2 * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
4 * Copyright (c) 2004 Poul-Henning Kamp
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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.
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
29 #include <sys/cdefs.h>
30 __FBSDID("$FreeBSD$");
32 #include <sys/param.h>
33 #include <sys/systm.h>
35 #include <sys/kernel.h>
37 #include <sys/malloc.h>
38 #include <sys/mutex.h>
40 #include <sys/vnode.h>
41 #include <sys/mount.h>
43 #include <geom/geom.h>
44 #include <geom/geom_vfs.h>
47 * subroutines for use by filesystems.
49 * XXX: should maybe live somewhere else ?
56 struct g_event *sc_event;
60 int sc_enxio_reported;
63 static struct buf_ops __g_vfs_bufops = {
64 .bop_name = "GEOM_VFS",
65 .bop_write = bufwrite,
66 .bop_strategy = g_vfs_strategy,
68 .bop_bdflush = bufbdflush
71 struct buf_ops *g_vfs_bufops = &__g_vfs_bufops;
73 static g_orphan_t g_vfs_orphan;
75 static struct g_class g_vfs_class = {
78 .orphan = g_vfs_orphan,
81 DECLARE_GEOM_CLASS(g_vfs_class, g_vfs);
84 g_vfs_destroy(void *arg, int flags __unused)
86 struct g_consumer *cp;
90 if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0)
91 g_access(cp, -cp->acr, -cp->acw, -cp->ace);
93 if (cp->geom->softc == NULL)
94 g_wither_geom(cp->geom, ENXIO);
98 g_vfs_done(struct bio *bip)
100 struct g_consumer *cp;
101 struct g_event *event;
102 struct g_vfs_softc *sc;
110 * Collect statistics on synchronous and asynchronous read
111 * and write counts for disks that have associated filesystems.
113 bp = bip->bio_caller2;
117 * If not a disk vnode, use its associated mount point
118 * otherwise use the mountpoint associated with the disk.
121 if (vp->v_type != VCHR ||
122 (cdevp = vp->v_rdev) == NULL ||
123 cdevp->si_devsw == NULL ||
124 (cdevp->si_devsw->d_flags & D_DISK) == 0)
127 mp = cdevp->si_mountpt;
129 if (bp->b_iocmd == BIO_READ) {
130 if (LK_HOLDER(bp->b_lock.lk_lock) == LK_KERNPROC)
131 mp->mnt_stat.f_asyncreads++;
133 mp->mnt_stat.f_syncreads++;
134 } else if (bp->b_iocmd == BIO_WRITE) {
135 if (LK_HOLDER(bp->b_lock.lk_lock) == LK_KERNPROC)
136 mp->mnt_stat.f_asyncwrites++;
138 mp->mnt_stat.f_syncwrites++;
145 sc = cp->geom->softc;
146 if (bip->bio_error != 0 && bip->bio_error != EOPNOTSUPP) {
147 if ((bp->b_xflags & BX_CVTENXIO) != 0) {
148 if (atomic_cmpset_int(&sc->sc_enxio_active, 0, 1))
149 printf("g_vfs_done(): %s converting all errors to ENXIO\n",
152 if (sc->sc_enxio_active)
153 bip->bio_error = ENXIO;
154 if (bip->bio_error != ENXIO ||
155 atomic_cmpset_int(&sc->sc_enxio_reported, 0, 1)) {
156 g_print_bio("g_vfs_done():", bip, "error = %d%s",
158 bip->bio_error != ENXIO ? "" :
159 " supressing further ENXIO");
162 bp->b_error = bip->bio_error;
163 bp->b_ioflags = bip->bio_flags;
165 bp->b_ioflags |= BIO_ERROR;
166 bp->b_resid = bp->b_bcount - bip->bio_completed;
169 mtx_lock(&sc->sc_mtx);
170 destroy = ((--sc->sc_active) == 0 && sc->sc_orphaned);
172 event = sc->sc_event;
176 mtx_unlock(&sc->sc_mtx);
178 g_post_event_ep(g_vfs_destroy, cp, event, NULL);
184 g_vfs_strategy(struct bufobj *bo, struct buf *bp)
186 struct g_vfs_softc *sc;
187 struct g_consumer *cp;
191 sc = cp->geom->softc;
194 * If the provider has orphaned us, just return ENXIO.
196 mtx_lock(&sc->sc_mtx);
197 if (sc->sc_orphaned || sc->sc_enxio_active) {
198 mtx_unlock(&sc->sc_mtx);
200 bp->b_ioflags |= BIO_ERROR;
205 mtx_unlock(&sc->sc_mtx);
208 bip->bio_cmd = bp->b_iocmd;
209 bip->bio_offset = bp->b_iooffset;
210 bip->bio_length = bp->b_bcount;
212 if ((bp->b_flags & B_BARRIER) != 0) {
213 bip->bio_flags |= BIO_ORDERED;
214 bp->b_flags &= ~B_BARRIER;
216 if (bp->b_iocmd == BIO_SPEEDUP)
217 bip->bio_flags |= bp->b_ioflags;
218 bip->bio_done = g_vfs_done;
219 bip->bio_caller2 = bp;
220 #if defined(BUF_TRACKING) || defined(FULL_BUF_TRACKING)
221 buf_track(bp, __func__);
222 bip->bio_track_bp = bp;
224 g_io_request(bip, cp);
228 g_vfs_orphan(struct g_consumer *cp)
231 struct g_event *event;
232 struct g_vfs_softc *sc;
238 g_trace(G_T_TOPOLOGY, "g_vfs_orphan(%p(%s))", cp, gp->name);
242 event = g_alloc_event(M_WAITOK);
243 mtx_lock(&sc->sc_mtx);
244 KASSERT(sc->sc_event == NULL, ("g_vfs %p already has an event", sc));
245 sc->sc_orphaned = true;
246 destroy = (sc->sc_active == 0);
248 sc->sc_event = event;
251 mtx_unlock(&sc->sc_mtx);
254 g_vfs_destroy(cp, 0);
258 * Do not destroy the geom. Filesystem will do that during unmount.
263 g_vfs_open(struct vnode *vp, struct g_consumer **cpp, const char *fsname, int wr)
266 struct g_provider *pp;
267 struct g_consumer *cp;
268 struct g_vfs_softc *sc;
276 if (bo->bo_private != vp)
279 pp = g_dev_getprovider(vp->v_rdev);
282 gp = g_new_geomf(&g_vfs_class, "%s.%s", fsname, pp->name);
283 sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO);
284 mtx_init(&sc->sc_mtx, "g_vfs", NULL, MTX_DEF);
287 cp = g_new_consumer(gp);
288 error = g_attach(cp, pp);
290 g_wither_geom(gp, ENXIO);
293 error = g_access(cp, 1, wr, wr);
295 g_wither_geom(gp, ENXIO);
298 vnode_create_vobject(vp, pp->mediasize, curthread);
301 cp->flags |= G_CF_DIRECT_SEND | G_CF_DIRECT_RECEIVE;
302 bo->bo_ops = g_vfs_bufops;
304 bo->bo_bsize = pp->sectorsize;
310 g_vfs_close(struct g_consumer *cp)
313 struct g_vfs_softc *sc;
319 bufobj_invalbuf(sc->sc_bo, V_SAVE, 0, 0);
320 sc->sc_bo->bo_private = cp->private;
322 mtx_destroy(&sc->sc_mtx);
323 if (!sc->sc_orphaned || cp->provider == NULL)
324 g_wither_geom_close(gp, ENXIO);
325 KASSERT(sc->sc_event == NULL, ("g_vfs %p event is non-NULL", sc));