1/*-
2 * Copyright (c) 2004 Poul-Henning Kamp
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 *
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24 * SUCH DAMAGE.
25 */
26
27#include <sys/cdefs.h>
28__FBSDID("$FreeBSD: releng/10.3/sys/geom/geom_vfs.c 260385 2014-01-07 01:32:23Z scottl $");
29
30#include <sys/param.h>
31#include <sys/systm.h>
32#include <sys/bio.h>
33#include <sys/kernel.h>
34#include <sys/lock.h>
35#include <sys/malloc.h>
36#include <sys/mutex.h>
37#include <sys/vnode.h>
38#include <sys/mount.h>
39
40#include <geom/geom.h>
41#include <geom/geom_vfs.h>
42
43/*
44 * subroutines for use by filesystems.
45 *
46 * XXX: should maybe live somewhere else ?
47 */
48#include <sys/buf.h>
49
50struct g_vfs_softc {
51	struct mtx	 sc_mtx;
52	struct bufobj	*sc_bo;
53	int		 sc_active;
54	int		 sc_orphaned;
55};
56
57static struct buf_ops __g_vfs_bufops = {
58	.bop_name =	"GEOM_VFS",
59	.bop_write =	bufwrite,
60	.bop_strategy =	g_vfs_strategy,
61	.bop_sync =	bufsync,
62	.bop_bdflush =	bufbdflush
63};
64
65struct buf_ops *g_vfs_bufops = &__g_vfs_bufops;
66
67static g_orphan_t g_vfs_orphan;
68
69static struct g_class g_vfs_class = {
70	.name =		"VFS",
71	.version =	G_VERSION,
72	.orphan =	g_vfs_orphan,
73};
74
75DECLARE_GEOM_CLASS(g_vfs_class, g_vfs);
76
77static void
78g_vfs_destroy(void *arg, int flags __unused)
79{
80	struct g_consumer *cp;
81
82	g_topology_assert();
83	cp = arg;
84	if (cp->acr > 0 || cp->acw > 0 || cp->ace > 0)
85		g_access(cp, -cp->acr, -cp->acw, -cp->ace);
86	g_detach(cp);
87	if (cp->geom->softc == NULL)
88		g_wither_geom(cp->geom, ENXIO);
89}
90
91static void
92g_vfs_done(struct bio *bip)
93{
94	struct g_consumer *cp;
95	struct g_vfs_softc *sc;
96	struct buf *bp;
97	int destroy;
98	struct mount *mp;
99	struct vnode *vp;
100	struct cdev *cdevp;
101
102	/*
103	 * Collect statistics on synchronous and asynchronous read
104	 * and write counts for disks that have associated filesystems.
105	 */
106	bp = bip->bio_caller2;
107	vp = bp->b_vp;
108	if (vp != NULL) {
109		/*
110		 * If not a disk vnode, use its associated mount point
111		 * otherwise use the mountpoint associated with the disk.
112		 */
113		VI_LOCK(vp);
114		if (vp->v_type != VCHR ||
115		    (cdevp = vp->v_rdev) == NULL ||
116		    cdevp->si_devsw == NULL ||
117		    (cdevp->si_devsw->d_flags & D_DISK) == 0)
118			mp = vp->v_mount;
119		else
120			mp = cdevp->si_mountpt;
121		if (mp != NULL) {
122			if (bp->b_iocmd == BIO_READ) {
123				if (LK_HOLDER(bp->b_lock.lk_lock) == LK_KERNPROC)
124					mp->mnt_stat.f_asyncreads++;
125				else
126					mp->mnt_stat.f_syncreads++;
127			} else if (bp->b_iocmd == BIO_WRITE) {
128				if (LK_HOLDER(bp->b_lock.lk_lock) == LK_KERNPROC)
129					mp->mnt_stat.f_asyncwrites++;
130				else
131					mp->mnt_stat.f_syncwrites++;
132			}
133		}
134		VI_UNLOCK(vp);
135	}
136
137	cp = bip->bio_from;
138	sc = cp->geom->softc;
139	if (bip->bio_error) {
140		printf("g_vfs_done():");
141		g_print_bio(bip);
142		printf("error = %d\n", bip->bio_error);
143	}
144	bp->b_error = bip->bio_error;
145	bp->b_ioflags = bip->bio_flags;
146	if (bip->bio_error)
147		bp->b_ioflags |= BIO_ERROR;
148	bp->b_resid = bp->b_bcount - bip->bio_completed;
149	g_destroy_bio(bip);
150
151	mtx_lock(&sc->sc_mtx);
152	destroy = ((--sc->sc_active) == 0 && sc->sc_orphaned);
153	mtx_unlock(&sc->sc_mtx);
154	if (destroy)
155		g_post_event(g_vfs_destroy, cp, M_WAITOK, NULL);
156
157	bufdone(bp);
158}
159
160void
161g_vfs_strategy(struct bufobj *bo, struct buf *bp)
162{
163	struct g_vfs_softc *sc;
164	struct g_consumer *cp;
165	struct bio *bip;
166
167	cp = bo->bo_private;
168	sc = cp->geom->softc;
169
170	/*
171	 * If the provider has orphaned us, just return EXIO.
172	 */
173	mtx_lock(&sc->sc_mtx);
174	if (sc->sc_orphaned) {
175		mtx_unlock(&sc->sc_mtx);
176		bp->b_error = ENXIO;
177		bp->b_ioflags |= BIO_ERROR;
178		bufdone(bp);
179		return;
180	}
181	sc->sc_active++;
182	mtx_unlock(&sc->sc_mtx);
183
184	bip = g_alloc_bio();
185	bip->bio_cmd = bp->b_iocmd;
186	bip->bio_offset = bp->b_iooffset;
187	bip->bio_length = bp->b_bcount;
188	bdata2bio(bp, bip);
189	if ((bp->b_flags & B_BARRIER) != 0) {
190		bip->bio_flags |= BIO_ORDERED;
191		bp->b_flags &= ~B_BARRIER;
192	}
193	bip->bio_done = g_vfs_done;
194	bip->bio_caller2 = bp;
195	g_io_request(bip, cp);
196}
197
198static void
199g_vfs_orphan(struct g_consumer *cp)
200{
201	struct g_geom *gp;
202	struct g_vfs_softc *sc;
203	int destroy;
204
205	g_topology_assert();
206
207	gp = cp->geom;
208	g_trace(G_T_TOPOLOGY, "g_vfs_orphan(%p(%s))", cp, gp->name);
209	sc = gp->softc;
210	if (sc == NULL)
211		return;
212	mtx_lock(&sc->sc_mtx);
213	sc->sc_orphaned = 1;
214	destroy = (sc->sc_active == 0);
215	mtx_unlock(&sc->sc_mtx);
216	if (destroy)
217		g_vfs_destroy(cp, 0);
218
219	/*
220	 * Do not destroy the geom.  Filesystem will do that during unmount.
221	 */
222}
223
224int
225g_vfs_open(struct vnode *vp, struct g_consumer **cpp, const char *fsname, int wr)
226{
227	struct g_geom *gp;
228	struct g_provider *pp;
229	struct g_consumer *cp;
230	struct g_vfs_softc *sc;
231	struct bufobj *bo;
232	int error;
233
234	g_topology_assert();
235
236	*cpp = NULL;
237	bo = &vp->v_bufobj;
238	if (bo->bo_private != vp)
239		return (EBUSY);
240
241	pp = g_dev_getprovider(vp->v_rdev);
242	if (pp == NULL)
243		return (ENOENT);
244	gp = g_new_geomf(&g_vfs_class, "%s.%s", fsname, pp->name);
245	sc = g_malloc(sizeof(*sc), M_WAITOK | M_ZERO);
246	mtx_init(&sc->sc_mtx, "g_vfs", NULL, MTX_DEF);
247	sc->sc_bo = bo;
248	gp->softc = sc;
249	cp = g_new_consumer(gp);
250	g_attach(cp, pp);
251	error = g_access(cp, 1, wr, wr);
252	if (error) {
253		g_wither_geom(gp, ENXIO);
254		return (error);
255	}
256	vnode_create_vobject(vp, pp->mediasize, curthread);
257	*cpp = cp;
258	cp->private = vp;
259	cp->flags |= G_CF_DIRECT_SEND | G_CF_DIRECT_RECEIVE;
260	bo->bo_ops = g_vfs_bufops;
261	bo->bo_private = cp;
262	bo->bo_bsize = pp->sectorsize;
263
264	return (error);
265}
266
267void
268g_vfs_close(struct g_consumer *cp)
269{
270	struct g_geom *gp;
271	struct g_vfs_softc *sc;
272
273	g_topology_assert();
274
275	gp = cp->geom;
276	sc = gp->softc;
277	bufobj_invalbuf(sc->sc_bo, V_SAVE, 0, 0);
278	sc->sc_bo->bo_private = cp->private;
279	gp->softc = NULL;
280	mtx_destroy(&sc->sc_mtx);
281	if (!sc->sc_orphaned || cp->provider == NULL)
282		g_wither_geom_close(gp, ENXIO);
283	g_free(sc);
284}
285