vp, lockreq)) != 0) {
1358 if (error == ENOENT) {
1359 MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
1364 error = VOP_FSYNC(vp, waitfor, td);
1373 vfs_stdnosync (mp, waitfor)
1382 vop_stdcopy_file_range(struct vop_copy_file_range_args *ap)
1386 error = vn_generic_copy_file_range(ap->a_invp, ap->a_inoffp,
1387 ap->a_outvp, ap->a_outoffp, ap->a_lenp, ap->a_flags, ap->a_incred,
1388 ap->a_outcred, ap->a_fsizetd);
1393 vfs_stdvget (mp, ino, flags, vpp)
1400 return (EOPNOTSUPP);
1404 vfs_stdfhtovp (mp, fhp, flags, vpp)
1411 return (EOPNOTSUPP);
1416 struct vfsconf *vfsp;
1423 vfs_stduninit (vfsp)
1424 struct vfsconf *vfsp;
1431 vfs_stdextattrctl(mp, cmd, filename_vp, attrnamespace, attrname)
1434 struct vnode *filename_vp;
1436 const char *attrname;
1439 if (filename_vp != NULL)
1440 VOP_UNLOCK(filename_vp);
1441 return (EOPNOTSUPP);
1445 vfs_stdsysctl(mp, op, req)
1448 struct sysctl_req *req;
1451 return (EOPNOTSUPP);
1454 static vop_bypass_t *
1455 bp_by_off(struct vop_vector *vop, struct vop_generic_args *a)
1458 return (*(vop_bypass_t **)((char *)vop + a->a_desc->vdesc_vop_offset));
1462 vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a)
1467 bp = bp_by_off(vop, a);
1470 prev_stops = sigdeferstop(SIGDEFERSTOP_SILENT);
1472 sigallowstop(prev_stops);
1477 vop_stdstat(struct vop_stat_args *a)
1489 error = vop_stat_helper_pre(a);
1496 * Initialize defaults for new and unusual fields, so that file
1497 * systems which don't support these fields don't need to know
1500 vap->va_birthtime.tv_sec = -1;
1501 vap->va_birthtime.tv_nsec = 0;
1502 vap->va_fsid = VNOVAL;
1503 vap->va_rdev = NODEV;
1505 error = VOP_GETATTR(vp, vap, a->a_active_cred);
1510 * Zero the spare stat fields
1512 bzero(sb, sizeof *sb);
1515 * Copy from vattr table
1517 if (vap->va_fsid != VNOVAL)
1518 sb->st_dev = vap->va_fsid;
1520 sb->st_dev = vp->v_mount->mnt_stat.f_fsid.val[0];
1521 sb->st_ino = vap->va_fileid;
1522 mode = vap->va_mode;
1523 switch (vap->va_type) {
1550 sb->st_nlink = vap->va_nlink;
1551 sb->st_uid = vap->va_uid;
1552 sb->st_gid = vap->va_gid;
1553 sb->st_rdev = vap->va_rdev;
1554 if (vap->va_size > OFF_MAX) {
1558 sb->st_size = vap->va_size;
1559 sb->st_atim.tv_sec = vap->va_atime.tv_sec;
1560 sb->st_atim.tv_nsec = vap->va_atime.tv_nsec;
1561 sb->st_mtim.tv_sec = vap->va_mtime.tv_sec;
1562 sb->st_mtim.tv_nsec = vap->va_mtime.tv_nsec;
1563 sb->st_ctim.tv_sec = vap->va_ctime.tv_sec;
1564 sb->st_ctim.tv_nsec = vap->va_ctime.tv_nsec;
1565 sb->st_birthtim.tv_sec = vap->va_birthtime.tv_sec;
1566 sb->st_birthtim.tv_nsec = vap->va_birthtime.tv_nsec;
1569 * According to www.opengroup.org, the meaning of st_blksize is
1570 * "a filesystem-specific preferred I/O block size for this
1571 * object. In some filesystem types, this may vary from file
1573 * Use minimum/default of PAGE_SIZE (e.g. for VCHR).
1576 sb->st_blksize = max(PAGE_SIZE, vap->va_blocksize);
1577 sb->st_flags = vap->va_flags;
1578 sb->st_blocks = vap->va_bytes / S_BLKSIZE;
1579 sb->st_gen = vap->va_gen;
1581 return (vop_stat_helper_post(a, error));
1585 vop_stdread_pgcache(struct vop_read_pgcache_args *ap __unused)
1587 return (EJUSTRETURN);
1591 vop_stdvput_pair(struct vop_vput_pair_args *ap)
1593 struct vnode *dvp, *vp, **vpp;
1598 if (vpp != NULL && ap->a_unlock_vp && (vp = *vpp) != NULL)