sbin/hammer/cmd_recover.c

   1 /*
   2  * Copyright (c) 2010 The DragonFly Project.  All rights reserved.
   3  *
   4  * This code is derived from software contributed to The DragonFly Project
   5  * by Matthew Dillon <dillon@backplane.com>
   6  *
   7  * Redistribution and use in source and binary forms, with or without
   8  * modification, are permitted provided that the following conditions
   9  * are met:
  10  *
  11  * 1. Redistributions of source code must retain the above copyright
  12  *    notice, this list of conditions and the following disclaimer.
  13  * 2. Redistributions in binary form must reproduce the above copyright
  14  *    notice, this list of conditions and the following disclaimer in
  15  *    the documentation and/or other materials provided with the
  16  *    distribution.
  17  * 3. Neither the name of The DragonFly Project nor the names of its
  18  *    contributors may be used to endorse or promote products derived
  19  *    from this software without specific, prior written permission.
  20  *
  21  * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  22  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  23  * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
  24  * FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE
  25  * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
  26  * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
  27  * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  28  * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
  29  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
  30  * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
  31  * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  32  * SUCH DAMAGE.
  33  */
  34
  35 #include "hammer.h"
  36
  37 struct recover_dict {
  38         struct recover_dict *next;
  39         struct recover_dict *parent;
  40         int64_t obj_id;
  41         uint8_t obj_type;
  42         uint8_t flags;
  43         uint16_t pfs_id;
  44         int64_t size;
  45         char    *name;
  46 };
  47
  48 #define DICTF_MADEDIR   0x01
  49 #define DICTF_MADEFILE  0x02
  50 #define DICTF_PARENT    0x04    /* parent attached for real */
  51 #define DICTF_TRAVERSED 0x80
  52
  53 static void recover_top(char *ptr);
  54 static void recover_elm(hammer_btree_leaf_elm_t leaf);
  55 static struct recover_dict *get_dict(int64_t obj_id, uint16_t pfs_id);
  56 static char *recover_path(struct recover_dict *dict);
  57 static void sanitize_string(char *str);
  58
  59 static const char *TargetDir;
  60 static int CachedFd = -1;
  61 static char *CachedPath;
  62
  63 void
  64 hammer_cmd_recover(const char *target_dir)
  65 {
  66         struct buffer_info *data_buffer;
  67         struct volume_info *volume;
  68         hammer_off_t off;
  69         hammer_off_t off_end;
  70         char *ptr;
  71
  72         AssertOnFailure = 0;
  73         TargetDir = target_dir;
  74
  75         printf("Running raw scan of HAMMER image, recovering to %s\n",
  76                 TargetDir);
  77         mkdir(TargetDir, 0777);
  78
  79         data_buffer = NULL;
  80         TAILQ_FOREACH(volume, &VolList, entry) {
  81                 off = HAMMER_ENCODE_RAW_BUFFER(volume->vol_no, 0);
  82                 off_end = off + HAMMER_VOL_BUF_SIZE(volume->ondisk);
  83                 while (off < off_end) {
  84                         ptr = get_buffer_data(off, &data_buffer, 0);
  85                         if (ptr) {
  86                                 recover_top(ptr);
  87                                 off += HAMMER_BUFSIZE;
  88                         }
  89                 }
  90         }
  91         rel_buffer(data_buffer);
  92
  93         if (CachedPath) {
  94                 free(CachedPath);
  95                 close(CachedFd);
  96                 CachedPath = NULL;
  97                 CachedFd = -1;
  98         }
  99
 100         AssertOnFailure = 1;
 101 }
 102
 103 /*
 104  * Top level recovery processor.  Assume the data is a B-Tree node.
 105  * If the CRC is good we attempt to process the node, building the
 106  * object space and creating the dictionary as we go.
 107  */
 108 static void
 109 recover_top(char *ptr)
 110 {
 111         struct hammer_node_ondisk *node;
 112         hammer_btree_elm_t elm;
 113         int maxcount;
 114         int i;
 115
 116         for (node = (void *)ptr; (char *)node < ptr + HAMMER_BUFSIZE; ++node) {
 117                 if (crc32(&node->crc + 1, HAMMER_BTREE_CRCSIZE) ==
 118                     node->crc &&
 119                     node->type == HAMMER_BTREE_TYPE_LEAF) {
 120                         /*
 121                          * Scan elements
 122                          */
 123                         maxcount = HAMMER_BTREE_LEAF_ELMS;
 124                         for (i = 0; i < node->count && i < maxcount; ++i) {
 125                                 elm = &node->elms[i];
 126                                 if (elm->base.btype != 'R')
 127                                         continue;
 128                                 recover_elm(&elm->leaf);
 129                         }
 130                 }
 131         }
 132 }
 133
 134 static void
 135 recover_elm(hammer_btree_leaf_elm_t leaf)
 136 {
 137         struct buffer_info *data_buffer = NULL;
 138         struct recover_dict *dict;
 139         struct recover_dict *dict2;
 140         hammer_data_ondisk_t ondisk;
 141         hammer_off_t data_offset;
 142         struct stat st;
 143         int chunk;
 144         int len;
 145         int zfill;
 146         int64_t file_offset;
 147         uint16_t pfs_id;
 148         size_t nlen;
 149         int fd;
 150         char *name;
 151         char *path1;
 152         char *path2;
 153
 154         /*
 155          * Ignore deleted records
 156          */
 157         if (leaf->delete_ts)
 158                 return;
 159         if ((data_offset = leaf->data_offset) != 0)
 160                 ondisk = get_buffer_data(data_offset, &data_buffer, 0);
 161         else
 162                 ondisk = NULL;
 163         if (ondisk == NULL)
 164                 goto done;
 165
 166         len = leaf->data_len;
 167         chunk = HAMMER_BUFSIZE - ((int)data_offset & HAMMER_BUFMASK);
 168         if (chunk > len)
 169                 chunk = len;
 170
 171         if (len < 0 || len > HAMMER_XBUFSIZE || len > chunk)
 172                 goto done;
 173
 174         pfs_id = lo_to_pfs(leaf->base.localization);
 175
 176         dict = get_dict(leaf->base.obj_id, pfs_id);
 177
 178         switch(leaf->base.rec_type) {
 179         case HAMMER_RECTYPE_INODE:
 180                 /*
 181                  * We found an inode which also tells us where the file
 182                  * or directory is in the directory hierarchy.
 183                  */
 184                 if (VerboseOpt) {
 185                         printf("file %016jx:%05d inode found\n",
 186                                 (uintmax_t)leaf->base.obj_id, pfs_id);
 187                 }
 188                 path1 = recover_path(dict);
 189
 190                 /*
 191                  * Attach the inode to its parent.  This isn't strictly
 192                  * necessary because the information is also in the
 193                  * directory entries, but if we do not find the directory
 194                  * entry this ensures that the files will still be
 195                  * reasonably well organized in their proper directories.
 196                  */
 197                 if ((dict->flags & DICTF_PARENT) == 0 &&
 198                     dict->obj_id != HAMMER_OBJID_ROOT &&
 199                     ondisk->inode.parent_obj_id != 0) {
 200                         dict->flags |= DICTF_PARENT;
 201                         dict->parent = get_dict(ondisk->inode.parent_obj_id,
 202                                                 pfs_id);
 203                         if (dict->parent &&
 204                             (dict->parent->flags & DICTF_MADEDIR) == 0) {
 205                                 dict->parent->flags |= DICTF_MADEDIR;
 206                                 path2 = recover_path(dict->parent);
 207                                 printf("mkdir %s\n", path2);
 208                                 mkdir(path2, 0777);
 209                                 free(path2);
 210                                 path2 = NULL;
 211                         }
 212                 }
 213                 if (dict->obj_type == 0)
 214                         dict->obj_type = ondisk->inode.obj_type;
 215                 dict->size = ondisk->inode.size;
 216                 path2 = recover_path(dict);
 217
 218                 if (lstat(path1, &st) == 0) {
 219                         if (ondisk->inode.obj_type == HAMMER_OBJTYPE_REGFILE) {
 220                                 truncate(path1, dict->size);
 221                                 /* chmod(path1, 0666); */
 222                         }
 223                         if (strcmp(path1, path2)) {
 224                                 printf("Rename %s -> %s\n", path1, path2);
 225                                 rename(path1, path2);
 226                         }
 227                 } else if (ondisk->inode.obj_type == HAMMER_OBJTYPE_REGFILE) {
 228                         printf("mkinode (file) %s\n", path2);
 229                         fd = open(path2, O_RDWR|O_CREAT, 0666);
 230                         if (fd > 0)
 231                                 close(fd);
 232                 } else if (ondisk->inode.obj_type == HAMMER_OBJTYPE_DIRECTORY) {
 233                         printf("mkinode (dir) %s\n", path2);
 234                         mkdir(path2, 0777);
 235                         dict->flags |= DICTF_MADEDIR;
 236                 }
 237                 free(path1);
 238                 free(path2);
 239                 break;
 240         case HAMMER_RECTYPE_DATA:
 241                 /*
 242                  * File record data
 243                  */
 244                 if (leaf->base.obj_id == 0)
 245                         break;
 246                 if (VerboseOpt) {
 247                         printf("file %016jx:%05d data %016jx,%d\n",
 248                                 (uintmax_t)leaf->base.obj_id,
 249                                 pfs_id,
 250                                 (uintmax_t)leaf->base.key - len,
 251                                 len);
 252                 }
 253
 254                 /*
 255                  * Update the dictionary entry
 256                  */
 257                 if (dict->obj_type == 0)
 258                         dict->obj_type = HAMMER_OBJTYPE_REGFILE;
 259
 260                 /*
 261                  * If the parent directory has not been created we
 262                  * have to create it (typically a PFS%05d)
 263                  */
 264                 if (dict->parent &&
 265                     (dict->parent->flags & DICTF_MADEDIR) == 0) {
 266                         dict->parent->flags |= DICTF_MADEDIR;
 267                         path2 = recover_path(dict->parent);
 268                         printf("mkdir %s\n", path2);
 269                         mkdir(path2, 0777);
 270                         free(path2);
 271                         path2 = NULL;
 272                 }
 273
 274                 /*
 275                  * Create the file if necessary, report file creations
 276                  */
 277                 path1 = recover_path(dict);
 278                 if (CachedPath && strcmp(CachedPath, path1) == 0) {
 279                         fd = CachedFd;
 280                 } else {
 281                         fd = open(path1, O_CREAT|O_RDWR, 0666);
 282                 }
 283                 if (fd < 0) {
 284                         printf("Unable to create %s: %s\n",
 285                                 path1, strerror(errno));
 286                         free(path1);
 287                         break;
 288                 }
 289                 if ((dict->flags & DICTF_MADEFILE) == 0) {
 290                         dict->flags |= DICTF_MADEFILE;
 291                         printf("mkfile %s\n", path1);
 292                 }
 293
 294                 /*
 295                  * And write the record.  A HAMMER data block is aligned
 296                  * and may contain trailing zeros after the file EOF.  The
 297                  * inode record is required to get the actual file size.
 298                  *
 299                  * However, when the inode record is not available
 300                  * we can do a sparse write and that will get it right
 301                  * most of the time even if the inode record is never
 302                  * found.
 303                  */
 304                 file_offset = (int64_t)leaf->base.key - len;
 305                 lseek(fd, (off_t)file_offset, SEEK_SET);
 306                 while (len) {
 307                         if (dict->size == -1) {
 308                                 for (zfill = chunk - 1; zfill >= 0; --zfill) {
 309                                         if (((char *)ondisk)[zfill])
 310                                                 break;
 311                                 }
 312                                 ++zfill;
 313                         } else {
 314                                 zfill = chunk;
 315                         }
 316
 317                         if (zfill)
 318                                 write(fd, ondisk, zfill);
 319                         if (zfill < chunk)
 320                                 lseek(fd, chunk - zfill, SEEK_CUR);
 321
 322                         len -= chunk;
 323                         data_offset += chunk;
 324                         file_offset += chunk;
 325                         ondisk = get_buffer_data(data_offset, &data_buffer, 0);
 326                         if (ondisk == NULL)
 327                                 break;
 328                         chunk = HAMMER_BUFSIZE -
 329                                 ((int)data_offset & HAMMER_BUFMASK);
 330                         if (chunk > len)
 331                                 chunk = len;
 332                 }
 333                 if (dict->size >= 0 && file_offset > dict->size) {
 334                         ftruncate(fd, dict->size);
 335                         /* fchmod(fd, 0666); */
 336                 }
 337
 338                 if (fd == CachedFd) {
 339                         free(path1);
 340                 } else if (CachedPath) {
 341                         free(CachedPath);
 342                         close(CachedFd);
 343                         CachedPath = path1;
 344                         CachedFd = fd;
 345                 } else {
 346                         CachedPath = path1;
 347                         CachedFd = fd;
 348                 }
 349                 break;
 350         case HAMMER_RECTYPE_DIRENTRY:
 351                 nlen = len - offsetof(struct hammer_direntry_data, name[0]);
 352                 if ((int)nlen < 0)      /* illegal length */
 353                         break;
 354                 if (ondisk->entry.obj_id == 0 ||
 355                     ondisk->entry.obj_id == HAMMER_OBJID_ROOT)
 356                         break;
 357                 name = malloc(nlen + 1);
 358                 bcopy(ondisk->entry.name, name, nlen);
 359                 name[nlen] = 0;
 360                 sanitize_string(name);
 361
 362                 /*
 363                  * We can't deal with hardlinks so if the object already
 364                  * has a name assigned to it we just keep using that name.
 365                  */
 366                 dict2 = get_dict(ondisk->entry.obj_id, pfs_id);
 367                 path1 = recover_path(dict2);
 368
 369                 if (dict2->name == NULL)
 370                         dict2->name = name;
 371                 else
 372                         free(name);
 373
 374                 /*
 375                  * Attach dict2 to its directory (dict), create the
 376                  * directory (dict) if necessary.  We must ensure
 377                  * that the directory entry exists in order to be
 378                  * able to properly rename() the file without creating
 379                  * a namespace conflict.
 380                  */
 381                 if ((dict2->flags & DICTF_PARENT) == 0) {
 382                         dict2->flags |= DICTF_PARENT;
 383                         dict2->parent = dict;
 384                         if ((dict->flags & DICTF_MADEDIR) == 0) {
 385                                 dict->flags |= DICTF_MADEDIR;
 386                                 path2 = recover_path(dict);
 387                                 printf("mkdir %s\n", path2);
 388                                 mkdir(path2, 0777);
 389                                 free(path2);
 390                                 path2 = NULL;
 391                         }
 392                 }
 393                 path2 = recover_path(dict2);
 394                 if (strcmp(path1, path2) != 0 && lstat(path1, &st) == 0) {
 395                         printf("Rename %s -> %s\n", path1, path2);
 396                         rename(path1, path2);
 397                 }
 398                 free(path1);
 399                 free(path2);
 400
 401                 printf("dir  %016jx:%05d entry %016jx \"%s\"\n",
 402                         (uintmax_t)leaf->base.obj_id,
 403                         pfs_id,
 404                         (uintmax_t)ondisk->entry.obj_id,
 405                         name);
 406                 break;
 407         default:
 408                 /*
 409                  * Ignore any other record types
 410                  */
 411                 break;
 412         }
 413 done:
 414         rel_buffer(data_buffer);
 415 }
 416
 417 #define RD_HSIZE        32768
 418 #define RD_HMASK        (RD_HSIZE - 1)
 419
 420 struct recover_dict *RDHash[RD_HSIZE];
 421
 422 static
 423 struct recover_dict *
 424 get_dict(int64_t obj_id, uint16_t pfs_id)
 425 {
 426         struct recover_dict *dict;
 427         int i;
 428
 429         if (obj_id == 0)
 430                 return(NULL);
 431
 432         i = crc32(&obj_id, sizeof(obj_id)) & RD_HMASK;
 433         for (dict = RDHash[i]; dict; dict = dict->next) {
 434                 if (dict->obj_id == obj_id &&
 435                     dict->pfs_id == pfs_id) {
 436                         break;
 437                 }
 438         }
 439         if (dict == NULL) {
 440                 dict = malloc(sizeof(*dict));
 441                 bzero(dict, sizeof(*dict));
 442                 dict->obj_id = obj_id;
 443                 dict->pfs_id = pfs_id;
 444                 dict->next = RDHash[i];
 445                 dict->size = -1;
 446                 RDHash[i] = dict;
 447
 448                 /*
 449                  * Always connect dangling dictionary entries to object 1
 450                  * (the root of the PFS).
 451                  *
 452                  * DICTF_PARENT will not be set until we know what the
 453                  * real parent directory object is.
 454                  */
 455                 if (dict->obj_id != HAMMER_OBJID_ROOT)
 456                         dict->parent = get_dict(1, pfs_id);
 457         }
 458         return(dict);
 459 }
 460
 461 struct path_info {
 462         enum { PI_FIGURE, PI_LOAD } state;
 463         uint16_t pfs_id;
 464         char *base;
 465         char *next;
 466         int len;
 467 };
 468
 469 static void recover_path_helper(struct recover_dict *, struct path_info *);
 470
 471 static
 472 char *
 473 recover_path(struct recover_dict *dict)
 474 {
 475         struct path_info info;
 476
 477         bzero(&info, sizeof(info));
 478         info.pfs_id = dict->pfs_id;
 479         info.state = PI_FIGURE;
 480         recover_path_helper(dict, &info);
 481         info.base = malloc(info.len);
 482         info.next = info.base;
 483         info.state = PI_LOAD;
 484         recover_path_helper(dict, &info);
 485
 486         return(info.base);
 487 }
 488
 489 static
 490 void
 491 recover_path_helper(struct recover_dict *dict, struct path_info *info)
 492 {
 493         /*
 494          * Calculate path element length
 495          */
 496         dict->flags |= DICTF_TRAVERSED;
 497
 498         switch(info->state) {
 499         case PI_FIGURE:
 500                 if (dict->obj_id == HAMMER_OBJID_ROOT)
 501                         info->len += 8;
 502                 else if (dict->name)
 503                         info->len += strlen(dict->name);
 504                 else
 505                         info->len += 6 + 16;
 506                 ++info->len;
 507
 508                 if (dict->parent &&
 509                     (dict->parent->flags & DICTF_TRAVERSED) == 0) {
 510                         recover_path_helper(dict->parent, info);
 511                 } else {
 512                         info->len += strlen(TargetDir) + 1;
 513                 }
 514                 break;
 515         case PI_LOAD:
 516                 if (dict->parent &&
 517                     (dict->parent->flags & DICTF_TRAVERSED) == 0) {
 518                         recover_path_helper(dict->parent, info);
 519                 } else {
 520                         strcpy(info->next, TargetDir);
 521                         info->next += strlen(info->next);
 522                 }
 523
 524                 *info->next++ = '/';
 525                 if (dict->obj_id == HAMMER_OBJID_ROOT) {
 526                         snprintf(info->next, 8+1, "PFS%05d", info->pfs_id);
 527                 } else if (dict->name) {
 528                         strcpy(info->next, dict->name);
 529                 } else {
 530                         snprintf(info->next, 6+16+1, "obj_0x%016jx",
 531                                 (uintmax_t)dict->obj_id);
 532                 }
 533                 info->next += strlen(info->next);
 534                 break;
 535         }
 536         dict->flags &= ~DICTF_TRAVERSED;
 537 }
 538
 539 static
 540 void
 541 sanitize_string(char *str)
 542 {
 543         while (*str) {
 544                 if (!isprint(*str))
 545                         *str = 'x';
 546                 ++str;
 547         }
 548 }