1/*-
2 * Copyright (c) 2002 Poul-Henning Kamp
3 * Copyright (c) 2002 Networks Associates Technology, Inc.
4 * All rights reserved.
5 *
6 * This software was developed for the FreeBSD Project by Poul-Henning Kamp
7 * and NAI Labs, the Security Research Division of Network Associates, Inc.
8 * under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the
9 * DARPA CHATS research program.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. The names of the authors may not be used to endorse or promote
20 * products derived from this software without specific prior written
21 * permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 */
35
36#include <sys/cdefs.h>
37__FBSDID("$FreeBSD: head/sys/geom/geom_disk.c 124864 2004-01-23 10:56:16Z phk $");
38
39#include "opt_geom.h"
40
41#include <sys/param.h>
42#include <sys/systm.h>
43#include <sys/kernel.h>
44#include <sys/sysctl.h>
45#include <sys/bio.h>
46#include <sys/conf.h>
47#include <sys/fcntl.h>
48#include <sys/malloc.h>
49#include <sys/sysctl.h>
50#include <sys/devicestat.h>
51#include <machine/md_var.h>
52
53#include <sys/lock.h>
54#include <sys/mutex.h>
55#include <geom/geom.h>
56#include <geom/geom_disk.h>
57#include <geom/geom_int.h>
58
59static struct mtx g_disk_done_mtx;
60
61static g_access_t g_disk_access;
62static g_init_t g_disk_init;
63static g_fini_t g_disk_fini;
64
65struct g_class g_disk_class = {
66 .name = "DISK",
67 .init = g_disk_init,
68 .fini = g_disk_fini,
69};
70
71static void
72g_disk_init(struct g_class *mp __unused)
73{
74
75 mtx_init(&g_disk_done_mtx, "g_disk_done", NULL, MTX_DEF);
76}
77
78static void
79g_disk_fini(struct g_class *mp __unused)
80{
81
82 mtx_destroy(&g_disk_done_mtx);
83}
84
85DECLARE_GEOM_CLASS(g_disk_class, g_disk);
86
87static void __inline
88g_disk_lock_giant(struct disk *dp)
89{
90 if (dp->d_flags & DISKFLAG_NOGIANT)
91 return;
92 mtx_lock(&Giant);
93}
94
95static void __inline
96g_disk_unlock_giant(struct disk *dp)
97{
98 if (dp->d_flags & DISKFLAG_NOGIANT)
99 return;
100 mtx_unlock(&Giant);
101}
102
103static int
104g_disk_access(struct g_provider *pp, int r, int w, int e)
105{
106 struct disk *dp;
107 int error;
108
109 g_trace(G_T_ACCESS, "g_disk_access(%s, %d, %d, %d)",
110 pp->name, r, w, e);
111 g_topology_assert();
112 r += pp->acr;
113 w += pp->acw;
114 e += pp->ace;
115 dp = pp->geom->softc;
116 if (dp == NULL)
117 return (ENXIO);
118 error = 0;
119 if ((pp->acr + pp->acw + pp->ace) == 0 && (r + w + e) > 0) {
120 if (dp->d_open != NULL) {
121 g_disk_lock_giant(dp);
122 error = dp->d_open(dp);
123 if (error != 0)
124 printf("Opened disk %s -> %d\n",
125 pp->name, error);
126 g_disk_unlock_giant(dp);
127 }
128 pp->mediasize = dp->d_mediasize;
129 pp->sectorsize = dp->d_sectorsize;
130 dp->d_flags |= DISKFLAG_OPEN;
131 if (dp->d_maxsize == 0) {
132 printf("WARNING: Disk drive %s%d has no d_maxsize\n",
133 dp->d_name, dp->d_unit);
134 dp->d_maxsize = DFLTPHYS;
135 }
136 } else if ((pp->acr + pp->acw + pp->ace) > 0 && (r + w + e) == 0) {
137 if (dp->d_close != NULL) {
138 g_disk_lock_giant(dp);
139 error = dp->d_close(dp);
140 if (error != 0)
141 printf("Closed disk %s -> %d\n",
142 pp->name, error);
143 g_disk_unlock_giant(dp);
144 }
145 dp->d_flags &= ~DISKFLAG_OPEN;
146 }
147 return (error);
148}
149
150static void
151g_disk_kerneldump(struct bio *bp, struct disk *dp)
152{
153 int error;
154 struct g_kerneldump *gkd;
155 struct dumperinfo di;
156 struct g_geom *gp;
157
158 gkd = (struct g_kerneldump*)bp->bio_data;
159 gp = bp->bio_to->geom;
160 g_trace(G_T_TOPOLOGY, "g_disk_kernedump(%s, %jd, %jd)",
161 gp->name, (intmax_t)gkd->offset, (intmax_t)gkd->length);
162 if (dp->d_dump == NULL) {
163 g_io_deliver(bp, ENODEV);
164 return;
165 }
166 di.dumper = dp->d_dump;
167 di.priv = dp;
168 di.blocksize = dp->d_sectorsize;
169 di.mediaoffset = gkd->offset;
170 di.mediasize = gkd->length;
171 error = set_dumper(&di);
172 g_io_deliver(bp, error);
173}
174
175static void
176g_disk_done(struct bio *bp)
177{
178 struct bio *bp2;
179 struct disk *dp;
180
181 /* See "notes" for why we need a mutex here */
182 /* XXX: will witness accept a mix of Giant/unGiant drivers here ? */
183 mtx_lock(&g_disk_done_mtx);
184 bp->bio_completed = bp->bio_length - bp->bio_resid;
185
186 bp2 = bp->bio_parent;
187 dp = bp2->bio_to->geom->softc;
188 if (bp2->bio_error == 0)
189 bp2->bio_error = bp->bio_error;
190 bp2->bio_completed += bp->bio_completed;
191 g_destroy_bio(bp);
192 bp2->bio_inbed++;
193 if (bp2->bio_children == bp2->bio_inbed) {
194 bp2->bio_resid = bp2->bio_bcount - bp2->bio_completed;
195 devstat_end_transaction_bio(dp->d_devstat, bp2);
196 g_io_deliver(bp2, bp2->bio_error);
197 }
198 mtx_unlock(&g_disk_done_mtx);
199}
200
201static int
202g_disk_ioctl(struct g_provider *pp, u_long cmd, void * data, struct thread *td)
203{
204 struct g_geom *gp;
205 struct disk *dp;
206 int error;
207
208 gp = pp->geom;
209 dp = gp->softc;
210
211 if (dp->d_ioctl == NULL)
212 return (ENOIOCTL);
213 g_disk_lock_giant(dp);
214 error = dp->d_ioctl(dp, cmd, data, 0, td);
215 g_disk_unlock_giant(dp);
216 return(error);
217}
218
219static void
220g_disk_start(struct bio *bp)
221{
222 struct bio *bp2, *bp3;
223 struct disk *dp;
224 int error;
225 off_t off;
226
227 dp = bp->bio_to->geom->softc;
228 if (dp == NULL)
229 g_io_deliver(bp, ENXIO);
230 error = EJUSTRETURN;
231 switch(bp->bio_cmd) {
232 case BIO_DELETE:
233 if (!(dp->d_flags & DISKFLAG_CANDELETE)) {
234 error = 0;
235 break;
236 }
237 /* fall-through */
238 case BIO_READ:
239 case BIO_WRITE:
240 off = 0;
241 bp3 = NULL;
242 bp2 = g_clone_bio(bp);
243 if (bp2 == NULL) {
244 error = ENOMEM;
245 break;
246 }
247 devstat_start_transaction_bio(dp->d_devstat, bp);
248 do {
249 bp2->bio_offset += off;
250 bp2->bio_length -= off;
251 bp2->bio_data += off;
252 if (bp2->bio_length > dp->d_maxsize) {
253 /*
254 * XXX: If we have a stripesize we should really
255 * use it here.
256 */
257 bp2->bio_length = dp->d_maxsize;
258 off += dp->d_maxsize;
259 /*
260 * To avoid a race, we need to grab the next bio
261 * before we schedule this one. See "notes".
262 */
263 bp3 = g_clone_bio(bp);
264 if (bp3 == NULL)
265 bp->bio_error = ENOMEM;
266 }
267 bp2->bio_done = g_disk_done;
268 bp2->bio_pblkno = bp2->bio_offset / dp->d_sectorsize;
269 bp2->bio_bcount = bp2->bio_length;
270 bp2->bio_disk = dp;
271 g_disk_lock_giant(dp);
272 dp->d_strategy(bp2);
273 g_disk_unlock_giant(dp);
274 bp2 = bp3;
275 bp3 = NULL;
276 } while (bp2 != NULL);
277 break;
278 case BIO_GETATTR:
279 if (g_handleattr_int(bp, "GEOM::fwsectors", dp->d_fwsectors))
280 break;
281 else if (g_handleattr_int(bp, "GEOM::fwheads", dp->d_fwheads))
282 break;
283 else if (g_handleattr_off_t(bp, "GEOM::frontstuff", 0))
284 break;
285 else if (!strcmp(bp->bio_attribute, "GEOM::kerneldump"))
286 g_disk_kerneldump(bp, dp);
287 else
288 error = ENOIOCTL;
289 break;
290 default:
291 error = EOPNOTSUPP;
292 break;
293 }
294 if (error != EJUSTRETURN)
295 g_io_deliver(bp, error);
296 return;
297}
298
299static void
300g_disk_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp)
301{
302 struct disk *dp;
303
304 dp = gp->softc;
305 if (dp == NULL)
306 return;
307 if (indent == NULL) {
308 sbuf_printf(sb, " hd %u", dp->d_fwheads);
309 sbuf_printf(sb, " sc %u", dp->d_fwsectors);
310 return;
311 }
312 if (pp != NULL) {
313 sbuf_printf(sb, "%s<fwheads>%u</fwheads>\n",
314 indent, dp->d_fwheads);
315 sbuf_printf(sb, "%s<fwsectors>%u</fwsectors>\n",
316 indent, dp->d_fwsectors);
317 }
318}
319
320static void
321g_disk_create(void *arg, int flag)
322{
323 struct g_geom *gp;
324 struct g_provider *pp;
325 struct disk *dp;
326
327 if (flag == EV_CANCEL)
328 return;
329 g_topology_assert();
330 dp = arg;
331 gp = g_new_geomf(&g_disk_class, "%s%d", dp->d_name, dp->d_unit);
332 gp->start = g_disk_start;
333 gp->access = g_disk_access;
334 gp->ioctl = g_disk_ioctl;
335 gp->softc = dp;
336 gp->dumpconf = g_disk_dumpconf;
337 pp = g_new_providerf(gp, "%s", gp->name);
338 pp->mediasize = dp->d_mediasize;
339 pp->sectorsize = dp->d_sectorsize;
340 if (dp->d_flags & DISKFLAG_CANDELETE)
341 pp->flags |= G_PF_CANDELETE;
342 pp->stripeoffset = dp->d_stripeoffset;
343 pp->stripesize = dp->d_stripesize;
344 if (bootverbose)
345 printf("GEOM: new disk %s\n", gp->name);
346 dp->d_geom = gp;
347 g_error_provider(pp, 0);
348}
349
350static void
351g_disk_destroy(void *ptr, int flag)
352{
353 struct g_geom *gp;
354
355 g_topology_assert();
356 gp = ptr;
357 gp->softc = NULL;
358 g_wither_geom(gp, ENXIO);
359}
360
361void
362disk_create(int unit, struct disk *dp, int flags, void *unused __unused, void * unused2 __unused)
363{
364
365 dp->d_unit = unit;
366 dp->d_flags = flags;
367 KASSERT(dp->d_strategy != NULL, ("disk_create need d_strategy"));
368 KASSERT(dp->d_name != NULL, ("disk_create need d_name"));
369 KASSERT(*dp->d_name != 0, ("disk_create need d_name"));
370 KASSERT(strlen(dp->d_name) < SPECNAMELEN - 4, ("disk name too long"));
371 if (dp->d_devstat == NULL)
372 dp->d_devstat = devstat_new_entry(dp->d_name, dp->d_unit,
373 dp->d_sectorsize, DEVSTAT_ALL_SUPPORTED,
374 DEVSTAT_TYPE_DIRECT, DEVSTAT_PRIORITY_MAX);
375 dp->d_geom = NULL;
376 g_post_event(g_disk_create, dp, M_WAITOK, dp, NULL);
377}
378
379void
380disk_destroy(struct disk *dp)
381{
382 struct g_geom *gp;
383
384 g_cancel_event(dp);
385 gp = dp->d_geom;
386 if (gp == NULL)
387 return;
388 gp->softc = NULL;
389 devstat_remove_entry(dp->d_devstat);
390 g_waitfor_event(g_disk_destroy, gp, M_WAITOK, NULL, NULL);
391 dp->d_geom = NULL;
392}
393
394static void
395g_kern_disks(void *p, int flag __unused)
396{
397 struct sbuf *sb;
398 struct g_geom *gp;
399 char *sp;
400
401 sb = p;
402 sp = "";
403 g_topology_assert();
404 LIST_FOREACH(gp, &g_disk_class.geom, geom) {
405 sbuf_printf(sb, "%s%s", sp, gp->name);
406 sp = " ";
407 }
408 sbuf_finish(sb);
409}
410
411static int
412sysctl_disks(SYSCTL_HANDLER_ARGS)
413{
414 int error;
415 struct sbuf *sb;
416
417 sb = sbuf_new(NULL, NULL, 0, SBUF_AUTOEXTEND);
418 sbuf_clear(sb);
419 g_waitfor_event(g_kern_disks, sb, M_WAITOK, NULL);
420 error = SYSCTL_OUT(req, sbuf_data(sb), sbuf_len(sb) + 1);
421 sbuf_delete(sb);
422 return error;
423}
424
425SYSCTL_PROC(_kern, OID_AUTO, disks, CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_NOLOCK, 0, 0,
426 sysctl_disks, "A", "names of available disks");
427