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);
97 * This vnode table stores what we want to do if the filesystem doesn't
98 * implement a particular VOP.
100 * If there is no specific entry here, we will return EOPNOTSUPP.
102 * Note that every filesystem has to implement either vop_access
103 * or vop_accessx; failing to do so will result in immediate crash
104 * due to stack overflow, as vop_stdaccess() calls vop_stdaccessx(),
105 * which calls vop_stdaccess() etc.
108 struct vop_vector default_vnodeops = {
110 .vop_bypass = VOP_EOPNOTSUPP,
112 .vop_access = vop_stdaccess,
113 .vop_accessx = vop_stdaccessx,
114 .vop_advise = vop_stdadvise,
115 .vop_advlock = vop_stdadvlock,
116 .vop_advlockasync = vop_stdadvlockasync,
117 .vop_advlockpurge = vop_stdadvlockpurge,
118 .vop_allocate = vop_stdallocate,
119 .vop_bmap = vop_stdbmap,
120 .vop_close = VOP_NULL,
121 .vop_fsync = VOP_NULL,
122 .vop_stat = vop_stdstat,
123 .vop_fdatasync = vop_stdfdatasync,
124 .vop_getpages = vop_stdgetpages,
125 .vop_getpages_async = vop_stdgetpages_async,
126 .vop_getwritemount = vop_stdgetwritemount,
127 .vop_inactive = VOP_NULL,
128 .vop_need_inactive = vop_stdneed_inactive,
129 .vop_ioctl = vop_stdioctl,
130 .vop_kqfilter = vop_stdkqfilter,
131 .vop_islocked = vop_stdislocked,
132 .vop_lock1 = vop_stdlock,
133 .vop_lookup = vop_nolookup,
134 .vop_open = VOP_NULL,
135 .vop_pathconf = VOP_EINVAL,
136 .vop_poll = vop_nopoll,
137 .vop_putpages = vop_stdputpages,
138 .vop_readlink = VOP_EINVAL,
139 .vop_read_pgcache = vop_stdread_pgcache,
140 .vop_rename = vop_norename,
141 .vop_revoke = VOP_PANIC,
142 .vop_strategy = vop_nostrategy,
143 .vop_unlock = vop_stdunlock,
144 .vop_vptocnp = vop_stdvptocnp,
145 .vop_vptofh = vop_stdvptofh,
146 .vop_unp_bind = vop_stdunp_bind,
147 .vop_unp_connect = vop_stdunp_connect,
148 .vop_unp_detach = vop_stdunp_detach,
149 .vop_is_text = vop_stdis_text,
150 .vop_set_text = vop_stdset_text,
151 .vop_unset_text = vop_stdunset_text,
152 .vop_add_writecount = vop_stdadd_writecount,
153 .vop_copy_file_range = vop_stdcopy_file_range,
155 VFS_VOP_VECTOR_REGISTER(default_vnodeops);
158 * Series of placeholder functions for various error returns for
163 vop_eopnotsupp(struct vop_generic_args *ap)
166 printf("vop_notsupp[%s]\n", ap->a_desc->vdesc_name);
173 vop_ebadf(struct vop_generic_args *ap)
180 vop_enotty(struct vop_generic_args *ap)
187 vop_einval(struct vop_generic_args *ap)
194 vop_enoent(struct vop_generic_args *ap)
201 vop_eagain(struct vop_generic_args *ap)
208 vop_null(struct vop_generic_args *ap)
215 * Helper function to panic on some bad VOPs in some filesystems.
218 vop_panic(struct vop_generic_args *ap)
221 panic("filesystem goof: vop_panic[%s]", ap->a_desc->vdesc_name);
225 * vop_std<something> and vop_no<something> are default functions for use by
226 * filesystems that need the "default reasonable" implementation for a
227 * particular operation.
229 * The documentation for the operations they implement exists (if it exists)
230 * in the VOP_<SOMETHING>(9) manpage (all uppercase).
234 * Default vop for filesystems that do not support name lookup
238 struct vop_lookup_args /* {
240 struct vnode **a_vpp;
241 struct componentname *a_cnp;
252 * Handle unlock and reference counting for arguments of vop_rename
253 * for filesystems that do not implement rename operation.
256 vop_norename(struct vop_rename_args *ap)
266 * Strategy routine for VFS devices that have none.
268 * BIO_ERROR and B_INVAL must be cleared prior to calling any strategy
269 * routine. Typically this is done for a BIO_READ strategy call.
270 * Typically B_INVAL is assumed to already be clear prior to a write
271 * and should not be cleared manually unless you just made the buffer
272 * invalid. BIO_ERROR should be cleared either way.
276 vop_nostrategy (struct vop_strategy_args *ap)
278 printf("No strategy for buffer at %p\n", ap->a_bp);
279 vn_printf(ap->a_vp, "vnode ");
280 ap->a_bp->b_ioflags |= BIO_ERROR;
281 ap->a_bp->b_error = EOPNOTSUPP;
287 get_next_dirent(struct vnode *vp, struct dirent **dpp, char *dirbuf,
288 int dirbuflen, off_t *off, char **cpos, int *len,
289 int *eofflag, struct thread *td)
296 KASSERT(VOP_ISLOCKED(vp), ("vp %p is not locked", vp));
297 KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp));
300 iov.iov_base = dirbuf;
301 iov.iov_len = dirbuflen;
305 uio.uio_offset = *off;
306 uio.uio_resid = dirbuflen;
307 uio.uio_segflg = UIO_SYSSPACE;
308 uio.uio_rw = UIO_READ;
314 error = mac_vnode_check_readdir(td->td_ucred, vp);
317 error = VOP_READDIR(vp, &uio, td->td_ucred, eofflag,
322 *off = uio.uio_offset;
325 *len = (dirbuflen - uio.uio_resid);
331 dp = (struct dirent *)(*cpos);
332 reclen = dp->d_reclen;
335 /* check for malformed directory.. */
336 if (reclen < DIRENT_MINSIZE)
346 * Check if a named file exists in a given directory vnode.
349 dirent_exists(struct vnode *vp, const char *dirname, struct thread *td)
352 int error, eofflag, dirbuflen, len, found;
357 KASSERT(VOP_ISLOCKED(vp), ("vp %p is not locked", vp));
358 KASSERT(vp->v_type == VDIR, ("vp %p is not a directory", vp));
362 error = VOP_GETATTR(vp, &va, td->td_ucred);
366 dirbuflen = DEV_BSIZE;
367 if (dirbuflen < va.va_blocksize)
368 dirbuflen = va.va_blocksize;
369 dirbuf = (char *)malloc(dirbuflen, M_TEMP, M_WAITOK);
374 error = get_next_dirent(vp, &dp, dirbuf, dirbuflen, &off,
375 &cpos, &len, &eofflag, td);
379 if (dp->d_type != DT_WHT && dp->d_fileno != 0 &&
380 strcmp(dp->d_name, dirname) == 0) {
384 } while (len > 0 || !eofflag);
387 free(dirbuf, M_TEMP);
392 vop_stdaccess(struct vop_access_args *ap)
395 KASSERT((ap->a_accmode & ~(VEXEC | VWRITE | VREAD | VADMIN |
396 VAPPEND)) == 0, ("invalid bit in accmode"));
398 return (VOP_ACCESSX(ap->a_vp, ap->a_accmode, ap->a_cred, ap->a_td));
402 vop_stdaccessx(struct vop_accessx_args *ap)
405 accmode_t accmode = ap->a_accmode;
407 error = vfs_unixify_accmode(&accmode);
414 return (VOP_ACCESS(ap->a_vp, accmode, ap->a_cred, ap->a_td));
418 * Advisory record locking support
421 vop_stdadvlock(struct vop_advlock_args *ap)
428 if (ap->a_fl->l_whence == SEEK_END) {
430 * The NFSv4 server must avoid doing a vn_lock() here, since it
431 * can deadlock the nfsd threads, due to a LOR. Fortunately
432 * the NFSv4 server always uses SEEK_SET and this code is
433 * only required for the SEEK_END case.
435 vn_lock(vp, LK_SHARED | LK_RETRY);
436 error = VOP_GETATTR(vp, &vattr, curthread->td_ucred);
443 return (lf_advlock(ap, &(vp->v_lockf), vattr.va_size));
447 vop_stdadvlockasync(struct vop_advlockasync_args *ap)
454 if (ap->a_fl->l_whence == SEEK_END) {
455 /* The size argument is only needed for SEEK_END. */
456 vn_lock(vp, LK_SHARED | LK_RETRY);
457 error = VOP_GETATTR(vp, &vattr, curthread->td_ucred);
464 return (lf_advlockasync(ap, &(vp->v_lockf), vattr.va_size));
468 vop_stdadvlockpurge(struct vop_advlockpurge_args *ap)
473 lf_purgelocks(vp, &vp->v_lockf);
480 * Standard implementation of POSIX pathconf, to get information about limits
482 * Override per filesystem for the case where the filesystem has smaller
487 struct vop_pathconf_args /* {
494 switch (ap->a_name) {
496 *ap->a_retval = _POSIX_ASYNCHRONOUS_IO;
499 *ap->a_retval = PATH_MAX;
501 case _PC_ACL_EXTENDED:
503 case _PC_CAP_PRESENT:
504 case _PC_INF_PRESENT:
505 case _PC_MAC_PRESENT:
515 * Standard lock, unlock and islocked functions.
519 struct vop_lock1_args /* {
526 struct vnode *vp = ap->a_vp;
530 return (lockmgr_lock_flags(vp->v_vnlock, ap->a_flags,
531 &ilk->lock_object, ap->a_file, ap->a_line));
537 struct vop_unlock_args /* {
541 struct vnode *vp = ap->a_vp;
543 return (lockmgr_unlock(vp->v_vnlock));
549 struct vop_islocked_args /* {
554 return (lockstatus(ap->a_vp->v_vnlock));
558 * Variants of the above set.
561 * - shared locking disablement is not supported
562 * - v_vnlock pointer is not honored
566 struct vop_lock1_args /* {
573 struct vnode *vp = ap->a_vp;
574 int flags = ap->a_flags;
577 MPASS(vp->v_vnlock == &vp->v_lock);
579 if (__predict_false((flags & ~(LK_TYPE_MASK | LK_NODDLKTREAT | LK_RETRY)) != 0))
582 switch (flags & LK_TYPE_MASK) {
584 return (lockmgr_slock(&vp->v_lock, flags, ap->a_file, ap->a_line));
586 return (lockmgr_xlock(&vp->v_lock, flags, ap->a_file, ap->a_line));
590 return (lockmgr_lock_flags(&vp->v_lock, flags,
591 &ilk->lock_object, ap->a_file, ap->a_line));
596 struct vop_unlock_args /* {
600 struct vnode *vp = ap->a_vp;
602 MPASS(vp->v_vnlock == &vp->v_lock);
604 return (lockmgr_unlock(&vp->v_lock));
609 struct vop_islocked_args /* {
613 struct vnode *vp = ap->a_vp;
615 MPASS(vp->v_vnlock == &vp->v_lock);
617 return (lockstatus(&vp->v_lock));
621 * Return true for select/poll.
625 struct vop_poll_args /* {
628 struct ucred *a_cred;
633 if (ap->a_events & ~POLLSTANDARD)
635 return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
639 * Implement poll for local filesystems that support it.
643 struct vop_poll_args /* {
646 struct ucred *a_cred;
650 if (ap->a_events & ~POLLSTANDARD)
651 return (vn_pollrecord(ap->a_vp, ap->a_td, ap->a_events));
652 return (ap->a_events & (POLLIN | POLLOUT | POLLRDNORM | POLLWRNORM));
656 * Return our mount point, as we will take charge of the writes.
659 vop_stdgetwritemount(ap)
660 struct vop_getwritemount_args /* {
662 struct mount **a_mpp;
669 * Note that having a reference does not prevent forced unmount from
670 * setting ->v_mount to NULL after the lock gets released. This is of
671 * no consequence for typical consumers (most notably vn_start_write)
672 * since in this case the vnode is VIRF_DOOMED. Unmount might have
673 * progressed far enough that its completion is only delayed by the
674 * reference obtained here. The consumer only needs to concern itself
683 if (vfs_op_thread_enter(mp)) {
684 if (mp == vp->v_mount) {
685 vfs_mp_count_add_pcpu(mp, ref, 1);
686 vfs_op_thread_exit(mp);
688 vfs_op_thread_exit(mp);
693 if (mp == vp->v_mount) {
706 * If the file system doesn't implement VOP_BMAP, then return sensible defaults:
707 * - Return the vnode's bufobj instead of any underlying device's bufobj
708 * - Calculate the physical block number as if there were equal size
709 * consecutive blocks, but
710 * - Report no contiguous runs of blocks.
714 struct vop_bmap_args /* {
717 struct bufobj **a_bop;
724 if (ap->a_bop != NULL)
725 *ap->a_bop = &ap->a_vp->v_bufobj;
726 if (ap->a_bnp != NULL)
727 *ap->a_bnp = ap->a_bn * btodb(ap->a_vp->v_mount->mnt_stat.f_iosize);
728 if (ap->a_runp != NULL)
730 if (ap->a_runb != NULL)
737 struct vop_fsync_args /* {
744 return (vn_fsync_buf(ap->a_vp, ap->a_waitfor));
748 vop_stdfdatasync(struct vop_fdatasync_args *ap)
751 return (VOP_FSYNC(ap->a_vp, MNT_WAIT, ap->a_td));
755 vop_stdfdatasync_buf(struct vop_fdatasync_args *ap)
758 return (vn_fsync_buf(ap->a_vp, MNT_WAIT));
761 /* XXX Needs good comment and more info in the manpage (VOP_GETPAGES(9)). */
764 struct vop_getpages_args /* {
773 return vnode_pager_generic_getpages(ap->a_vp, ap->a_m,
774 ap->a_count, ap->a_rbehind, ap->a_rahead, NULL, NULL);
778 vop_stdgetpages_async(struct vop_getpages_async_args *ap)
782 error = VOP_GETPAGES(ap->a_vp, ap->a_m, ap->a_count, ap->a_rbehind,
784 if (ap->a_iodone != NULL)
785 ap->a_iodone(ap->a_arg, ap->a_m, ap->a_count, error);
790 vop_stdkqfilter(struct vop_kqfilter_args *ap)
792 return vfs_kqfilter(ap);
795 /* XXX Needs good comment and more info in the manpage (VOP_PUTPAGES(9)). */
798 struct vop_putpages_args /* {
807 return vnode_pager_generic_putpages(ap->a_vp, ap->a_m, ap->a_count,
808 ap->a_sync, ap->a_rtvals);
812 vop_stdvptofh(struct vop_vptofh_args *ap)
818 vop_stdvptocnp(struct vop_vptocnp_args *ap)
820 struct vnode *vp = ap->a_vp;
821 struct vnode **dvp = ap->a_vpp;
822 struct ucred *cred = ap->a_cred;
823 char *buf = ap->a_buf;
824 size_t *buflen = ap->a_buflen;
826 int i, error, eofflag, dirbuflen, flags, locked, len, covered;
840 if (vp->v_type != VDIR)
843 error = VOP_GETATTR(vp, &va, cred);
848 locked = VOP_ISLOCKED(vp);
850 NDINIT_ATVP(&nd, LOOKUP, FOLLOW | LOCKSHARED | LOCKLEAF, UIO_SYSSPACE,
853 error = vn_open_cred(&nd, &flags, 0, VN_OPEN_NOAUDIT, cred, NULL);
855 vn_lock(vp, locked | LK_RETRY);
858 NDFREE(&nd, NDF_ONLY_PNBUF);
860 mvp = *dvp = nd.ni_vp;
862 if (vp->v_mount != (*dvp)->v_mount &&
863 ((*dvp)->v_vflag & VV_ROOT) &&
864 ((*dvp)->v_mount->mnt_flag & MNT_UNION)) {
865 *dvp = (*dvp)->v_mount->mnt_vnodecovered;
868 vn_close(mvp, FREAD, cred, td);
870 vn_lock(*dvp, LK_SHARED | LK_RETRY);
874 fileno = va.va_fileid;
876 dirbuflen = DEV_BSIZE;
877 if (dirbuflen < va.va_blocksize)
878 dirbuflen = va.va_blocksize;
879 dirbuf = (char *)malloc(dirbuflen, M_TEMP, M_WAITOK);
881 if ((*dvp)->v_type != VDIR) {
889 /* call VOP_READDIR of parent */
890 error = get_next_dirent(*dvp, &dp, dirbuf, dirbuflen, &off,
891 &cpos, &len, &eofflag, td);
895 if ((dp->d_type != DT_WHT) &&
896 (dp->d_fileno == fileno)) {
899 vn_lock(mvp, LK_SHARED | LK_RETRY);
900 if (dirent_exists(mvp, dp->d_name, td)) {
903 vn_lock(*dvp, LK_SHARED | LK_RETRY);
907 vn_lock(*dvp, LK_SHARED | LK_RETRY);
915 if (dp->d_namlen == 1 && dp->d_name[0] == '.') {
918 bcopy(dp->d_name, buf + i, dp->d_namlen);
923 } while (len > 0 || !eofflag);
927 free(dirbuf, M_TEMP);
937 vn_close(mvp, FREAD, cred, td);
939 vn_lock(vp, locked | LK_RETRY);
944 vop_stdallocate(struct vop_allocate_args *ap)
948 off_t maxfilesize = 0;
951 struct vattr vattr, *vap;
953 off_t fsize, len, cur, offset;
966 offset = *ap->a_offset;
968 error = VOP_GETATTR(vp, vap, td->td_ucred);
971 fsize = vap->va_size;
972 iosize = vap->va_blocksize;
974 iosize = BLKDEV_IOSIZE;
975 if (iosize > MAXPHYS)
977 buf = malloc(iosize, M_TEMP, M_WAITOK);
981 * Check if the filesystem sets f_maxfilesize; if not use
982 * VOP_SETATTR to perform the check.
984 sfs = malloc(sizeof(struct statfs), M_STATFS, M_WAITOK);
985 error = VFS_STATFS(vp->v_mount, sfs, td);
987 maxfilesize = sfs->f_maxfilesize;
992 if (offset > maxfilesize || len > maxfilesize ||
993 offset + len > maxfilesize) {
999 if (offset + len > vap->va_size) {
1001 * Test offset + len against the filesystem's maxfilesize.
1004 vap->va_size = offset + len;
1005 error = VOP_SETATTR(vp, vap, td->td_ucred);
1009 vap->va_size = fsize;
1010 error = VOP_SETATTR(vp, vap, td->td_ucred);
1017 * Read and write back anything below the nominal file
1018 * size. There's currently no way outside the filesystem
1019 * to know whether this area is sparse or not.
1022 if ((offset % iosize) != 0)
1023 cur -= (offset % iosize);
1026 if (offset < fsize) {
1027 aiov.iov_base = buf;
1029 auio.uio_iov = &aiov;
1030 auio.uio_iovcnt = 1;
1031 auio.uio_offset = offset;
1032 auio.uio_resid = cur;
1033 auio.uio_segflg = UIO_SYSSPACE;
1034 auio.uio_rw = UIO_READ;
1036 error = VOP_READ(vp, &auio, 0, td->td_ucred);
1039 if (auio.uio_resid > 0) {
1040 bzero(buf + cur - auio.uio_resid,
1047 aiov.iov_base = buf;
1049 auio.uio_iov = &aiov;
1050 auio.uio_iovcnt = 1;
1051 auio.uio_offset = offset;
1052 auio.uio_resid = cur;
1053 auio.uio_segflg = UIO_SYSSPACE;
1054 auio.uio_rw = UIO_WRITE;
1057 error = VOP_WRITE(vp, &auio, 0, td->td_ucred);
1071 *ap->a_offset = offset;
1077 vop_stdadvise(struct vop_advise_args *ap)
1081 daddr_t startn, endn;
1082 off_t bstart, bend, start, end;
1086 switch (ap->a_advice) {
1087 case POSIX_FADV_WILLNEED:
1089 * Do nothing for now. Filesystems should provide a
1090 * custom method which starts an asynchronous read of
1091 * the requested region.
1095 case POSIX_FADV_DONTNEED:
1097 vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
1098 if (VN_IS_DOOMED(vp)) {
1104 * Round to block boundaries (and later possibly further to
1105 * page boundaries). Applications cannot reasonably be aware
1106 * of the boundaries, and the rounding must be to expand at
1107 * both extremities to cover enough. It still doesn't cover
1108 * read-ahead. For partial blocks, this gives unnecessary
1109 * discarding of buffers but is efficient enough since the
1110 * pages usually remain in VMIO for some time.
1112 bsize = vp->v_bufobj.bo_bsize;
1113 bstart = rounddown(ap->a_start, bsize);
1114 bend = roundup(ap->a_end, bsize);
1117 * Deactivate pages in the specified range from the backing VM
1118 * object. Pages that are resident in the buffer cache will
1119 * remain wired until their corresponding buffers are released
1122 if (vp->v_object != NULL) {
1123 start = trunc_page(bstart);
1124 end = round_page(bend);
1125 VM_OBJECT_RLOCK(vp->v_object);
1126 vm_object_page_noreuse(vp->v_object, OFF_TO_IDX(start),
1128 VM_OBJECT_RUNLOCK(vp->v_object);
1133 startn = bstart / bsize;
1134 endn = bend / bsize;
1135 error = bnoreuselist(&bo->bo_clean, bo, startn, endn);
1137 error = bnoreuselist(&bo->bo_dirty, bo, startn, endn);
1149 vop_stdunp_bind(struct vop_unp_bind_args *ap)
1152 ap->a_vp->v_unpcb = ap->a_unpcb;
1157 vop_stdunp_connect(struct vop_unp_connect_args *ap)
1160 *ap->a_unpcb = ap->a_vp->v_unpcb;
1165 vop_stdunp_detach(struct vop_unp_detach_args *ap)
1168 ap->a_vp->v_unpcb = NULL;
1173 vop_stdis_text(struct vop_is_text_args *ap)
1176 return (ap->a_vp->v_writecount < 0);
1180 vop_stdset_text(struct vop_set_text_args *ap)
1188 if (vp->v_writecount > 0) {
1192 * If requested by fs, keep a use reference to the
1193 * vnode until the last text reference is released.
1196 if (mp != NULL && (mp->mnt_kern_flag & MNTK_TEXT_REFS) != 0 &&
1197 vp->v_writecount == 0) {
1198 VNPASS((vp->v_iflag & VI_TEXT_REF) == 0, vp);
1199 vp->v_iflag |= VI_TEXT_REF;
1211 vop_stdunset_text(struct vop_unset_text_args *ap)
1220 if (vp->v_writecount < 0) {
1221 if ((vp->v_iflag & VI_TEXT_REF) != 0 &&
1222 vp->v_writecount == -1) {
1224 vp->v_iflag &= ~VI_TEXT_REF;
1238 vop_stdadd_writecount(struct vop_add_writecount_args *ap)
1245 VI_LOCK_FLAGS(vp, MTX_DUPOK);
1246 if (vp->v_writecount < 0) {
1249 VNASSERT(vp->v_writecount + ap->a_inc >= 0, vp,
1250 ("neg writecount increment %d", ap->a_inc));
1251 if (vp->v_writecount == 0) {
1253 if (mp != NULL && (mp->mnt_kern_flag & MNTK_NOMSYNC) == 0)
1256 vp->v_writecount += ap->a_inc;
1264 vop_stdneed_inactive(struct vop_need_inactive_args *ap)
1271 vop_stdioctl(struct vop_ioctl_args *ap)
1278 switch (ap->a_command) {
1282 error = vn_lock(vp, LK_SHARED);
1285 if (vp->v_type == VREG)
1286 error = VOP_GETATTR(vp, &va, ap->a_cred);
1291 if (*offp < 0 || *offp >= va.va_size)
1293 else if (ap->a_command == FIOSEEKHOLE)
1307 * used to fill the vfs function table to get reasonable default return values.
1310 vfs_stdroot (mp, flags, vpp)
1316 return (EOPNOTSUPP);
1320 vfs_stdstatfs (mp, sbp)
1325 return (EOPNOTSUPP);
1329 vfs_stdquotactl (mp, cmds, uid, arg)
1336 return (EOPNOTSUPP);
1340 vfs_stdsync(mp, waitfor)
1344 struct vnode *vp, *mvp;
1346 int error, lockreq, allerror = 0;
1349 lockreq = LK_EXCLUSIVE | LK_INTERLOCK;
1350 if (waitfor != MNT_WAIT)
1351 lockreq |= LK_NOWAIT;
1353 * Force stale buffer cache information to be flushed.
1356 MNT_VNODE_FOREACH_ALL(vp, mp, mvp) {
1357 if (vp->v_bufobj.bo_dirty.bv_cnt == 0) {
1361 if ((error = vget(vp, lockreq)) != 0) {
1362 if (error == ENOENT) {
1363 MNT_VNODE_FOREACH_ALL_ABORT(mp, mvp);
1368 error = VOP_FSYNC(vp, waitfor, td);
1377 vfs_stdnosync (mp, waitfor)
1386 vop_stdcopy_file_range(struct vop_copy_file_range_args *ap)
1390 error = vn_generic_copy_file_range(ap->a_invp, ap->a_inoffp,
1391 ap->a_outvp, ap->a_outoffp, ap->a_lenp, ap->a_flags, ap->a_incred,
1392 ap->a_outcred, ap->a_fsizetd);
1397 vfs_stdvget (mp, ino, flags, vpp)
1404 return (EOPNOTSUPP);
1408 vfs_stdfhtovp (mp, fhp, flags, vpp)
1415 return (EOPNOTSUPP);
1420 struct vfsconf *vfsp;
1427 vfs_stduninit (vfsp)
1428 struct vfsconf *vfsp;
1435 vfs_stdextattrctl(mp, cmd, filename_vp, attrnamespace, attrname)
1438 struct vnode *filename_vp;
1440 const char *attrname;
1443 if (filename_vp != NULL)
1444 VOP_UNLOCK(filename_vp);
1445 return (EOPNOTSUPP);
1449 vfs_stdsysctl(mp, op, req)
1452 struct sysctl_req *req;
1455 return (EOPNOTSUPP);
1458 static vop_bypass_t *
1459 bp_by_off(struct vop_vector *vop, struct vop_generic_args *a)
1462 return (*(vop_bypass_t **)((char *)vop + a->a_desc->vdesc_vop_offset));
1466 vop_sigdefer(struct vop_vector *vop, struct vop_generic_args *a)
1471 bp = bp_by_off(vop, a);
1474 prev_stops = sigdeferstop(SIGDEFERSTOP_SILENT);
1476 sigallowstop(prev_stops);
1481 vop_stdstat(struct vop_stat_args *a)
1493 error = vop_stat_helper_pre(a);
1500 * Initialize defaults for new and unusual fields, so that file
1501 * systems which don't support these fields don't need to know
1504 vap->va_birthtime.tv_sec = -1;
1505 vap->va_birthtime.tv_nsec = 0;
1506 vap->va_fsid = VNOVAL;
1507 vap->va_rdev = NODEV;
1509 error = VOP_GETATTR(vp, vap, a->a_active_cred);
1514 * Zero the spare stat fields
1516 bzero(sb, sizeof *sb);
1519 * Copy from vattr table
1521 if (vap->va_fsid != VNOVAL)
1522 sb->st_dev = vap->va_fsid;
1524 sb->st_dev = vp->v_mount->mnt_stat.f_fsid.val[0];
1525 sb->st_ino = vap->va_fileid;
1526 mode = vap->va_mode;
1527 switch (vap->va_type) {
1554 sb->st_nlink = vap->va_nlink;
1555 sb->st_uid = vap->va_uid;
1556 sb->st_gid = vap->va_gid;
1557 sb->st_rdev = vap->va_rdev;
1558 if (vap->va_size > OFF_MAX) {
1562 sb->st_size = vap->va_size;
1563 sb->st_atim.tv_sec = vap->va_atime.tv_sec;
1564 sb->st_atim.tv_nsec = vap->va_atime.tv_nsec;
1565 sb->st_mtim.tv_sec = vap->va_mtime.tv_sec;
1566 sb->st_mtim.tv_nsec = vap->va_mtime.tv_nsec;
1567 sb->st_ctim.tv_sec = vap->va_ctime.tv_sec;
1568 sb->st_ctim.tv_nsec = vap->va_ctime.tv_nsec;
1569 sb->st_birthtim.tv_sec = vap->va_birthtime.tv_sec;
1570 sb->st_birthtim.tv_nsec = vap->va_birthtime.tv_nsec;
1573 * According to www.opengroup.org, the meaning of st_blksize is
1574 * "a filesystem-specific preferred I/O block size for this
1575 * object. In some filesystem types, this may vary from file
1577 * Use minimum/default of PAGE_SIZE (e.g. for VCHR).
1580 sb->st_blksize = max(PAGE_SIZE, vap->va_blocksize);
1581 sb->st_flags = vap->va_flags;
1582 sb->st_blocks = vap->va_bytes / S_BLKSIZE;
1583 sb->st_gen = vap->va_gen;
1585 return (vop_stat_helper_post(a, error));
1589 vop_stdread_pgcache(struct vop_read_pgcache_args *ap __unused)
1591 return (EJUSTRETURN);