]> git.ipfire.org Git - thirdparty/git.git/blame - object-file.c
refs: print error message in debug output
[thirdparty/git.git] / object-file.c
CommitLineData
0fcfd160
LT
1/*
2 * GIT - The information manager from hell
3 *
4 * Copyright (C) Linus Torvalds, 2005
5 *
e5afd444 6 * This handles basic git object files - packing, unpacking,
0fcfd160
LT
7 * creation etc.
8 */
0fcfd160 9#include "cache.h"
b2141fc1 10#include "config.h"
6eac50d8 11#include "string-list.h"
697cc8ef 12#include "lockfile.h"
1f688557 13#include "delta.h"
a733cb60 14#include "pack.h"
8e440259
PE
15#include "blob.h"
16#include "commit.h"
4dd1fbc7 17#include "run-command.h"
8e440259
PE
18#include "tag.h"
19#include "tree.h"
c879daa2 20#include "tree-walk.h"
f35a6d3b 21#include "refs.h"
70f5d5d3 22#include "pack-revindex.h"
bc626927 23#include "hash-lookup.h"
568508e7 24#include "bulk-checkin.h"
031dc927 25#include "repository.h"
47f351e9 26#include "replace-object.h"
090ea126 27#include "streaming.h"
543c5caa 28#include "dir.h"
12d95ef6 29#include "list.h"
c4c6effa 30#include "mergesort.h"
cf3c6352 31#include "quote.h"
4f39cd82 32#include "packfile.h"
90c62155 33#include "object-store.h"
b14ed5ad 34#include "promisor-remote.h"
a35e03de 35#include "submodule.h"
e05db0fd 36
1af64f73 37/* The maximum size for an object header. */
38#define MAX_HEADER_LEN 32
39
e1ccd7e2 40
41#define EMPTY_TREE_SHA1_BIN_LITERAL \
42 "\x4b\x82\x5d\xc6\x42\xcb\x6e\xb9\xa0\x60" \
43 "\xe5\x4b\xf8\xd6\x92\x88\xfb\xee\x49\x04"
13eeedb5 44#define EMPTY_TREE_SHA256_BIN_LITERAL \
45 "\x6e\xf1\x9b\x41\x22\x5c\x53\x69\xf1\xc1" \
46 "\x04\xd4\x5d\x8d\x85\xef\xa9\xb0\x57\xb5" \
47 "\x3b\x14\xb4\xb9\xb9\x39\xdd\x74\xde\xcc" \
48 "\x53\x21"
e1ccd7e2 49
50#define EMPTY_BLOB_SHA1_BIN_LITERAL \
51 "\xe6\x9d\xe2\x9b\xb2\xd1\xd6\x43\x4b\x8b" \
52 "\x29\xae\x77\x5a\xd8\xc2\xe4\x8c\x53\x91"
13eeedb5 53#define EMPTY_BLOB_SHA256_BIN_LITERAL \
54 "\x47\x3a\x0f\x4c\x3b\xe8\xa9\x36\x81\xa2" \
55 "\x67\xe3\xb1\xe9\xa7\xdc\xda\x11\x85\x43" \
56 "\x6f\xe1\x41\xf7\x74\x91\x20\xa3\x03\x72" \
57 "\x18\x13"
e1ccd7e2 58
e1ccd7e2 59static const struct object_id empty_tree_oid = {
5a6dce70 60 .hash = EMPTY_TREE_SHA1_BIN_LITERAL,
61 .algo = GIT_HASH_SHA1,
8576fde6 62};
e1ccd7e2 63static const struct object_id empty_blob_oid = {
5a6dce70 64 .hash = EMPTY_BLOB_SHA1_BIN_LITERAL,
65 .algo = GIT_HASH_SHA1,
8576fde6 66};
14228447 67static const struct object_id null_oid_sha1 = {
68 .hash = {0},
69 .algo = GIT_HASH_SHA1,
70};
13eeedb5 71static const struct object_id empty_tree_oid_sha256 = {
5a6dce70 72 .hash = EMPTY_TREE_SHA256_BIN_LITERAL,
73 .algo = GIT_HASH_SHA256,
13eeedb5 74};
75static const struct object_id empty_blob_oid_sha256 = {
5a6dce70 76 .hash = EMPTY_BLOB_SHA256_BIN_LITERAL,
77 .algo = GIT_HASH_SHA256,
13eeedb5 78};
14228447 79static const struct object_id null_oid_sha256 = {
80 .hash = {0},
81 .algo = GIT_HASH_SHA256,
82};
88cd621d 83
ac73cedf 84static void git_hash_sha1_init(git_hash_ctx *ctx)
f50e766b 85{
ac73cedf 86 git_SHA1_Init(&ctx->sha1);
f50e766b 87}
88
768e30ea 89static void git_hash_sha1_clone(git_hash_ctx *dst, const git_hash_ctx *src)
90{
91 git_SHA1_Clone(&dst->sha1, &src->sha1);
92}
93
ac73cedf 94static void git_hash_sha1_update(git_hash_ctx *ctx, const void *data, size_t len)
f50e766b 95{
ac73cedf 96 git_SHA1_Update(&ctx->sha1, data, len);
f50e766b 97}
98
ac73cedf 99static void git_hash_sha1_final(unsigned char *hash, git_hash_ctx *ctx)
f50e766b 100{
ac73cedf 101 git_SHA1_Final(hash, &ctx->sha1);
f50e766b 102}
103
ab795f0d 104static void git_hash_sha1_final_oid(struct object_id *oid, git_hash_ctx *ctx)
105{
106 git_SHA1_Final(oid->hash, &ctx->sha1);
107 memset(oid->hash + GIT_SHA1_RAWSZ, 0, GIT_MAX_RAWSZ - GIT_SHA1_RAWSZ);
5a6dce70 108 oid->algo = GIT_HASH_SHA1;
ab795f0d 109}
110
13eeedb5 111
112static void git_hash_sha256_init(git_hash_ctx *ctx)
113{
114 git_SHA256_Init(&ctx->sha256);
115}
116
768e30ea 117static void git_hash_sha256_clone(git_hash_ctx *dst, const git_hash_ctx *src)
118{
119 git_SHA256_Clone(&dst->sha256, &src->sha256);
120}
121
13eeedb5 122static void git_hash_sha256_update(git_hash_ctx *ctx, const void *data, size_t len)
123{
124 git_SHA256_Update(&ctx->sha256, data, len);
125}
126
127static void git_hash_sha256_final(unsigned char *hash, git_hash_ctx *ctx)
128{
129 git_SHA256_Final(hash, &ctx->sha256);
130}
131
ab795f0d 132static void git_hash_sha256_final_oid(struct object_id *oid, git_hash_ctx *ctx)
133{
134 git_SHA256_Final(oid->hash, &ctx->sha256);
135 /*
136 * This currently does nothing, so the compiler should optimize it out,
137 * but keep it in case we extend the hash size again.
138 */
139 memset(oid->hash + GIT_SHA256_RAWSZ, 0, GIT_MAX_RAWSZ - GIT_SHA256_RAWSZ);
5a6dce70 140 oid->algo = GIT_HASH_SHA256;
ab795f0d 141}
142
ac73cedf 143static void git_hash_unknown_init(git_hash_ctx *ctx)
f50e766b 144{
1a07e59c 145 BUG("trying to init unknown hash");
f50e766b 146}
147
768e30ea 148static void git_hash_unknown_clone(git_hash_ctx *dst, const git_hash_ctx *src)
149{
150 BUG("trying to clone unknown hash");
151}
152
ac73cedf 153static void git_hash_unknown_update(git_hash_ctx *ctx, const void *data, size_t len)
f50e766b 154{
1a07e59c 155 BUG("trying to update unknown hash");
f50e766b 156}
157
ac73cedf 158static void git_hash_unknown_final(unsigned char *hash, git_hash_ctx *ctx)
f50e766b 159{
1a07e59c 160 BUG("trying to finalize unknown hash");
f50e766b 161}
162
ab795f0d 163static void git_hash_unknown_final_oid(struct object_id *oid, git_hash_ctx *ctx)
164{
165 BUG("trying to finalize unknown hash");
166}
167
f50e766b 168const struct git_hash_algo hash_algos[GIT_HASH_NALGOS] = {
169 {
170 NULL,
171 0x00000000,
172 0,
173 0,
a2ce0a75 174 0,
f50e766b 175 git_hash_unknown_init,
768e30ea 176 git_hash_unknown_clone,
f50e766b 177 git_hash_unknown_update,
178 git_hash_unknown_final,
ab795f0d 179 git_hash_unknown_final_oid,
f50e766b 180 NULL,
181 NULL,
14228447 182 NULL,
f50e766b 183 },
184 {
1ccf07cb 185 "sha1",
27f3796a 186 GIT_SHA1_FORMAT_ID,
f50e766b 187 GIT_SHA1_RAWSZ,
188 GIT_SHA1_HEXSZ,
a2ce0a75 189 GIT_SHA1_BLKSZ,
f50e766b 190 git_hash_sha1_init,
768e30ea 191 git_hash_sha1_clone,
f50e766b 192 git_hash_sha1_update,
193 git_hash_sha1_final,
ab795f0d 194 git_hash_sha1_final_oid,
f50e766b 195 &empty_tree_oid,
196 &empty_blob_oid,
14228447 197 &null_oid_sha1,
f50e766b 198 },
13eeedb5 199 {
200 "sha256",
27f3796a 201 GIT_SHA256_FORMAT_ID,
13eeedb5 202 GIT_SHA256_RAWSZ,
203 GIT_SHA256_HEXSZ,
204 GIT_SHA256_BLKSZ,
205 git_hash_sha256_init,
768e30ea 206 git_hash_sha256_clone,
13eeedb5 207 git_hash_sha256_update,
208 git_hash_sha256_final,
ab795f0d 209 git_hash_sha256_final_oid,
13eeedb5 210 &empty_tree_oid_sha256,
211 &empty_blob_oid_sha256,
14228447 212 &null_oid_sha256,
13eeedb5 213 }
f50e766b 214};
215
14228447 216const struct object_id *null_oid(void)
217{
218 return the_hash_algo->null_oid;
219}
220
d8a92ced 221const char *empty_tree_oid_hex(void)
222{
223 static char buf[GIT_MAX_HEXSZ + 1];
224 return oid_to_hex_r(buf, the_hash_algo->empty_tree);
225}
226
227const char *empty_blob_oid_hex(void)
228{
229 static char buf[GIT_MAX_HEXSZ + 1];
230 return oid_to_hex_r(buf, the_hash_algo->empty_blob);
231}
232
2f90b9d9 233int hash_algo_by_name(const char *name)
234{
235 int i;
236 if (!name)
237 return GIT_HASH_UNKNOWN;
238 for (i = 1; i < GIT_HASH_NALGOS; i++)
239 if (!strcmp(name, hash_algos[i].name))
240 return i;
241 return GIT_HASH_UNKNOWN;
242}
243
244int hash_algo_by_id(uint32_t format_id)
245{
246 int i;
247 for (i = 1; i < GIT_HASH_NALGOS; i++)
248 if (format_id == hash_algos[i].format_id)
249 return i;
250 return GIT_HASH_UNKNOWN;
251}
252
95399788 253int hash_algo_by_length(int len)
254{
255 int i;
256 for (i = 1; i < GIT_HASH_NALGOS; i++)
257 if (len == hash_algos[i].rawsz)
258 return i;
259 return GIT_HASH_UNKNOWN;
260}
2f90b9d9 261
c597ba80
NTND
262/*
263 * This is meant to hold a *small* number of objects that you would
cb1c8d1d 264 * want read_object_file() to be able to return, but yet you do not want
c597ba80
NTND
265 * to write them into the object store (e.g. a browse-only
266 * application).
267 */
268static struct cached_object {
62ba93ea 269 struct object_id oid;
c597ba80
NTND
270 enum object_type type;
271 void *buf;
272 unsigned long size;
273} *cached_objects;
274static int cached_object_nr, cached_object_alloc;
275
276static struct cached_object empty_tree = {
62ba93ea 277 { EMPTY_TREE_SHA1_BIN_LITERAL },
c597ba80
NTND
278 OBJ_TREE,
279 "",
280 0
281};
282
62ba93ea 283static struct cached_object *find_cached_object(const struct object_id *oid)
c597ba80
NTND
284{
285 int i;
286 struct cached_object *co = cached_objects;
287
288 for (i = 0; i < cached_object_nr; i++, co++) {
4a7e27e9 289 if (oideq(&co->oid, oid))
c597ba80
NTND
290 return co;
291 }
4a7e27e9 292 if (oideq(oid, the_hash_algo->empty_tree))
c597ba80
NTND
293 return &empty_tree;
294 return NULL;
295}
296
9472935d 297
8462ff43 298static int get_conv_flags(unsigned flags)
9472935d
TB
299{
300 if (flags & HASH_RENORMALIZE)
8462ff43 301 return CONV_EOL_RENORMALIZE;
9472935d 302 else if (flags & HASH_WRITE_OBJECT)
107642fe 303 return global_conv_flags_eol | CONV_WRITE_OBJECT;
9472935d 304 else
8462ff43 305 return 0;
9472935d
TB
306}
307
308
90a6464b
JH
309int mkdir_in_gitdir(const char *path)
310{
311 if (mkdir(path, 0777)) {
312 int saved_errno = errno;
313 struct stat st;
314 struct strbuf sb = STRBUF_INIT;
315
316 if (errno != EEXIST)
317 return -1;
318 /*
319 * Are we looking at a path in a symlinked worktree
320 * whose original repository does not yet have it?
321 * e.g. .git/rr-cache pointing at its original
322 * repository in which the user hasn't performed any
323 * conflict resolution yet?
324 */
325 if (lstat(path, &st) || !S_ISLNK(st.st_mode) ||
326 strbuf_readlink(&sb, path, st.st_size) ||
327 !is_absolute_path(sb.buf) ||
328 mkdir(sb.buf, 0777)) {
329 strbuf_release(&sb);
330 errno = saved_errno;
331 return -1;
332 }
333 strbuf_release(&sb);
334 }
335 return adjust_shared_perm(path);
336}
337
eb3c027e 338static enum scld_error safe_create_leading_directories_1(char *path, int share)
b2cb9425 339{
26c8ae2a 340 char *next_component = path + offset_1st_component(path);
0be0521b 341 enum scld_error ret = SCLD_OK;
67d42212 342
0be0521b 343 while (ret == SCLD_OK && next_component) {
f0502332 344 struct stat st;
0f527403 345 char *slash = next_component, slash_character;
f0502332 346
0f527403
MH
347 while (*slash && !is_dir_sep(*slash))
348 slash++;
349
350 if (!*slash)
b2cb9425 351 break;
bf10cf70 352
26c8ae2a 353 next_component = slash + 1;
0f527403 354 while (is_dir_sep(*next_component))
bf10cf70 355 next_component++;
26c8ae2a 356 if (!*next_component)
5f0bdf50 357 break;
831651fd 358
0f527403 359 slash_character = *slash;
831651fd 360 *slash = '\0';
67d42212
JR
361 if (!stat(path, &st)) {
362 /* path exists */
204a047f
MH
363 if (!S_ISDIR(st.st_mode)) {
364 errno = ENOTDIR;
0be0521b 365 ret = SCLD_EXISTS;
204a047f 366 }
53a39721 367 } else if (mkdir(path, 0777)) {
928734d9 368 if (errno == EEXIST &&
9e6f885d 369 !stat(path, &st) && S_ISDIR(st.st_mode))
928734d9 370 ; /* somebody created it since we checked */
18d37e86
MH
371 else if (errno == ENOENT)
372 /*
373 * Either mkdir() failed because
374 * somebody just pruned the containing
375 * directory, or stat() failed because
376 * the file that was in our way was
377 * just removed. Either way, inform
378 * the caller that it might be worth
379 * trying again:
380 */
381 ret = SCLD_VANISHED;
9e6f885d 382 else
0be0521b 383 ret = SCLD_FAILED;
eb3c027e 384 } else if (share && adjust_shared_perm(path)) {
0be0521b 385 ret = SCLD_PERMS;
457f06d6 386 }
0f527403 387 *slash = slash_character;
b2cb9425 388 }
9e6f885d 389 return ret;
b2cb9425 390}
723c31fe 391
eb3c027e
MT
392enum scld_error safe_create_leading_directories(char *path)
393{
394 return safe_create_leading_directories_1(path, 1);
395}
396
397enum scld_error safe_create_leading_directories_no_share(char *path)
398{
399 return safe_create_leading_directories_1(path, 0);
400}
401
0be0521b 402enum scld_error safe_create_leading_directories_const(const char *path)
8e21d63b 403{
02944307 404 int save_errno;
8e21d63b
JK
405 /* path points to cache entries, so xstrdup before messing with it */
406 char *buf = xstrdup(path);
0be0521b 407 enum scld_error result = safe_create_leading_directories(buf);
02944307
MH
408
409 save_errno = errno;
8e21d63b 410 free(buf);
02944307 411 errno = save_errno;
8e21d63b
JK
412 return result;
413}
414
514c5fdd 415static void fill_loose_path(struct strbuf *buf, const struct object_id *oid)
ace1534d
JH
416{
417 int i;
94b5e093 418 for (i = 0; i < the_hash_algo->rawsz; i++) {
ace1534d 419 static char hex[] = "0123456789abcdef";
514c5fdd 420 unsigned int val = oid->hash[i];
f7b7774f
JK
421 strbuf_addch(buf, hex[val >> 4]);
422 strbuf_addch(buf, hex[val & 0xf]);
afbba2f0 423 if (!i)
f7b7774f 424 strbuf_addch(buf, '/');
ace1534d
JH
425 }
426}
427
f0eaf638
JK
428static const char *odb_loose_path(struct object_directory *odb,
429 struct strbuf *buf,
514c5fdd 430 const struct object_id *oid)
0fcfd160 431{
b69fb867 432 strbuf_reset(buf);
f0eaf638 433 strbuf_addstr(buf, odb->path);
34498471 434 strbuf_addch(buf, '/');
514c5fdd 435 fill_loose_path(buf, oid);
f3f043a1 436 return buf->buf;
0fcfd160
LT
437}
438
f3f043a1 439const char *loose_object_path(struct repository *r, struct strbuf *buf,
514c5fdd 440 const struct object_id *oid)
29ec6af2 441{
514c5fdd 442 return odb_loose_path(r->objects->odb, buf, oid);
0fcfd160
LT
443}
444
4ea82473
JK
445/*
446 * Return non-zero iff the path is usable as an alternate object database.
447 */
13313fc3
SB
448static int alt_odb_usable(struct raw_object_store *o,
449 struct strbuf *path,
cf2dc1c2 450 const char *normalized_objdir, khiter_t *pos)
4ea82473 451{
cf2dc1c2 452 int r;
4ea82473
JK
453
454 /* Detect cases where alternate disappeared */
455 if (!is_directory(path->buf)) {
259328b7
NTND
456 error(_("object directory %s does not exist; "
457 "check .git/objects/info/alternates"),
4ea82473
JK
458 path->buf);
459 return 0;
460 }
461
462 /*
463 * Prevent the common mistake of listing the same
464 * thing twice, or object directory itself.
465 */
cf2dc1c2
EW
466 if (!o->odb_by_path) {
467 khiter_t p;
468
469 o->odb_by_path = kh_init_odb_path_map();
470 assert(!o->odb->next);
471 p = kh_put_odb_path_map(o->odb_by_path, o->odb->path, &r);
472 assert(r == 1); /* never used */
473 kh_value(o->odb_by_path, p) = o->odb;
4ea82473 474 }
cf2dc1c2 475 if (fspatheq(path->buf, normalized_objdir))
4ea82473 476 return 0;
cf2dc1c2
EW
477 *pos = kh_put_odb_path_map(o->odb_by_path, path->buf, &r);
478 /* r: 0 = exists, 1 = never used, 2 = deleted */
479 return r == 0 ? 0 : 1;
4ea82473
JK
480}
481
ddd5d056
JH
482/*
483 * Prepare alternate object database registry.
d5a63b99
JH
484 *
485 * The variable alt_odb_list points at the list of struct
263db403 486 * object_directory. The elements on this list come from
d5a63b99
JH
487 * non-empty elements from colon separated ALTERNATE_DB_ENVIRONMENT
488 * environment variable, and $GIT_OBJECT_DIRECTORY/info/alternates,
1494e038
JH
489 * whose contents is similar to that environment variable but can be
490 * LF separated. Its base points at a statically allocated buffer that
d5a63b99
JH
491 * contains "/the/directory/corresponding/to/.git/objects/...", while
492 * its name points just after the slash at the end of ".git/objects/"
e5afd444
493 * in the example above, and has enough space to hold all hex characters
494 * of the object ID, an extra slash for the first level indirection, and
495 * the terminating NUL.
ddd5d056 496 */
77f012e8
SB
497static void read_info_alternates(struct repository *r,
498 const char *relative_base,
499 int depth);
407532f8 500static int link_alt_odb_entry(struct repository *r, const struct strbuf *entry,
cfc62fc9 501 const char *relative_base, int depth, const char *normalized_objdir)
ace1534d 502{
263db403 503 struct object_directory *ent;
5bdf0a84 504 struct strbuf pathbuf = STRBUF_INIT;
cf2dc1c2 505 khiter_t pos;
d5a63b99 506
407532f8 507 if (!is_absolute_path(entry->buf) && relative_base) {
4ac9006f 508 strbuf_realpath(&pathbuf, relative_base, 1);
5bdf0a84 509 strbuf_addch(&pathbuf, '/');
c2f493a4 510 }
407532f8 511 strbuf_addbuf(&pathbuf, entry);
c2f493a4 512
37a95862 513 if (strbuf_normalize_path(&pathbuf) < 0 && relative_base) {
259328b7 514 error(_("unable to normalize alternate object path: %s"),
670c359d
JK
515 pathbuf.buf);
516 strbuf_release(&pathbuf);
517 return -1;
518 }
5bdf0a84
HW
519
520 /*
521 * The trailing slash after the directory name is given by
522 * this function at the end. Remove duplicates.
523 */
4ea82473
JK
524 while (pathbuf.len && pathbuf.buf[pathbuf.len - 1] == '/')
525 strbuf_setlen(&pathbuf, pathbuf.len - 1);
5bdf0a84 526
cf2dc1c2 527 if (!alt_odb_usable(r->objects, &pathbuf, normalized_objdir, &pos)) {
4ea82473 528 strbuf_release(&pathbuf);
c2f493a4
MW
529 return -1;
530 }
531
ca56dadb 532 CALLOC_ARRAY(ent, 1);
cf2dc1c2
EW
533 /* pathbuf.buf is already in r->objects->odb_by_path */
534 ent->path = strbuf_detach(&pathbuf, NULL);
c2f493a4
MW
535
536 /* add the alternate entry */
f0eaf638
JK
537 *r->objects->odb_tail = ent;
538 r->objects->odb_tail = &(ent->next);
c2f493a4 539 ent->next = NULL;
cf2dc1c2
EW
540 assert(r->objects->odb_by_path);
541 kh_value(r->objects->odb_by_path, pos) = ent;
c2f493a4
MW
542
543 /* recursively add alternates */
cf2dc1c2 544 read_info_alternates(r, ent->path, depth + 1);
c2f493a4
MW
545
546 return 0;
547}
548
cf3c6352
JK
549static const char *parse_alt_odb_entry(const char *string,
550 int sep,
551 struct strbuf *out)
552{
553 const char *end;
554
555 strbuf_reset(out);
556
557 if (*string == '#') {
558 /* comment; consume up to next separator */
559 end = strchrnul(string, sep);
560 } else if (*string == '"' && !unquote_c_style(out, string, &end)) {
561 /*
562 * quoted path; unquote_c_style has copied the
563 * data for us and set "end". Broken quoting (e.g.,
564 * an entry that doesn't end with a quote) falls
565 * back to the unquoted case below.
566 */
567 } else {
568 /* normal, unquoted path */
569 end = strchrnul(string, sep);
570 strbuf_add(out, string, end - string);
571 }
572
573 if (*end)
574 end++;
575 return end;
576}
577
77f012e8
SB
578static void link_alt_odb_entries(struct repository *r, const char *alt,
579 int sep, const char *relative_base, int depth)
c2f493a4 580{
539e7506 581 struct strbuf objdirbuf = STRBUF_INIT;
cf3c6352 582 struct strbuf entry = STRBUF_INIT;
c2f493a4 583
f28e3668
JK
584 if (!alt || !*alt)
585 return;
586
c2f493a4 587 if (depth > 5) {
259328b7 588 error(_("%s: ignoring alternate object stores, nesting too deep"),
c2f493a4
MW
589 relative_base);
590 return;
591 }
592
f0eaf638 593 strbuf_add_absolute_path(&objdirbuf, r->objects->odb->path);
670c359d 594 if (strbuf_normalize_path(&objdirbuf) < 0)
259328b7 595 die(_("unable to normalize object directory: %s"),
670c359d 596 objdirbuf.buf);
539e7506 597
cf3c6352
JK
598 while (*alt) {
599 alt = parse_alt_odb_entry(alt, sep, &entry);
600 if (!entry.len)
9577e7e3 601 continue;
407532f8 602 link_alt_odb_entry(r, &entry,
cfc62fc9 603 relative_base, depth, objdirbuf.buf);
9577e7e3 604 }
cf3c6352 605 strbuf_release(&entry);
539e7506 606 strbuf_release(&objdirbuf);
d5a63b99
JH
607}
608
77f012e8
SB
609static void read_info_alternates(struct repository *r,
610 const char *relative_base,
611 int depth)
d5a63b99 612{
5015f01c 613 char *path;
dc732bd5 614 struct strbuf buf = STRBUF_INIT;
d5a63b99 615
5015f01c 616 path = xstrfmt("%s/info/alternates", relative_base);
dc732bd5 617 if (strbuf_read_file(&buf, path, 1024) < 0) {
f0f7bebe 618 warn_on_fopen_errors(path);
dc732bd5 619 free(path);
9a217f2a 620 return;
ace1534d 621 }
d5a63b99 622
77f012e8 623 link_alt_odb_entries(r, buf.buf, '\n', relative_base, depth);
dc732bd5
JK
624 strbuf_release(&buf);
625 free(path);
ace1534d
JH
626}
627
bef70b22
DB
628void add_to_alternates_file(const char *reference)
629{
f132a127 630 struct lock_file lock = LOCK_INIT;
77b9b1d1
JK
631 char *alts = git_pathdup("objects/info/alternates");
632 FILE *in, *out;
f132a127 633 int found = 0;
77b9b1d1 634
f132a127
635 hold_lock_file_for_update(&lock, alts, LOCK_DIE_ON_ERROR);
636 out = fdopen_lock_file(&lock, "w");
77b9b1d1 637 if (!out)
259328b7 638 die_errno(_("unable to fdopen alternates lockfile"));
77b9b1d1
JK
639
640 in = fopen(alts, "r");
641 if (in) {
642 struct strbuf line = STRBUF_INIT;
77b9b1d1 643
3f163962 644 while (strbuf_getline(&line, in) != EOF) {
77b9b1d1
JK
645 if (!strcmp(reference, line.buf)) {
646 found = 1;
647 break;
648 }
649 fprintf_or_die(out, "%s\n", line.buf);
650 }
651
652 strbuf_release(&line);
653 fclose(in);
77b9b1d1
JK
654 }
655 else if (errno != ENOENT)
259328b7 656 die_errno(_("unable to read alternates file"));
77b9b1d1 657
f132a127
658 if (found) {
659 rollback_lock_file(&lock);
660 } else {
77b9b1d1 661 fprintf_or_die(out, "%s\n", reference);
f132a127 662 if (commit_lock_file(&lock))
259328b7 663 die_errno(_("unable to move new alternates file into place"));
f0eaf638 664 if (the_repository->objects->loaded_alternates)
93d8d1e2
SB
665 link_alt_odb_entries(the_repository, reference,
666 '\n', NULL, 0);
77b9b1d1
JK
667 }
668 free(alts);
bef70b22
DB
669}
670
a5b34d21
JK
671void add_to_alternates_memory(const char *reference)
672{
673 /*
674 * Make sure alternates are initialized, or else our entry may be
675 * overwritten when they are.
676 */
0b209034 677 prepare_alt_odb(the_repository);
a5b34d21 678
93d8d1e2
SB
679 link_alt_odb_entries(the_repository, reference,
680 '\n', NULL, 0);
a5b34d21
JK
681}
682
9eeea7d2
SB
683/*
684 * Compute the exact path an alternate is at and returns it. In case of
685 * error NULL is returned and the human readable error is added to `err`
efde7b72 686 * `path` may be relative and should point to $GIT_DIR.
9eeea7d2
SB
687 * `err` must not be null.
688 */
689char *compute_alternate_path(const char *path, struct strbuf *err)
690{
691 char *ref_git = NULL;
4530a85b 692 const char *repo;
9eeea7d2
SB
693 int seen_error = 0;
694
4530a85b
AM
695 ref_git = real_pathdup(path, 0);
696 if (!ref_git) {
9eeea7d2
SB
697 seen_error = 1;
698 strbuf_addf(err, _("path '%s' does not exist"), path);
699 goto out;
4530a85b 700 }
9eeea7d2
SB
701
702 repo = read_gitfile(ref_git);
703 if (!repo)
704 repo = read_gitfile(mkpath("%s/.git", ref_git));
705 if (repo) {
706 free(ref_git);
707 ref_git = xstrdup(repo);
708 }
709
710 if (!repo && is_directory(mkpath("%s/.git/objects", ref_git))) {
711 char *ref_git_git = mkpathdup("%s/.git", ref_git);
712 free(ref_git);
713 ref_git = ref_git_git;
714 } else if (!is_directory(mkpath("%s/objects", ref_git))) {
715 struct strbuf sb = STRBUF_INIT;
716 seen_error = 1;
717 if (get_common_dir(&sb, ref_git)) {
718 strbuf_addf(err,
719 _("reference repository '%s' as a linked "
720 "checkout is not supported yet."),
721 path);
722 goto out;
723 }
724
725 strbuf_addf(err, _("reference repository '%s' is not a "
726 "local repository."), path);
727 goto out;
728 }
729
730 if (!access(mkpath("%s/shallow", ref_git), F_OK)) {
731 strbuf_addf(err, _("reference repository '%s' is shallow"),
732 path);
733 seen_error = 1;
734 goto out;
735 }
736
737 if (!access(mkpath("%s/info/grafts", ref_git), F_OK)) {
738 strbuf_addf(err,
739 _("reference repository '%s' is grafted"),
740 path);
741 seen_error = 1;
742 goto out;
743 }
744
745out:
746 if (seen_error) {
6a83d902 747 FREE_AND_NULL(ref_git);
9eeea7d2
SB
748 }
749
750 return ref_git;
751}
752
f57a7396
TB
753struct object_directory *find_odb(struct repository *r, const char *obj_dir)
754{
755 struct object_directory *odb;
756 char *obj_dir_real = real_pathdup(obj_dir, 1);
757 struct strbuf odb_path_real = STRBUF_INIT;
758
759 prepare_alt_odb(r);
760 for (odb = r->objects->odb; odb; odb = odb->next) {
761 strbuf_realpath(&odb_path_real, odb->path, 1);
762 if (!strcmp(obj_dir_real, odb_path_real.buf))
763 break;
764 }
765
766 free(obj_dir_real);
767 strbuf_release(&odb_path_real);
768
769 if (!odb)
770 die(_("could not find object directory matching %s"), obj_dir);
771 return odb;
772}
773
709dfa69
JK
774static void fill_alternate_refs_command(struct child_process *cmd,
775 const char *repo_path)
776{
777 const char *value;
778
779 if (!git_config_get_value("core.alternateRefsCommand", &value)) {
780 cmd->use_shell = 1;
781
c972bf4c
JK
782 strvec_push(&cmd->args, value);
783 strvec_push(&cmd->args, repo_path);
709dfa69
JK
784 } else {
785 cmd->git_cmd = 1;
786
c972bf4c
JK
787 strvec_pushf(&cmd->args, "--git-dir=%s", repo_path);
788 strvec_push(&cmd->args, "for-each-ref");
789 strvec_push(&cmd->args, "--format=%(objectname)");
709dfa69
JK
790
791 if (!git_config_get_value("core.alternateRefsPrefixes", &value)) {
c972bf4c
JK
792 strvec_push(&cmd->args, "--");
793 strvec_split(&cmd->args, value);
709dfa69
JK
794 }
795 }
796
c7c4bdec 797 strvec_pushv(&cmd->env_array, (const char **)local_repo_env);
709dfa69
JK
798 cmd->out = -1;
799}
800
801static void read_alternate_refs(const char *path,
802 alternate_ref_fn *cb,
803 void *data)
804{
805 struct child_process cmd = CHILD_PROCESS_INIT;
806 struct strbuf line = STRBUF_INIT;
807 FILE *fh;
808
809 fill_alternate_refs_command(&cmd, path);
810
811 if (start_command(&cmd))
812 return;
813
814 fh = xfdopen(cmd.out, "r");
815 while (strbuf_getline_lf(&line, fh) != EOF) {
816 struct object_id oid;
817 const char *p;
818
819 if (parse_oid_hex(line.buf, &oid, &p) || *p) {
820 warning(_("invalid line while parsing alternate refs: %s"),
821 line.buf);
822 break;
823 }
824
825 cb(&oid, data);
826 }
827
828 fclose(fh);
829 finish_command(&cmd);
86ad3ea5 830 strbuf_release(&line);
709dfa69
JK
831}
832
833struct alternate_refs_data {
834 alternate_ref_fn *fn;
835 void *data;
836};
837
838static int refs_from_alternate_cb(struct object_directory *e,
839 void *data)
840{
841 struct strbuf path = STRBUF_INIT;
842 size_t base_len;
843 struct alternate_refs_data *cb = data;
844
845 if (!strbuf_realpath(&path, e->path, 0))
846 goto out;
847 if (!strbuf_strip_suffix(&path, "/objects"))
848 goto out;
849 base_len = path.len;
850
851 /* Is this a git repository with refs? */
852 strbuf_addstr(&path, "/refs");
853 if (!is_directory(path.buf))
854 goto out;
855 strbuf_setlen(&path, base_len);
856
857 read_alternate_refs(path.buf, cb->fn, cb->data);
858
859out:
860 strbuf_release(&path);
861 return 0;
862}
863
864void for_each_alternate_ref(alternate_ref_fn fn, void *data)
865{
866 struct alternate_refs_data cb;
867 cb.fn = fn;
868 cb.data = data;
869 foreach_alt_odb(refs_from_alternate_cb, &cb);
870}
871
fe1b2268 872int foreach_alt_odb(alt_odb_fn fn, void *cb)
d79796bc 873{
263db403 874 struct object_directory *ent;
fe1b2268 875 int r = 0;
d79796bc 876
0b209034 877 prepare_alt_odb(the_repository);
f0eaf638 878 for (ent = the_repository->objects->odb->next; ent; ent = ent->next) {
fe1b2268
JK
879 r = fn(ent, cb);
880 if (r)
881 break;
882 }
883 return r;
d79796bc
JH
884}
885
13068bf0 886void prepare_alt_odb(struct repository *r)
c2f493a4 887{
f0eaf638 888 if (r->objects->loaded_alternates)
7dc24aa5
SP
889 return;
890
13068bf0 891 link_alt_odb_entries(r, r->objects->alternate_db, PATH_SEP, NULL, 0);
c2f493a4 892
f0eaf638
JK
893 read_info_alternates(r, r->objects->odb->path, 0);
894 r->objects->loaded_alternates = 1;
c2f493a4
MW
895}
896
3096b2ec 897/* Returns 1 if we have successfully freshened the file, 0 otherwise. */
33d4221c 898static int freshen_file(const char *fn)
ace1534d 899{
312cd761 900 return !utime(fn, NULL);
0f4dc14a 901}
ace1534d 902
3096b2ec
JK
903/*
904 * All of the check_and_freshen functions return 1 if the file exists and was
905 * freshened (if freshening was requested), 0 otherwise. If they return
906 * 0, you should not assume that it is safe to skip a write of the object (it
907 * either does not exist on disk, or has a stale mtime and may be subject to
908 * pruning).
909 */
6a5e6f5e 910int check_and_freshen_file(const char *fn, int freshen)
33d4221c
JK
911{
912 if (access(fn, F_OK))
913 return 0;
3096b2ec 914 if (freshen && !freshen_file(fn))
33d4221c
JK
915 return 0;
916 return 1;
917}
918
f0eaf638
JK
919static int check_and_freshen_odb(struct object_directory *odb,
920 const struct object_id *oid,
921 int freshen)
33d4221c 922{
f0eaf638 923 static struct strbuf path = STRBUF_INIT;
514c5fdd 924 odb_loose_path(odb, &path, oid);
f0eaf638
JK
925 return check_and_freshen_file(path.buf, freshen);
926}
ea657730 927
f0eaf638
JK
928static int check_and_freshen_local(const struct object_id *oid, int freshen)
929{
930 return check_and_freshen_odb(the_repository->objects->odb, oid, freshen);
33d4221c
JK
931}
932
6862ebbf 933static int check_and_freshen_nonlocal(const struct object_id *oid, int freshen)
0f4dc14a 934{
263db403 935 struct object_directory *odb;
f3f043a1 936
0b209034 937 prepare_alt_odb(the_repository);
f0eaf638
JK
938 for (odb = the_repository->objects->odb->next; odb; odb = odb->next) {
939 if (check_and_freshen_odb(odb, oid, freshen))
c529d75a 940 return 1;
ace1534d 941 }
c529d75a 942 return 0;
ace1534d
JH
943}
944
6862ebbf 945static int check_and_freshen(const struct object_id *oid, int freshen)
33d4221c 946{
6862ebbf 947 return check_and_freshen_local(oid, freshen) ||
948 check_and_freshen_nonlocal(oid, freshen);
33d4221c
JK
949}
950
6862ebbf 951int has_loose_object_nonlocal(const struct object_id *oid)
33d4221c 952{
6862ebbf 953 return check_and_freshen_nonlocal(oid, 0);
33d4221c
JK
954}
955
6862ebbf 956static int has_loose_object(const struct object_id *oid)
0f4dc14a 957{
6862ebbf 958 return check_and_freshen(oid, 0);
0f4dc14a
BC
959}
960
02710228
SP
961static void mmap_limit_check(size_t length)
962{
963 static size_t limit = 0;
964 if (!limit) {
965 limit = git_env_ulong("GIT_MMAP_LIMIT", 0);
966 if (!limit)
967 limit = SIZE_MAX;
968 }
969 if (length > limit)
259328b7 970 die(_("attempting to mmap %"PRIuMAX" over limit %"PRIuMAX),
02710228
SP
971 (uintmax_t)length, (uintmax_t)limit);
972}
973
1570856b
JK
974void *xmmap_gently(void *start, size_t length,
975 int prot, int flags, int fd, off_t offset)
58ecbd5e 976{
02710228
SP
977 void *ret;
978
979 mmap_limit_check(length);
980 ret = mmap(start, length, prot, flags, fd, offset);
9827d4c1
JK
981 if (ret == MAP_FAILED && !length)
982 ret = NULL;
58ecbd5e
JN
983 return ret;
984}
985
dc059294
EW
986const char *mmap_os_err(void)
987{
988 static const char blank[] = "";
989#if defined(__linux__)
990 if (errno == ENOMEM) {
991 /* this continues an existing error message: */
992 static const char enomem[] =
993", check sys.vm.max_map_count and/or RLIMIT_DATA";
994 return enomem;
995 }
996#endif /* OS-specific bits */
997 return blank;
998}
999
1570856b
JK
1000void *xmmap(void *start, size_t length,
1001 int prot, int flags, int fd, off_t offset)
1002{
1003 void *ret = xmmap_gently(start, length, prot, flags, fd, offset);
1004 if (ret == MAP_FAILED)
dc059294 1005 die_errno(_("mmap failed%s"), mmap_os_err());
1570856b
JK
1006 return ret;
1007}
1008
88d0db55 1009/*
090ea126 1010 * With an in-core object data in "map", rehash it to make sure the
cb1c8d1d
JK
1011 * object name actually matches "oid" to detect object corruption.
1012 * With "map" == NULL, try reading the object named with "oid" using
090ea126 1013 * the streaming interface and rehash it to do the same.
88d0db55 1014 */
b98d1885 1015int check_object_signature(struct repository *r, const struct object_id *oid,
96e41f58
ÆAB
1016 void *map, unsigned long size, const char *type,
1017 struct object_id *real_oidp)
88d0db55 1018{
96e41f58
ÆAB
1019 struct object_id tmp;
1020 struct object_id *real_oid = real_oidp ? real_oidp : &tmp;
090ea126
NTND
1021 enum object_type obj_type;
1022 struct git_istream *st;
18e2588e 1023 git_hash_ctx c;
1af64f73 1024 char hdr[MAX_HEADER_LEN];
090ea126 1025 int hdrlen;
88d0db55 1026
090ea126 1027 if (map) {
96e41f58
ÆAB
1028 hash_object_file(r->hash_algo, map, size, type, real_oid);
1029 return !oideq(oid, real_oid) ? -1 : 0;
88d0db55
BC
1030 }
1031
b98d1885 1032 st = open_istream(r, oid, &obj_type, &size, NULL);
090ea126
NTND
1033 if (!st)
1034 return -1;
88d0db55 1035
090ea126 1036 /* Generate the header */
ca473cef 1037 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %"PRIuMAX , type_name(obj_type), (uintmax_t)size) + 1;
88d0db55 1038
090ea126 1039 /* Sha1.. */
b98d1885
MT
1040 r->hash_algo->init_fn(&c);
1041 r->hash_algo->update_fn(&c, hdr, hdrlen);
090ea126
NTND
1042 for (;;) {
1043 char buf[1024 * 16];
1044 ssize_t readlen = read_istream(st, buf, sizeof(buf));
1f688557 1045
f54fac53
JK
1046 if (readlen < 0) {
1047 close_istream(st);
1048 return -1;
1049 }
090ea126
NTND
1050 if (!readlen)
1051 break;
b98d1885 1052 r->hash_algo->update_fn(&c, buf, readlen);
fa5fc15d 1053 }
96e41f58 1054 r->hash_algo->final_oid_fn(real_oid, &c);
090ea126 1055 close_istream(st);
96e41f58 1056 return !oideq(oid, real_oid) ? -1 : 0;
fa5fc15d
SP
1057}
1058
1b8ac5ea 1059int git_open_cloexec(const char *name, int flags)
a0788266 1060{
1e3001a8
JH
1061 int fd;
1062 static int o_cloexec = O_CLOEXEC;
a0788266 1063
1e3001a8
JH
1064 fd = open(name, flags | o_cloexec);
1065 if ((o_cloexec & O_CLOEXEC) && fd < 0 && errno == EINVAL) {
cd66ada0 1066 /* Try again w/o O_CLOEXEC: the kernel might not support it */
1e3001a8
JH
1067 o_cloexec &= ~O_CLOEXEC;
1068 fd = open(name, flags | o_cloexec);
491a8dec 1069 }
491a8dec 1070
9fb9495d 1071#if defined(F_GETFD) && defined(F_SETFD) && defined(FD_CLOEXEC)
491a8dec 1072 {
1e3001a8 1073 static int fd_cloexec = FD_CLOEXEC;
a0788266 1074
1e3001a8
JH
1075 if (!o_cloexec && 0 <= fd && fd_cloexec) {
1076 /* Opened w/o O_CLOEXEC? try with fcntl(2) to add it */
9fb9495d
EW
1077 int flags = fcntl(fd, F_GETFD);
1078 if (fcntl(fd, F_SETFD, flags | fd_cloexec))
1e3001a8 1079 fd_cloexec = 0;
cd66ada0 1080 }
44d1c19e 1081 }
a0788266 1082#endif
1b8ac5ea 1083 return fd;
a0788266
JS
1084}
1085
3cf8b462 1086/*
514c5fdd 1087 * Find "oid" as a loose object in the local repository or in an alternate.
771e7d57
JK
1088 * Returns 0 on success, negative on failure.
1089 *
1090 * The "path" out-parameter will give the path of the object we found (if any).
1091 * Note that it may point to static storage and is only valid until another
514c5fdd 1092 * call to stat_loose_object().
3cf8b462 1093 */
514c5fdd
JK
1094static int stat_loose_object(struct repository *r, const struct object_id *oid,
1095 struct stat *st, const char **path)
1f688557 1096{
263db403 1097 struct object_directory *odb;
ea657730
CC
1098 static struct strbuf buf = STRBUF_INIT;
1099
d2607fa0 1100 prepare_alt_odb(r);
f0eaf638 1101 for (odb = r->objects->odb; odb; odb = odb->next) {
514c5fdd 1102 *path = odb_loose_path(odb, &buf, oid);
771e7d57 1103 if (!lstat(*path, st))
052fe5ea 1104 return 0;
c7934306
SP
1105 }
1106
3cf8b462
SP
1107 return -1;
1108}
1109
771e7d57 1110/*
514c5fdd 1111 * Like stat_loose_object(), but actually open the object and return the
771e7d57
JK
1112 * descriptor. See the caveats on the "path" parameter above.
1113 */
514c5fdd
JK
1114static int open_loose_object(struct repository *r,
1115 const struct object_id *oid, const char **path)
60bb8b14 1116{
44d1c19e 1117 int fd;
263db403 1118 struct object_directory *odb;
f0eaf638 1119 int most_interesting_errno = ENOENT;
ea657730
CC
1120 static struct strbuf buf = STRBUF_INIT;
1121
ec7283e5 1122 prepare_alt_odb(r);
f0eaf638 1123 for (odb = r->objects->odb; odb; odb = odb->next) {
514c5fdd 1124 *path = odb_loose_path(odb, &buf, oid);
771e7d57 1125 fd = git_open(*path);
44d1c19e
LT
1126 if (fd >= 0)
1127 return fd;
f0eaf638 1128
d6c8a05b
JK
1129 if (most_interesting_errno == ENOENT)
1130 most_interesting_errno = errno;
03e79c88 1131 }
d6c8a05b 1132 errno = most_interesting_errno;
44d1c19e 1133 return -1;
1f688557
JH
1134}
1135
61c7711c 1136static int quick_has_loose(struct repository *r,
d7a24573 1137 const struct object_id *oid)
61c7711c 1138{
61c7711c
JK
1139 struct object_directory *odb;
1140
61c7711c
JK
1141 prepare_alt_odb(r);
1142 for (odb = r->objects->odb; odb; odb = odb->next) {
92d8ed8a 1143 if (oidtree_contains(odb_loose_cache(odb, oid), oid))
61c7711c
JK
1144 return 1;
1145 }
1146 return 0;
1147}
1148
f6371f92
JK
1149/*
1150 * Map the loose object at "path" if it is not NULL, or the path found by
514c5fdd 1151 * searching for a loose object named "oid".
f6371f92 1152 */
514c5fdd
JK
1153static void *map_loose_object_1(struct repository *r, const char *path,
1154 const struct object_id *oid, unsigned long *size)
27d69a46 1155{
0fcfd160 1156 void *map;
144bde78 1157 int fd;
84dd81c1 1158
f6371f92
JK
1159 if (path)
1160 fd = git_open(path);
8261e1f1 1161 else
514c5fdd 1162 fd = open_loose_object(r, oid, &path);
44d1c19e
LT
1163 map = NULL;
1164 if (fd >= 0) {
1165 struct stat st;
1f688557 1166
44d1c19e
LT
1167 if (!fstat(fd, &st)) {
1168 *size = xsize_t(st.st_size);
33e42de0
MM
1169 if (!*size) {
1170 /* mmap() is forbidden on empty files */
259328b7 1171 error(_("object file %s is empty"), path);
6881925e 1172 close(fd);
33e42de0
MM
1173 return NULL;
1174 }
44d1c19e 1175 map = xmmap(NULL, *size, PROT_READ, MAP_PRIVATE, fd, 0);
144bde78 1176 }
44d1c19e 1177 close(fd);
74e34e1f 1178 }
0fcfd160 1179 return map;
74e34e1f
NP
1180}
1181
514c5fdd
JK
1182void *map_loose_object(struct repository *r,
1183 const struct object_id *oid,
1184 unsigned long *size)
068f85e3 1185{
514c5fdd 1186 return map_loose_object_1(r, NULL, oid, size);
068f85e3 1187}
1188
3b6a8db3
ÆAB
1189enum unpack_loose_header_result unpack_loose_header(git_zstream *stream,
1190 unsigned char *map,
1191 unsigned long mapsize,
1192 void *buffer,
1193 unsigned long bufsiz,
1194 struct strbuf *header)
47fe3f6e 1195{
01cab976 1196 int status;
31877c9a 1197
c4483576
LT
1198 /* Get the data stream */
1199 memset(stream, 0, sizeof(*stream));
1200 stream->next_in = map;
1201 stream->avail_in = mapsize;
1202 stream->next_out = buffer;
93821bd9 1203 stream->avail_out = bufsiz;
47fe3f6e 1204
39c68542 1205 git_inflate_init(stream);
31877c9a 1206 obj_read_unlock();
01cab976 1207 status = git_inflate(stream, 0);
31877c9a 1208 obj_read_lock();
d21f8426 1209 if (status < Z_OK)
3b6a8db3 1210 return ULHR_BAD;
d131b7af 1211
46f03448
KN
1212 /*
1213 * Check if entire header is unpacked in the first iteration.
d131b7af 1214 */
46f03448 1215 if (memchr(buffer, '\0', stream->next_out - (unsigned char *)buffer))
3b6a8db3 1216 return ULHR_OK;
d131b7af 1217
01cab976
ÆAB
1218 /*
1219 * We have a header longer than MAX_HEADER_LEN. The "header"
1220 * here is only non-NULL when we run "cat-file
1221 * --allow-unknown-type".
1222 */
1223 if (!header)
5848fb11 1224 return ULHR_TOO_LONG;
d131b7af 1225
46f03448
KN
1226 /*
1227 * buffer[0..bufsiz] was not large enough. Copy the partial
1228 * result out to header, and then append the result of further
1229 * reading the stream.
1230 */
1231 strbuf_add(header, buffer, stream->next_out - (unsigned char *)buffer);
1232 stream->next_out = buffer;
1233 stream->avail_out = bufsiz;
d131b7af 1234
46f03448 1235 do {
31877c9a 1236 obj_read_unlock();
46f03448 1237 status = git_inflate(stream, 0);
31877c9a 1238 obj_read_lock();
46f03448
KN
1239 strbuf_add(header, buffer, stream->next_out - (unsigned char *)buffer);
1240 if (memchr(buffer, '\0', stream->next_out - (unsigned char *)buffer))
1241 return 0;
1242 stream->next_out = buffer;
1243 stream->avail_out = bufsiz;
1244 } while (status != Z_STREAM_END);
5848fb11 1245 return ULHR_TOO_LONG;
d131b7af
SP
1246}
1247
00a7760e
JK
1248static void *unpack_loose_rest(git_zstream *stream,
1249 void *buffer, unsigned long size,
1250 const struct object_id *oid)
95099731 1251{
5180cacc 1252 int bytes = strlen(buffer) + 1;
3aee68aa 1253 unsigned char *buf = xmallocz(size);
93821bd9 1254 unsigned long n;
7efbff75 1255 int status = Z_OK;
95099731 1256
93821bd9
LT
1257 n = stream->total_out - bytes;
1258 if (n > size)
1259 n = size;
1260 memcpy(buf, (char *) buffer + bytes, n);
1261 bytes = n;
456cdf6e
LT
1262 if (bytes <= size) {
1263 /*
1264 * The above condition must be (bytes <= size), not
1265 * (bytes < size). In other words, even though we
ccf5ace0 1266 * expect no more output and set avail_out to zero,
456cdf6e
LT
1267 * the input zlib stream may have bytes that express
1268 * "this concludes the stream", and we *do* want to
1269 * eat that input.
1270 *
1271 * Otherwise we would not be able to test that we
1272 * consumed all the input to reach the expected size;
1273 * we also want to check that zlib tells us that all
1274 * went well with status == Z_STREAM_END at the end.
1275 */
5180cacc
LT
1276 stream->next_out = buf + bytes;
1277 stream->avail_out = size - bytes;
31877c9a
MT
1278 while (status == Z_OK) {
1279 obj_read_unlock();
39c68542 1280 status = git_inflate(stream, Z_FINISH);
31877c9a
MT
1281 obj_read_lock();
1282 }
5180cacc 1283 }
456cdf6e 1284 if (status == Z_STREAM_END && !stream->avail_in) {
39c68542 1285 git_inflate_end(stream);
7efbff75 1286 return buf;
95099731
NTND
1287 }
1288
7efbff75 1289 if (status < 0)
00a7760e 1290 error(_("corrupt loose object '%s'"), oid_to_hex(oid));
7efbff75 1291 else if (stream->avail_in)
259328b7 1292 error(_("garbage at end of loose object '%s'"),
00a7760e 1293 oid_to_hex(oid));
7efbff75
JH
1294 free(buf);
1295 return NULL;
95099731
NTND
1296}
1297
d40d535b 1298/*
5180cacc
LT
1299 * We used to just use "sscanf()", but that's actually way
1300 * too permissive for what we want to check. So do an anal
1301 * object header parse by hand.
d40d535b 1302 */
dccb32bf 1303int parse_loose_header(const char *hdr, struct object_info *oi)
1f688557 1304{
46f03448 1305 const char *type_buf = hdr;
d6a09e79 1306 size_t size;
46f03448 1307 int type, type_len = 0;
43057304 1308
5180cacc 1309 /*
46f03448 1310 * The type can be of any size but is followed by
21666f1a 1311 * a space.
5180cacc 1312 */
5180cacc
LT
1313 for (;;) {
1314 char c = *hdr++;
d21f8426
JH
1315 if (!c)
1316 return -1;
5180cacc
LT
1317 if (c == ' ')
1318 break;
46f03448 1319 type_len++;
5180cacc 1320 }
1f688557 1321
46f03448 1322 type = type_from_string_gently(type_buf, type_len, 1);
6ca32f47
BW
1323 if (oi->type_name)
1324 strbuf_add(oi->type_name, type_buf, type_len);
46f03448
KN
1325 if (oi->typep)
1326 *oi->typep = type;
5180cacc
LT
1327
1328 /*
1329 * The length must follow immediately, and be in canonical
1330 * decimal format (ie "010" is not valid).
1331 */
1332 size = *hdr++ - '0';
1333 if (size > 9)
1334 return -1;
1335 if (size) {
1336 for (;;) {
1337 unsigned long c = *hdr - '0';
1338 if (c > 9)
1339 break;
1340 hdr++;
d6a09e79 1341 size = st_add(st_mult(size, 10), c);
1b1005d1 1342 }
c01f51cc 1343 }
46f03448
KN
1344
1345 if (oi->sizep)
d6a09e79 1346 *oi->sizep = cast_size_t_to_ulong(size);
5180cacc
LT
1347
1348 /*
1349 * The length must be followed by a zero byte
1350 */
dccb32bf
ÆAB
1351 if (*hdr)
1352 return -1;
21666f1a 1353
dccb32bf
ÆAB
1354 /*
1355 * The format is valid, but the type may still be bogus. The
1356 * Caller needs to check its oi->typep.
1357 */
1358 return 0;
bf592c50
DB
1359}
1360
514c5fdd
JK
1361static int loose_object_info(struct repository *r,
1362 const struct object_id *oid,
1363 struct object_info *oi, int flags)
65c2e0c3 1364{
46f03448
KN
1365 int status = 0;
1366 unsigned long mapsize;
65c2e0c3 1367 void *map;
ef49a7a0 1368 git_zstream stream;
1af64f73 1369 char hdr[MAX_HEADER_LEN];
46f03448 1370 struct strbuf hdrbuf = STRBUF_INIT;
c84a1f3e 1371 unsigned long size_scratch;
dccb32bf 1372 enum object_type type_scratch;
01cab976 1373 int allow_unknown = flags & OBJECT_INFO_ALLOW_UNKNOWN_TYPE;
65c2e0c3 1374
b99b6bcc
JK
1375 if (oi->delta_base_oid)
1376 oidclr(oi->delta_base_oid);
5d642e75 1377
052fe5ea
JK
1378 /*
1379 * If we don't care about type or size, then we don't
4ef8d1dd
JH
1380 * need to look inside the object at all. Note that we
1381 * do not optimize out the stat call, even if the
1382 * caller doesn't care about the disk-size, since our
1383 * return value implicitly indicates whether the
1384 * object even exists.
052fe5ea 1385 */
6ca32f47 1386 if (!oi->typep && !oi->type_name && !oi->sizep && !oi->contentp) {
771e7d57 1387 const char *path;
4ef8d1dd 1388 struct stat st;
61c7711c 1389 if (!oi->disk_sizep && (flags & OBJECT_INFO_QUICK))
d7a24573 1390 return quick_has_loose(r, oid) ? 0 : -1;
514c5fdd 1391 if (stat_loose_object(r, oid, &st, &path) < 0)
4ef8d1dd
JH
1392 return -1;
1393 if (oi->disk_sizep)
23c339c0 1394 *oi->disk_sizep = st.st_size;
052fe5ea
JK
1395 return 0;
1396 }
1397
514c5fdd 1398 map = map_loose_object(r, oid, &mapsize);
f0df4ed5 1399 if (!map)
dbea72a8 1400 return -1;
c84a1f3e
JT
1401
1402 if (!oi->sizep)
1403 oi->sizep = &size_scratch;
dccb32bf
ÆAB
1404 if (!oi->typep)
1405 oi->typep = &type_scratch;
c84a1f3e 1406
23c339c0
JK
1407 if (oi->disk_sizep)
1408 *oi->disk_sizep = mapsize;
01cab976 1409
3b6a8db3
ÆAB
1410 switch (unpack_loose_header(&stream, map, mapsize, hdr, sizeof(hdr),
1411 allow_unknown ? &hdrbuf : NULL)) {
1412 case ULHR_OK:
dccb32bf
ÆAB
1413 if (parse_loose_header(hdrbuf.len ? hdrbuf.buf : hdr, oi) < 0)
1414 status = error(_("unable to parse %s header"), oid_to_hex(oid));
1415 else if (!allow_unknown && *oi->typep < 0)
1416 die(_("invalid object type"));
1417
1418 if (!oi->contentp)
1419 break;
1420 *oi->contentp = unpack_loose_rest(&stream, hdr, *oi->sizep, oid);
1421 if (*oi->contentp)
1422 goto cleanup;
1423
1424 status = -1;
3b6a8db3
ÆAB
1425 break;
1426 case ULHR_BAD:
259328b7 1427 status = error(_("unable to unpack %s header"),
514c5fdd 1428 oid_to_hex(oid));
3b6a8db3 1429 break;
5848fb11
ÆAB
1430 case ULHR_TOO_LONG:
1431 status = error(_("header for %s too long, exceeds %d bytes"),
1432 oid_to_hex(oid), MAX_HEADER_LEN);
1433 break;
3b6a8db3 1434 }
c84a1f3e 1435
dccb32bf
ÆAB
1436 git_inflate_end(&stream);
1437cleanup:
65c2e0c3 1438 munmap(map, mapsize);
c84a1f3e
JT
1439 if (oi->sizep == &size_scratch)
1440 oi->sizep = NULL;
46f03448 1441 strbuf_release(&hdrbuf);
dccb32bf
ÆAB
1442 if (oi->typep == &type_scratch)
1443 oi->typep = NULL;
3ab0fb06 1444 oi->whence = OI_LOOSE;
dccb32bf 1445 return status;
65c2e0c3
JH
1446}
1447
31877c9a
MT
1448int obj_read_use_lock = 0;
1449pthread_mutex_t obj_read_mutex;
1450
1451void enable_obj_read_lock(void)
1452{
1453 if (obj_read_use_lock)
1454 return;
1455
1456 obj_read_use_lock = 1;
1457 init_recursive_mutex(&obj_read_mutex);
1458}
1459
1460void disable_obj_read_lock(void)
1461{
1462 if (!obj_read_use_lock)
1463 return;
1464
1465 obj_read_use_lock = 0;
1466 pthread_mutex_destroy(&obj_read_mutex);
1467}
1468
8b4c0103
JT
1469int fetch_if_missing = 1;
1470
31877c9a
MT
1471static int do_oid_object_info_extended(struct repository *r,
1472 const struct object_id *oid,
1473 struct object_info *oi, unsigned flags)
f0df4ed5 1474{
cd585e2a 1475 static struct object_info blank_oi = OBJECT_INFO_INIT;
9c8a294a 1476 struct cached_object *co;
f0df4ed5 1477 struct pack_entry e;
5b086407 1478 int rtype;
b383a13c 1479 const struct object_id *real = oid;
8b4c0103 1480 int already_retried = 0;
f0df4ed5 1481
31877c9a 1482
b383a13c 1483 if (flags & OBJECT_INFO_LOOKUP_REPLACE)
9d98354f 1484 real = lookup_replace_object(r, oid);
f0df4ed5 1485
b383a13c 1486 if (is_null_oid(real))
87b5e236
JK
1487 return -1;
1488
cd585e2a
JT
1489 if (!oi)
1490 oi = &blank_oi;
1491
9c8a294a
JT
1492 co = find_cached_object(real);
1493 if (co) {
1494 if (oi->typep)
1495 *(oi->typep) = co->type;
1496 if (oi->sizep)
1497 *(oi->sizep) = co->size;
1498 if (oi->disk_sizep)
1499 *(oi->disk_sizep) = 0;
b99b6bcc
JK
1500 if (oi->delta_base_oid)
1501 oidclr(oi->delta_base_oid);
9c8a294a
JT
1502 if (oi->type_name)
1503 strbuf_addstr(oi->type_name, type_name(co->type));
1504 if (oi->contentp)
1505 *oi->contentp = xmemdupz(co->buf, co->size);
1506 oi->whence = OI_CACHED;
1507 return 0;
c4d9986f
NTND
1508 }
1509
8b4c0103 1510 while (1) {
42c8ce1c 1511 if (find_pack_entry(r, real, &e))
8b4c0103
JT
1512 break;
1513
024aa469
TI
1514 if (flags & OBJECT_INFO_IGNORE_LOOSE)
1515 return -1;
1516
ddd63e64 1517 /* Most likely it's a loose object. */
514c5fdd 1518 if (!loose_object_info(r, real, oi, flags))
5b086407 1519 return 0;
ddd63e64
SG
1520
1521 /* Not a loose object; someone else may have just packed it. */
2b7750c9 1522 if (!(flags & OBJECT_INFO_QUICK)) {
9d98354f 1523 reprepare_packed_git(r);
42c8ce1c 1524 if (find_pack_entry(r, real, &e))
2b7750c9
JT
1525 break;
1526 }
8b4c0103 1527
eef71904
JT
1528 /*
1529 * If r is the_repository, this might be an attempt at
1530 * accessing a submodule object as if it were in the_repository
1531 * (having called add_submodule_odb() on that submodule's ODB).
1532 * If any such ODBs exist, register them and try again.
1533 */
1534 if (r == the_repository &&
1535 register_all_submodule_odb_as_alternates())
a35e03de
JT
1536 /* We added some alternates; retry */
1537 continue;
1538
8b4c0103 1539 /* Check if it is a missing object */
ef830cc4
JT
1540 if (fetch_if_missing && repo_has_promisor_remote(r) &&
1541 !already_retried &&
31f5256c 1542 !(flags & OBJECT_INFO_SKIP_FETCH_OBJECT)) {
8b4c0103 1543 /*
b14ed5ad
CC
1544 * TODO Investigate checking promisor_remote_get_direct()
1545 * TODO return value and stopping on error here.
8b4c0103 1546 */
b14ed5ad 1547 promisor_remote_get_direct(r, real, 1);
8b4c0103
JT
1548 already_retried = 1;
1549 continue;
dfdd4afc 1550 }
8b4c0103
JT
1551
1552 return -1;
f0df4ed5 1553 }
3d77d877 1554
cd585e2a
JT
1555 if (oi == &blank_oi)
1556 /*
1557 * We know that the caller doesn't actually need the
1558 * information below, so return early.
1559 */
1560 return 0;
9d98354f 1561 rtype = packed_object_info(r, e.p, e.offset, oi);
412916ee 1562 if (rtype < 0) {
751530de 1563 mark_bad_packed_object(e.p, real);
31877c9a 1564 return do_oid_object_info_extended(r, real, oi, 0);
3ab0fb06 1565 } else if (oi->whence == OI_PACKED) {
9a490590
JH
1566 oi->u.packed.offset = e.offset;
1567 oi->u.packed.pack = e.p;
1568 oi->u.packed.is_delta = (rtype == OBJ_REF_DELTA ||
1569 rtype == OBJ_OFS_DELTA);
3d77d877
NP
1570 }
1571
5b086407 1572 return 0;
f0df4ed5
JS
1573}
1574
31877c9a
MT
1575int oid_object_info_extended(struct repository *r, const struct object_id *oid,
1576 struct object_info *oi, unsigned flags)
1577{
1578 int ret;
1579 obj_read_lock();
1580 ret = do_oid_object_info_extended(r, oid, oi, flags);
1581 obj_read_unlock();
1582 return ret;
1583}
1584
1585
3fc0dca9 1586/* returns enum object_type or negative */
9d98354f
SB
1587int oid_object_info(struct repository *r,
1588 const struct object_id *oid,
1589 unsigned long *sizep)
9a490590 1590{
5b086407 1591 enum object_type type;
27b5c1a0 1592 struct object_info oi = OBJECT_INFO_INIT;
9a490590 1593
5b086407 1594 oi.typep = &type;
9a490590 1595 oi.sizep = sizep;
9d98354f
SB
1596 if (oid_object_info_extended(r, oid, &oi,
1597 OBJECT_INFO_LOOKUP_REPLACE) < 0)
5b086407
JK
1598 return -1;
1599 return type;
9a490590
JH
1600}
1601
1b9b5c69 1602static void *read_object(struct repository *r,
cba595ab 1603 const struct object_id *oid, enum object_type *type,
f1d8130b
JT
1604 unsigned long *size)
1605{
1606 struct object_info oi = OBJECT_INFO_INIT;
1607 void *content;
1608 oi.typep = type;
1609 oi.sizep = size;
1610 oi.contentp = &content;
1611
cba595ab 1612 if (oid_object_info_extended(r, oid, &oi, 0) < 0)
f1d8130b
JT
1613 return NULL;
1614 return content;
1615}
1616
829e5c3b
PO
1617int pretend_object_file(void *buf, unsigned long len, enum object_type type,
1618 struct object_id *oid)
d66b37bb
JH
1619{
1620 struct cached_object *co;
1621
2dcde20e 1622 hash_object_file(the_hash_algo, buf, len, type_name(type), oid);
a64d2aae
JT
1623 if (has_object_file_with_flags(oid, OBJECT_INFO_QUICK | OBJECT_INFO_SKIP_FETCH_OBJECT) ||
1624 find_cached_object(oid))
d66b37bb 1625 return 0;
c7353967 1626 ALLOC_GROW(cached_objects, cached_object_nr + 1, cached_object_alloc);
d66b37bb
JH
1627 co = &cached_objects[cached_object_nr++];
1628 co->size = len;
21666f1a 1629 co->type = type;
efa13f7b
JH
1630 co->buf = xmalloc(len);
1631 memcpy(co->buf, buf, len);
62ba93ea 1632 oidcpy(&co->oid, oid);
d66b37bb
JH
1633 return 0;
1634}
1635
b6c4cecc
JH
1636/*
1637 * This function dies on corrupt objects; the callers who want to
1638 * deal with them should arrange to call read_object() and give error
1639 * messages themselves.
1640 */
a3b72c89
SB
1641void *read_object_file_extended(struct repository *r,
1642 const struct object_id *oid,
b4f5aca4 1643 enum object_type *type,
1644 unsigned long *size,
1645 int lookup_replace)
ac939109 1646{
3ba7a065 1647 void *data;
b6c4cecc 1648 const struct packed_git *p;
771e7d57
JK
1649 const char *path;
1650 struct stat st;
1f2e7cea 1651 const struct object_id *repl = lookup_replace ?
a3b72c89 1652 lookup_replace_object(r, oid) : oid;
b6c4cecc 1653
3ba7a065 1654 errno = 0;
cba595ab 1655 data = read_object(r, repl, type, size);
4bbf5a26 1656 if (data)
b6c4cecc 1657 return data;
68095570 1658
31877c9a 1659 obj_read_lock();
25f3af3f 1660 if (errno && errno != ENOENT)
259328b7 1661 die_errno(_("failed to read object %s"), oid_to_hex(oid));
3ba7a065 1662
68095570 1663 /* die if we replaced an object with one that does not exist */
b383a13c 1664 if (repl != oid)
259328b7 1665 die(_("replacement %s not found for %s"),
b383a13c 1666 oid_to_hex(repl), oid_to_hex(oid));
68095570 1667
cba595ab 1668 if (!stat_loose_object(r, repl, &st, &path))
259328b7 1669 die(_("loose object %s (stored in %s) is corrupt"),
b383a13c 1670 oid_to_hex(repl), path);
68095570 1671
7407d733 1672 if ((p = has_packed_and_bad(r, repl)) != NULL)
259328b7 1673 die(_("packed object %s (stored in %s) is corrupt"),
b383a13c 1674 oid_to_hex(repl), p->pack_name);
31877c9a 1675 obj_read_unlock();
f5552aee 1676
b6c4cecc 1677 return NULL;
ac939109
NP
1678}
1679
d3b4705a
NTND
1680void *read_object_with_reference(struct repository *r,
1681 const struct object_id *oid,
21666f1a 1682 const char *required_type_name,
40469ee9 1683 unsigned long *size,
02f0547e 1684 struct object_id *actual_oid_return)
f4913f91 1685{
21666f1a 1686 enum object_type type, required_type;
f4913f91
JH
1687 void *buffer;
1688 unsigned long isize;
02f0547e 1689 struct object_id actual_oid;
f4913f91 1690
21666f1a 1691 required_type = type_from_string(required_type_name);
02f0547e 1692 oidcpy(&actual_oid, oid);
40469ee9
JH
1693 while (1) {
1694 int ref_length = -1;
1695 const char *ref_type = NULL;
f4913f91 1696
d3b4705a 1697 buffer = repo_read_object_file(r, &actual_oid, &type, &isize);
40469ee9
JH
1698 if (!buffer)
1699 return NULL;
21666f1a 1700 if (type == required_type) {
40469ee9 1701 *size = isize;
02f0547e 1702 if (actual_oid_return)
1703 oidcpy(actual_oid_return, &actual_oid);
40469ee9
JH
1704 return buffer;
1705 }
1706 /* Handle references */
21666f1a 1707 else if (type == OBJ_COMMIT)
40469ee9 1708 ref_type = "tree ";
21666f1a 1709 else if (type == OBJ_TAG)
40469ee9
JH
1710 ref_type = "object ";
1711 else {
1712 free(buffer);
1713 return NULL;
1714 }
1715 ref_length = strlen(ref_type);
f4913f91 1716
94b5e093 1717 if (ref_length + the_hash_algo->hexsz > isize ||
50974ec9 1718 memcmp(buffer, ref_type, ref_length) ||
02f0547e 1719 get_oid_hex((char *) buffer + ref_length, &actual_oid)) {
40469ee9
JH
1720 free(buffer);
1721 return NULL;
1722 }
1cf58e72 1723 free(buffer);
40469ee9 1724 /* Now we have the ID of the referred-to object in
02f0547e 1725 * actual_oid. Check again. */
f4913f91 1726 }
f4913f91
JH
1727}
1728
7ad5c44d
MT
1729static void write_object_file_prepare(const struct git_hash_algo *algo,
1730 const void *buf, unsigned long len,
a09c985e
PO
1731 const char *type, struct object_id *oid,
1732 char *hdr, int *hdrlen)
d410c0f5 1733{
18e2588e 1734 git_hash_ctx c;
d410c0f5
JH
1735
1736 /* Generate the header */
ca473cef 1737 *hdrlen = xsnprintf(hdr, *hdrlen, "%s %"PRIuMAX , type, (uintmax_t)len)+1;
d410c0f5
JH
1738
1739 /* Sha1.. */
7ad5c44d
MT
1740 algo->init_fn(&c);
1741 algo->update_fn(&c, hdr, *hdrlen);
1742 algo->update_fn(&c, buf, len);
5951bf46 1743 algo->final_oid_fn(oid, &c);
d410c0f5
JH
1744}
1745
230f1322 1746/*
5a688fe4 1747 * Move the just written object into its final resting place.
230f1322 1748 */
cb5add58 1749int finalize_object_file(const char *tmpfile, const char *filename)
230f1322 1750{
e32c0a9c 1751 int ret = 0;
5a688fe4 1752
348df166 1753 if (object_creation_mode == OBJECT_CREATION_USES_RENAMES)
be66a6c4
JS
1754 goto try_rename;
1755 else if (link(tmpfile, filename))
e32c0a9c 1756 ret = errno;
7ebb6fca
LT
1757
1758 /*
1759 * Coda hack - coda doesn't like cross-directory links,
1760 * so we fall back to a rename, which will mean that it
1761 * won't be able to check collisions, but that's not a
1762 * big deal.
1763 *
1764 * The same holds for FAT formatted media.
1765 *
3be1f18e 1766 * When this succeeds, we just return. We have nothing
7ebb6fca
LT
1767 * left to unlink.
1768 */
1769 if (ret && ret != EEXIST) {
be66a6c4 1770 try_rename:
7ebb6fca 1771 if (!rename(tmpfile, filename))
3be1f18e 1772 goto out;
9e48b389 1773 ret = errno;
230f1322 1774 }
691f1a28 1775 unlink_or_warn(tmpfile);
230f1322
LT
1776 if (ret) {
1777 if (ret != EEXIST) {
2c319886 1778 return error_errno(_("unable to write file %s"), filename);
230f1322
LT
1779 }
1780 /* FIXME!!! Collision check here ? */
1781 }
1782
3be1f18e 1783out:
5256b006 1784 if (adjust_shared_perm(filename))
259328b7 1785 return error(_("unable to set permission to '%s'"), filename);
230f1322
LT
1786 return 0;
1787}
1788
4d548150
LT
1789static int write_buffer(int fd, const void *buf, size_t len)
1790{
d34cf19b 1791 if (write_in_full(fd, buf, len) < 0)
259328b7 1792 return error_errno(_("file write error"));
4d548150
LT
1793 return 0;
1794}
1795
2dcde20e
MT
1796int hash_object_file(const struct git_hash_algo *algo, const void *buf,
1797 unsigned long len, const char *type,
f070facc 1798 struct object_id *oid)
abdc3fc8 1799{
1af64f73 1800 char hdr[MAX_HEADER_LEN];
ef1286d3 1801 int hdrlen = sizeof(hdr);
2dcde20e 1802 write_object_file_prepare(algo, buf, len, type, oid, hdr, &hdrlen);
abdc3fc8
RS
1803 return 0;
1804}
1805
e9039dd3 1806/* Finalize a file on disk, and close it. */
514c5fdd 1807static void close_loose_object(int fd)
e9039dd3 1808{
aafe9fba 1809 if (fsync_object_files)
76011357 1810 fsync_or_die(fd, "loose object file");
e9039dd3 1811 if (close(fd) != 0)
76011357 1812 die_errno(_("error when closing loose object file"));
e9039dd3
LT
1813}
1814
5723fe7e
LT
1815/* Size of directory component, including the ending '/' */
1816static inline int directory_size(const char *filename)
1817{
1818 const char *s = strrchr(filename, '/');
1819 if (!s)
1820 return 0;
1821 return s - filename + 1;
1822}
1823
1824/*
1825 * This creates a temporary file in the same directory as the final
1826 * 'filename'
1827 *
1828 * We want to avoid cross-directory filename renames, because those
1829 * can have problems on various filesystems (FAT, NFS, Coda).
1830 */
d4b3d11a 1831static int create_tmpfile(struct strbuf *tmp, const char *filename)
5723fe7e
LT
1832{
1833 int fd, dirlen = directory_size(filename);
1834
d4b3d11a
JK
1835 strbuf_reset(tmp);
1836 strbuf_add(tmp, filename, dirlen);
1837 strbuf_addstr(tmp, "tmp_obj_XXXXXX");
1838 fd = git_mkstemp_mode(tmp->buf, 0444);
cbacbf4e 1839 if (fd < 0 && dirlen && errno == ENOENT) {
d4b3d11a
JK
1840 /*
1841 * Make sure the directory exists; note that the contents
1842 * of the buffer are undefined after mkstemp returns an
1843 * error, so we have to rewrite the whole buffer from
1844 * scratch.
1845 */
1846 strbuf_reset(tmp);
1847 strbuf_add(tmp, filename, dirlen - 1);
1848 if (mkdir(tmp->buf, 0777) && errno != EEXIST)
b2476a60 1849 return -1;
d4b3d11a 1850 if (adjust_shared_perm(tmp->buf))
5723fe7e
LT
1851 return -1;
1852
1853 /* Try again */
d4b3d11a
JK
1854 strbuf_addstr(tmp, "/tmp_obj_XXXXXX");
1855 fd = git_mkstemp_mode(tmp->buf, 0444);
5723fe7e
LT
1856 }
1857 return fd;
1858}
1859
3fc7281f
PO
1860static int write_loose_object(const struct object_id *oid, char *hdr,
1861 int hdrlen, const void *buf, unsigned long len,
4ef91a2d 1862 time_t mtime, unsigned flags)
0fcfd160 1863{
915308b1 1864 int fd, ret;
9892beba 1865 unsigned char compressed[4096];
ef49a7a0 1866 git_zstream stream;
18e2588e 1867 git_hash_ctx c;
3fc7281f 1868 struct object_id parano_oid;
d4b3d11a 1869 static struct strbuf tmp_file = STRBUF_INIT;
ea657730
CC
1870 static struct strbuf filename = STRBUF_INIT;
1871
514c5fdd 1872 loose_object_path(the_repository, &filename, oid);
a44c9a5e 1873
ea657730 1874 fd = create_tmpfile(&tmp_file, filename.buf);
aac17941 1875 if (fd < 0) {
4ef91a2d
ÆAB
1876 if (flags & HASH_SILENT)
1877 return -1;
1878 else if (errno == EACCES)
259328b7 1879 return error(_("insufficient permission for adding an object to repository database %s"), get_object_directory());
916d081b 1880 else
259328b7 1881 return error_errno(_("unable to create temporary file"));
aac17941
LT
1882 }
1883
0fcfd160 1884 /* Set it up */
55bb5c91 1885 git_deflate_init(&stream, zlib_compression_level);
0fcfd160 1886 stream.next_out = compressed;
9892beba 1887 stream.avail_out = sizeof(compressed);
18e2588e 1888 the_hash_algo->init_fn(&c);
a44c9a5e
LT
1889
1890 /* First header.. */
d65a16f6 1891 stream.next_in = (unsigned char *)hdr;
a44c9a5e 1892 stream.avail_in = hdrlen;
55bb5c91
JH
1893 while (git_deflate(&stream, 0) == Z_OK)
1894 ; /* nothing */
18e2588e 1895 the_hash_algo->update_fn(&c, hdr, hdrlen);
a44c9a5e
LT
1896
1897 /* Then the data itself.. */
c00e657d 1898 stream.next_in = (void *)buf;
a44c9a5e 1899 stream.avail_in = len;
9892beba 1900 do {
748af44c 1901 unsigned char *in0 = stream.next_in;
55bb5c91 1902 ret = git_deflate(&stream, Z_FINISH);
18e2588e 1903 the_hash_algo->update_fn(&c, in0, stream.next_in - in0);
9892beba 1904 if (write_buffer(fd, compressed, stream.next_out - compressed) < 0)
76011357 1905 die(_("unable to write loose object file"));
9892beba
NP
1906 stream.next_out = compressed;
1907 stream.avail_out = sizeof(compressed);
1908 } while (ret == Z_OK);
1909
ac54c277 1910 if (ret != Z_STREAM_END)
259328b7 1911 die(_("unable to deflate new object %s (%d)"), oid_to_hex(oid),
3fc7281f 1912 ret);
55bb5c91 1913 ret = git_deflate_end_gently(&stream);
ac54c277 1914 if (ret != Z_OK)
259328b7 1915 die(_("deflateEnd on object %s failed (%d)"), oid_to_hex(oid),
3fc7281f 1916 ret);
5951bf46 1917 the_hash_algo->final_oid_fn(&parano_oid, &c);
9001dc2a 1918 if (!oideq(oid, &parano_oid))
259328b7 1919 die(_("confused by unstable object source data for %s"),
3fc7281f 1920 oid_to_hex(oid));
ac54c277 1921
514c5fdd 1922 close_loose_object(fd);
0fcfd160 1923
bbac7311
NP
1924 if (mtime) {
1925 struct utimbuf utb;
1926 utb.actime = mtime;
1927 utb.modtime = mtime;
4ef91a2d
ÆAB
1928 if (utime(tmp_file.buf, &utb) < 0 &&
1929 !(flags & HASH_SILENT))
259328b7 1930 warning_errno(_("failed utime() on %s"), tmp_file.buf);
bbac7311
NP
1931 }
1932
ea657730 1933 return finalize_object_file(tmp_file.buf, filename.buf);
0fcfd160 1934}
8237b185 1935
6862ebbf 1936static int freshen_loose_object(const struct object_id *oid)
33d4221c 1937{
6862ebbf 1938 return check_and_freshen(oid, 1);
33d4221c
JK
1939}
1940
6862ebbf 1941static int freshen_packed_object(const struct object_id *oid)
33d4221c
JK
1942{
1943 struct pack_entry e;
544443cb 1944 if (!find_pack_entry(the_repository, oid, &e))
ee1c6c34
JK
1945 return 0;
1946 if (e.p->freshened)
1947 return 1;
1948 if (!freshen_file(e.p->pack_name))
1949 return 0;
1950 e.p->freshened = 1;
1951 return 1;
33d4221c
JK
1952}
1953
4ef91a2d
ÆAB
1954int write_object_file_flags(const void *buf, unsigned long len,
1955 const char *type, struct object_id *oid,
1956 unsigned flags)
bbac7311 1957{
1af64f73 1958 char hdr[MAX_HEADER_LEN];
ef1286d3 1959 int hdrlen = sizeof(hdr);
bbac7311
NP
1960
1961 /* Normally if we have it in the pack then we do not bother writing
1962 * it out into .git/objects/??/?{38} file.
1963 */
7ad5c44d
MT
1964 write_object_file_prepare(the_hash_algo, buf, len, type, oid, hdr,
1965 &hdrlen);
6862ebbf 1966 if (freshen_packed_object(oid) || freshen_loose_object(oid))
bbac7311 1967 return 0;
4ef91a2d 1968 return write_loose_object(oid, hdr, hdrlen, buf, len, 0, flags);
bbac7311
NP
1969}
1970
1752cbbc
PO
1971int hash_object_file_literally(const void *buf, unsigned long len,
1972 const char *type, struct object_id *oid,
1973 unsigned flags)
0c3db67c
ES
1974{
1975 char *header;
1976 int hdrlen, status = 0;
1977
1978 /* type string, SP, %lu of the length plus NUL must fit this */
1af64f73 1979 hdrlen = strlen(type) + MAX_HEADER_LEN;
ef1286d3 1980 header = xmalloc(hdrlen);
7ad5c44d
MT
1981 write_object_file_prepare(the_hash_algo, buf, len, type, oid, header,
1982 &hdrlen);
0c3db67c
ES
1983
1984 if (!(flags & HASH_WRITE_OBJECT))
1985 goto cleanup;
6862ebbf 1986 if (freshen_packed_object(oid) || freshen_loose_object(oid))
0c3db67c 1987 goto cleanup;
4ef91a2d 1988 status = write_loose_object(oid, header, hdrlen, buf, len, 0, 0);
0c3db67c
ES
1989
1990cleanup:
1991 free(header);
1992 return status;
1993}
1994
4bdb70a4 1995int force_object_loose(const struct object_id *oid, time_t mtime)
bbac7311 1996{
bbac7311
NP
1997 void *buf;
1998 unsigned long len;
1999 enum object_type type;
1af64f73 2000 char hdr[MAX_HEADER_LEN];
bbac7311 2001 int hdrlen;
1fb23e65 2002 int ret;
bbac7311 2003
6862ebbf 2004 if (has_loose_object(oid))
bbac7311 2005 return 0;
cba595ab 2006 buf = read_object(the_repository, oid, &type, &len);
bbac7311 2007 if (!buf)
2c319886 2008 return error(_("cannot read object for %s"), oid_to_hex(oid));
ca473cef 2009 hdrlen = xsnprintf(hdr, sizeof(hdr), "%s %"PRIuMAX , type_name(type), (uintmax_t)len) + 1;
4ef91a2d 2010 ret = write_loose_object(oid, hdr, hdrlen, buf, len, mtime, 0);
1fb23e65
BS
2011 free(buf);
2012
2013 return ret;
bbac7311
NP
2014}
2015
1d8d9cb6
JT
2016int has_object(struct repository *r, const struct object_id *oid,
2017 unsigned flags)
2018{
2019 int quick = !(flags & HAS_OBJECT_RECHECK_PACKED);
2020 unsigned object_info_flags = OBJECT_INFO_SKIP_FETCH_OBJECT |
2021 (quick ? OBJECT_INFO_QUICK : 0);
2022
2023 if (!startup_info->have_repository)
2024 return 0;
2025 return oid_object_info_extended(r, oid, NULL, object_info_flags) >= 0;
2026}
2027
cba595ab
JH
2028int repo_has_object_file_with_flags(struct repository *r,
2029 const struct object_id *oid, int flags)
8237b185 2030{
3e8b7d3c
JN
2031 if (!startup_info->have_repository)
2032 return 0;
9c8a294a 2033 return oid_object_info_extended(r, oid, NULL, flags) >= 0;
b419aa25 2034}
2035
9b45f499
SB
2036int repo_has_object_file(struct repository *r,
2037 const struct object_id *oid)
b419aa25 2038{
cba595ab 2039 return repo_has_object_file_with_flags(r, oid, 0);
5827a035
JK
2040}
2041
c879daa2
NTND
2042static void check_tree(const void *buf, size_t size)
2043{
2044 struct tree_desc desc;
2045 struct name_entry entry;
2046
2047 init_tree_desc(&desc, buf, size);
2048 while (tree_entry(&desc, &entry))
2049 /* do nothing
2050 * tree_entry() will die() on malformed entries */
2051 ;
2052}
2053
2054static void check_commit(const void *buf, size_t size)
2055{
2056 struct commit c;
2057 memset(&c, 0, sizeof(c));
08f4f445 2058 if (parse_commit_buffer(the_repository, &c, buf, size, 0))
259328b7 2059 die(_("corrupt commit"));
c879daa2
NTND
2060}
2061
2062static void check_tag(const void *buf, size_t size)
2063{
2064 struct tag t;
2065 memset(&t, 0, sizeof(t));
0e740fed 2066 if (parse_tag_buffer(the_repository, &t, buf, size))
259328b7 2067 die(_("corrupt tag"));
c879daa2
NTND
2068}
2069
58bf2a4c
NTND
2070static int index_mem(struct index_state *istate,
2071 struct object_id *oid, void *buf, size_t size,
c4ce46fc
JH
2072 enum object_type type,
2073 const char *path, unsigned flags)
e7332f96 2074{
6c510bee 2075 int ret, re_allocated = 0;
c4ce46fc 2076 int write_object = flags & HASH_WRITE_OBJECT;
74400e71 2077
7672db20 2078 if (!type)
edaec3fb 2079 type = OBJ_BLOB;
6c510bee
LT
2080
2081 /*
2082 * Convert blobs to git internal format
2083 */
43df4f86 2084 if ((type == OBJ_BLOB) && path) {
f285a2d7 2085 struct strbuf nbuf = STRBUF_INIT;
58bf2a4c 2086 if (convert_to_git(istate, path, buf, size, &nbuf,
8462ff43 2087 get_conv_flags(flags))) {
b315c5c0 2088 buf = strbuf_detach(&nbuf, &size);
6c510bee
LT
2089 re_allocated = 1;
2090 }
2091 }
c4ce46fc 2092 if (flags & HASH_FORMAT_CHECK) {
c879daa2
NTND
2093 if (type == OBJ_TREE)
2094 check_tree(buf, size);
2095 if (type == OBJ_COMMIT)
2096 check_commit(buf, size);
2097 if (type == OBJ_TAG)
2098 check_tag(buf, size);
2099 }
6c510bee 2100
7672db20 2101 if (write_object)
169c9c01 2102 ret = write_object_file(buf, size, type_name(type), oid);
abdc3fc8 2103 else
2dcde20e
MT
2104 ret = hash_object_file(the_hash_algo, buf, size,
2105 type_name(type), oid);
43df4f86 2106 if (re_allocated)
6c510bee 2107 free(buf);
43df4f86
DP
2108 return ret;
2109}
2110
58bf2a4c
NTND
2111static int index_stream_convert_blob(struct index_state *istate,
2112 struct object_id *oid,
2113 int fd,
2114 const char *path,
2115 unsigned flags)
9035d75a
SP
2116{
2117 int ret;
2118 const int write_object = flags & HASH_WRITE_OBJECT;
2119 struct strbuf sbuf = STRBUF_INIT;
2120
2121 assert(path);
58bf2a4c 2122 assert(would_convert_to_git_filter_fd(istate, path));
9035d75a 2123
58bf2a4c 2124 convert_to_git_filter_fd(istate, path, fd, &sbuf,
8462ff43 2125 get_conv_flags(flags));
9035d75a
SP
2126
2127 if (write_object)
169c9c01 2128 ret = write_object_file(sbuf.buf, sbuf.len, type_name(OBJ_BLOB),
a09c985e 2129 oid);
9035d75a 2130 else
2dcde20e
MT
2131 ret = hash_object_file(the_hash_algo, sbuf.buf, sbuf.len,
2132 type_name(OBJ_BLOB), oid);
9035d75a
SP
2133 strbuf_release(&sbuf);
2134 return ret;
2135}
2136
58bf2a4c
NTND
2137static int index_pipe(struct index_state *istate, struct object_id *oid,
2138 int fd, enum object_type type,
7b41e1e1
JH
2139 const char *path, unsigned flags)
2140{
2141 struct strbuf sbuf = STRBUF_INIT;
2142 int ret;
2143
2144 if (strbuf_read(&sbuf, fd, 4096) >= 0)
58bf2a4c 2145 ret = index_mem(istate, oid, sbuf.buf, sbuf.len, type, path, flags);
7b41e1e1
JH
2146 else
2147 ret = -1;
2148 strbuf_release(&sbuf);
2149 return ret;
2150}
2151
ea68b0ce
DP
2152#define SMALL_FILE_SIZE (32*1024)
2153
58bf2a4c
NTND
2154static int index_core(struct index_state *istate,
2155 struct object_id *oid, int fd, size_t size,
7b41e1e1
JH
2156 enum object_type type, const char *path,
2157 unsigned flags)
43df4f86
DP
2158{
2159 int ret;
43df4f86 2160
7b41e1e1 2161 if (!size) {
58bf2a4c 2162 ret = index_mem(istate, oid, "", size, type, path, flags);
ea68b0ce
DP
2163 } else if (size <= SMALL_FILE_SIZE) {
2164 char *buf = xmalloc(size);
90dca671
JK
2165 ssize_t read_result = read_in_full(fd, buf, size);
2166 if (read_result < 0)
259328b7 2167 ret = error_errno(_("read error while indexing %s"),
90dca671
JK
2168 path ? path : "<unknown>");
2169 else if (read_result != size)
259328b7 2170 ret = error(_("short read while indexing %s"),
90dca671 2171 path ? path : "<unknown>");
ea68b0ce 2172 else
58bf2a4c 2173 ret = index_mem(istate, oid, buf, size, type, path, flags);
ea68b0ce 2174 free(buf);
08bda208 2175 } else {
43df4f86 2176 void *buf = xmmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
58bf2a4c 2177 ret = index_mem(istate, oid, buf, size, type, path, flags);
aac17941 2178 munmap(buf, size);
08bda208 2179 }
7b41e1e1
JH
2180 return ret;
2181}
2182
4dd1fbc7 2183/*
568508e7
JH
2184 * This creates one packfile per large blob unless bulk-checkin
2185 * machinery is "plugged".
4dd1fbc7
JH
2186 *
2187 * This also bypasses the usual "convert-to-git" dance, and that is on
2188 * purpose. We could write a streaming version of the converting
2189 * functions and insert that before feeding the data to fast-import
4f22b101
JK
2190 * (or equivalent in-core API described above). However, that is
2191 * somewhat complicated, as we do not know the size of the filter
2192 * result, which we need to know beforehand when writing a git object.
2193 * Since the primary motivation for trying to stream from the working
2194 * tree file and to avoid mmaping it in core is to deal with large
2195 * binary blobs, they generally do not want to get any conversion, and
2196 * callers should avoid this code path when filters are requested.
4dd1fbc7 2197 */
7d5e1dc3 2198static int index_stream(struct object_id *oid, int fd, size_t size,
4dd1fbc7
JH
2199 enum object_type type, const char *path,
2200 unsigned flags)
2201{
68ee6dfc 2202 return index_bulk_checkin(oid, fd, size, type, path, flags);
4dd1fbc7
JH
2203}
2204
58bf2a4c
NTND
2205int index_fd(struct index_state *istate, struct object_id *oid,
2206 int fd, struct stat *st,
7b41e1e1
JH
2207 enum object_type type, const char *path, unsigned flags)
2208{
2209 int ret;
7b41e1e1 2210
9079ab7c
SP
2211 /*
2212 * Call xsize_t() only when needed to avoid potentially unnecessary
2213 * die() for large files.
2214 */
58bf2a4c
NTND
2215 if (type == OBJ_BLOB && path && would_convert_to_git_filter_fd(istate, path))
2216 ret = index_stream_convert_blob(istate, oid, fd, path, flags);
9035d75a 2217 else if (!S_ISREG(st->st_mode))
58bf2a4c 2218 ret = index_pipe(istate, oid, fd, type, path, flags);
9079ab7c 2219 else if (st->st_size <= big_file_threshold || type != OBJ_BLOB ||
58bf2a4c
NTND
2220 (path && would_convert_to_git(istate, path)))
2221 ret = index_core(istate, oid, fd, xsize_t(st->st_size),
2222 type, path, flags);
4dd1fbc7 2223 else
7d5e1dc3 2224 ret = index_stream(oid, fd, xsize_t(st->st_size), type, path,
9079ab7c 2225 flags);
43df4f86 2226 close(fd);
aac17941 2227 return ret;
74400e71 2228}
ec1fcc16 2229
58bf2a4c
NTND
2230int index_path(struct index_state *istate, struct object_id *oid,
2231 const char *path, struct stat *st, unsigned flags)
ec1fcc16
JH
2232{
2233 int fd;
b760d3aa 2234 struct strbuf sb = STRBUF_INIT;
ea8e0297 2235 int rc = 0;
ec1fcc16
JH
2236
2237 switch (st->st_mode & S_IFMT) {
2238 case S_IFREG:
2239 fd = open(path, O_RDONLY);
2240 if (fd < 0)
7616c6ca 2241 return error_errno("open(\"%s\")", path);
58bf2a4c 2242 if (index_fd(istate, oid, fd, st, OBJ_BLOB, path, flags) < 0)
259328b7 2243 return error(_("%s: failed to insert into database"),
ec1fcc16
JH
2244 path);
2245 break;
2246 case S_IFLNK:
7616c6ca
NTND
2247 if (strbuf_readlink(&sb, path, st->st_size))
2248 return error_errno("readlink(\"%s\")", path);
c4ce46fc 2249 if (!(flags & HASH_WRITE_OBJECT))
2dcde20e
MT
2250 hash_object_file(the_hash_algo, sb.buf, sb.len,
2251 blob_type, oid);
a09c985e 2252 else if (write_object_file(sb.buf, sb.len, blob_type, oid))
259328b7 2253 rc = error(_("%s: failed to insert into database"), path);
b760d3aa 2254 strbuf_release(&sb);
ec1fcc16 2255 break;
f35a6d3b 2256 case S_IFDIR:
a98e6101 2257 return resolve_gitlink_ref(path, "HEAD", oid);
ec1fcc16 2258 default:
259328b7 2259 return error(_("%s: unsupported file type"), path);
ec1fcc16 2260 }
ea8e0297 2261 return rc;
ec1fcc16 2262}
a69e5429
JH
2263
2264int read_pack_header(int fd, struct pack_header *header)
2265{
f48ecd38 2266 if (read_in_full(fd, header, sizeof(*header)) != sizeof(*header))
c697ad14
HO
2267 /* "eof before pack header was fully read" */
2268 return PH_ERROR_EOF;
2269
a69e5429
JH
2270 if (header->hdr_signature != htonl(PACK_SIGNATURE))
2271 /* "protocol error (pack signature mismatch detected)" */
2272 return PH_ERROR_PACK_SIGNATURE;
2273 if (!pack_version_ok(header->hdr_version))
2274 /* "protocol error (pack version unsupported)" */
2275 return PH_ERROR_PROTOCOL;
2276 return 0;
2277}
40d52ff7 2278
e816caa0 2279void assert_oid_type(const struct object_id *oid, enum object_type expect)
40d52ff7 2280{
0df8e965 2281 enum object_type type = oid_object_info(the_repository, oid, NULL);
40d52ff7 2282 if (type < 0)
259328b7 2283 die(_("%s is not a valid object"), oid_to_hex(oid));
40d52ff7 2284 if (type != expect)
259328b7 2285 die(_("%s is not a valid '%s' object"), oid_to_hex(oid),
debca9d2 2286 type_name(expect));
40d52ff7 2287}
27e1e22d 2288
70c49050 2289int for_each_file_in_obj_subdir(unsigned int subdir_nr,
cc817ca3
RS
2290 struct strbuf *path,
2291 each_loose_object_fn obj_cb,
2292 each_loose_cruft_fn cruft_cb,
2293 each_loose_subdir_fn subdir_cb,
2294 void *data)
27e1e22d 2295{
0375f472
RS
2296 size_t origlen, baselen;
2297 DIR *dir;
27e1e22d
JK
2298 struct dirent *de;
2299 int r = 0;
62a24c89 2300 struct object_id oid;
27e1e22d 2301
70c49050
RS
2302 if (subdir_nr > 0xff)
2303 BUG("invalid loose object subdirectory: %x", subdir_nr);
2304
0375f472
RS
2305 origlen = path->len;
2306 strbuf_complete(path, '/');
2307 strbuf_addf(path, "%02x", subdir_nr);
0375f472
RS
2308
2309 dir = opendir(path->buf);
27e1e22d 2310 if (!dir) {
0375f472 2311 if (errno != ENOENT)
259328b7 2312 r = error_errno(_("unable to open %s"), path->buf);
0375f472
RS
2313 strbuf_setlen(path, origlen);
2314 return r;
27e1e22d
JK
2315 }
2316
62a24c89 2317 oid.hash[0] = subdir_nr;
163ee5e6
DS
2318 strbuf_addch(path, '/');
2319 baselen = path->len;
62a24c89 2320
b548f0f1 2321 while ((de = readdir_skip_dot_and_dotdot(dir))) {
163ee5e6 2322 size_t namelen;
27e1e22d 2323
163ee5e6 2324 namelen = strlen(de->d_name);
27e1e22d 2325 strbuf_setlen(path, baselen);
163ee5e6 2326 strbuf_add(path, de->d_name, namelen);
94b5e093 2327 if (namelen == the_hash_algo->hexsz - 2 &&
62a24c89 2328 !hex_to_bytes(oid.hash + 1, de->d_name,
94b5e093 2329 the_hash_algo->rawsz - 1)) {
5a6dce70 2330 oid_set_algo(&oid, the_hash_algo);
62a24c89
RS
2331 if (obj_cb) {
2332 r = obj_cb(&oid, path->buf, data);
2333 if (r)
2334 break;
27e1e22d 2335 }
62a24c89 2336 continue;
27e1e22d
JK
2337 }
2338
2339 if (cruft_cb) {
2340 r = cruft_cb(de->d_name, path->buf, data);
2341 if (r)
2342 break;
2343 }
2344 }
094c7e63 2345 closedir(dir);
27e1e22d 2346
163ee5e6 2347 strbuf_setlen(path, baselen - 1);
27e1e22d
JK
2348 if (!r && subdir_cb)
2349 r = subdir_cb(subdir_nr, path->buf, data);
2350
0375f472
RS
2351 strbuf_setlen(path, origlen);
2352
27e1e22d
JK
2353 return r;
2354}
2355
e6f875e0 2356int for_each_loose_file_in_objdir_buf(struct strbuf *path,
27e1e22d
JK
2357 each_loose_object_fn obj_cb,
2358 each_loose_cruft_fn cruft_cb,
2359 each_loose_subdir_fn subdir_cb,
2360 void *data)
2361{
27e1e22d
JK
2362 int r = 0;
2363 int i;
2364
27e1e22d 2365 for (i = 0; i < 256; i++) {
e6f875e0 2366 r = for_each_file_in_obj_subdir(i, path, obj_cb, cruft_cb,
27e1e22d 2367 subdir_cb, data);
27e1e22d
JK
2368 if (r)
2369 break;
2370 }
2371
e6f875e0
JK
2372 return r;
2373}
2374
2375int for_each_loose_file_in_objdir(const char *path,
2376 each_loose_object_fn obj_cb,
2377 each_loose_cruft_fn cruft_cb,
2378 each_loose_subdir_fn subdir_cb,
2379 void *data)
2380{
2381 struct strbuf buf = STRBUF_INIT;
2382 int r;
2383
2384 strbuf_addstr(&buf, path);
2385 r = for_each_loose_file_in_objdir_buf(&buf, obj_cb, cruft_cb,
2386 subdir_cb, data);
27e1e22d 2387 strbuf_release(&buf);
e6f875e0 2388
27e1e22d
JK
2389 return r;
2390}
660c889e 2391
a7ff6f5a
JK
2392int for_each_loose_object(each_loose_object_fn cb, void *data,
2393 enum for_each_object_flags flags)
660c889e 2394{
f0eaf638 2395 struct object_directory *odb;
b0a42642 2396
f0eaf638
JK
2397 prepare_alt_odb(the_repository);
2398 for (odb = the_repository->objects->odb; odb; odb = odb->next) {
2399 int r = for_each_loose_file_in_objdir(odb->path, cb, NULL,
2400 NULL, data);
2401 if (r)
2402 return r;
660c889e 2403
f0eaf638
JK
2404 if (flags & FOR_EACH_OBJECT_LOCAL_ONLY)
2405 break;
2406 }
1385bb7b 2407
f0eaf638 2408 return 0;
660c889e
JK
2409}
2410
3a2e0824
JK
2411static int append_loose_object(const struct object_id *oid, const char *path,
2412 void *data)
660c889e 2413{
92d8ed8a 2414 oidtree_insert(data, oid);
3a2e0824
JK
2415 return 0;
2416}
660c889e 2417
92d8ed8a 2418struct oidtree *odb_loose_cache(struct object_directory *odb,
0000d654
RS
2419 const struct object_id *oid)
2420{
2421 int subdir_nr = oid->hash[0];
3a2e0824 2422 struct strbuf buf = STRBUF_INIT;
33f379ee
EW
2423 size_t word_bits = bitsizeof(odb->loose_objects_subdir_seen[0]);
2424 size_t word_index = subdir_nr / word_bits;
26de1fc0 2425 size_t mask = (size_t)1u << (subdir_nr % word_bits);
33f379ee 2426 uint32_t *bitmap;
660c889e 2427
3a2e0824 2428 if (subdir_nr < 0 ||
33f379ee 2429 subdir_nr >= bitsizeof(odb->loose_objects_subdir_seen))
3a2e0824 2430 BUG("subdir_nr out of range");
1385bb7b 2431
33f379ee
EW
2432 bitmap = &odb->loose_objects_subdir_seen[word_index];
2433 if (*bitmap & mask)
92d8ed8a
EW
2434 return odb->loose_objects_cache;
2435 if (!odb->loose_objects_cache) {
2436 ALLOC_ARRAY(odb->loose_objects_cache, 1);
2437 oidtree_init(odb->loose_objects_cache);
2438 }
3a2e0824
JK
2439 strbuf_addstr(&buf, odb->path);
2440 for_each_file_in_obj_subdir(subdir_nr, &buf,
2441 append_loose_object,
2442 NULL, NULL,
92d8ed8a 2443 odb->loose_objects_cache);
33f379ee 2444 *bitmap |= mask;
7317aa71 2445 strbuf_release(&buf);
92d8ed8a 2446 return odb->loose_objects_cache;
660c889e
JK
2447}
2448
d4e19e51
RS
2449void odb_clear_loose_cache(struct object_directory *odb)
2450{
92d8ed8a
EW
2451 oidtree_clear(odb->loose_objects_cache);
2452 FREE_AND_NULL(odb->loose_objects_cache);
d4e19e51
RS
2453 memset(&odb->loose_objects_subdir_seen, 0,
2454 sizeof(odb->loose_objects_subdir_seen));
2455}
2456
00a7760e
JK
2457static int check_stream_oid(git_zstream *stream,
2458 const char *hdr,
2459 unsigned long size,
2460 const char *path,
2461 const struct object_id *expected_oid)
f6371f92 2462{
18e2588e 2463 git_hash_ctx c;
00a7760e 2464 struct object_id real_oid;
f6371f92
JK
2465 unsigned char buf[4096];
2466 unsigned long total_read;
2467 int status = Z_OK;
2468
18e2588e 2469 the_hash_algo->init_fn(&c);
2470 the_hash_algo->update_fn(&c, hdr, stream->total_out);
f6371f92
JK
2471
2472 /*
2473 * We already read some bytes into hdr, but the ones up to the NUL
2474 * do not count against the object's content size.
2475 */
2476 total_read = stream->total_out - strlen(hdr) - 1;
2477
2478 /*
2479 * This size comparison must be "<=" to read the final zlib packets;
00a7760e 2480 * see the comment in unpack_loose_rest for details.
f6371f92
JK
2481 */
2482 while (total_read <= size &&
18ad13e5
JH
2483 (status == Z_OK ||
2484 (status == Z_BUF_ERROR && !stream->avail_out))) {
f6371f92
JK
2485 stream->next_out = buf;
2486 stream->avail_out = sizeof(buf);
2487 if (size - total_read < stream->avail_out)
2488 stream->avail_out = size - total_read;
2489 status = git_inflate(stream, Z_FINISH);
18e2588e 2490 the_hash_algo->update_fn(&c, buf, stream->next_out - buf);
f6371f92
JK
2491 total_read += stream->next_out - buf;
2492 }
2493 git_inflate_end(stream);
2494
2495 if (status != Z_STREAM_END) {
00a7760e 2496 error(_("corrupt loose object '%s'"), oid_to_hex(expected_oid));
f6371f92
JK
2497 return -1;
2498 }
cce044df 2499 if (stream->avail_in) {
259328b7 2500 error(_("garbage at end of loose object '%s'"),
00a7760e 2501 oid_to_hex(expected_oid));
cce044df
JK
2502 return -1;
2503 }
f6371f92 2504
5951bf46 2505 the_hash_algo->final_oid_fn(&real_oid, &c);
00a7760e 2506 if (!oideq(expected_oid, &real_oid)) {
01f8d594 2507 error(_("hash mismatch for %s (expected %s)"), path,
00a7760e 2508 oid_to_hex(expected_oid));
f6371f92
JK
2509 return -1;
2510 }
2511
2512 return 0;
2513}
2514
2515int read_loose_object(const char *path,
d61d87bd 2516 const struct object_id *expected_oid,
96e41f58 2517 struct object_id *real_oid,
31deb28f
ÆAB
2518 void **contents,
2519 struct object_info *oi)
f6371f92
JK
2520{
2521 int ret = -1;
f6371f92
JK
2522 void *map = NULL;
2523 unsigned long mapsize;
2524 git_zstream stream;
1af64f73 2525 char hdr[MAX_HEADER_LEN];
31deb28f 2526 unsigned long *size = oi->sizep;
f6371f92 2527
514c5fdd 2528 map = map_loose_object_1(the_repository, path, NULL, &mapsize);
f6371f92 2529 if (!map) {
259328b7 2530 error_errno(_("unable to mmap %s"), path);
f6371f92
JK
2531 goto out;
2532 }
2533
01cab976
ÆAB
2534 if (unpack_loose_header(&stream, map, mapsize, hdr, sizeof(hdr),
2535 NULL) < 0) {
259328b7 2536 error(_("unable to unpack header of %s"), path);
f6371f92
JK
2537 goto out;
2538 }
2539
31deb28f 2540 if (parse_loose_header(hdr, oi) < 0) {
259328b7 2541 error(_("unable to parse header of %s"), path);
f6371f92
JK
2542 git_inflate_end(&stream);
2543 goto out;
2544 }
2545
31deb28f 2546 if (*oi->typep == OBJ_BLOB && *size > big_file_threshold) {
00a7760e 2547 if (check_stream_oid(&stream, hdr, *size, path, expected_oid) < 0)
f6371f92
JK
2548 goto out;
2549 } else {
00a7760e 2550 *contents = unpack_loose_rest(&stream, hdr, *size, expected_oid);
f6371f92 2551 if (!*contents) {
259328b7 2552 error(_("unable to unpack contents of %s"), path);
f6371f92
JK
2553 git_inflate_end(&stream);
2554 goto out;
2555 }
b98d1885 2556 if (check_object_signature(the_repository, expected_oid,
16235e3b
ÆAB
2557 *contents, *size,
2558 oi->type_name->buf, real_oid))
f6371f92 2559 goto out;
f6371f92
JK
2560 }
2561
2562 ret = 0; /* everything checks out */
2563
2564out:
2565 if (map)
2566 munmap(map, mapsize);
f6371f92
JK
2567 return ret;
2568}