Merge branch 'ds/bloom-cleanup'
[git] / 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 "prune-packed.h"
14 #include "object-store.h"
15 #include "promisor-remote.h"
16 #include "shallow.h"
17
18 static int delta_base_offset = 1;
19 static int pack_kept_objects = -1;
20 static int write_bitmaps = -1;
21 static int use_delta_islands;
22 static char *packdir, *packtmp;
23
24 static const char *const git_repack_usage[] = {
25         N_("git repack [<options>]"),
26         NULL
27 };
28
29 static const char incremental_bitmap_conflict_error[] = N_(
30 "Incremental repacks are incompatible with bitmap indexes.  Use\n"
31 "--no-write-bitmap-index or disable the pack.writebitmaps configuration."
32 );
33
34
35 static int repack_config(const char *var, const char *value, void *cb)
36 {
37         if (!strcmp(var, "repack.usedeltabaseoffset")) {
38                 delta_base_offset = git_config_bool(var, value);
39                 return 0;
40         }
41         if (!strcmp(var, "repack.packkeptobjects")) {
42                 pack_kept_objects = git_config_bool(var, value);
43                 return 0;
44         }
45         if (!strcmp(var, "repack.writebitmaps") ||
46             !strcmp(var, "pack.writebitmaps")) {
47                 write_bitmaps = git_config_bool(var, value);
48                 return 0;
49         }
50         if (!strcmp(var, "repack.usedeltaislands")) {
51                 use_delta_islands = git_config_bool(var, value);
52                 return 0;
53         }
54         return git_default_config(var, value, cb);
55 }
56
57 /*
58  * Remove temporary $GIT_OBJECT_DIRECTORY/pack/.tmp-$$-pack-* files.
59  */
60 static void remove_temporary_files(void)
61 {
62         struct strbuf buf = STRBUF_INIT;
63         size_t dirlen, prefixlen;
64         DIR *dir;
65         struct dirent *e;
66
67         dir = opendir(packdir);
68         if (!dir)
69                 return;
70
71         /* Point at the slash at the end of ".../objects/pack/" */
72         dirlen = strlen(packdir) + 1;
73         strbuf_addstr(&buf, packtmp);
74         /* Hold the length of  ".tmp-%d-pack-" */
75         prefixlen = buf.len - dirlen;
76
77         while ((e = readdir(dir))) {
78                 if (strncmp(e->d_name, buf.buf + dirlen, prefixlen))
79                         continue;
80                 strbuf_setlen(&buf, dirlen);
81                 strbuf_addstr(&buf, e->d_name);
82                 unlink(buf.buf);
83         }
84         closedir(dir);
85         strbuf_release(&buf);
86 }
87
88 static void remove_pack_on_signal(int signo)
89 {
90         remove_temporary_files();
91         sigchain_pop(signo);
92         raise(signo);
93 }
94
95 /*
96  * Adds all packs hex strings to the fname list, which do not
97  * have a corresponding .keep file. These packs are not to
98  * be kept if we are going to pack everything into one file.
99  */
100 static void get_non_kept_pack_filenames(struct string_list *fname_list,
101                                         const struct string_list *extra_keep)
102 {
103         DIR *dir;
104         struct dirent *e;
105         char *fname;
106
107         if (!(dir = opendir(packdir)))
108                 return;
109
110         while ((e = readdir(dir)) != NULL) {
111                 size_t len;
112                 int i;
113
114                 for (i = 0; i < extra_keep->nr; i++)
115                         if (!fspathcmp(e->d_name, extra_keep->items[i].string))
116                                 break;
117                 if (extra_keep->nr > 0 && i < extra_keep->nr)
118                         continue;
119
120                 if (!strip_suffix(e->d_name, ".pack", &len))
121                         continue;
122
123                 fname = xmemdupz(e->d_name, len);
124
125                 if (!file_exists(mkpath("%s/%s.keep", packdir, fname)))
126                         string_list_append_nodup(fname_list, fname);
127                 else
128                         free(fname);
129         }
130         closedir(dir);
131 }
132
133 static void remove_redundant_pack(const char *dir_name, const char *base_name)
134 {
135         struct strbuf buf = STRBUF_INIT;
136         strbuf_addf(&buf, "%s/%s.pack", dir_name, base_name);
137         unlink_pack_path(buf.buf, 1);
138         strbuf_release(&buf);
139 }
140
141 struct pack_objects_args {
142         const char *window;
143         const char *window_memory;
144         const char *depth;
145         const char *threads;
146         const char *max_pack_size;
147         int no_reuse_delta;
148         int no_reuse_object;
149         int quiet;
150         int local;
151 };
152
153 static void prepare_pack_objects(struct child_process *cmd,
154                                  const struct pack_objects_args *args)
155 {
156         argv_array_push(&cmd->args, "pack-objects");
157         if (args->window)
158                 argv_array_pushf(&cmd->args, "--window=%s", args->window);
159         if (args->window_memory)
160                 argv_array_pushf(&cmd->args, "--window-memory=%s", args->window_memory);
161         if (args->depth)
162                 argv_array_pushf(&cmd->args, "--depth=%s", args->depth);
163         if (args->threads)
164                 argv_array_pushf(&cmd->args, "--threads=%s", args->threads);
165         if (args->max_pack_size)
166                 argv_array_pushf(&cmd->args, "--max-pack-size=%s", args->max_pack_size);
167         if (args->no_reuse_delta)
168                 argv_array_pushf(&cmd->args, "--no-reuse-delta");
169         if (args->no_reuse_object)
170                 argv_array_pushf(&cmd->args, "--no-reuse-object");
171         if (args->local)
172                 argv_array_push(&cmd->args,  "--local");
173         if (args->quiet)
174                 argv_array_push(&cmd->args,  "--quiet");
175         if (delta_base_offset)
176                 argv_array_push(&cmd->args,  "--delta-base-offset");
177         argv_array_push(&cmd->args, packtmp);
178         cmd->git_cmd = 1;
179         cmd->out = -1;
180 }
181
182 /*
183  * Write oid to the given struct child_process's stdin, starting it first if
184  * necessary.
185  */
186 static int write_oid(const struct object_id *oid, struct packed_git *pack,
187                      uint32_t pos, void *data)
188 {
189         struct child_process *cmd = data;
190
191         if (cmd->in == -1) {
192                 if (start_command(cmd))
193                         die(_("could not start pack-objects to repack promisor objects"));
194         }
195
196         xwrite(cmd->in, oid_to_hex(oid), the_hash_algo->hexsz);
197         xwrite(cmd->in, "\n", 1);
198         return 0;
199 }
200
201 static void repack_promisor_objects(const struct pack_objects_args *args,
202                                     struct string_list *names)
203 {
204         struct child_process cmd = CHILD_PROCESS_INIT;
205         FILE *out;
206         struct strbuf line = STRBUF_INIT;
207
208         prepare_pack_objects(&cmd, args);
209         cmd.in = -1;
210
211         /*
212          * NEEDSWORK: Giving pack-objects only the OIDs without any ordering
213          * hints may result in suboptimal deltas in the resulting pack. See if
214          * the OIDs can be sent with fake paths such that pack-objects can use a
215          * {type -> existing pack order} ordering when computing deltas instead
216          * of a {type -> size} ordering, which may produce better deltas.
217          */
218         for_each_packed_object(write_oid, &cmd,
219                                FOR_EACH_OBJECT_PROMISOR_ONLY);
220
221         if (cmd.in == -1)
222                 /* No packed objects; cmd was never started */
223                 return;
224
225         close(cmd.in);
226
227         out = xfdopen(cmd.out, "r");
228         while (strbuf_getline_lf(&line, out) != EOF) {
229                 char *promisor_name;
230                 int fd;
231                 if (line.len != the_hash_algo->hexsz)
232                         die(_("repack: Expecting full hex object ID lines only from pack-objects."));
233                 string_list_append(names, line.buf);
234
235                 /*
236                  * pack-objects creates the .pack and .idx files, but not the
237                  * .promisor file. Create the .promisor file, which is empty.
238                  *
239                  * NEEDSWORK: fetch-pack sometimes generates non-empty
240                  * .promisor files containing the ref names and associated
241                  * hashes at the point of generation of the corresponding
242                  * packfile, but this would not preserve their contents. Maybe
243                  * concatenate the contents of all .promisor files instead of
244                  * just creating a new empty file.
245                  */
246                 promisor_name = mkpathdup("%s-%s.promisor", packtmp,
247                                           line.buf);
248                 fd = open(promisor_name, O_CREAT|O_EXCL|O_WRONLY, 0600);
249                 if (fd < 0)
250                         die_errno(_("unable to create '%s'"), promisor_name);
251                 close(fd);
252                 free(promisor_name);
253         }
254         fclose(out);
255         if (finish_command(&cmd))
256                 die(_("could not finish pack-objects to repack promisor objects"));
257 }
258
259 #define ALL_INTO_ONE 1
260 #define LOOSEN_UNREACHABLE 2
261
262 int cmd_repack(int argc, const char **argv, const char *prefix)
263 {
264         struct {
265                 const char *name;
266                 unsigned optional:1;
267         } exts[] = {
268                 {".pack"},
269                 {".idx"},
270                 {".bitmap", 1},
271                 {".promisor", 1},
272         };
273         struct child_process cmd = CHILD_PROCESS_INIT;
274         struct string_list_item *item;
275         struct string_list names = STRING_LIST_INIT_DUP;
276         struct string_list rollback = STRING_LIST_INIT_NODUP;
277         struct string_list existing_packs = STRING_LIST_INIT_DUP;
278         struct strbuf line = STRBUF_INIT;
279         int i, ext, ret, failed;
280         FILE *out;
281
282         /* variables to be filled by option parsing */
283         int pack_everything = 0;
284         int delete_redundant = 0;
285         const char *unpack_unreachable = NULL;
286         int keep_unreachable = 0;
287         struct string_list keep_pack_list = STRING_LIST_INIT_NODUP;
288         int no_update_server_info = 0;
289         int midx_cleared = 0;
290         struct pack_objects_args po_args = {NULL};
291
292         struct option builtin_repack_options[] = {
293                 OPT_BIT('a', NULL, &pack_everything,
294                                 N_("pack everything in a single pack"), ALL_INTO_ONE),
295                 OPT_BIT('A', NULL, &pack_everything,
296                                 N_("same as -a, and turn unreachable objects loose"),
297                                    LOOSEN_UNREACHABLE | ALL_INTO_ONE),
298                 OPT_BOOL('d', NULL, &delete_redundant,
299                                 N_("remove redundant packs, and run git-prune-packed")),
300                 OPT_BOOL('f', NULL, &po_args.no_reuse_delta,
301                                 N_("pass --no-reuse-delta to git-pack-objects")),
302                 OPT_BOOL('F', NULL, &po_args.no_reuse_object,
303                                 N_("pass --no-reuse-object to git-pack-objects")),
304                 OPT_BOOL('n', NULL, &no_update_server_info,
305                                 N_("do not run git-update-server-info")),
306                 OPT__QUIET(&po_args.quiet, N_("be quiet")),
307                 OPT_BOOL('l', "local", &po_args.local,
308                                 N_("pass --local to git-pack-objects")),
309                 OPT_BOOL('b', "write-bitmap-index", &write_bitmaps,
310                                 N_("write bitmap index")),
311                 OPT_BOOL('i', "delta-islands", &use_delta_islands,
312                                 N_("pass --delta-islands to git-pack-objects")),
313                 OPT_STRING(0, "unpack-unreachable", &unpack_unreachable, N_("approxidate"),
314                                 N_("with -A, do not loosen objects older than this")),
315                 OPT_BOOL('k', "keep-unreachable", &keep_unreachable,
316                                 N_("with -a, repack unreachable objects")),
317                 OPT_STRING(0, "window", &po_args.window, N_("n"),
318                                 N_("size of the window used for delta compression")),
319                 OPT_STRING(0, "window-memory", &po_args.window_memory, N_("bytes"),
320                                 N_("same as the above, but limit memory size instead of entries count")),
321                 OPT_STRING(0, "depth", &po_args.depth, N_("n"),
322                                 N_("limits the maximum delta depth")),
323                 OPT_STRING(0, "threads", &po_args.threads, N_("n"),
324                                 N_("limits the maximum number of threads")),
325                 OPT_STRING(0, "max-pack-size", &po_args.max_pack_size, N_("bytes"),
326                                 N_("maximum size of each packfile")),
327                 OPT_BOOL(0, "pack-kept-objects", &pack_kept_objects,
328                                 N_("repack objects in packs marked with .keep")),
329                 OPT_STRING_LIST(0, "keep-pack", &keep_pack_list, N_("name"),
330                                 N_("do not repack this pack")),
331                 OPT_END()
332         };
333
334         git_config(repack_config, NULL);
335
336         argc = parse_options(argc, argv, prefix, builtin_repack_options,
337                                 git_repack_usage, 0);
338
339         if (delete_redundant && repository_format_precious_objects)
340                 die(_("cannot delete packs in a precious-objects repo"));
341
342         if (keep_unreachable &&
343             (unpack_unreachable || (pack_everything & LOOSEN_UNREACHABLE)))
344                 die(_("--keep-unreachable and -A are incompatible"));
345
346         if (write_bitmaps < 0) {
347                 if (!(pack_everything & ALL_INTO_ONE) ||
348                     !is_bare_repository())
349                         write_bitmaps = 0;
350         }
351         if (pack_kept_objects < 0)
352                 pack_kept_objects = write_bitmaps > 0;
353
354         if (write_bitmaps && !(pack_everything & ALL_INTO_ONE))
355                 die(_(incremental_bitmap_conflict_error));
356
357         packdir = mkpathdup("%s/pack", get_object_directory());
358         packtmp = mkpathdup("%s/.tmp-%d-pack", packdir, (int)getpid());
359
360         sigchain_push_common(remove_pack_on_signal);
361
362         prepare_pack_objects(&cmd, &po_args);
363
364         argv_array_push(&cmd.args, "--keep-true-parents");
365         if (!pack_kept_objects)
366                 argv_array_push(&cmd.args, "--honor-pack-keep");
367         for (i = 0; i < keep_pack_list.nr; i++)
368                 argv_array_pushf(&cmd.args, "--keep-pack=%s",
369                                  keep_pack_list.items[i].string);
370         argv_array_push(&cmd.args, "--non-empty");
371         argv_array_push(&cmd.args, "--all");
372         argv_array_push(&cmd.args, "--reflog");
373         argv_array_push(&cmd.args, "--indexed-objects");
374         if (has_promisor_remote())
375                 argv_array_push(&cmd.args, "--exclude-promisor-objects");
376         if (write_bitmaps > 0)
377                 argv_array_push(&cmd.args, "--write-bitmap-index");
378         else if (write_bitmaps < 0)
379                 argv_array_push(&cmd.args, "--write-bitmap-index-quiet");
380         if (use_delta_islands)
381                 argv_array_push(&cmd.args, "--delta-islands");
382
383         if (pack_everything & ALL_INTO_ONE) {
384                 get_non_kept_pack_filenames(&existing_packs, &keep_pack_list);
385
386                 repack_promisor_objects(&po_args, &names);
387
388                 if (existing_packs.nr && delete_redundant) {
389                         if (unpack_unreachable) {
390                                 argv_array_pushf(&cmd.args,
391                                                 "--unpack-unreachable=%s",
392                                                 unpack_unreachable);
393                                 argv_array_push(&cmd.env_array, "GIT_REF_PARANOIA=1");
394                         } else if (pack_everything & LOOSEN_UNREACHABLE) {
395                                 argv_array_push(&cmd.args,
396                                                 "--unpack-unreachable");
397                         } else if (keep_unreachable) {
398                                 argv_array_push(&cmd.args, "--keep-unreachable");
399                                 argv_array_push(&cmd.args, "--pack-loose-unreachable");
400                         } else {
401                                 argv_array_push(&cmd.env_array, "GIT_REF_PARANOIA=1");
402                         }
403                 }
404         } else {
405                 argv_array_push(&cmd.args, "--unpacked");
406                 argv_array_push(&cmd.args, "--incremental");
407         }
408
409         cmd.no_stdin = 1;
410
411         ret = start_command(&cmd);
412         if (ret)
413                 return ret;
414
415         out = xfdopen(cmd.out, "r");
416         while (strbuf_getline_lf(&line, out) != EOF) {
417                 if (line.len != the_hash_algo->hexsz)
418                         die(_("repack: Expecting full hex object ID lines only from pack-objects."));
419                 string_list_append(&names, line.buf);
420         }
421         fclose(out);
422         ret = finish_command(&cmd);
423         if (ret)
424                 return ret;
425
426         if (!names.nr && !po_args.quiet)
427                 printf_ln(_("Nothing new to pack."));
428
429         close_object_store(the_repository->objects);
430
431         /*
432          * Ok we have prepared all new packfiles.
433          * First see if there are packs of the same name and if so
434          * if we can move them out of the way (this can happen if we
435          * repacked immediately after packing fully.
436          */
437         failed = 0;
438         for_each_string_list_item(item, &names) {
439                 for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
440                         char *fname, *fname_old;
441
442                         if (!midx_cleared) {
443                                 clear_midx_file(the_repository);
444                                 midx_cleared = 1;
445                         }
446
447                         fname = mkpathdup("%s/pack-%s%s", packdir,
448                                                 item->string, exts[ext].name);
449                         if (!file_exists(fname)) {
450                                 free(fname);
451                                 continue;
452                         }
453
454                         fname_old = mkpathdup("%s/old-%s%s", packdir,
455                                                 item->string, exts[ext].name);
456                         if (file_exists(fname_old))
457                                 if (unlink(fname_old))
458                                         failed = 1;
459
460                         if (!failed && rename(fname, fname_old)) {
461                                 free(fname);
462                                 free(fname_old);
463                                 failed = 1;
464                                 break;
465                         } else {
466                                 string_list_append(&rollback, fname);
467                                 free(fname_old);
468                         }
469                 }
470                 if (failed)
471                         break;
472         }
473         if (failed) {
474                 struct string_list rollback_failure = STRING_LIST_INIT_DUP;
475                 for_each_string_list_item(item, &rollback) {
476                         char *fname, *fname_old;
477                         fname = mkpathdup("%s/%s", packdir, item->string);
478                         fname_old = mkpathdup("%s/old-%s", packdir, item->string);
479                         if (rename(fname_old, fname))
480                                 string_list_append(&rollback_failure, fname);
481                         free(fname);
482                         free(fname_old);
483                 }
484
485                 if (rollback_failure.nr) {
486                         int i;
487                         fprintf(stderr,
488                                 _("WARNING: Some packs in use have been renamed by\n"
489                                   "WARNING: prefixing old- to their name, in order to\n"
490                                   "WARNING: replace them with the new version of the\n"
491                                   "WARNING: file.  But the operation failed, and the\n"
492                                   "WARNING: attempt to rename them back to their\n"
493                                   "WARNING: original names also failed.\n"
494                                   "WARNING: Please rename them in %s manually:\n"), packdir);
495                         for (i = 0; i < rollback_failure.nr; i++)
496                                 fprintf(stderr, "WARNING:   old-%s -> %s\n",
497                                         rollback_failure.items[i].string,
498                                         rollback_failure.items[i].string);
499                 }
500                 exit(1);
501         }
502
503         /* Now the ones with the same name are out of the way... */
504         for_each_string_list_item(item, &names) {
505                 for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
506                         char *fname, *fname_old;
507                         struct stat statbuffer;
508                         int exists = 0;
509                         fname = mkpathdup("%s/pack-%s%s",
510                                         packdir, item->string, exts[ext].name);
511                         fname_old = mkpathdup("%s-%s%s",
512                                         packtmp, item->string, exts[ext].name);
513                         if (!stat(fname_old, &statbuffer)) {
514                                 statbuffer.st_mode &= ~(S_IWUSR | S_IWGRP | S_IWOTH);
515                                 chmod(fname_old, statbuffer.st_mode);
516                                 exists = 1;
517                         }
518                         if (exists || !exts[ext].optional) {
519                                 if (rename(fname_old, fname))
520                                         die_errno(_("renaming '%s' failed"), fname_old);
521                         }
522                         free(fname);
523                         free(fname_old);
524                 }
525         }
526
527         /* Remove the "old-" files */
528         for_each_string_list_item(item, &names) {
529                 for (ext = 0; ext < ARRAY_SIZE(exts); ext++) {
530                         char *fname;
531                         fname = mkpathdup("%s/old-%s%s",
532                                           packdir,
533                                           item->string,
534                                           exts[ext].name);
535                         if (remove_path(fname))
536                                 warning(_("failed to remove '%s'"), fname);
537                         free(fname);
538                 }
539         }
540
541         /* End of pack replacement. */
542
543         reprepare_packed_git(the_repository);
544
545         if (delete_redundant) {
546                 const int hexsz = the_hash_algo->hexsz;
547                 int opts = 0;
548                 string_list_sort(&names);
549                 for_each_string_list_item(item, &existing_packs) {
550                         char *sha1;
551                         size_t len = strlen(item->string);
552                         if (len < hexsz)
553                                 continue;
554                         sha1 = item->string + len - hexsz;
555                         if (!string_list_has_string(&names, sha1))
556                                 remove_redundant_pack(packdir, item->string);
557                 }
558                 if (!po_args.quiet && isatty(2))
559                         opts |= PRUNE_PACKED_VERBOSE;
560                 prune_packed_objects(opts);
561
562                 if (!keep_unreachable &&
563                     (!(pack_everything & LOOSEN_UNREACHABLE) ||
564                      unpack_unreachable) &&
565                     is_repository_shallow(the_repository))
566                         prune_shallow(PRUNE_QUICK);
567         }
568
569         if (!no_update_server_info)
570                 update_server_info(0);
571         remove_temporary_files();
572
573         if (git_env_bool(GIT_TEST_MULTI_PACK_INDEX, 0))
574                 write_midx_file(get_object_directory(), 0);
575
576         string_list_clear(&names, 0);
577         string_list_clear(&rollback, 0);
578         string_list_clear(&existing_packs, 0);
579         strbuf_release(&line);
580
581         return 0;
582 }