sys/kern/vfs_hash.c

   1 /*-
   2  * SPDX-License-Identifier: BSD-2-Clause
   3  *
   4  * Copyright (c) 2005 Poul-Henning Kamp
   5  * All rights reserved.
   6  *
   7  * Redistribution and use in source and binary forms, with or without
   8  * modification, are permitted provided that the following conditions
   9  * are met:
  10  * 1. Redistributions of source code must retain the above copyright
  11  *    notice, this list of conditions and the following disclaimer.
  12  * 2. Redistributions in binary form must reproduce the above copyright
  13  *    notice, this list of conditions and the following disclaimer in the
  14  *    documentation and/or other materials provided with the distribution.
  15  *
  16  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  17  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  18  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  19  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  20  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  21  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  22  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  23  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  24  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  25  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  26  * SUCH DAMAGE.
  27  *
  28  */
  29
  30 #include <sys/cdefs.h>
  31 #include <sys/param.h>
  32 #include <sys/systm.h>
  33 #include <sys/kernel.h>
  34 #include <sys/malloc.h>
  35 #include <sys/mount.h>
  36 #include <sys/rwlock.h>
  37 #include <sys/vnode.h>
  38
  39 static MALLOC_DEFINE(M_VFS_HASH, "vfs_hash", "VFS hash table");
  40
  41 static LIST_HEAD(vfs_hash_head, vnode)  *vfs_hash_tbl;
  42 static LIST_HEAD(,vnode)                vfs_hash_side;
  43 static u_long                           vfs_hash_mask;
  44 static struct rwlock __exclusive_cache_line vfs_hash_lock;
  45
  46 static void
  47 vfs_hashinit(void *dummy __unused)
  48 {
  49
  50         vfs_hash_tbl = hashinit(desiredvnodes, M_VFS_HASH, &vfs_hash_mask);
  51         rw_init(&vfs_hash_lock, "vfs hash");
  52         LIST_INIT(&vfs_hash_side);
  53 }
  54
  55 /* Must be SI_ORDER_SECOND so desiredvnodes is available */
  56 SYSINIT(vfs_hash, SI_SUB_VFS, SI_ORDER_SECOND, vfs_hashinit, NULL);
  57
  58 u_int
  59 vfs_hash_index(struct vnode *vp)
  60 {
  61
  62         return (vp->v_hash + vp->v_mount->mnt_hashseed);
  63 }
  64
  65 static struct vfs_hash_head *
  66 vfs_hash_bucket(const struct mount *mp, u_int hash)
  67 {
  68
  69         return (&vfs_hash_tbl[(hash + mp->mnt_hashseed) & vfs_hash_mask]);
  70 }
  71
  72 int
  73 vfs_hash_get(const struct mount *mp, u_int hash, int flags, struct thread *td,
  74     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg)
  75 {
  76         struct vnode *vp;
  77         enum vgetstate vs;
  78         int error;
  79
  80         while (1) {
  81                 rw_rlock(&vfs_hash_lock);
  82                 LIST_FOREACH(vp, vfs_hash_bucket(mp, hash), v_hashlist) {
  83                         if (vp->v_hash != hash)
  84                                 continue;
  85                         if (vp->v_mount != mp)
  86                                 continue;
  87                         if (fn != NULL && fn(vp, arg))
  88                                 continue;
  89                         vs = vget_prep(vp);
  90                         rw_runlock(&vfs_hash_lock);
  91                         error = vget_finish(vp, flags, vs);
  92                         if (error == ENOENT && (flags & LK_NOWAIT) == 0)
  93                                 break;
  94                         if (error != 0)
  95                                 return (error);
  96                         if (vp->v_hash != hash ||
  97                             (fn != NULL && fn(vp, arg))) {
  98                                 vput(vp);
  99                                 /* Restart the bucket walk. */
 100                                 break;
 101                         }
 102                         *vpp = vp;
 103                         return (0);
 104                 }
 105                 if (vp == NULL) {
 106                         rw_runlock(&vfs_hash_lock);
 107                         *vpp = NULL;
 108                         return (0);
 109                 }
 110         }
 111 }
 112
 113 void
 114 vfs_hash_ref(const struct mount *mp, u_int hash, struct thread *td,
 115     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg)
 116 {
 117         struct vnode *vp;
 118
 119         while (1) {
 120                 rw_rlock(&vfs_hash_lock);
 121                 LIST_FOREACH(vp, vfs_hash_bucket(mp, hash), v_hashlist) {
 122                         if (vp->v_hash != hash)
 123                                 continue;
 124                         if (vp->v_mount != mp)
 125                                 continue;
 126                         if (fn != NULL && fn(vp, arg))
 127                                 continue;
 128                         vhold(vp);
 129                         rw_runlock(&vfs_hash_lock);
 130                         vref(vp);
 131                         vdrop(vp);
 132                         *vpp = vp;
 133                         return;
 134                 }
 135                 if (vp == NULL) {
 136                         rw_runlock(&vfs_hash_lock);
 137                         *vpp = NULL;
 138                         return;
 139                 }
 140         }
 141 }
 142
 143 void
 144 vfs_hash_remove(struct vnode *vp)
 145 {
 146
 147         rw_wlock(&vfs_hash_lock);
 148         LIST_REMOVE(vp, v_hashlist);
 149         rw_wunlock(&vfs_hash_lock);
 150 }
 151
 152 int
 153 vfs_hash_insert(struct vnode *vp, u_int hash, int flags, struct thread *td,
 154     struct vnode **vpp, vfs_hash_cmp_t *fn, void *arg)
 155 {
 156         struct vnode *vp2;
 157         enum vgetstate vs;
 158         int error;
 159
 160         *vpp = NULL;
 161         while (1) {
 162                 rw_wlock(&vfs_hash_lock);
 163                 LIST_FOREACH(vp2,
 164                     vfs_hash_bucket(vp->v_mount, hash), v_hashlist) {
 165                         if (vp2->v_hash != hash)
 166                                 continue;
 167                         if (vp2->v_mount != vp->v_mount)
 168                                 continue;
 169                         if (fn != NULL && fn(vp2, arg))
 170                                 continue;
 171                         vs = vget_prep(vp2);
 172                         rw_wunlock(&vfs_hash_lock);
 173                         error = vget_finish(vp2, flags, vs);
 174                         if (error == ENOENT && (flags & LK_NOWAIT) == 0)
 175                                 break;
 176                         rw_wlock(&vfs_hash_lock);
 177                         LIST_INSERT_HEAD(&vfs_hash_side, vp, v_hashlist);
 178                         rw_wunlock(&vfs_hash_lock);
 179                         vgone(vp);
 180                         vput(vp);
 181                         if (!error)
 182                                 *vpp = vp2;
 183                         return (error);
 184                 }
 185                 if (vp2 == NULL)
 186                         break;
 187         }
 188         vp->v_hash = hash;
 189         LIST_INSERT_HEAD(vfs_hash_bucket(vp->v_mount, hash), vp, v_hashlist);
 190         rw_wunlock(&vfs_hash_lock);
 191         return (0);
 192 }
 193
 194 void
 195 vfs_hash_rehash(struct vnode *vp, u_int hash)
 196 {
 197         ASSERT_VOP_ELOCKED(vp, "rehash requires excl lock");
 198
 199         rw_wlock(&vfs_hash_lock);
 200         LIST_REMOVE(vp, v_hashlist);
 201         LIST_INSERT_HEAD(vfs_hash_bucket(vp->v_mount, hash), vp, v_hashlist);
 202         vp->v_hash = hash;
 203         rw_wunlock(&vfs_hash_lock);
 204 }
 205
 206 void
 207 vfs_hash_changesize(u_long newmaxvnodes)
 208 {
 209         struct vfs_hash_head *vfs_hash_newtbl, *vfs_hash_oldtbl;
 210         u_long vfs_hash_newmask, vfs_hash_oldmask;
 211         struct vnode *vp;
 212         int i;
 213
 214         vfs_hash_newtbl = hashinit(newmaxvnodes, M_VFS_HASH,
 215                 &vfs_hash_newmask);
 216         /* If same hash table size, nothing to do */
 217         if (vfs_hash_mask == vfs_hash_newmask) {
 218                 free(vfs_hash_newtbl, M_VFS_HASH);
 219                 return;
 220         }
 221         /*
 222          * Move everything from the old hash table to the new table.
 223          * None of the vnodes in the table can be recycled because to
 224          * do so, they have to be removed from the hash table.
 225          */
 226         rw_wlock(&vfs_hash_lock);
 227         vfs_hash_oldtbl = vfs_hash_tbl;
 228         vfs_hash_oldmask = vfs_hash_mask;
 229         vfs_hash_tbl = vfs_hash_newtbl;
 230         vfs_hash_mask = vfs_hash_newmask;
 231         for (i = 0; i <= vfs_hash_oldmask; i++) {
 232                 while ((vp = LIST_FIRST(&vfs_hash_oldtbl[i])) != NULL) {
 233                         LIST_REMOVE(vp, v_hashlist);
 234                         LIST_INSERT_HEAD(
 235                             vfs_hash_bucket(vp->v_mount, vp->v_hash),
 236                             vp, v_hashlist);
 237                 }
 238         }
 239         rw_wunlock(&vfs_hash_lock);
 240         free(vfs_hash_oldtbl, M_VFS_HASH);
 241 }