1 /*- 2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD 3 * 4 * Copyright (c) 2004 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 #include <sys/cdefs.h> 30 __FBSDID("$FreeBSD$"); 31 32 #include <sys/param.h> 33 #include <sys/systm.h> 34 #include <sys/bio.h> 35 #include <sys/kernel.h> 36 #include <sys/lock.h> 37 #include <sys/malloc.h> 38 #include <sys/mutex.h> 39 #include <sys/sbuf.h> 40 #include <sys/vnode.h> 41 #include <sys/mount.h> 42 43 #include <geom/geom.h> 44 #include <geom/geom_vfs.h> 45 46 /* 47 * subroutines for use by filesystems. 48 * 49 * XXX: should maybe live somewhere else ? 50 */ 51 #include <sys/buf.h> 52 53 struct g_vfs_softc { 54 struct mtx sc_mtx; 55 struct bufobj *sc_bo; 56 int sc_active; 57 int sc_orphaned; 58 }; 59 60 static struct buf_ops __g_vfs_bufops = { 61 .bop_name = "GEOM_VFS", 62 .bop_write = bufwrite, 63 .bop_strategy = g_vfs_strategy, 64 .bop_sync = bufsync, 65 .bop_bdflush = bufbdflush 66 }; 67 68 struct buf_ops *g_vfs_bufops = &__g_vfs_bufops; 69 70 static g_orphan_t g_vfs_orphan; 71 72 static struct g_class g_vfs_class = { 73 .name = "VFS", 74 .version = G_VERSION, 75 .orphan = g_vfs_orphan, 76 }; 77 78 DECLARE_GEOM_CLASS(g_vfs_class, g_vfs); 79 80 static void 81 g_vfs_destroy(void *arg, int flags __unused) 82 { 83 struct g_consumer *cp; 84 85 g_topology_assert(); 86 cp = arg; 87 if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0) 88 g_access(cp, -cp->acr, -cp->acw, -cp->ace); 89 g_detach(cp); 90 if (cp->geom->softc == NULL) 91 g_wither_geom(cp->geom, ENXIO); 92 } 93 94 static void 95 g_vfs_done(struct bio *bip) 96 { 97 struct g_consumer *cp; 98 struct g_vfs_softc *sc; 99 struct buf *bp; 100 int destroy; 101 struct mount *mp; 102 struct vnode *vp; 103 struct cdev *cdevp; 104 105 /* 106 * Collect statistics on synchronous and asynchronous read 107 * and write counts for disks that have associated filesystems. 108 */ 109 bp = bip->bio_caller2; 110 vp = bp->b_vp; 111 if (vp != NULL) { 112 /* 113 * If not a disk vnode, use its associated mount point 114 * otherwise use the mountpoint associated with the disk. 115 */ 116 VI_LOCK(vp); 117 if (vp->v_type != VCHR || 118 (cdevp = vp->v_rdev) == NULL || 119 cdevp->si_devsw == NULL || 120 (cdevp->si_devsw->d_flags & D_DISK) == 0) 121 mp = vp->v_mount; 122 else 123 mp = cdevp->si_mountpt; 124 if (mp != NULL) { 125 if (bp->b_iocmd == BIO_READ) { 126 if (LK_HOLDER(bp->b_lock.lk_lock) == LK_KERNPROC) 127 mp->mnt_stat.f_asyncreads++; 128 else 129 mp->mnt_stat.f_syncreads++; 130 } else if (bp->b_iocmd == BIO_WRITE) { 131 if (LK_HOLDER(bp->b_lock.lk_lock) == LK_KERNPROC) 132 mp->mnt_stat.f_asyncwrites++; 133 else 134 mp->mnt_stat.f_syncwrites++; 135 } 136 } 137 VI_UNLOCK(vp); 138 } 139 140 cp = bip->bio_from; 141 sc = cp->geom->softc; 142 if (bip->bio_error) 143 g_print_bio("g_vfs_done():", bip, "error = %d", 144 bip->bio_error); 145 bp->b_error = bip->bio_error; 146 bp->b_ioflags = bip->bio_flags; 147 if (bip->bio_error) 148 bp->b_ioflags |= BIO_ERROR; 149 bp->b_resid = bp->b_bcount - bip->bio_completed; 150 g_destroy_bio(bip); 151 152 mtx_lock(&sc->sc_mtx); 153 destroy = ((--sc->sc_active) == 0 && sc->sc_orphaned); 154 mtx_unlock(&sc->sc_mtx); 155 if (destroy) 156 g_post_event(g_vfs_destroy, cp, M_WAITOK, NULL); 157 158 bufdone(bp); 159 } 160 161 void 162 g_vfs_strategy(struct bufobj *bo, struct buf *bp) 163 { 164 struct g_vfs_softc *sc; 165 struct g_consumer *cp; 166 struct bio *bip; 167 168 cp = bo->bo_private; 169 sc = cp->geom->softc; 170 171 /* 172 * If the provider has orphaned us, just return ENXIO. 173 */ 174 mtx_lock(&sc->sc_mtx); 175 if (sc->sc_orphaned) { 176 mtx_unlock(&sc->sc_mtx); 177 bp->b_error = ENXIO; 178 bp->b_ioflags |= BIO_ERROR; 179 bufdone(bp); 180 return; 181 } 182 sc->sc_active++; 183 mtx_unlock(&sc->sc_mtx); 184 185 bip = g_alloc_bio(); 186 bip->bio_cmd = bp->b_iocmd; 187 bip->bio_offset = bp->b_iooffset; 188 bip->bio_length = bp->b_bcount; 189 bdata2bio(bp, bip); 190 if ((bp->b_flags & B_BARRIER) != 0) { 191 bip->bio_flags |= BIO_ORDERED; 192 bp->b_flags &= ~B_BARRIER; 193 } 194 if (bp->b_iocmd == BIO_SPEEDUP) 195 bip->bio_flags |= bp->b_ioflags; 196 bip->bio_done = g_vfs_done; 197 bip->bio_caller2 = bp; 198 #if defined(BUF_TRACKING) || defined(FULL_BUF_TRACKING) 199 buf_track(bp, __func__); 200 bip->bio_track_bp = bp; 201 #endif 202 g_io_request(bip, cp); 203 } 204 205 static void 206 g_vfs_orphan(struct g_consumer *cp) 207 { 208 struct g_geom *gp; 209 struct g_vfs_softc *sc; 210 int destroy; 211 212 g_topology_assert(); 213 214 gp = cp->geom; 215 g_trace(G_T_TOPOLOGY, "g_vfs_orphan(%p(%s))", cp, gp->name); 216 sc = gp->softc; 217 if (sc == NULL) 218 return; 219 mtx_lock(&sc->sc_mtx); 220 sc->sc_orphaned = 1; 221 destroy = (sc->sc_active == 0); 222 mtx_unlock(&sc->sc_mtx); 223 if (destroy) 224 g_vfs_destroy(cp, 0); 225 226 /* 227 * Do not destroy the geom. Filesystem will do that during unmount. 228 */ 229 } 230 231 int 232 g_vfs_open(struct vnode *vp, struct g_consumer **cpp, const char *fsname, int wr) 233 { 234 struct g_geom *gp; 235 struct g_provider *pp; 236 struct g_consumer *cp; 237 struct g_vfs_softc *sc; 238 struct bufobj *bo; 239 int error; 240 241 g_topology_assert(); 242 243 *cpp = NULL; 244 bo = &vp->v_bufobj; 245 if (bo->bo_private != vp) 246 return (EBUSY); 247 248 pp = g_dev_getprovider(vp->v_rdev); 249 if (pp == NULL) 250 return (ENOENT); 251 gp = g_new_geomf(&g_vfs_class, "%s.%s", fsname, pp->name); 252 sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO); 253 mtx_init(&sc->sc_mtx, "g_vfs", NULL, MTX_DEF); 254 sc->sc_bo = bo; 255 gp->softc = sc; 256 cp = g_new_consumer(gp); 257 g_attach(cp, pp); 258 error = g_access(cp, 1, wr, wr); 259 if (error) { 260 g_wither_geom(gp, ENXIO); 261 return (error); 262 } 263 vnode_create_vobject(vp, pp->mediasize, curthread); 264 *cpp = cp; 265 cp->private = vp; 266 cp->flags |= G_CF_DIRECT_SEND | G_CF_DIRECT_RECEIVE; 267 bo->bo_ops = g_vfs_bufops; 268 bo->bo_private = cp; 269 bo->bo_bsize = pp->sectorsize; 270 271 return (error); 272 } 273 274 void 275 g_vfs_close(struct g_consumer *cp) 276 { 277 struct g_geom *gp; 278 struct g_vfs_softc *sc; 279 280 g_topology_assert(); 281 282 gp = cp->geom; 283 sc = gp->softc; 284 bufobj_invalbuf(sc->sc_bo, V_SAVE, 0, 0); 285 sc->sc_bo->bo_private = cp->private; 286 gp->softc = NULL; 287 mtx_destroy(&sc->sc_mtx); 288 if (!sc->sc_orphaned || cp->provider == NULL) 289 g_wither_geom_close(gp, ENXIO); 290 g_free(sc); 291 } 292