sys/dev/disk/dm/device-mapper.c

   1 /*        $NetBSD: device-mapper.c,v 1.22 2010/03/26 15:46:04 jakllsch Exp $ */
   2
   3 /*
   4  * Copyright (c) 2010-2011 Alex Hornung <alex@alexhornung.com>
   5  * Copyright (c) 2010 The NetBSD Foundation, Inc.
   6  * All rights reserved.
   7  *
   8  * This code is derived from software contributed to The NetBSD Foundation
   9  * by Adam Hamsik.
  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  *
  20  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
  21  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
  22  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
  23  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
  24  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
  25  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
  26  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
  27  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
  28  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
  29  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  30  * POSSIBILITY OF SUCH DAMAGE.
  31  */
  32
  33 /*
  34  * I want to say thank you to all people who helped me with this project.
  35  */
  36
  37 #include <sys/types.h>
  38 #include <sys/ctype.h>
  39
  40 #include <sys/buf.h>
  41 #include <sys/conf.h>
  42 #include <sys/device.h>
  43 #include <sys/disk.h>
  44 #include <sys/disklabel.h>
  45 #include <sys/dtype.h>
  46 #include <sys/malloc.h>
  47 #include <sys/module.h>
  48 #include <sys/sysctl.h>
  49 #include <dev/disk/dm/dm.h>
  50
  51 #include "netbsd-dm.h"
  52
  53 static  d_ioctl_t       dmioctl;
  54 static  d_open_t        dmopen;
  55 static  d_close_t       dmclose;
  56 static  d_psize_t       dmsize;
  57 static  d_strategy_t    dmstrategy;
  58 static  d_dump_t        dmdump;
  59
  60 /* New module handle and destroy routines */
  61 static int dm_modcmd(module_t mod, int cmd, void *unused);
  62 static int dmdestroy(void);
  63
  64 static void dm_doinit(void);
  65
  66 static int dm_cmd_to_fun(prop_dictionary_t);
  67 static int disk_ioctl_switch(cdev_t, u_long, void *);
  68 static int dm_ioctl_switch(u_long);
  69 #if 0
  70 static void dmminphys(struct buf *);
  71 #endif
  72
  73 /* ***Variable-definitions*** */
  74 struct dev_ops dm_ops = {
  75         { "dm", 0, D_DISK | D_MPSAFE },
  76         .d_open         = dmopen,
  77         .d_close        = dmclose,
  78         .d_read         = physread,
  79         .d_write        = physwrite,
  80         .d_ioctl        = dmioctl,
  81         .d_strategy     = dmstrategy,
  82         .d_psize        = dmsize,
  83         .d_dump         = dmdump,
  84 /* D_DISK */
  85 };
  86
  87 MALLOC_DEFINE(M_DM, "dm", "Device Mapper allocations");
  88
  89 int dm_debug_level = 0;
  90
  91 extern uint64_t dm_dev_counter;
  92
  93 static cdev_t dmcdev;
  94
  95 static moduledata_t dm_mod = {
  96     "dm",
  97     dm_modcmd,
  98     NULL
  99 };
 100 DECLARE_MODULE(dm, dm_mod, SI_SUB_RAID, SI_ORDER_ANY);
 101 MODULE_VERSION(dm, 1);
 102
 103 /*
 104  * This array is used to translate cmd to function pointer.
 105  *
 106  * Interface between libdevmapper and lvm2tools uses different
 107  * names for one IOCTL call because libdevmapper do another thing
 108  * then. When I run "info" or "mknodes" libdevmapper will send same
 109  * ioctl to kernel but will do another things in userspace.
 110  *
 111  */
 112 static struct cmd_function cmd_fn[] = {
 113                 { .cmd = "version", .fn = dm_get_version_ioctl},
 114                 { .cmd = "targets", .fn = dm_list_versions_ioctl},
 115                 { .cmd = "create",  .fn = dm_dev_create_ioctl},
 116                 { .cmd = "info",    .fn = dm_dev_status_ioctl},
 117                 { .cmd = "mknodes", .fn = dm_dev_status_ioctl},
 118                 { .cmd = "names",   .fn = dm_dev_list_ioctl},
 119                 { .cmd = "suspend", .fn = dm_dev_suspend_ioctl},
 120                 { .cmd = "remove",  .fn = dm_dev_remove_ioctl},
 121                 { .cmd = "remove_all", .fn = dm_dev_remove_all_ioctl},
 122                 { .cmd = "rename",  .fn = dm_dev_rename_ioctl},
 123                 { .cmd = "resume",  .fn = dm_dev_resume_ioctl},
 124                 { .cmd = "clear",   .fn = dm_table_clear_ioctl},
 125                 { .cmd = "deps",    .fn = dm_table_deps_ioctl},
 126                 { .cmd = "reload",  .fn = dm_table_load_ioctl},
 127                 { .cmd = "status",  .fn = dm_table_status_ioctl},
 128                 { .cmd = "table",   .fn = dm_table_status_ioctl},
 129                 { .cmd = "message", .fn = dm_message_ioctl},
 130                 {NULL, NULL}
 131 };
 132
 133 /* New module handle routine */
 134 static int
 135 dm_modcmd(module_t mod, int cmd, void *unused)
 136 {
 137         int error;
 138
 139         error = 0;
 140
 141         switch (cmd) {
 142         case MOD_LOAD:
 143                 dm_doinit();
 144                 kprintf("Device Mapper version %d.%d.%d loaded\n",
 145                     DM_VERSION_MAJOR, DM_VERSION_MINOR, DM_VERSION_PATCHLEVEL);
 146                 break;
 147
 148         case MOD_UNLOAD:
 149                 /*
 150                  * Disable unloading of dm module if there are any devices
 151                  * defined in driver. This is probably too strong we need
 152                  * to disable auto-unload only if there is mounted dm device
 153                  * present.
 154                  */
 155                 if (dm_dev_counter > 0)
 156                         return EBUSY;
 157
 158                 error = dmdestroy();
 159                 if (error)
 160                         break;
 161                 kprintf("Device Mapper unloaded\n");
 162                 break;
 163
 164         default:
 165                 break;
 166         }
 167
 168         return error;
 169 }
 170
 171 static void
 172 dm_doinit(void)
 173 {
 174         dm_target_init();
 175         dm_dev_init();
 176         dm_pdev_init();
 177         dmcdev = make_dev(&dm_ops, 0, UID_ROOT, GID_OPERATOR, 0640, "mapper/control");
 178 }
 179
 180 /* Destroy routine */
 181 static int
 182 dmdestroy(void)
 183 {
 184         destroy_dev(dmcdev);
 185
 186         dm_dev_uninit();
 187         dm_pdev_uninit();
 188         dm_target_uninit();
 189
 190         return 0;
 191 }
 192
 193 static int
 194 dmopen(struct dev_open_args *ap)
 195 {
 196         cdev_t dev = ap->a_head.a_dev;
 197         dm_dev_t *dmv;
 198
 199         /* Shortcut for the control device */
 200         if (minor(dev) == 0)
 201                 return 0;
 202
 203         if ((dmv = dm_dev_lookup(NULL, NULL, minor(dev))) == NULL)
 204                 return ENXIO;
 205
 206         dmv->is_open = 1;
 207         dm_dev_unbusy(dmv);
 208
 209         aprint_debug("dm open routine called %" PRIu32 "\n",
 210             minor(ap->a_head.a_dev));
 211         return 0;
 212 }
 213
 214 static int
 215 dmclose(struct dev_close_args *ap)
 216 {
 217         cdev_t dev = ap->a_head.a_dev;
 218         dm_dev_t *dmv;
 219
 220         /* Shortcut for the control device */
 221         if (minor(dev) == 0)
 222                 return 0;
 223
 224         if ((dmv = dm_dev_lookup(NULL, NULL, minor(dev))) == NULL)
 225                 return ENXIO;
 226
 227         dmv->is_open = 0;
 228         dm_dev_unbusy(dmv);
 229
 230         aprint_debug("dm close routine called %" PRIu32 "\n",
 231             minor(ap->a_head.a_dev));
 232         return 0;
 233 }
 234
 235
 236 static int
 237 dmioctl(struct dev_ioctl_args *ap)
 238 {
 239         cdev_t dev = ap->a_head.a_dev;
 240         u_long cmd = ap->a_cmd;
 241         void *data = ap->a_data;
 242
 243         int r, err;
 244         prop_dictionary_t dm_dict_in;
 245
 246         err = r = 0;
 247
 248         aprint_debug("dmioctl called\n");
 249
 250         KKASSERT(data != NULL);
 251
 252         if ((r = disk_ioctl_switch(dev, cmd, data)) == ENOTTY) {
 253                 struct plistref *pref = (struct plistref *) data;
 254
 255                 /* Check if we were called with NETBSD_DM_IOCTL ioctl
 256                    otherwise quit. */
 257                 if ((r = dm_ioctl_switch(cmd)) != 0)
 258                         return r;
 259
 260                 if ((r = prop_dictionary_copyin_ioctl(pref, cmd, &dm_dict_in)) != 0)
 261                         return r;
 262
 263                 if ((r = dm_check_version(dm_dict_in)) != 0)
 264                         goto cleanup_exit;
 265
 266                 /* run ioctl routine */
 267                 if ((err = dm_cmd_to_fun(dm_dict_in)) != 0)
 268                         goto cleanup_exit;
 269
 270 cleanup_exit:
 271                 r = prop_dictionary_copyout_ioctl(pref, cmd, dm_dict_in);
 272                 prop_object_release(dm_dict_in);
 273         }
 274
 275         /*
 276          * Return the error of the actual command if one one has
 277          * happened. Otherwise return 'r' which indicates errors
 278          * that occurred during helper operations.
 279          */
 280         return (err != 0)?err:r;
 281 }
 282
 283 /*
 284  * Translate command sent from libdevmapper to func.
 285  */
 286 static int
 287 dm_cmd_to_fun(prop_dictionary_t dm_dict)
 288 {
 289         int i, r;
 290         prop_string_t command;
 291
 292         r = 0;
 293
 294         if ((command = prop_dictionary_get(dm_dict, DM_IOCTL_COMMAND)) == NULL)
 295                 return EINVAL;
 296
 297         for (i = 0; cmd_fn[i].cmd != NULL; i++)
 298                 if (prop_string_equals_cstring(command, cmd_fn[i].cmd))
 299                         break;
 300
 301         if (cmd_fn[i].cmd == NULL)
 302                 return EINVAL;
 303
 304         aprint_debug("ioctl %s called\n", cmd_fn[i].cmd);
 305         r = cmd_fn[i].fn(dm_dict);
 306
 307         return r;
 308 }
 309
 310 /* Call apropriate ioctl handler function. */
 311 static int
 312 dm_ioctl_switch(u_long cmd)
 313 {
 314
 315         switch(cmd) {
 316
 317         case NETBSD_DM_IOCTL:
 318                 aprint_debug("dm NETBSD_DM_IOCTL called\n");
 319                 break;
 320         default:
 321                 aprint_debug("dm unknown ioctl called\n");
 322                 return ENOTTY;
 323                 break; /* NOT REACHED */
 324         }
 325
 326         return 0;
 327 }
 328
 329  /*
 330   * Check for disk specific ioctls.
 331   */
 332
 333 static int
 334 disk_ioctl_switch(cdev_t dev, u_long cmd, void *data)
 335 {
 336         dm_dev_t *dmv;
 337
 338         /* disk ioctls make sense only on block devices */
 339         if (minor(dev) == 0)
 340                 return ENOTTY;
 341
 342         switch(cmd) {
 343         case DIOCGPART:
 344         {
 345                 struct partinfo *dpart;
 346                 u_int64_t size;
 347                 dpart = data;
 348                 bzero(dpart, sizeof(*dpart));
 349
 350                 if ((dmv = dev->si_drv1) == NULL)
 351                         return ENODEV;
 352                 if (dmv->diskp->d_info.d_media_blksize == 0) {
 353                         return ENOTSUP;
 354                 } else {
 355                         size = dm_table_size(&dmv->table_head);
 356                         dpart->media_offset  = 0;
 357                         dpart->media_size    = size * DEV_BSIZE;
 358                         dpart->media_blocks  = size;
 359                         dpart->media_blksize = DEV_BSIZE;
 360                         dpart->fstype = FS_BSDFFS;
 361                 }
 362                 break;
 363         }
 364
 365         default:
 366                 aprint_debug("unknown disk_ioctl called\n");
 367                 return ENOTTY;
 368                 break; /* NOT REACHED */
 369         }
 370
 371         return 0;
 372 }
 373
 374 /*
 375  * Do all IO operations on dm logical devices.
 376  */
 377 static int
 378 dmstrategy(struct dev_strategy_args *ap)
 379 {
 380         cdev_t dev = ap->a_head.a_dev;
 381         struct bio *bio = ap->a_bio;
 382         struct buf *bp = bio->bio_buf;
 383         int bypass;
 384
 385         dm_dev_t *dmv;
 386         dm_table_t  *tbl;
 387         dm_table_entry_t *table_en;
 388         struct buf *nestbuf;
 389
 390         uint64_t buf_start, buf_len, issued_len;
 391         uint64_t table_start, table_end;
 392         uint64_t start, end;
 393
 394         buf_start = bio->bio_offset;
 395         buf_len = bp->b_bcount;
 396
 397         tbl = NULL;
 398
 399         table_end = 0;
 400         issued_len = 0;
 401
 402         dmv = dev->si_drv1;
 403
 404         switch(bp->b_cmd) {
 405         case BUF_CMD_READ:
 406         case BUF_CMD_WRITE:
 407         case BUF_CMD_FREEBLKS:
 408                 bypass = 0;
 409                 break;
 410         case BUF_CMD_FLUSH:
 411                 bypass = 1;
 412                 KKASSERT(buf_len == 0);
 413                 break;
 414         default:
 415                 bp->b_error = EIO;
 416                 bp->b_resid = bp->b_bcount;
 417                 biodone(bio);
 418                 return 0;
 419         }
 420
 421         if (bypass == 0 &&
 422             bounds_check_with_mediasize(bio, DEV_BSIZE,
 423                                         dm_table_size(&dmv->table_head)) <= 0) {
 424                 bp->b_resid = bp->b_bcount;
 425                 biodone(bio);
 426                 return 0;
 427         }
 428
 429         /* Select active table */
 430         tbl = dm_table_get_entry(&dmv->table_head, DM_TABLE_ACTIVE);
 431
 432         nestiobuf_init(bio);
 433         devstat_start_transaction(&dmv->stats);
 434
 435         /*
 436          * Find out what tables I want to select.
 437          */
 438         SLIST_FOREACH(table_en, tbl, next) {
 439                 /*
 440                  * I need need number of bytes not blocks.
 441                  */
 442                 table_start = table_en->start * DEV_BSIZE;
 443                 table_end = table_start + (table_en->length) * DEV_BSIZE;
 444
 445                 /*
 446                  * Calculate the start and end
 447                  */
 448                 start = MAX(table_start, buf_start);
 449                 end = MIN(table_end, buf_start + buf_len);
 450
 451                 if (dm_debug_level) {
 452                         aprint_normal("----------------------------------------\n");
 453                         aprint_normal("table_start %010" PRIu64", table_end %010"
 454                             PRIu64 "\n", table_start, table_end);
 455                         aprint_normal("buf_start %010" PRIu64", buf_len %010"
 456                             PRIu64"\n", buf_start, buf_len);
 457                         aprint_normal("start-buf_start %010"PRIu64", end %010"
 458                             PRIu64"\n", start - buf_start, end);
 459                         aprint_normal("start %010" PRIu64", end %010"
 460                             PRIu64"\n", start, end);
 461                         aprint_normal("----------------------------------------\n");
 462                 }
 463
 464                 if (bypass) {
 465                         nestbuf = getpbuf(NULL);
 466                         nestbuf->b_flags |= bio->bio_buf->b_flags & B_HASBOGUS;
 467
 468                         nestiobuf_add(bio, nestbuf, 0, 0, &dmv->stats);
 469                         nestbuf->b_bio1.bio_offset = 0;
 470                         table_en->target->strategy(table_en, nestbuf);
 471                 } else if (start < end) {
 472                         nestbuf = getpbuf(NULL);
 473                         nestbuf->b_flags |= bio->bio_buf->b_flags & B_HASBOGUS;
 474
 475                         nestiobuf_add(bio, nestbuf,
 476                                       start - buf_start, (end - start),
 477                                       &dmv->stats);
 478                         issued_len += end - start;
 479
 480                         nestbuf->b_bio1.bio_offset = (start - table_start);
 481                         table_en->target->strategy(table_en, nestbuf);
 482                 }
 483         }
 484
 485         if (issued_len < buf_len)
 486                 nestiobuf_error(bio, EINVAL);
 487         nestiobuf_start(bio);
 488         dm_table_release(&dmv->table_head, DM_TABLE_ACTIVE);
 489
 490         return 0;
 491 }
 492
 493 static int
 494 dmdump(struct dev_dump_args *ap)
 495 {
 496         cdev_t dev = ap->a_head.a_dev;
 497         dm_dev_t *dmv;
 498         dm_table_t  *tbl;
 499         dm_table_entry_t *table_en;
 500         uint64_t buf_start, buf_len, issued_len;
 501         uint64_t table_start, table_end;
 502         uint64_t start, end, data_offset;
 503         off_t offset;
 504         size_t length;
 505         int error = 0;
 506
 507         buf_start = ap->a_offset;
 508         buf_len = ap->a_length;
 509
 510         tbl = NULL;
 511
 512         table_end = 0;
 513         issued_len = 0;
 514
 515         dmv = dev->si_drv1;
 516
 517         /* Select active table */
 518         tbl = dm_table_get_entry(&dmv->table_head, DM_TABLE_ACTIVE);
 519
 520
 521         /*
 522          * Find out what tables I want to select.
 523          */
 524         SLIST_FOREACH(table_en, tbl, next) {
 525                 /*
 526                  * I need need number of bytes not blocks.
 527                  */
 528                 table_start = table_en->start * DEV_BSIZE;
 529                 table_end = table_start + (table_en->length) * DEV_BSIZE;
 530
 531                 /*
 532                  * Calculate the start and end
 533                  */
 534                 start = MAX(table_start, buf_start);
 535                 end = MIN(table_end, buf_start + buf_len);
 536
 537                 if (ap->a_length == 0) {
 538                         if (table_en->target->dump == NULL) {
 539                                 error = ENXIO;
 540                                 goto out;
 541                         }
 542
 543                         table_en->target->dump(table_en, NULL, 0, 0);
 544                 } else if (start < end) {
 545                         data_offset = start - buf_start;
 546                         offset = start - table_start;
 547                         length = end - start;
 548
 549                         if (table_en->target->dump == NULL) {
 550                                 error = ENXIO;
 551                                 goto out;
 552                         }
 553
 554                         table_en->target->dump(table_en,
 555                             (char *)ap->a_virtual + data_offset,
 556                             length, offset);
 557
 558                         issued_len += end - start;
 559                 }
 560         }
 561
 562         if (issued_len < buf_len)
 563                 error = EINVAL;
 564
 565 out:
 566         dm_table_release(&dmv->table_head, DM_TABLE_ACTIVE);
 567
 568         return error;
 569 }
 570
 571 static int
 572 dmsize(struct dev_psize_args *ap)
 573 {
 574         cdev_t dev = ap->a_head.a_dev;
 575         dm_dev_t *dmv;
 576         uint64_t size;
 577
 578         size = 0;
 579
 580         if ((dmv = dev->si_drv1) == NULL)
 581                 return ENXIO;
 582
 583         size = dm_table_size(&dmv->table_head);
 584         ap->a_result = (int64_t)size;
 585
 586         return 0;
 587 }
 588
 589 #if 0
 590 static void
 591 dmminphys(struct buf *bp)
 592 {
 593
 594         bp->b_bcount = MIN(bp->b_bcount, MAXPHYS);
 595 }
 596 #endif
 597
 598 void
 599 dmsetdiskinfo(struct disk *disk, dm_table_head_t *head)
 600 {
 601         struct disk_info info;
 602         uint64_t dmp_size;
 603
 604         dmp_size = dm_table_size(head);
 605
 606         bzero(&info, sizeof(struct disk_info));
 607         info.d_media_blksize = DEV_BSIZE;
 608         info.d_media_blocks = dmp_size;
 609 #if 0
 610         /* this is set by disk_setdiskinfo */
 611         info.d_media_size = dmp_size * DEV_BSIZE;
 612 #endif
 613         info.d_dsflags = DSO_MBRQUIET | DSO_DEVICEMAPPER | DSO_RAWPSIZE;
 614
 615         info.d_secpertrack = 32;
 616         info.d_nheads = 64;
 617         info.d_secpercyl = info.d_secpertrack * info.d_nheads;
 618         info.d_ncylinders = dmp_size / info.d_secpercyl;
 619
 620         disk_setdiskinfo(disk, &info);
 621 }
 622
 623 /*
 624  * Transform char s to uint64_t offset number.
 625  */
 626 uint64_t
 627 atoi64(const char *s)
 628 {
 629         uint64_t n;
 630         n = 0;
 631
 632         while (*s != '\0') {
 633                 if (!isdigit(*s))
 634                         break;
 635
 636                 n = (10 * n) + (*s - '0');
 637                 s++;
 638         }
 639
 640         return n;
 641 }
 642
 643 void
 644 dm_builtin_init(void *arg)
 645 {
 646         modeventhand_t evh = (modeventhand_t)arg;
 647
 648         KKASSERT(evh != NULL);
 649         evh(NULL, MOD_LOAD, NULL);
 650 }
 651
 652 void
 653 dm_builtin_uninit(void *arg)
 654 {
 655         modeventhand_t evh = (modeventhand_t)arg;
 656
 657         KKASSERT(evh != NULL);
 658         evh(NULL, MOD_UNLOAD, NULL);
 659 }
 660
 661 TUNABLE_INT("debug.dm_debug", &dm_debug_level);
 662 SYSCTL_INT(_debug, OID_AUTO, dm_debug, CTLFLAG_RW, &dm_debug_level,
 663                0, "Enable device mapper debugging");
 664