btrfs-calc-size.c

   1 /*
   2  * Copyright (C) 2011 Red Hat.  All rights reserved.
   3  *
   4  * This program is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU General Public
   6  * License v2 as published by the Free Software Foundation.
   7  *
   8  * This program is distributed in the hope that it will be useful,
   9  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  10  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  11  * General Public License for more details.
  12  *
  13  * You should have received a copy of the GNU General Public
  14  * License along with this program; if not, write to the
  15  * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  16  * Boston, MA 021110-1307, USA.
  17  */
  18
  19 #include <ctype.h>
  20 #include <stdio.h>
  21 #include <stdlib.h>
  22 #include <unistd.h>
  23 #include <fcntl.h>
  24 #include <sys/stat.h>
  25 #include <sys/time.h>
  26 #include <sys/types.h>
  27 #include <zlib.h>
  28 #include "kerncompat.h"
  29 #include "ctree.h"
  30 #include "disk-io.h"
  31 #include "print-tree.h"
  32 #include "transaction.h"
  33 #include "list.h"
  34 #include "version.h"
  35 #include "volumes.h"
  36 #include "utils.h"
  37
  38 static int verbose = 0;
  39 static int no_pretty = 0;
  40
  41 struct seek {
  42         u64 distance;
  43         u64 count;
  44         struct rb_node n;
  45 };
  46
  47 struct root_stats {
  48         u64 total_nodes;
  49         u64 total_leaves;
  50         u64 total_bytes;
  51         u64 total_inline;
  52         u64 total_seeks;
  53         u64 forward_seeks;
  54         u64 backward_seeks;
  55         u64 total_seek_len;
  56         u64 max_seek_len;
  57         u64 total_clusters;
  58         u64 total_cluster_size;
  59         u64 min_cluster_size;
  60         u64 max_cluster_size;
  61         u64 lowest_bytenr;
  62         u64 highest_bytenr;
  63         struct rb_root seek_root;
  64         int total_levels;
  65 };
  66
  67 struct fs_root {
  68         struct btrfs_key key;
  69         struct btrfs_key *snaps;
  70 };
  71
  72 static int add_seek(struct rb_root *root, u64 dist)
  73 {
  74         struct rb_node **p = &root->rb_node;
  75         struct rb_node *parent = NULL;
  76         struct seek *seek = NULL;
  77
  78         while (*p) {
  79                 parent = *p;
  80                 seek = rb_entry(parent, struct seek, n);
  81
  82                 if (dist < seek->distance) {
  83                         p = &(*p)->rb_left;
  84                 } else if (dist > seek->distance) {
  85                         p = &(*p)->rb_right;
  86                 } else {
  87                         seek->count++;
  88                         return 0;
  89                 }
  90         }
  91
  92         seek = malloc(sizeof(struct seek));
  93         if (!seek)
  94                 return -ENOMEM;
  95         seek->distance = dist;
  96         seek->count = 1;
  97         rb_link_node(&seek->n, parent, p);
  98         rb_insert_color(&seek->n, root);
  99         return 0;
 100 }
 101
 102 static int walk_leaf(struct btrfs_root *root, struct btrfs_path *path,
 103                      struct root_stats *stat, int find_inline)
 104 {
 105         struct extent_buffer *b = path->nodes[0];
 106         struct btrfs_file_extent_item *fi;
 107         struct btrfs_key found_key;
 108         int i;
 109
 110         stat->total_bytes += root->leafsize;
 111         stat->total_leaves++;
 112
 113         if (!find_inline)
 114                 return 0;
 115
 116         for (i = 0; i < btrfs_header_nritems(b); i++) {
 117                 btrfs_item_key_to_cpu(b, &found_key, i);
 118                 if (found_key.type != BTRFS_EXTENT_DATA_KEY)
 119                         continue;
 120
 121                 fi = btrfs_item_ptr(b, i, struct btrfs_file_extent_item);
 122                 if (btrfs_file_extent_type(b, fi) == BTRFS_FILE_EXTENT_INLINE)
 123                         stat->total_inline +=
 124                                 btrfs_file_extent_inline_item_len(b,
 125                                                         btrfs_item_nr(i));
 126         }
 127
 128         return 0;
 129 }
 130
 131 static u64 calc_distance(u64 block1, u64 block2)
 132 {
 133         if (block1 < block2)
 134                 return block2 - block1;
 135         return block1 - block2;
 136 }
 137
 138 static int walk_nodes(struct btrfs_root *root, struct btrfs_path *path,
 139                       struct root_stats *stat, int level, int find_inline)
 140 {
 141         struct extent_buffer *b = path->nodes[level];
 142         u64 last_block;
 143         u64 cluster_size = root->leafsize;
 144         int i;
 145         int ret = 0;
 146
 147         stat->total_bytes += root->nodesize;
 148         stat->total_nodes++;
 149
 150         last_block = btrfs_header_bytenr(b);
 151         for (i = 0; i < btrfs_header_nritems(b); i++) {
 152                 struct extent_buffer *tmp = NULL;
 153                 u64 cur_blocknr = btrfs_node_blockptr(b, i);
 154
 155                 path->slots[level] = i;
 156                 if ((level - 1) > 0 || find_inline) {
 157                         tmp = read_tree_block(root, cur_blocknr,
 158                                               btrfs_level_size(root, level - 1),
 159                                               btrfs_node_ptr_generation(b, i));
 160                         if (!tmp) {
 161                                 fprintf(stderr, "Failed to read blocknr %Lu\n",
 162                                         btrfs_node_blockptr(b, i));
 163                                 continue;
 164                         }
 165                         path->nodes[level - 1] = tmp;
 166                 }
 167                 if (level - 1)
 168                         ret = walk_nodes(root, path, stat, level - 1,
 169                                          find_inline);
 170                 else
 171                         ret = walk_leaf(root, path, stat, find_inline);
 172                 if (last_block + root->leafsize != cur_blocknr) {
 173                         u64 distance = calc_distance(last_block +
 174                                                      root->leafsize,
 175                                                      cur_blocknr);
 176                         stat->total_seeks++;
 177                         stat->total_seek_len += distance;
 178                         if (stat->max_seek_len < distance)
 179                                 stat->max_seek_len = distance;
 180                         if (add_seek(&stat->seek_root, distance)) {
 181                                 fprintf(stderr, "Error adding new seek\n");
 182                                 ret = -ENOMEM;
 183                                 break;
 184                         }
 185
 186                         if (last_block < cur_blocknr)
 187                                 stat->forward_seeks++;
 188                         else
 189                                 stat->backward_seeks++;
 190                         if (cluster_size != root->leafsize) {
 191                                 stat->total_cluster_size += cluster_size;
 192                                 stat->total_clusters++;
 193                                 if (cluster_size < stat->min_cluster_size)
 194                                         stat->min_cluster_size = cluster_size;
 195                                 if (cluster_size > stat->max_cluster_size)
 196                                         stat->max_cluster_size = cluster_size;
 197                         }
 198                         cluster_size = root->leafsize;
 199                 } else {
 200                         cluster_size += root->leafsize;
 201                 }
 202                 last_block = cur_blocknr;
 203                 if (cur_blocknr < stat->lowest_bytenr)
 204                         stat->lowest_bytenr = cur_blocknr;
 205                 if (cur_blocknr > stat->highest_bytenr)
 206                         stat->highest_bytenr = cur_blocknr;
 207                 free_extent_buffer(tmp);
 208                 if (ret) {
 209                         fprintf(stderr, "Error walking down path\n");
 210                         break;
 211                 }
 212         }
 213
 214         return ret;
 215 }
 216
 217 static void print_seek_histogram(struct root_stats *stat)
 218 {
 219         struct rb_node *n = rb_first(&stat->seek_root);
 220         struct seek *seek;
 221         u64 tick_interval;
 222         u64 group_start;
 223         u64 group_count = 0;
 224         u64 group_end;
 225         u64 i;
 226         u64 max_seek = stat->max_seek_len;
 227         int digits = 1;
 228
 229         if (stat->total_seeks < 20)
 230                 return;
 231
 232         while ((max_seek /= 10))
 233                 digits++;
 234
 235         /* Make a tick count as 5% of the total seeks */
 236         tick_interval = stat->total_seeks / 20;
 237         printf("\tSeek histogram\n");
 238         for (; n; n = rb_next(n)) {
 239                 u64 ticks, gticks = 0;
 240
 241                 seek = rb_entry(n, struct seek, n);
 242                 ticks = seek->count / tick_interval;
 243                 if (group_count)
 244                         gticks = group_count / tick_interval;
 245
 246                 if (ticks <= 2 && gticks <= 2) {
 247                         if (group_count == 0)
 248                                 group_start = seek->distance;
 249                         group_end = seek->distance;
 250                         group_count += seek->count;
 251                         continue;
 252                 }
 253
 254                 if (group_count) {
 255
 256                         gticks = group_count / tick_interval;
 257                         printf("\t\t%*Lu - %*Lu: %*Lu ", digits, group_start,
 258                                digits, group_end, digits, group_count);
 259                         if (gticks) {
 260                                 for (i = 0; i < gticks; i++)
 261                                         printf("#");
 262                                 printf("\n");
 263                         } else {
 264                                 printf("|\n");
 265                         }
 266                         group_count = 0;
 267                 }
 268
 269                 if (ticks <= 2)
 270                         continue;
 271
 272                 printf("\t\t%*Lu - %*Lu: %*Lu ", digits, seek->distance,
 273                        digits, seek->distance, digits, seek->count);
 274                 for (i = 0; i < ticks; i++)
 275                         printf("#");
 276                 printf("\n");
 277         }
 278         if (group_count) {
 279                 u64 gticks;
 280
 281                 gticks = group_count / tick_interval;
 282                 printf("\t\t%*Lu - %*Lu: %*Lu ", digits, group_start,
 283                        digits, group_end, digits, group_count);
 284                 if (gticks) {
 285                         for (i = 0; i < gticks; i++)
 286                                 printf("#");
 287                         printf("\n");
 288                 } else {
 289                         printf("|\n");
 290                 }
 291                 group_count = 0;
 292         }
 293 }
 294
 295 static void timeval_subtract(struct timeval *result,struct timeval *x,
 296                              struct timeval *y)
 297 {
 298         if (x->tv_usec < y->tv_usec) {
 299                 int nsec = (y->tv_usec - x->tv_usec) / 1000000 + 1;
 300                 y->tv_usec -= 1000000 * nsec;
 301                 y->tv_sec += nsec;
 302         }
 303
 304         if (x->tv_usec - y->tv_usec > 1000000) {
 305                 int nsec = (x->tv_usec - y->tv_usec) / 1000000;
 306                 y->tv_usec += 1000000 * nsec;
 307                 y->tv_sec -= nsec;
 308         }
 309
 310         result->tv_sec = x->tv_sec - y->tv_sec;
 311         result->tv_usec = x->tv_usec - y->tv_usec;
 312 }
 313
 314 static int calc_root_size(struct btrfs_root *tree_root, struct btrfs_key *key,
 315                           int find_inline)
 316 {
 317         struct btrfs_root *root;
 318         struct btrfs_path *path;
 319         struct rb_node *n;
 320         struct timeval start, end, diff = {0};
 321         struct root_stats stat;
 322         int level;
 323         int ret = 0;
 324         int size_fail = 0;
 325
 326         root = btrfs_read_fs_root(tree_root->fs_info, key);
 327         if (IS_ERR(root)) {
 328                 fprintf(stderr, "Failed to read root %Lu\n", key->objectid);
 329                 return 1;
 330         }
 331
 332         path = btrfs_alloc_path();
 333         if (!path) {
 334                 fprintf(stderr, "Could not allocate path\n");
 335                 return 1;
 336         }
 337
 338         memset(&stat, 0, sizeof(stat));
 339         level = btrfs_header_level(root->node);
 340         stat.lowest_bytenr = btrfs_header_bytenr(root->node);
 341         stat.highest_bytenr = stat.lowest_bytenr;
 342         stat.min_cluster_size = (u64)-1;
 343         stat.max_cluster_size = root->leafsize;
 344         path->nodes[level] = root->node;
 345         if (gettimeofday(&start, NULL)) {
 346                 fprintf(stderr, "Error getting time: %d\n", errno);
 347                 goto out;
 348         }
 349         if (!level) {
 350                 ret = walk_leaf(root, path, &stat, find_inline);
 351                 if (ret)
 352                         goto out;
 353                 goto out_print;
 354         }
 355
 356         ret = walk_nodes(root, path, &stat, level, find_inline);
 357         if (ret)
 358                 goto out;
 359         if (gettimeofday(&end, NULL)) {
 360                 fprintf(stderr, "Error getting time: %d\n", errno);
 361                 goto out;
 362         }
 363         timeval_subtract(&diff, &end, &start);
 364 out_print:
 365         if (stat.min_cluster_size == (u64)-1) {
 366                 stat.min_cluster_size = 0;
 367                 stat.total_clusters = 1;
 368         }
 369
 370         if (no_pretty || size_fail) {
 371                 printf("\tTotal size: %Lu\n", stat.total_bytes);
 372                 printf("\t\tInline data: %Lu\n", stat.total_inline);
 373                 printf("\tTotal seeks: %Lu\n", stat.total_seeks);
 374                 printf("\t\tForward seeks: %Lu\n", stat.forward_seeks);
 375                 printf("\t\tBackward seeks: %Lu\n", stat.backward_seeks);
 376                 printf("\t\tAvg seek len: %Lu\n", stat.total_seek_len /
 377                        stat.total_seeks);
 378                 print_seek_histogram(&stat);
 379                 printf("\tTotal clusters: %Lu\n", stat.total_clusters);
 380                 printf("\t\tAvg cluster size: %Lu\n", stat.total_cluster_size /
 381                        stat.total_clusters);
 382                 printf("\t\tMin cluster size: %Lu\n", stat.min_cluster_size);
 383                 printf("\t\tMax cluster size: %Lu\n", stat.max_cluster_size);
 384                 printf("\tTotal disk spread: %Lu\n", stat.highest_bytenr -
 385                        stat.lowest_bytenr);
 386                 printf("\tTotal read time: %d s %d us\n", (int)diff.tv_sec,
 387                        (int)diff.tv_usec);
 388                 printf("\tLevels: %d\n", level + 1);
 389         } else {
 390                 printf("\tTotal size: %s\n", pretty_size(stat.total_bytes));
 391                 printf("\t\tInline data: %s\n", pretty_size(stat.total_inline));
 392                 printf("\tTotal seeks: %Lu\n", stat.total_seeks);
 393                 printf("\t\tForward seeks: %Lu\n", stat.forward_seeks);
 394                 printf("\t\tBackward seeks: %Lu\n", stat.backward_seeks);
 395                 printf("\t\tAvg seek len: %s\n", stat.total_seeks ?
 396                         pretty_size(stat.total_seek_len / stat.total_seeks) :
 397                         pretty_size(0));
 398                 print_seek_histogram(&stat);
 399                 printf("\tTotal clusters: %Lu\n", stat.total_clusters);
 400                 printf("\t\tAvg cluster size: %s\n",
 401                                 pretty_size((stat.total_cluster_size /
 402                                                 stat.total_clusters)));
 403                 printf("\t\tMin cluster size: %s\n",
 404                                 pretty_size(stat.min_cluster_size));
 405                 printf("\t\tMax cluster size: %s\n",
 406                                 pretty_size(stat.max_cluster_size));
 407                 printf("\tTotal disk spread: %s\n",
 408                                 pretty_size(stat.highest_bytenr -
 409                                         stat.lowest_bytenr));
 410                 printf("\tTotal read time: %d s %d us\n", (int)diff.tv_sec,
 411                        (int)diff.tv_usec);
 412                 printf("\tLevels: %d\n", level + 1);
 413         }
 414 out:
 415         while ((n = rb_first(&stat.seek_root)) != NULL) {
 416                 struct seek *seek = rb_entry(n, struct seek, n);
 417                 rb_erase(n, &stat.seek_root);
 418                 free(seek);
 419         }
 420
 421         btrfs_free_path(path);
 422         return ret;
 423 }
 424
 425 static void usage()
 426 {
 427         fprintf(stderr, "Usage: calc-size [-v] [-b] <device>\n");
 428 }
 429
 430 int main(int argc, char **argv)
 431 {
 432         struct btrfs_key key;
 433         struct fs_root *roots;
 434         struct btrfs_root *root;
 435         size_t fs_roots_size = sizeof(struct fs_root);
 436         int opt;
 437         int ret = 0;
 438
 439         while ((opt = getopt(argc, argv, "vb")) != -1) {
 440                 switch (opt) {
 441                         case 'v':
 442                                 verbose++;
 443                                 break;
 444                         case 'b':
 445                                 no_pretty = 1;
 446                                 break;
 447                         default:
 448                                 usage();
 449                                 exit(1);
 450                 }
 451         }
 452
 453         set_argv0(argv);
 454         argc = argc - optind;
 455         if (check_argc_min(argc, 1)) {
 456                 usage();
 457                 exit(1);
 458         }
 459
 460         /*
 461         if ((ret = check_mounted(argv[optind])) < 0) {
 462                 fprintf(stderr, "Could not check mount status: %d\n", ret);
 463                 if (ret == -EACCES)
 464                         fprintf(stderr, "Maybe you need to run as root?\n");
 465                 return ret;
 466         } else if (ret) {
 467                 fprintf(stderr, "%s is currently mounted.  Aborting.\n",
 468                         argv[optind]);
 469                 return -EBUSY;
 470         }
 471         */
 472
 473         root = open_ctree(argv[optind], 0, 0);
 474         if (!root) {
 475                 fprintf(stderr, "Couldn't open ctree\n");
 476                 exit(1);
 477         }
 478
 479         roots = malloc(fs_roots_size);
 480         if (!roots) {
 481                 fprintf(stderr, "No memory\n");
 482                 goto out;
 483         }
 484
 485         printf("Calculating size of root tree\n");
 486         key.objectid = BTRFS_ROOT_TREE_OBJECTID;
 487         ret = calc_root_size(root, &key, 0);
 488         if (ret)
 489                 goto out;
 490
 491         printf("Calculating size of extent tree\n");
 492         key.objectid = BTRFS_EXTENT_TREE_OBJECTID;
 493         ret = calc_root_size(root, &key, 0);
 494         if (ret)
 495                 goto out;
 496
 497         printf("Calculating size of csum tree\n");
 498         key.objectid = BTRFS_CSUM_TREE_OBJECTID;
 499         ret = calc_root_size(root, &key, 0);
 500         if (ret)
 501                 goto out;
 502
 503         roots[0].key.objectid = BTRFS_FS_TREE_OBJECTID;
 504         roots[0].key.offset = (u64)-1;
 505         printf("Calculatin' size of fs tree\n");
 506         ret = calc_root_size(root, &roots[0].key, 1);
 507         if (ret)
 508                 goto out;
 509 out:
 510         close_ctree(root);
 511         free(roots);
 512         return ret;
 513 }