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