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