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("$FreeBSD$"); 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 59 static struct mtx g_disk_done_mtx; 60 61 static g_access_t g_disk_access; 62 static g_init_t g_disk_init; 63 static g_fini_t g_disk_fini; 64 static g_start_t g_disk_start; 65 static g_ioctl_t g_disk_ioctl; 66 static g_dumpconf_t g_disk_dumpconf; 67 68 static struct g_class g_disk_class = { 69 .name = "DISK", 70 .version = G_VERSION, 71 .init = g_disk_init, 72 .fini = g_disk_fini, 73 .start = g_disk_start, 74 .access = g_disk_access, 75 .ioctl = g_disk_ioctl, 76 .dumpconf = g_disk_dumpconf, 77 }; 78 79 static void 80 g_disk_init(struct g_class *mp __unused) 81 { 82 83 mtx_init(&g_disk_done_mtx, "g_disk_done", NULL, MTX_DEF); 84 } 85 86 static void 87 g_disk_fini(struct g_class *mp __unused) 88 { 89 90 mtx_destroy(&g_disk_done_mtx); 91 } 92 93 DECLARE_GEOM_CLASS(g_disk_class, g_disk); 94 95 static void __inline 96 g_disk_lock_giant(struct disk *dp) 97 { 98 if (dp->d_flags & DISKFLAG_NEEDSGIANT) 99 mtx_lock(&Giant); 100 } 101 102 static void __inline 103 g_disk_unlock_giant(struct disk *dp) 104 { 105 if (dp->d_flags & DISKFLAG_NEEDSGIANT) 106 mtx_unlock(&Giant); 107 } 108 109 static int 110 g_disk_access(struct g_provider *pp, int r, int w, int e) 111 { 112 struct disk *dp; 113 int error; 114 115 g_trace(G_T_ACCESS, "g_disk_access(%s, %d, %d, %d)", 116 pp->name, r, w, e); 117 g_topology_assert(); 118 dp = pp->geom->softc; 119 if (dp == NULL || dp->d_destroyed) { 120 /* 121 * Allow decreasing access count even if disk is not 122 * avaliable anymore. 123 */ 124 if (r <= 0 && w <= 0 && e <= 0) 125 return (0); 126 return (ENXIO); 127 } 128 r += pp->acr; 129 w += pp->acw; 130 e += pp->ace; 131 error = 0; 132 if ((pp->acr + pp->acw + pp->ace) == 0 && (r + w + e) > 0) { 133 if (dp->d_open != NULL) { 134 g_disk_lock_giant(dp); 135 error = dp->d_open(dp); 136 if (error != 0) 137 printf("Opened disk %s -> %d\n", 138 pp->name, error); 139 g_disk_unlock_giant(dp); 140 } 141 pp->mediasize = dp->d_mediasize; 142 pp->sectorsize = dp->d_sectorsize; 143 dp->d_flags |= DISKFLAG_OPEN; 144 if (dp->d_maxsize == 0) { 145 printf("WARNING: Disk drive %s%d has no d_maxsize\n", 146 dp->d_name, dp->d_unit); 147 dp->d_maxsize = DFLTPHYS; 148 } 149 } else if ((pp->acr + pp->acw + pp->ace) > 0 && (r + w + e) == 0) { 150 if (dp->d_close != NULL) { 151 g_disk_lock_giant(dp); 152 error = dp->d_close(dp); 153 if (error != 0) 154 printf("Closed disk %s -> %d\n", 155 pp->name, error); 156 g_disk_unlock_giant(dp); 157 } 158 dp->d_flags &= ~DISKFLAG_OPEN; 159 } 160 return (error); 161 } 162 163 static void 164 g_disk_kerneldump(struct bio *bp, struct disk *dp) 165 { 166 int error; 167 struct g_kerneldump *gkd; 168 struct dumperinfo di; 169 struct g_geom *gp; 170 171 gkd = (struct g_kerneldump*)bp->bio_data; 172 gp = bp->bio_to->geom; 173 g_trace(G_T_TOPOLOGY, "g_disk_kernedump(%s, %jd, %jd)", 174 gp->name, (intmax_t)gkd->offset, (intmax_t)gkd->length); 175 if (dp->d_dump == NULL) { 176 g_io_deliver(bp, ENODEV); 177 return; 178 } 179 di.dumper = dp->d_dump; 180 di.priv = dp; 181 di.blocksize = dp->d_sectorsize; 182 di.mediaoffset = gkd->offset; 183 if ((gkd->offset + gkd->length) > dp->d_mediasize) 184 gkd->length = dp->d_mediasize - gkd->offset; 185 di.mediasize = gkd->length; 186 error = set_dumper(&di); 187 g_io_deliver(bp, error); 188 } 189 190 static void 191 g_disk_done(struct bio *bp) 192 { 193 struct bio *bp2; 194 struct disk *dp; 195 196 /* See "notes" for why we need a mutex here */ 197 /* XXX: will witness accept a mix of Giant/unGiant drivers here ? */ 198 mtx_lock(&g_disk_done_mtx); 199 bp->bio_completed = bp->bio_length - bp->bio_resid; 200 201 bp2 = bp->bio_parent; 202 if (bp2->bio_error == 0) 203 bp2->bio_error = bp->bio_error; 204 bp2->bio_completed += bp->bio_completed; 205 g_destroy_bio(bp); 206 bp2->bio_inbed++; 207 if (bp2->bio_children == bp2->bio_inbed) { 208 bp2->bio_resid = bp2->bio_bcount - bp2->bio_completed; 209 if ((dp = bp2->bio_to->geom->softc)) 210 devstat_end_transaction_bio(dp->d_devstat, bp2); 211 g_io_deliver(bp2, bp2->bio_error); 212 } 213 mtx_unlock(&g_disk_done_mtx); 214 } 215 216 static int 217 g_disk_ioctl(struct g_provider *pp, u_long cmd, void * data, int fflag, struct thread *td) 218 { 219 struct g_geom *gp; 220 struct disk *dp; 221 int error; 222 223 gp = pp->geom; 224 dp = gp->softc; 225 226 if (dp->d_ioctl == NULL) 227 return (ENOIOCTL); 228 g_disk_lock_giant(dp); 229 error = dp->d_ioctl(dp, cmd, data, fflag, td); 230 g_disk_unlock_giant(dp); 231 return(error); 232 } 233 234 static void 235 g_disk_start(struct bio *bp) 236 { 237 struct bio *bp2, *bp3; 238 struct disk *dp; 239 int error; 240 off_t off; 241 242 dp = bp->bio_to->geom->softc; 243 if (dp == NULL || dp->d_destroyed) 244 g_io_deliver(bp, ENXIO); 245 error = EJUSTRETURN; 246 switch(bp->bio_cmd) { 247 case BIO_DELETE: 248 if (!(dp->d_flags & DISKFLAG_CANDELETE)) { 249 error = 0; 250 break; 251 } 252 /* fall-through */ 253 case BIO_READ: 254 case BIO_WRITE: 255 off = 0; 256 bp3 = NULL; 257 bp2 = g_clone_bio(bp); 258 if (bp2 == NULL) { 259 error = ENOMEM; 260 break; 261 } 262 devstat_start_transaction_bio(dp->d_devstat, bp); 263 do { 264 bp2->bio_offset += off; 265 bp2->bio_length -= off; 266 bp2->bio_data += off; 267 if (bp2->bio_length > dp->d_maxsize) { 268 /* 269 * XXX: If we have a stripesize we should really 270 * use it here. 271 */ 272 bp2->bio_length = dp->d_maxsize; 273 off += dp->d_maxsize; 274 /* 275 * To avoid a race, we need to grab the next bio 276 * before we schedule this one. See "notes". 277 */ 278 bp3 = g_clone_bio(bp); 279 if (bp3 == NULL) 280 bp->bio_error = ENOMEM; 281 } 282 bp2->bio_done = g_disk_done; 283 bp2->bio_pblkno = bp2->bio_offset / dp->d_sectorsize; 284 bp2->bio_bcount = bp2->bio_length; 285 bp2->bio_disk = dp; 286 g_disk_lock_giant(dp); 287 dp->d_strategy(bp2); 288 g_disk_unlock_giant(dp); 289 bp2 = bp3; 290 bp3 = NULL; 291 } while (bp2 != NULL); 292 break; 293 case BIO_GETATTR: 294 if (g_handleattr_int(bp, "GEOM::fwsectors", dp->d_fwsectors)) 295 break; 296 else if (g_handleattr_int(bp, "GEOM::fwheads", dp->d_fwheads)) 297 break; 298 else if (g_handleattr_off_t(bp, "GEOM::frontstuff", 0)) 299 break; 300 else if (!strcmp(bp->bio_attribute, "GEOM::kerneldump")) 301 g_disk_kerneldump(bp, dp); 302 else 303 error = ENOIOCTL; 304 break; 305 default: 306 error = EOPNOTSUPP; 307 break; 308 } 309 if (error != EJUSTRETURN) 310 g_io_deliver(bp, error); 311 return; 312 } 313 314 static void 315 g_disk_dumpconf(struct sbuf *sb, const char *indent, struct g_geom *gp, struct g_consumer *cp, struct g_provider *pp) 316 { 317 struct disk *dp; 318 319 dp = gp->softc; 320 if (dp == NULL) 321 return; 322 if (indent == NULL) { 323 sbuf_printf(sb, " hd %u", dp->d_fwheads); 324 sbuf_printf(sb, " sc %u", dp->d_fwsectors); 325 return; 326 } 327 if (pp != NULL) { 328 sbuf_printf(sb, "%s<fwheads>%u</fwheads>\n", 329 indent, dp->d_fwheads); 330 sbuf_printf(sb, "%s<fwsectors>%u</fwsectors>\n", 331 indent, dp->d_fwsectors); 332 } 333 } 334 335 static void 336 g_disk_create(void *arg, int flag) 337 { 338 struct g_geom *gp; 339 struct g_provider *pp; 340 struct disk *dp; 341 342 if (flag == EV_CANCEL) 343 return; 344 g_topology_assert(); 345 dp = arg; 346 gp = g_new_geomf(&g_disk_class, "%s%d", dp->d_name, dp->d_unit); 347 gp->softc = dp; 348 pp = g_new_providerf(gp, "%s", gp->name); 349 pp->mediasize = dp->d_mediasize; 350 pp->sectorsize = dp->d_sectorsize; 351 if (dp->d_flags & DISKFLAG_CANDELETE) 352 pp->flags |= G_PF_CANDELETE; 353 pp->stripeoffset = dp->d_stripeoffset; 354 pp->stripesize = dp->d_stripesize; 355 if (bootverbose) 356 printf("GEOM: new disk %s\n", gp->name); 357 dp->d_geom = gp; 358 g_error_provider(pp, 0); 359 } 360 361 static void 362 g_disk_destroy(void *ptr, int flag) 363 { 364 struct disk *dp; 365 struct g_geom *gp; 366 367 g_topology_assert(); 368 dp = ptr; 369 gp = dp->d_geom; 370 if (gp != NULL) { 371 gp->softc = NULL; 372 g_wither_geom(gp, ENXIO); 373 } 374 g_free(dp); 375 } 376 377 struct disk * 378 disk_alloc() 379 { 380 struct disk *dp; 381 382 dp = g_malloc(sizeof *dp, M_WAITOK | M_ZERO); 383 return (dp); 384 } 385 386 void 387 disk_create(struct disk *dp, int version) 388 { 389 if (version != DISK_VERSION_00) { 390 printf("WARNING: Attempt to add disk %s%d %s", 391 dp->d_name, dp->d_unit, 392 " using incompatible ABI version of disk(9)\n"); 393 printf("WARNING: Ignoring disk %s%d\n", 394 dp->d_name, dp->d_unit); 395 return; 396 } 397 KASSERT(dp->d_strategy != NULL, ("disk_create need d_strategy")); 398 KASSERT(dp->d_name != NULL, ("disk_create need d_name")); 399 KASSERT(*dp->d_name != 0, ("disk_create need d_name")); 400 KASSERT(strlen(dp->d_name) < SPECNAMELEN - 4, ("disk name too long")); 401 if (dp->d_devstat == NULL) 402 dp->d_devstat = devstat_new_entry(dp->d_name, dp->d_unit, 403 dp->d_sectorsize, DEVSTAT_ALL_SUPPORTED, 404 DEVSTAT_TYPE_DIRECT, DEVSTAT_PRIORITY_MAX); 405 dp->d_geom = NULL; 406 g_post_event(g_disk_create, dp, M_WAITOK, dp, NULL); 407 } 408 409 void 410 disk_destroy(struct disk *dp) 411 { 412 413 g_cancel_event(dp); 414 dp->d_destroyed = 1; 415 if (dp->d_devstat != NULL) 416 devstat_remove_entry(dp->d_devstat); 417 g_post_event(g_disk_destroy, dp, M_WAITOK, NULL); 418 } 419 420 static void 421 g_kern_disks(void *p, int flag __unused) 422 { 423 struct sbuf *sb; 424 struct g_geom *gp; 425 char *sp; 426 427 sb = p; 428 sp = ""; 429 g_topology_assert(); 430 LIST_FOREACH(gp, &g_disk_class.geom, geom) { 431 sbuf_printf(sb, "%s%s", sp, gp->name); 432 sp = " "; 433 } 434 sbuf_finish(sb); 435 } 436 437 static int 438 sysctl_disks(SYSCTL_HANDLER_ARGS) 439 { 440 int error; 441 struct sbuf *sb; 442 443 sb = sbuf_new(NULL, NULL, 0, SBUF_AUTOEXTEND); 444 g_waitfor_event(g_kern_disks, sb, M_WAITOK, NULL); 445 error = SYSCTL_OUT(req, sbuf_data(sb), sbuf_len(sb) + 1); 446 sbuf_delete(sb); 447 return error; 448 } 449 450 SYSCTL_PROC(_kern, OID_AUTO, disks, CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_NOLOCK, 0, 0, 451 sysctl_disks, "A", "names of available disks"); 452 453