builtin/repack.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "config.h"
   4 #include "dir.h"
   5 #include "parse-options.h"
   6 #include "run-command.h"
   7 #include "sigchain.h"
   8 #include "strbuf.h"
   9 #include "string-list.h"
  10 #include "argv-array.h"
  11 #include "midx.h"
  12 #include "packfile.h"
  13 #include "object-store.h"
  14
  15 static int delta_base_offset = 1;
  16 static int pack_kept_objects = -1;
  17 static int write_bitmaps;
  18 static char *packdir, *packtmp;
  19
  20 static const char *const git_repack_usage[] = {
  21         N_("git repack [<options>]"),
  22         NULL
  23 };
  24
  25 static const char incremental_bitmap_conflict_error[] = N_(
  26 "Incremental repacks are incompatible with bitmap indexes.  Use\n"
  27 "--no-write-bitmap-index or disable the pack.writebitmaps configuration."
  28 );
  29
  30
  31 static int repack_config(const char *var, const char *value, void *cb)
  32 {
  33         if (!strcmp(var, "repack.usedeltabaseoffset")) {
  34                 delta_base_offset = git_config_bool(var, value);
  35                 return 0;
  36         }
  37         if (!strcmp(var, "repack.packkeptobjects")) {
  38                 pack_kept_objects = git_config_bool(var, value);
  39                 return 0;
  40         }
  41         if (!strcmp(var, "repack.writebitmaps") ||
  42             !strcmp(var, "pack.writebitmaps")) {
  43                 write_bitmaps = git_config_bool(var, value);
  44                 return 0;
  45         }
  46         return git_default_config(var, value, cb);
  47 }
  48
  49 /*
  50  * Remove temporary $GIT_OBJECT_DIRECTORY/pack/.tmp-$$-pack-* files.
  51  */
  52 static void remove_temporary_files(void)
  53 {
  54         struct strbuf buf = STRBUF_INIT;
  55         size_t dirlen, prefixlen;
  56         DIR *dir;
  57         struct dirent *e;
  58
  59         dir = opendir(packdir);
  60         if (!dir)
  61                 return;
  62
  63         /* Point at the slash at the end of ".../objects/pack/" */
  64         dirlen = strlen(packdir) + 1;
  65         strbuf_addstr(&buf, packtmp);
  66         /* Hold the length of  ".tmp-%d-pack-" */
  67         prefixlen = buf.len - dirlen;
  68
  69         while ((e = readdir(dir))) {
  70                 if (strncmp(e->d_name, buf.buf + dirlen, prefixlen))
  71                         continue;
  72                 strbuf_setlen(&buf, dirlen);
  73                 strbuf_addstr(&buf, e->d_name);
  74                 unlink(buf.buf);
  75         }
  76         closedir(dir);
  77         strbuf_release(&buf);
  78 }
  79
  80 static void remove_pack_on_signal(int signo)
  81 {
  82         remove_temporary_files();
  83         sigchain_pop(signo);
  84         raise(signo);
  85 }
  86
  87 /*
  88  * Adds all packs hex strings to the fname list, which do not
  89  * have a corresponding .keep file. These packs are not to
  90  * be kept if we are going to pack everything into one file.
  91  */
  92 static void get_non_kept_pack_filenames(struct string_list *fname_list,
  93                                         const struct string_list *extra_keep)
  94 {
  95         DIR *dir;
  96         struct dirent *e;
  97         char *fname;
  98
  99         if (!(dir = opendir(packdir)))
 100                 return;
 101
 102         while ((e = readdir(dir)) != NULL) {
 103                 size_t len;
 104                 int i;
 105
 106                 for (i = 0; i < extra_keep->nr; i++)
 107                         if (!fspathcmp(e->d_name, extra_keep->items[i].string))
 108                                 break;
 109                 if (extra_keep->nr > 0 && i < extra_keep->nr)
 110                         continue;
 111
 112                 if (!strip_suffix(e->d_name, ".pack", &len))
 113                         continue;
 114
 115                 fname = xmemdupz(e->d_name, len);
 116
 117                 if (!file_exists(mkpath("%s/%s.keep", packdir, fname)))
 118                         string_list_append_nodup(fname_list, fname);
 119                 else
 120                         free(fname);
 121         }
 122         closedir(dir);
 123 }
 124
 125 static void remove_redundant_pack(const char *dir_name, const char *base_name)
 126 {
 127         const char *exts[] = {".pack", ".idx", ".keep", ".bitmap", ".promisor"};
 128         int i;
 129         struct strbuf buf = STRBUF_INIT;
 130         size_t plen;
 131
 132         strbuf_addf(&buf, "%s/%s", dir_name, base_name);
 133         plen = buf.len;
 134
 135         for (i = 0; i < ARRAY_SIZE(exts); i++) {
 136                 strbuf_setlen(&buf, plen);
 137                 strbuf_addstr(&buf, exts[i]);
 138                 unlink(buf.buf);
 139         }
 140         strbuf_release(&buf);
 141 }
 142
 143 struct pack_objects_args {
 144         const char *window;
 145         const char *window_memory;
 146         const char *depth;
 147         const char *threads;
 148         const char *max_pack_size;
 149         int no_reuse_delta;
 150         int no_reuse_object;
 151         int quiet;
 152         int local;
 153 };
 154
 155 static void prepare_pack_objects(struct child_process *cmd,
 156                                  const struct pack_objects_args *args)
 157 {
 158         argv_array_push(&cmd->args, "pack-objects");
 159         if (args->window)
 160                 argv_array_pushf(&cmd->args, "--window=%s", args->window);
 161         if (args->window_memory)
 162                 argv_array_pushf(&cmd->args, "--window-memory=%s", args->window_memory);
 163         if (args->depth)
 164                 argv_array_pushf(&cmd->args, "--depth=%s", args->depth);
 165         if (args->threads)
 166                 argv_array_pushf(&cmd->args, "--threads=%s", args->threads);
 167         if (args->max_pack_size)
 168                 argv_array_pushf(&cmd->args, "--max-pack-size=%s", args->max_pack_size);
 169         if (args->no_reuse_delta)
 170                 argv_array_pushf(&cmd->args, "--no-reuse-delta");
 171         if (args->no_reuse_object)
 172                 argv_array_pushf(&cmd->args, "--no-reuse-object");
 173         if (args->local)
 174                 argv_array_push(&cmd->args,  "--local");
 175         if (args->quiet)
 176                 argv_array_push(&cmd->args,  "--quiet");
 177         if (delta_base_offset)
 178                 argv_array_push(&cmd->args,  "--delta-base-offset");
 179         argv_array_push(&cmd->args, packtmp);
 180         cmd->git_cmd = 1;
 181         cmd->out = -1;
 182 }
 183
 184 /*
 185  * Write oid to the given struct child_process's stdin, starting it first if
 186  * necessary.
 187  */
 188 static int write_oid(const struct object_id *oid, struct packed_git *pack,
 189                      uint32_t pos, void *data)
 190 {
 191         struct child_process *cmd = data;
 192
 193         if (cmd->in == -1) {
 194                 if (start_command(cmd))
 195                         die("Could not start pack-objects to repack promisor objects");
 196         }
 197
 198         xwrite(cmd->in, oid_to_hex(oid), GIT_SHA1_HEXSZ);
 199         xwrite(cmd->in, "\n", 1);
 200         return 0;
 201 }
 202
 203 static void repack_promisor_objects(const struct pack_objects_args *args,
 204                                     struct string_list *names)
 205 {
 206         struct child_process cmd = CHILD_PROCESS_INIT;
 207         FILE *out;
 208         struct strbuf line = STRBUF_INIT;
 209
 210         prepare_pack_objects(&cmd, args);
 211         cmd.in = -1;
 212
 213         /*
 214          * NEEDSWORK: Giving pack-objects only the OIDs without any ordering
 215          * hints may result in suboptimal deltas in the resulting pack. See if
 216          * the OIDs can be sent with fake paths such that pack-objects can use a
 217          * {type -> existing pack order} ordering when computing deltas instead
 218          * of a {type -> size} ordering, which may produce better deltas.
 219          */
 220         for_each_packed_object(write_oid, &cmd,
 221                                FOR_EACH_OBJECT_PROMISOR_ONLY);
 222
 223         if (cmd.in == -1)
 224                 /* No packed objects; cmd was never started */
 225                 return;
 226
 227         close(cmd.in);
 228
 229         out = xfdopen(cmd.out, "r");
 230         while (strbuf_getline_lf(&line, out) != EOF) {
 231                 char *promisor_name;
 232                 int fd;
 233                 if (line.len != 40)
 234                         die("repack: Expecting 40 character sha1 lines only from pack-objects.");
 235                 string_list_append(names, line.buf);
 236
 237                 /*
 238                  * pack-objects creates the .pack and .idx files, but not the
 239                  * .promisor file. Create the .promisor file, which is empty.
 240                  */
 241                 promisor_name = mkpathdup("%s-%s.promisor", packtmp,
 242                                           line.buf);
 243                 fd = open(promisor_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
 244                 if (fd < 0)
 245                         die_errno("unable to create '%s'", promisor_name);
 246                 close(fd);
 247                 free(promisor_name);
 248         }
 249         fclose(out);
 250         if (finish_command(&cmd))
 251                 die("Could not finish pack-objects to repack promisor objects");
 252 }
 253
 254 #define ALL_INTO_ONE 1
 255 #define LOOSEN_UNREACHABLE 2
 256
 257 int cmd_repack(int argc, const char **argv, const char *prefix)
 258 {
 259         struct {
 260                 const char *name;
 261                 unsigned optional:1;
 262         } exts[] = {
 263                 {".pack"},
 264                 {".idx"},
 265                 {".bitmap", 1},
 266                 {".promisor", 1},
 267         };
 268         struct child_process cmd = CHILD_PROCESS_INIT;
 269         struct string_list_item *item;
 270         struct string_list names = STRING_LIST_INIT_DUP;
 271         struct string_list rollback = STRING_LIST_INIT_NODUP;
 272         struct string_list existing_packs = STRING_LIST_INIT_DUP;
 273         struct strbuf line = STRBUF_INIT;
 274         int i, ext, ret, failed;
 275         FILE *out;
 276
 277         /* variables to be filled by option parsing */
 278         int pack_everything = 0;
 279         int delete_redundant = 0;
 280         const char *unpack_unreachable = NULL;
 281         int keep_unreachable = 0;
 282         struct string_list keep_pack_list = STRING_LIST_INIT_NODUP;
 283         int no_update_server_info = 0;
 284         int midx_cleared = 0;
 285         struct pack_objects_args po_args = {NULL};
 286
 287         struct option builtin_repack_options[] = {
 288                 OPT_BIT('a', NULL, &pack_everything,
 289                                 N_("pack everything in a single pack"), ALL_INTO_ONE),
 290                 OPT_BIT('A', NULL, &pack_everything,
 291                                 N_("same as -a, and turn unreachable objects loose"),
 292                                    LOOSEN_UNREACHABLE | ALL_INTO_ONE),
 293                 OPT_BOOL('d', NULL, &delete_redundant,
 294                                 N_("remove redundant packs, and run git-prune-packed")),
 295                 OPT_BOOL('f', NULL, &po_args.no_reuse_delta,
 296                                 N_("pass --no-reuse-delta to git-pack-objects")),
 297                 OPT_BOOL('F', NULL, &po_args.no_reuse_object,
 298                                 N_("pass --no-reuse-object to git-pack-objects")),
 299                 OPT_BOOL('n', NULL, &no_update_server_info,
 300                                 N_("do not run git-update-server-info")),
 301                 OPT__QUIET(&po_args.quiet, N_("be quiet")),
 302                 OPT_BOOL('l', "local", &po_args.local,
 303                                 N_("pass --local to git-pack-objects")),
 304                 OPT_BOOL('b', "write-bitmap-index", &write_bitmaps,
 305                                 N_("write bitmap index")),
 306                 OPT_STRING(0, "unpack-unreachable", &unpack_unreachable, N_("approxidate"),
 307                                 N_("with -A, do not loosen objects older than this")),
 308                 OPT_BOOL('k', "keep-unreachable", &keep_unreachable,
 309                                 N_("with -a, repack unreachable objects")),
 310                 OPT_STRING(0, "window", &po_args.window, N_("n"),
 311                                 N_("size of the window used for delta compression")),
 312                 OPT_STRING(0, "window-memory", &po_args.window_memory, N_("bytes"),
 313                                 N_("same as the above, but limit memory size instead of entries count")),
 314                 OPT_STRING(0, "depth", &po_args.depth, N_("n"),
 315                                 N_("limits the maximum delta depth")),
 316                 OPT_STRING(0, "threads", &po_args.threads, N_("n"),
 317                                 N_("limits the maximum number of threads")),
 318                 OPT_STRING(0, "max-pack-size", &po_args.max_pack_size, N_("bytes"),
 319                                 N_("maximum size of each packfile")),
 320                 OPT_BOOL(0, "pack-kept-objects", &pack_kept_objects,
 321                                 N_("repack objects in packs marked with .keep")),
 322                 OPT_STRING_LIST(0, "keep-pack", &keep_pack_list, N_("name"),
 323                                 N_("do not repack this pack")),
 324                 OPT_END()
 325         };
 326
 327         git_config(repack_config, NULL);
 328
 329         argc = parse_options(argc, argv, prefix, builtin_repack_options,
 330                                 git_repack_usage, 0);
 331
 332         if (delete_redundant && repository_format_precious_objects)
 333                 die(_("cannot delete packs in a precious-objects repo"));
 334
 335         if (keep_unreachable &&
 336             (unpack_unreachable || (pack_everything & LOOSEN_UNREACHABLE)))
 337                 die(_("--keep-unreachable and -A are incompatible"));
 338
 339         if (pack_kept_objects < 0)
 340                 pack_kept_objects = write_bitmaps;
 341
 342         if (write_bitmaps && !(pack_everything & ALL_INTO_ONE))
 343                 die(_(incremental_bitmap_conflict_error));
 344
 345         packdir = mkpathdup("%s/pack", get_object_directory());
 346         packtmp = mkpathdup("%s/.tmp-%d-pack", packdir, (int)getpid());
 347
 348         sigchain_push_common(remove_pack_on_signal);
 349
 350         prepare_pack_objects(&cmd, &po_args);
 351
 352         argv_array_push(&cmd.args, "--keep-true-parents");
 353         if (!pack_kept_objects)
 354                 argv_array_push(&cmd.args, "--honor-pack-keep");
 355         for (i = 0; i < keep_pack_list.nr; i++)
 356                 argv_array_pushf(&cmd.args, "--keep-pack=%s",
 357                                  keep_pack_list.items[i].string);
 358         argv_array_push(&cmd.args, "--non-empty");
 359         argv_array_push(&cmd.args, "--all");
 360         argv_array_push(&cmd.args, "--reflog");
 361         argv_array_push(&cmd.args, "--indexed-objects");
 362         if (repository_format_partial_clone)
 363                 argv_array_push(&cmd.args, "--exclude-promisor-objects");
 364         if (write_bitmaps)
 365                 argv_array_push(&cmd.args, "--write-bitmap-index");
 366
 367         if (pack_everything & ALL_INTO_ONE) {
 368                 get_non_kept_pack_filenames(&existing_packs, &keep_pack_list);
 369
 370                 repack_promisor_objects(&po_args, &names);
 371
 372                 if (existing_packs.nr && delete_redundant) {
 373                         if (unpack_unreachable) {
 374                                 argv_array_pushf(&cmd.args,
 375                                                 "--unpack-unreachable=%s",
 376                                                 unpack_unreachable);
 377                                 argv_array_push(&cmd.env_array, "GIT_REF_PARANOIA=1");
 378                         } else if (pack_everything & LOOSEN_UNREACHABLE) {
 379                                 argv_array_push(&cmd.args,
 380                                                 "--unpack-unreachable");
 381                         } else if (keep_unreachable) {
 382                                 argv_array_push(&cmd.args, "--keep-unreachable");
 383                                 argv_array_push(&cmd.args, "--pack-loose-unreachable");
 384                         } else {
 385                                 argv_array_push(&cmd.env_array, "GIT_REF_PARANOIA=1");
 386                         }
 387                 }
 388         } else {
 389                 argv_array_push(&cmd.args, "--unpacked");
 390                 argv_array_push(&cmd.args, "--incremental");
 391         }
 392
 393         cmd.no_stdin = 1;
 394
 395         ret = start_command(&cmd);
 396         if (ret)
 397                 return ret;
 398
 399         out = xfdopen(cmd.out, "r");
 400         while (strbuf_getline_lf(&line, out) != EOF) {
 401                 if (line.len != 40)
 402                         die("repack: Expecting 40 character sha1 lines only from pack-objects.");
 403                 string_list_append(&names, line.buf);
 404         }
 405         fclose(out);
 406         ret = finish_command(&cmd);
 407         if (ret)
 408                 return ret;
 409
 410         if (!names.nr && !po_args.quiet)
 411                 printf("Nothing new to pack.\n");
 412
 413         /*
 414          * Ok we have prepared all new packfiles.
 415          * First see if there are packs of the same name and if so
 416          * if we can move them out of the way (this can happen if we
 417          * repacked immediately after packing fully.
 418          */
 419         failed = 0;
 420         for_each_string_list_item(item, &names) {
 421                 for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
 422                         char *fname, *fname_old;
 423
 424                         if (!midx_cleared) {
 425                                 /* if we move a packfile, it will invalidated the midx */
 426                                 clear_midx_file(get_object_directory());
 427                                 midx_cleared = 1;
 428                         }
 429
 430                         fname = mkpathdup("%s/pack-%s%s", packdir,
 431                                                 item->string, exts[ext].name);
 432                         if (!file_exists(fname)) {
 433                                 free(fname);
 434                                 continue;
 435                         }
 436
 437                         fname_old = mkpathdup("%s/old-%s%s", packdir,
 438                                                 item->string, exts[ext].name);
 439                         if (file_exists(fname_old))
 440                                 if (unlink(fname_old))
 441                                         failed = 1;
 442
 443                         if (!failed && rename(fname, fname_old)) {
 444                                 free(fname);
 445                                 free(fname_old);
 446                                 failed = 1;
 447                                 break;
 448                         } else {
 449                                 string_list_append(&rollback, fname);
 450                                 free(fname_old);
 451                         }
 452                 }
 453                 if (failed)
 454                         break;
 455         }
 456         if (failed) {
 457                 struct string_list rollback_failure = STRING_LIST_INIT_DUP;
 458                 for_each_string_list_item(item, &rollback) {
 459                         char *fname, *fname_old;
 460                         fname = mkpathdup("%s/%s", packdir, item->string);
 461                         fname_old = mkpathdup("%s/old-%s", packdir, item->string);
 462                         if (rename(fname_old, fname))
 463                                 string_list_append(&rollback_failure, fname);
 464                         free(fname);
 465                         free(fname_old);
 466                 }
 467
 468                 if (rollback_failure.nr) {
 469                         int i;
 470                         fprintf(stderr,
 471                                 "WARNING: Some packs in use have been renamed by\n"
 472                                 "WARNING: prefixing old- to their name, in order to\n"
 473                                 "WARNING: replace them with the new version of the\n"
 474                                 "WARNING: file.  But the operation failed, and the\n"
 475                                 "WARNING: attempt to rename them back to their\n"
 476                                 "WARNING: original names also failed.\n"
 477                                 "WARNING: Please rename them in %s manually:\n", packdir);
 478                         for (i = 0; i < rollback_failure.nr; i++)
 479                                 fprintf(stderr, "WARNING:   old-%s -> %s\n",
 480                                         rollback_failure.items[i].string,
 481                                         rollback_failure.items[i].string);
 482                 }
 483                 exit(1);
 484         }
 485
 486         /* Now the ones with the same name are out of the way... */
 487         for_each_string_list_item(item, &names) {
 488                 for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
 489                         char *fname, *fname_old;
 490                         struct stat statbuffer;
 491                         int exists = 0;
 492                         fname = mkpathdup("%s/pack-%s%s",
 493                                         packdir, item->string, exts[ext].name);
 494                         fname_old = mkpathdup("%s-%s%s",
 495                                         packtmp, item->string, exts[ext].name);
 496                         if (!stat(fname_old, &statbuffer)) {
 497                                 statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
 498                                 chmod(fname_old, statbuffer.st_mode);
 499                                 exists = 1;
 500                         }
 501                         if (exists || !exts[ext].optional) {
 502                                 if (rename(fname_old, fname))
 503                                         die_errno(_("renaming '%s' failed"), fname_old);
 504                         }
 505                         free(fname);
 506                         free(fname_old);
 507                 }
 508         }
 509
 510         /* Remove the "old-" files */
 511         for_each_string_list_item(item, &names) {
 512                 for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
 513                         char *fname;
 514                         fname = mkpathdup("%s/old-%s%s",
 515                                           packdir,
 516                                           item->string,
 517                                           exts[ext].name);
 518                         if (remove_path(fname))
 519                                 warning(_("failed to remove '%s'"), fname);
 520                         free(fname);
 521                 }
 522         }
 523
 524         /* End of pack replacement. */
 525
 526         reprepare_packed_git(the_repository);
 527
 528         if (delete_redundant) {
 529                 int opts = 0;
 530                 string_list_sort(&names);
 531                 for_each_string_list_item(item, &existing_packs) {
 532                         char *sha1;
 533                         size_t len = strlen(item->string);
 534                         if (len < 40)
 535                                 continue;
 536                         sha1 = item->string + len - 40;
 537                         if (!string_list_has_string(&names, sha1))
 538                                 remove_redundant_pack(packdir, item->string);
 539                 }
 540                 if (!po_args.quiet && isatty(2))
 541                         opts |= PRUNE_PACKED_VERBOSE;
 542                 prune_packed_objects(opts);
 543         }
 544
 545         if (!no_update_server_info)
 546                 update_server_info(0);
 547         remove_temporary_files();
 548         string_list_clear(&names, 0);
 549         string_list_clear(&rollback, 0);
 550         string_list_clear(&existing_packs, 0);
 551         strbuf_release(&line);
 552
 553         return 0;
 554 }