![[tokkee]](http://tokkee.org/images/avatar.png){kind=avatar}
1 #include "builtin.h"
2 #include "cache.h"
3 #include "object.h"
4 #include "blob.h"
5 #include "commit.h"
6 #include "tag.h"
7 #include "tree.h"
8 #include "delta.h"
9 #include "pack.h"
10 #include "csum-file.h"
11 #include "tree-walk.h"
12 #include "diff.h"
13 #include "revision.h"
14 #include "list-objects.h"
15 #include "progress.h"
17 static const char pack_usage[] = "\
18 git-pack-objects [{ -q | --progress | --all-progress }] \n\
19 [--local] [--incremental] [--window=N] [--depth=N] \n\
20 [--no-reuse-delta] [--delta-base-offset] [--non-empty] \n\
21 [--revs [--unpacked | --all]*] [--reflog] [--stdout | base-name] \n\
22 [<ref-list | <object-list]";
24 struct object_entry {
25 unsigned char sha1[20];
26 uint32_t crc32; /* crc of raw pack data for this object */
27 off_t offset; /* offset into the final pack file */
28 unsigned long size; /* uncompressed size */
29 unsigned int hash; /* name hint hash */
30 unsigned int depth; /* delta depth */
31 struct packed_git *in_pack; /* already in pack */
32 off_t in_pack_offset;
33 struct object_entry *delta; /* delta base object */
34 struct object_entry *delta_child; /* deltified objects who bases me */
35 struct object_entry *delta_sibling; /* other deltified objects who
36 * uses the same base as me
37 */
38 unsigned long delta_size; /* delta data size (uncompressed) */
39 enum object_type type;
40 enum object_type in_pack_type; /* could be delta */
41 unsigned char in_pack_header_size;
42 unsigned char preferred_base; /* we do not pack this, but is available
43 * to be used as the base objectto delta
44 * objects against.
45 */
46 };
48 /*
49 * Objects we are going to pack are collected in objects array (dynamically
50 * expanded). nr_objects & nr_alloc controls this array. They are stored
51 * in the order we see -- typically rev-list --objects order that gives us
52 * nice "minimum seek" order.
53 */
54 static struct object_entry *objects;
55 static uint32_t nr_objects, nr_alloc, nr_result;
57 static int non_empty;
58 static int no_reuse_delta;
59 static int local;
60 static int incremental;
61 static int allow_ofs_delta;
62 static const char *pack_tmp_name, *idx_tmp_name;
63 static char tmpname[PATH_MAX];
64 static unsigned char pack_file_sha1[20];
65 static int progress = 1;
66 static int window = 10;
67 static int depth = 50;
68 static int pack_to_stdout;
69 static int num_preferred_base;
70 static struct progress progress_state;
72 /*
73 * The object names in objects array are hashed with this hashtable,
74 * to help looking up the entry by object name.
75 * This hashtable is built after all the objects are seen.
76 */
77 static int *object_ix;
78 static int object_ix_hashsz;
80 /*
81 * Pack index for existing packs give us easy access to the offsets into
82 * corresponding pack file where each object's data starts, but the entries
83 * do not store the size of the compressed representation (uncompressed
84 * size is easily available by examining the pack entry header). It is
85 * also rather expensive to find the sha1 for an object given its offset.
86 *
87 * We build a hashtable of existing packs (pack_revindex), and keep reverse
88 * index here -- pack index file is sorted by object name mapping to offset;
89 * this pack_revindex[].revindex array is a list of offset/index_nr pairs
90 * ordered by offset, so if you know the offset of an object, next offset
91 * is where its packed representation ends and the index_nr can be used to
92 * get the object sha1 from the main index.
93 */
94 struct revindex_entry {
95 off_t offset;
96 unsigned int nr;
97 };
98 struct pack_revindex {
99 struct packed_git *p;
100 struct revindex_entry *revindex;
101 };
102 static struct pack_revindex *pack_revindex;
103 static int pack_revindex_hashsz;
105 /*
106 * stats
107 */
108 static uint32_t written, written_delta;
109 static uint32_t reused, reused_delta;
111 static int pack_revindex_ix(struct packed_git *p)
112 {
113 unsigned long ui = (unsigned long)p;
114 int i;
116 ui = ui ^ (ui >> 16); /* defeat structure alignment */
117 i = (int)(ui % pack_revindex_hashsz);
118 while (pack_revindex[i].p) {
119 if (pack_revindex[i].p == p)
120 return i;
121 if (++i == pack_revindex_hashsz)
122 i = 0;
123 }
124 return -1 - i;
125 }
127 static void prepare_pack_ix(void)
128 {
129 int num;
130 struct packed_git *p;
131 for (num = 0, p = packed_git; p; p = p->next)
132 num++;
133 if (!num)
134 return;
135 pack_revindex_hashsz = num * 11;
136 pack_revindex = xcalloc(sizeof(*pack_revindex), pack_revindex_hashsz);
137 for (p = packed_git; p; p = p->next) {
138 num = pack_revindex_ix(p);
139 num = - 1 - num;
140 pack_revindex[num].p = p;
141 }
142 /* revindex elements are lazily initialized */
143 }
145 static int cmp_offset(const void *a_, const void *b_)
146 {
147 const struct revindex_entry *a = a_;
148 const struct revindex_entry *b = b_;
149 return (a->offset < b->offset) ? -1 : (a->offset > b->offset) ? 1 : 0;
150 }
152 /*
153 * Ordered list of offsets of objects in the pack.
154 */
155 static void prepare_pack_revindex(struct pack_revindex *rix)
156 {
157 struct packed_git *p = rix->p;
158 int num_ent = p->num_objects;
159 int i;
160 const char *index = p->index_data;
162 rix->revindex = xmalloc(sizeof(*rix->revindex) * (num_ent + 1));
163 index += 4 * 256;
165 if (p->index_version > 1) {
166 const uint32_t *off_32 =
167 (uint32_t *)(index + 8 + p->num_objects * (20 + 4));
168 const uint32_t *off_64 = off_32 + p->num_objects;
169 for (i = 0; i < num_ent; i++) {
170 uint32_t off = ntohl(*off_32++);
171 if (!(off & 0x80000000)) {
172 rix->revindex[i].offset = off;
173 } else {
174 rix->revindex[i].offset =
175 ((uint64_t)ntohl(*off_64++)) << 32;
176 rix->revindex[i].offset |=
177 ntohl(*off_64++);
178 }
179 rix->revindex[i].nr = i;
180 }
181 } else {
182 for (i = 0; i < num_ent; i++) {
183 uint32_t hl = *((uint32_t *)(index + 24 * i));
184 rix->revindex[i].offset = ntohl(hl);
185 rix->revindex[i].nr = i;
186 }
187 }
189 /* This knows the pack format -- the 20-byte trailer
190 * follows immediately after the last object data.
191 */
192 rix->revindex[num_ent].offset = p->pack_size - 20;
193 rix->revindex[num_ent].nr = -1;
194 qsort(rix->revindex, num_ent, sizeof(*rix->revindex), cmp_offset);
195 }
197 static struct revindex_entry * find_packed_object(struct packed_git *p,
198 off_t ofs)
199 {
200 int num;
201 int lo, hi;
202 struct pack_revindex *rix;
203 struct revindex_entry *revindex;
204 num = pack_revindex_ix(p);
205 if (num < 0)
206 die("internal error: pack revindex uninitialized");
207 rix = &pack_revindex[num];
208 if (!rix->revindex)
209 prepare_pack_revindex(rix);
210 revindex = rix->revindex;
211 lo = 0;
212 hi = p->num_objects + 1;
213 do {
214 int mi = (lo + hi) / 2;
215 if (revindex[mi].offset == ofs) {
216 return revindex + mi;
217 }
218 else if (ofs < revindex[mi].offset)
219 hi = mi;
220 else
221 lo = mi + 1;
222 } while (lo < hi);
223 die("internal error: pack revindex corrupt");
224 }
226 static const unsigned char *find_packed_object_name(struct packed_git *p,
227 off_t ofs)
228 {
229 struct revindex_entry *entry = find_packed_object(p, ofs);
230 return nth_packed_object_sha1(p, entry->nr);
231 }
233 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
234 {
235 unsigned long othersize, delta_size;
236 enum object_type type;
237 void *otherbuf = read_sha1_file(entry->delta->sha1, &type, &othersize);
238 void *delta_buf;
240 if (!otherbuf)
241 die("unable to read %s", sha1_to_hex(entry->delta->sha1));
242 delta_buf = diff_delta(otherbuf, othersize,
243 buf, size, &delta_size, 0);
244 if (!delta_buf || delta_size != entry->delta_size)
245 die("delta size changed");
246 free(buf);
247 free(otherbuf);
248 return delta_buf;
249 }
251 /*
252 * The per-object header is a pretty dense thing, which is
253 * - first byte: low four bits are "size", then three bits of "type",
254 * and the high bit is "size continues".
255 * - each byte afterwards: low seven bits are size continuation,
256 * with the high bit being "size continues"
257 */
258 static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr)
259 {
260 int n = 1;
261 unsigned char c;
263 if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
264 die("bad type %d", type);
266 c = (type << 4) | (size & 15);
267 size >>= 4;
268 while (size) {
269 *hdr++ = c | 0x80;
270 c = size & 0x7f;
271 size >>= 7;
272 n++;
273 }
274 *hdr = c;
275 return n;
276 }
278 /*
279 * we are going to reuse the existing object data as is. make
280 * sure it is not corrupt.
281 */
282 static int check_pack_inflate(struct packed_git *p,
283 struct pack_window **w_curs,
284 off_t offset,
285 off_t len,
286 unsigned long expect)
287 {
288 z_stream stream;
289 unsigned char fakebuf[4096], *in;
290 int st;
292 memset(&stream, 0, sizeof(stream));
293 inflateInit(&stream);
294 do {
295 in = use_pack(p, w_curs, offset, &stream.avail_in);
296 stream.next_in = in;
297 stream.next_out = fakebuf;
298 stream.avail_out = sizeof(fakebuf);
299 st = inflate(&stream, Z_FINISH);
300 offset += stream.next_in - in;
301 } while (st == Z_OK || st == Z_BUF_ERROR);
302 inflateEnd(&stream);
303 return (st == Z_STREAM_END &&
304 stream.total_out == expect &&
305 stream.total_in == len) ? 0 : -1;
306 }
308 static int check_pack_crc(struct packed_git *p, struct pack_window **w_curs,
309 off_t offset, off_t len, unsigned int nr)
310 {
311 const uint32_t *index_crc;
312 uint32_t data_crc = crc32(0, Z_NULL, 0);
314 do {
315 unsigned int avail;
316 void *data = use_pack(p, w_curs, offset, &avail);
317 if (avail > len)
318 avail = len;
319 data_crc = crc32(data_crc, data, avail);
320 offset += avail;
321 len -= avail;
322 } while (len);
324 index_crc = p->index_data;
325 index_crc += 2 + 256 + p->num_objects * (20/4) + nr;
327 return data_crc != ntohl(*index_crc);
328 }
330 static void copy_pack_data(struct sha1file *f,
331 struct packed_git *p,
332 struct pack_window **w_curs,
333 off_t offset,
334 off_t len)
335 {
336 unsigned char *in;
337 unsigned int avail;
339 while (len) {
340 in = use_pack(p, w_curs, offset, &avail);
341 if (avail > len)
342 avail = (unsigned int)len;
343 sha1write(f, in, avail);
344 offset += avail;
345 len -= avail;
346 }
347 }
349 static int check_loose_inflate(unsigned char *data, unsigned long len, unsigned long expect)
350 {
351 z_stream stream;
352 unsigned char fakebuf[4096];
353 int st;
355 memset(&stream, 0, sizeof(stream));
356 stream.next_in = data;
357 stream.avail_in = len;
358 stream.next_out = fakebuf;
359 stream.avail_out = sizeof(fakebuf);
360 inflateInit(&stream);
362 while (1) {
363 st = inflate(&stream, Z_FINISH);
364 if (st == Z_STREAM_END || st == Z_OK) {
365 st = (stream.total_out == expect &&
366 stream.total_in == len) ? 0 : -1;
367 break;
368 }
369 if (st != Z_BUF_ERROR) {
370 st = -1;
371 break;
372 }
373 stream.next_out = fakebuf;
374 stream.avail_out = sizeof(fakebuf);
375 }
376 inflateEnd(&stream);
377 return st;
378 }
380 static int revalidate_loose_object(struct object_entry *entry,
381 unsigned char *map,
382 unsigned long mapsize)
383 {
384 /* we already know this is a loose object with new type header. */
385 enum object_type type;
386 unsigned long size, used;
388 if (pack_to_stdout)
389 return 0;
391 used = unpack_object_header_gently(map, mapsize, &type, &size);
392 if (!used)
393 return -1;
394 map += used;
395 mapsize -= used;
396 return check_loose_inflate(map, mapsize, size);
397 }
399 static unsigned long write_object(struct sha1file *f,
400 struct object_entry *entry)
401 {
402 unsigned long size;
403 enum object_type type;
404 void *buf;
405 unsigned char header[10];
406 unsigned hdrlen;
407 off_t datalen;
408 enum object_type obj_type;
409 int to_reuse = 0;
411 if (!pack_to_stdout)
412 crc32_begin(f);
414 obj_type = entry->type;
415 if (! entry->in_pack)
416 to_reuse = 0; /* can't reuse what we don't have */
417 else if (obj_type == OBJ_REF_DELTA || obj_type == OBJ_OFS_DELTA)
418 to_reuse = 1; /* check_object() decided it for us */
419 else if (obj_type != entry->in_pack_type)
420 to_reuse = 0; /* pack has delta which is unusable */
421 else if (entry->delta)
422 to_reuse = 0; /* we want to pack afresh */
423 else
424 to_reuse = 1; /* we have it in-pack undeltified,
425 * and we do not need to deltify it.
426 */
428 if (!entry->in_pack && !entry->delta) {
429 unsigned char *map;
430 unsigned long mapsize;
431 map = map_sha1_file(entry->sha1, &mapsize);
432 if (map && !legacy_loose_object(map)) {
433 /* We can copy straight into the pack file */
434 if (revalidate_loose_object(entry, map, mapsize))
435 die("corrupt loose object %s",
436 sha1_to_hex(entry->sha1));
437 sha1write(f, map, mapsize);
438 munmap(map, mapsize);
439 written++;
440 reused++;
441 return mapsize;
442 }
443 if (map)
444 munmap(map, mapsize);
445 }
447 if (!to_reuse) {
448 buf = read_sha1_file(entry->sha1, &type, &size);
449 if (!buf)
450 die("unable to read %s", sha1_to_hex(entry->sha1));
451 if (size != entry->size)
452 die("object %s size inconsistency (%lu vs %lu)",
453 sha1_to_hex(entry->sha1), size, entry->size);
454 if (entry->delta) {
455 buf = delta_against(buf, size, entry);
456 size = entry->delta_size;
457 obj_type = (allow_ofs_delta && entry->delta->offset) ?
458 OBJ_OFS_DELTA : OBJ_REF_DELTA;
459 }
460 /*
461 * The object header is a byte of 'type' followed by zero or
462 * more bytes of length.
463 */
464 hdrlen = encode_header(obj_type, size, header);
465 sha1write(f, header, hdrlen);
467 if (obj_type == OBJ_OFS_DELTA) {
468 /*
469 * Deltas with relative base contain an additional
470 * encoding of the relative offset for the delta
471 * base from this object's position in the pack.
472 */
473 off_t ofs = entry->offset - entry->delta->offset;
474 unsigned pos = sizeof(header) - 1;
475 header[pos] = ofs & 127;
476 while (ofs >>= 7)
477 header[--pos] = 128 | (--ofs & 127);
478 sha1write(f, header + pos, sizeof(header) - pos);
479 hdrlen += sizeof(header) - pos;
480 } else if (obj_type == OBJ_REF_DELTA) {
481 /*
482 * Deltas with a base reference contain
483 * an additional 20 bytes for the base sha1.
484 */
485 sha1write(f, entry->delta->sha1, 20);
486 hdrlen += 20;
487 }
488 datalen = sha1write_compressed(f, buf, size);
489 free(buf);
490 }
491 else {
492 struct packed_git *p = entry->in_pack;
493 struct pack_window *w_curs = NULL;
494 struct revindex_entry *revidx;
495 off_t offset;
497 if (entry->delta) {
498 obj_type = (allow_ofs_delta && entry->delta->offset) ?
499 OBJ_OFS_DELTA : OBJ_REF_DELTA;
500 reused_delta++;
501 }
502 hdrlen = encode_header(obj_type, entry->size, header);
503 sha1write(f, header, hdrlen);
504 if (obj_type == OBJ_OFS_DELTA) {
505 off_t ofs = entry->offset - entry->delta->offset;
506 unsigned pos = sizeof(header) - 1;
507 header[pos] = ofs & 127;
508 while (ofs >>= 7)
509 header[--pos] = 128 | (--ofs & 127);
510 sha1write(f, header + pos, sizeof(header) - pos);
511 hdrlen += sizeof(header) - pos;
512 } else if (obj_type == OBJ_REF_DELTA) {
513 sha1write(f, entry->delta->sha1, 20);
514 hdrlen += 20;
515 }
517 offset = entry->in_pack_offset;
518 revidx = find_packed_object(p, offset);
519 datalen = revidx[1].offset - offset;
520 if (!pack_to_stdout && p->index_version > 1 &&
521 check_pack_crc(p, &w_curs, offset, datalen, revidx->nr))
522 die("bad packed object CRC for %s", sha1_to_hex(entry->sha1));
523 offset += entry->in_pack_header_size;
524 datalen -= entry->in_pack_header_size;
525 if (!pack_to_stdout && p->index_version == 1 &&
526 check_pack_inflate(p, &w_curs, offset, datalen, entry->size))
527 die("corrupt packed object for %s", sha1_to_hex(entry->sha1));
528 copy_pack_data(f, p, &w_curs, offset, datalen);
529 unuse_pack(&w_curs);
530 reused++;
531 }
532 if (entry->delta)
533 written_delta++;
534 written++;
535 if (!pack_to_stdout)
536 entry->crc32 = crc32_end(f);
537 return hdrlen + datalen;
538 }
540 static off_t write_one(struct sha1file *f,
541 struct object_entry *e,
542 off_t offset)
543 {
544 unsigned long size;
546 /* offset is non zero if object is written already. */
547 if (e->offset || e->preferred_base)
548 return offset;
550 /* if we are deltified, write out base object first. */
551 if (e->delta)
552 offset = write_one(f, e->delta, offset);
554 e->offset = offset;
555 size = write_object(f, e);
557 /* make sure off_t is sufficiently large not to wrap */
558 if (offset > offset + size)
559 die("pack too large for current definition of off_t");
560 return offset + size;
561 }
563 static int open_object_dir_tmp(const char *path)
564 {
565 snprintf(tmpname, sizeof(tmpname), "%s/%s", get_object_directory(), path);
566 return mkstemp(tmpname);
567 }
569 static off_t write_pack_file(void)
570 {
571 uint32_t i;
572 struct sha1file *f;
573 off_t offset, last_obj_offset = 0;
574 struct pack_header hdr;
575 int do_progress = progress;
577 if (pack_to_stdout) {
578 f = sha1fd(1, "<stdout>");
579 do_progress >>= 1;
580 } else {
581 int fd = open_object_dir_tmp("tmp_pack_XXXXXX");
582 if (fd < 0)
583 die("unable to create %s: %s\n", tmpname, strerror(errno));
584 pack_tmp_name = xstrdup(tmpname);
585 f = sha1fd(fd, pack_tmp_name);
586 }
588 if (do_progress)
589 start_progress(&progress_state, "Writing %u objects...", "", nr_result);
591 hdr.hdr_signature = htonl(PACK_SIGNATURE);
592 hdr.hdr_version = htonl(PACK_VERSION);
593 hdr.hdr_entries = htonl(nr_result);
594 sha1write(f, &hdr, sizeof(hdr));
595 offset = sizeof(hdr);
596 if (!nr_result)
597 goto done;
598 for (i = 0; i < nr_objects; i++) {
599 last_obj_offset = offset;
600 offset = write_one(f, objects + i, offset);
601 if (do_progress)
602 display_progress(&progress_state, written);
603 }
604 if (do_progress)
605 stop_progress(&progress_state);
606 done:
607 if (written != nr_result)
608 die("wrote %u objects while expecting %u", written, nr_result);
609 sha1close(f, pack_file_sha1, 1);
611 return last_obj_offset;
612 }
614 static int sha1_sort(const void *_a, const void *_b)
615 {
616 const struct object_entry *a = *(struct object_entry **)_a;
617 const struct object_entry *b = *(struct object_entry **)_b;
618 return hashcmp(a->sha1, b->sha1);
619 }
621 static uint32_t index_default_version = 1;
622 static uint32_t index_off32_limit = 0x7fffffff;
624 static void write_index_file(off_t last_obj_offset, unsigned char *sha1)
625 {
626 struct sha1file *f;
627 struct object_entry **sorted_by_sha, **list, **last;
628 uint32_t array[256];
629 uint32_t i, index_version;
630 SHA_CTX ctx;
632 int fd = open_object_dir_tmp("tmp_idx_XXXXXX");
633 if (fd < 0)
634 die("unable to create %s: %s\n", tmpname, strerror(errno));
635 idx_tmp_name = xstrdup(tmpname);
636 f = sha1fd(fd, idx_tmp_name);
638 if (nr_result) {
639 uint32_t j = 0;
640 sorted_by_sha =
641 xcalloc(nr_result, sizeof(struct object_entry *));
642 for (i = 0; i < nr_objects; i++)
643 if (!objects[i].preferred_base)
644 sorted_by_sha[j++] = objects + i;
645 if (j != nr_result)
646 die("listed %u objects while expecting %u", j, nr_result);
647 qsort(sorted_by_sha, nr_result, sizeof(*sorted_by_sha), sha1_sort);
648 list = sorted_by_sha;
649 last = sorted_by_sha + nr_result;
650 } else
651 sorted_by_sha = list = last = NULL;
653 /* if last object's offset is >= 2^31 we should use index V2 */
654 index_version = (last_obj_offset >> 31) ? 2 : index_default_version;
656 /* index versions 2 and above need a header */
657 if (index_version >= 2) {
658 struct pack_idx_header hdr;
659 hdr.idx_signature = htonl(PACK_IDX_SIGNATURE);
660 hdr.idx_version = htonl(index_version);
661 sha1write(f, &hdr, sizeof(hdr));
662 }
664 /*
665 * Write the first-level table (the list is sorted,
666 * but we use a 256-entry lookup to be able to avoid
667 * having to do eight extra binary search iterations).
668 */
669 for (i = 0; i < 256; i++) {
670 struct object_entry **next = list;
671 while (next < last) {
672 struct object_entry *entry = *next;
673 if (entry->sha1[0] != i)
674 break;
675 next++;
676 }
677 array[i] = htonl(next - sorted_by_sha);
678 list = next;
679 }
680 sha1write(f, array, 256 * 4);
682 /* Compute the SHA1 hash of sorted object names. */
683 SHA1_Init(&ctx);
685 /* Write the actual SHA1 entries. */
686 list = sorted_by_sha;
687 for (i = 0; i < nr_result; i++) {
688 struct object_entry *entry = *list++;
689 if (index_version < 2) {
690 uint32_t offset = htonl(entry->offset);
691 sha1write(f, &offset, 4);
692 }
693 sha1write(f, entry->sha1, 20);
694 SHA1_Update(&ctx, entry->sha1, 20);
695 }
697 if (index_version >= 2) {
698 unsigned int nr_large_offset = 0;
700 /* write the crc32 table */
701 list = sorted_by_sha;
702 for (i = 0; i < nr_objects; i++) {
703 struct object_entry *entry = *list++;
704 uint32_t crc32_val = htonl(entry->crc32);
705 sha1write(f, &crc32_val, 4);
706 }
708 /* write the 32-bit offset table */
709 list = sorted_by_sha;
710 for (i = 0; i < nr_objects; i++) {
711 struct object_entry *entry = *list++;
712 uint32_t offset = (entry->offset <= index_off32_limit) ?
713 entry->offset : (0x80000000 | nr_large_offset++);
714 offset = htonl(offset);
715 sha1write(f, &offset, 4);
716 }
718 /* write the large offset table */
719 list = sorted_by_sha;
720 while (nr_large_offset) {
721 struct object_entry *entry = *list++;
722 uint64_t offset = entry->offset;
723 if (offset > index_off32_limit) {
724 uint32_t split[2];
725 split[0] = htonl(offset >> 32);
726 split[1] = htonl(offset & 0xffffffff);
727 sha1write(f, split, 8);
728 nr_large_offset--;
729 }
730 }
731 }
733 sha1write(f, pack_file_sha1, 20);
734 sha1close(f, NULL, 1);
735 free(sorted_by_sha);
736 SHA1_Final(sha1, &ctx);
737 }
739 static int locate_object_entry_hash(const unsigned char *sha1)
740 {
741 int i;
742 unsigned int ui;
743 memcpy(&ui, sha1, sizeof(unsigned int));
744 i = ui % object_ix_hashsz;
745 while (0 < object_ix[i]) {
746 if (!hashcmp(sha1, objects[object_ix[i] - 1].sha1))
747 return i;
748 if (++i == object_ix_hashsz)
749 i = 0;
750 }
751 return -1 - i;
752 }
754 static struct object_entry *locate_object_entry(const unsigned char *sha1)
755 {
756 int i;
758 if (!object_ix_hashsz)
759 return NULL;
761 i = locate_object_entry_hash(sha1);
762 if (0 <= i)
763 return &objects[object_ix[i]-1];
764 return NULL;
765 }
767 static void rehash_objects(void)
768 {
769 uint32_t i;
770 struct object_entry *oe;
772 object_ix_hashsz = nr_objects * 3;
773 if (object_ix_hashsz < 1024)
774 object_ix_hashsz = 1024;
775 object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz);
776 memset(object_ix, 0, sizeof(int) * object_ix_hashsz);
777 for (i = 0, oe = objects; i < nr_objects; i++, oe++) {
778 int ix = locate_object_entry_hash(oe->sha1);
779 if (0 <= ix)
780 continue;
781 ix = -1 - ix;
782 object_ix[ix] = i + 1;
783 }
784 }
786 static unsigned name_hash(const char *name)
787 {
788 unsigned char c;
789 unsigned hash = 0;
791 /*
792 * This effectively just creates a sortable number from the
793 * last sixteen non-whitespace characters. Last characters
794 * count "most", so things that end in ".c" sort together.
795 */
796 while ((c = *name++) != 0) {
797 if (isspace(c))
798 continue;
799 hash = (hash >> 2) + (c << 24);
800 }
801 return hash;
802 }
804 static int add_object_entry(const unsigned char *sha1, enum object_type type,
805 unsigned hash, int exclude)
806 {
807 struct object_entry *entry;
808 struct packed_git *p, *found_pack = NULL;
809 off_t found_offset = 0;
810 int ix;
812 ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
813 if (ix >= 0) {
814 if (exclude) {
815 entry = objects + object_ix[ix] - 1;
816 if (!entry->preferred_base)
817 nr_result--;
818 entry->preferred_base = 1;
819 }
820 return 0;
821 }
823 for (p = packed_git; p; p = p->next) {
824 off_t offset = find_pack_entry_one(sha1, p);
825 if (offset) {
826 if (!found_pack) {
827 found_offset = offset;
828 found_pack = p;
829 }
830 if (exclude)
831 break;
832 if (incremental)
833 return 0;
834 if (local && !p->pack_local)
835 return 0;
836 }
837 }
839 if (nr_objects >= nr_alloc) {
840 nr_alloc = (nr_alloc + 1024) * 3 / 2;
841 objects = xrealloc(objects, nr_alloc * sizeof(*entry));
842 }
844 entry = objects + nr_objects++;
845 memset(entry, 0, sizeof(*entry));
846 hashcpy(entry->sha1, sha1);
847 entry->hash = hash;
848 if (type)
849 entry->type = type;
850 if (exclude)
851 entry->preferred_base = 1;
852 else
853 nr_result++;
854 if (found_pack) {
855 entry->in_pack = found_pack;
856 entry->in_pack_offset = found_offset;
857 }
859 if (object_ix_hashsz * 3 <= nr_objects * 4)
860 rehash_objects();
861 else
862 object_ix[-1 - ix] = nr_objects;
864 if (progress)
865 display_progress(&progress_state, nr_objects);
867 return 1;
868 }
870 struct pbase_tree_cache {
871 unsigned char sha1[20];
872 int ref;
873 int temporary;
874 void *tree_data;
875 unsigned long tree_size;
876 };
878 static struct pbase_tree_cache *(pbase_tree_cache[256]);
879 static int pbase_tree_cache_ix(const unsigned char *sha1)
880 {
881 return sha1[0] % ARRAY_SIZE(pbase_tree_cache);
882 }
883 static int pbase_tree_cache_ix_incr(int ix)
884 {
885 return (ix+1) % ARRAY_SIZE(pbase_tree_cache);
886 }
888 static struct pbase_tree {
889 struct pbase_tree *next;
890 /* This is a phony "cache" entry; we are not
891 * going to evict it nor find it through _get()
892 * mechanism -- this is for the toplevel node that
893 * would almost always change with any commit.
894 */
895 struct pbase_tree_cache pcache;
896 } *pbase_tree;
898 static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1)
899 {
900 struct pbase_tree_cache *ent, *nent;
901 void *data;
902 unsigned long size;
903 enum object_type type;
904 int neigh;
905 int my_ix = pbase_tree_cache_ix(sha1);
906 int available_ix = -1;
908 /* pbase-tree-cache acts as a limited hashtable.
909 * your object will be found at your index or within a few
910 * slots after that slot if it is cached.
911 */
912 for (neigh = 0; neigh < 8; neigh++) {
913 ent = pbase_tree_cache[my_ix];
914 if (ent && !hashcmp(ent->sha1, sha1)) {
915 ent->ref++;
916 return ent;
917 }
918 else if (((available_ix < 0) && (!ent || !ent->ref)) ||
919 ((0 <= available_ix) &&
920 (!ent && pbase_tree_cache[available_ix])))
921 available_ix = my_ix;
922 if (!ent)
923 break;
924 my_ix = pbase_tree_cache_ix_incr(my_ix);
925 }
927 /* Did not find one. Either we got a bogus request or
928 * we need to read and perhaps cache.
929 */
930 data = read_sha1_file(sha1, &type, &size);
931 if (!data)
932 return NULL;
933 if (type != OBJ_TREE) {
934 free(data);
935 return NULL;
936 }
938 /* We need to either cache or return a throwaway copy */
940 if (available_ix < 0)
941 ent = NULL;
942 else {
943 ent = pbase_tree_cache[available_ix];
944 my_ix = available_ix;
945 }
947 if (!ent) {
948 nent = xmalloc(sizeof(*nent));
949 nent->temporary = (available_ix < 0);
950 }
951 else {
952 /* evict and reuse */
953 free(ent->tree_data);
954 nent = ent;
955 }
956 hashcpy(nent->sha1, sha1);
957 nent->tree_data = data;
958 nent->tree_size = size;
959 nent->ref = 1;
960 if (!nent->temporary)
961 pbase_tree_cache[my_ix] = nent;
962 return nent;
963 }
965 static void pbase_tree_put(struct pbase_tree_cache *cache)
966 {
967 if (!cache->temporary) {
968 cache->ref--;
969 return;
970 }
971 free(cache->tree_data);
972 free(cache);
973 }
975 static int name_cmp_len(const char *name)
976 {
977 int i;
978 for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++)
979 ;
980 return i;
981 }
983 static void add_pbase_object(struct tree_desc *tree,
984 const char *name,
985 int cmplen,
986 const char *fullname)
987 {
988 struct name_entry entry;
989 int cmp;
991 while (tree_entry(tree,&entry)) {
992 cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 :
993 memcmp(name, entry.path, cmplen);
994 if (cmp > 0)
995 continue;
996 if (cmp < 0)
997 return;
998 if (name[cmplen] != '/') {
999 unsigned hash = name_hash(fullname);
1000 add_object_entry(entry.sha1,
1001 S_ISDIR(entry.mode) ? OBJ_TREE : OBJ_BLOB,
1002 hash, 1);
1003 return;
1004 }
1005 if (S_ISDIR(entry.mode)) {
1006 struct tree_desc sub;
1007 struct pbase_tree_cache *tree;
1008 const char *down = name+cmplen+1;
1009 int downlen = name_cmp_len(down);
1011 tree = pbase_tree_get(entry.sha1);
1012 if (!tree)
1013 return;
1014 init_tree_desc(&sub, tree->tree_data, tree->tree_size);
1016 add_pbase_object(&sub, down, downlen, fullname);
1017 pbase_tree_put(tree);
1018 }
1019 }
1020 }
1022 static unsigned *done_pbase_paths;
1023 static int done_pbase_paths_num;
1024 static int done_pbase_paths_alloc;
1025 static int done_pbase_path_pos(unsigned hash)
1026 {
1027 int lo = 0;
1028 int hi = done_pbase_paths_num;
1029 while (lo < hi) {
1030 int mi = (hi + lo) / 2;
1031 if (done_pbase_paths[mi] == hash)
1032 return mi;
1033 if (done_pbase_paths[mi] < hash)
1034 hi = mi;
1035 else
1036 lo = mi + 1;
1037 }
1038 return -lo-1;
1039 }
1041 static int check_pbase_path(unsigned hash)
1042 {
1043 int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash);
1044 if (0 <= pos)
1045 return 1;
1046 pos = -pos - 1;
1047 if (done_pbase_paths_alloc <= done_pbase_paths_num) {
1048 done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc);
1049 done_pbase_paths = xrealloc(done_pbase_paths,
1050 done_pbase_paths_alloc *
1051 sizeof(unsigned));
1052 }
1053 done_pbase_paths_num++;
1054 if (pos < done_pbase_paths_num)
1055 memmove(done_pbase_paths + pos + 1,
1056 done_pbase_paths + pos,
1057 (done_pbase_paths_num - pos - 1) * sizeof(unsigned));
1058 done_pbase_paths[pos] = hash;
1059 return 0;
1060 }
1062 static void add_preferred_base_object(const char *name, unsigned hash)
1063 {
1064 struct pbase_tree *it;
1065 int cmplen;
1067 if (!num_preferred_base || check_pbase_path(hash))
1068 return;
1070 cmplen = name_cmp_len(name);
1071 for (it = pbase_tree; it; it = it->next) {
1072 if (cmplen == 0) {
1073 add_object_entry(it->pcache.sha1, OBJ_TREE, 0, 1);
1074 }
1075 else {
1076 struct tree_desc tree;
1077 init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size);
1078 add_pbase_object(&tree, name, cmplen, name);
1079 }
1080 }
1081 }
1083 static void add_preferred_base(unsigned char *sha1)
1084 {
1085 struct pbase_tree *it;
1086 void *data;
1087 unsigned long size;
1088 unsigned char tree_sha1[20];
1090 if (window <= num_preferred_base++)
1091 return;
1093 data = read_object_with_reference(sha1, tree_type, &size, tree_sha1);
1094 if (!data)
1095 return;
1097 for (it = pbase_tree; it; it = it->next) {
1098 if (!hashcmp(it->pcache.sha1, tree_sha1)) {
1099 free(data);
1100 return;
1101 }
1102 }
1104 it = xcalloc(1, sizeof(*it));
1105 it->next = pbase_tree;
1106 pbase_tree = it;
1108 hashcpy(it->pcache.sha1, tree_sha1);
1109 it->pcache.tree_data = data;
1110 it->pcache.tree_size = size;
1111 }
1113 static void check_object(struct object_entry *entry)
1114 {
1115 if (entry->in_pack) {
1116 struct packed_git *p = entry->in_pack;
1117 struct pack_window *w_curs = NULL;
1118 const unsigned char *base_ref = NULL;
1119 struct object_entry *base_entry;
1120 unsigned long used, used_0;
1121 unsigned int avail;
1122 off_t ofs;
1123 unsigned char *buf, c;
1125 buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
1127 /*
1128 * We want in_pack_type even if we do not reuse delta.
1129 * There is no point not reusing non-delta representations.
1130 */
1131 used = unpack_object_header_gently(buf, avail,
1132 &entry->in_pack_type,
1133 &entry->size);
1135 /*
1136 * Determine if this is a delta and if so whether we can
1137 * reuse it or not. Otherwise let's find out as cheaply as
1138 * possible what the actual type and size for this object is.
1139 */
1140 switch (entry->in_pack_type) {
1141 default:
1142 /* Not a delta hence we've already got all we need. */
1143 entry->type = entry->in_pack_type;
1144 entry->in_pack_header_size = used;
1145 unuse_pack(&w_curs);
1146 return;
1147 case OBJ_REF_DELTA:
1148 if (!no_reuse_delta && !entry->preferred_base)
1149 base_ref = use_pack(p, &w_curs,
1150 entry->in_pack_offset + used, NULL);
1151 entry->in_pack_header_size = used + 20;
1152 break;
1153 case OBJ_OFS_DELTA:
1154 buf = use_pack(p, &w_curs,
1155 entry->in_pack_offset + used, NULL);
1156 used_0 = 0;
1157 c = buf[used_0++];
1158 ofs = c & 127;
1159 while (c & 128) {
1160 ofs += 1;
1161 if (!ofs || MSB(ofs, 7))
1162 die("delta base offset overflow in pack for %s",
1163 sha1_to_hex(entry->sha1));
1164 c = buf[used_0++];
1165 ofs = (ofs << 7) + (c & 127);
1166 }
1167 if (ofs >= entry->in_pack_offset)
1168 die("delta base offset out of bound for %s",
1169 sha1_to_hex(entry->sha1));
1170 ofs = entry->in_pack_offset - ofs;
1171 if (!no_reuse_delta && !entry->preferred_base)
1172 base_ref = find_packed_object_name(p, ofs);
1173 entry->in_pack_header_size = used + used_0;
1174 break;
1175 }
1177 if (base_ref && (base_entry = locate_object_entry(base_ref))) {
1178 /*
1179 * If base_ref was set above that means we wish to
1180 * reuse delta data, and we even found that base
1181 * in the list of objects we want to pack. Goodie!
1182 *
1183 * Depth value does not matter - find_deltas() will
1184 * never consider reused delta as the base object to
1185 * deltify other objects against, in order to avoid
1186 * circular deltas.
1187 */
1188 entry->type = entry->in_pack_type;
1189 entry->delta = base_entry;
1190 entry->delta_sibling = base_entry->delta_child;
1191 base_entry->delta_child = entry;
1192 unuse_pack(&w_curs);
1193 return;
1194 }
1196 if (entry->type) {
1197 /*
1198 * This must be a delta and we already know what the
1199 * final object type is. Let's extract the actual
1200 * object size from the delta header.
1201 */
1202 entry->size = get_size_from_delta(p, &w_curs,
1203 entry->in_pack_offset + entry->in_pack_header_size);
1204 unuse_pack(&w_curs);
1205 return;
1206 }
1208 /*
1209 * No choice but to fall back to the recursive delta walk
1210 * with sha1_object_info() to find about the object type
1211 * at this point...
1212 */
1213 unuse_pack(&w_curs);
1214 }
1216 entry->type = sha1_object_info(entry->sha1, &entry->size);
1217 if (entry->type < 0)
1218 die("unable to get type of object %s",
1219 sha1_to_hex(entry->sha1));
1220 }
1222 static int pack_offset_sort(const void *_a, const void *_b)
1223 {
1224 const struct object_entry *a = *(struct object_entry **)_a;
1225 const struct object_entry *b = *(struct object_entry **)_b;
1227 /* avoid filesystem trashing with loose objects */
1228 if (!a->in_pack && !b->in_pack)
1229 return hashcmp(a->sha1, b->sha1);
1231 if (a->in_pack < b->in_pack)
1232 return -1;
1233 if (a->in_pack > b->in_pack)
1234 return 1;
1235 return a->in_pack_offset < b->in_pack_offset ? -1 :
1236 (a->in_pack_offset > b->in_pack_offset);
1237 }
1239 static void get_object_details(void)
1240 {
1241 uint32_t i;
1242 struct object_entry **sorted_by_offset;
1244 sorted_by_offset = xcalloc(nr_objects, sizeof(struct object_entry *));
1245 for (i = 0; i < nr_objects; i++)
1246 sorted_by_offset[i] = objects + i;
1247 qsort(sorted_by_offset, nr_objects, sizeof(*sorted_by_offset), pack_offset_sort);
1249 prepare_pack_ix();
1250 for (i = 0; i < nr_objects; i++)
1251 check_object(sorted_by_offset[i]);
1252 free(sorted_by_offset);
1253 }
1255 static int type_size_sort(const void *_a, const void *_b)
1256 {
1257 const struct object_entry *a = *(struct object_entry **)_a;
1258 const struct object_entry *b = *(struct object_entry **)_b;
1260 if (a->type < b->type)
1261 return -1;
1262 if (a->type > b->type)
1263 return 1;
1264 if (a->hash < b->hash)
1265 return -1;
1266 if (a->hash > b->hash)
1267 return 1;
1268 if (a->preferred_base < b->preferred_base)
1269 return -1;
1270 if (a->preferred_base > b->preferred_base)
1271 return 1;
1272 if (a->size < b->size)
1273 return -1;
1274 if (a->size > b->size)
1275 return 1;
1276 return a > b ? -1 : (a < b); /* newest last */
1277 }
1279 struct unpacked {
1280 struct object_entry *entry;
1281 void *data;
1282 struct delta_index *index;
1283 };
1285 /*
1286 * We search for deltas _backwards_ in a list sorted by type and
1287 * by size, so that we see progressively smaller and smaller files.
1288 * That's because we prefer deltas to be from the bigger file
1289 * to the smaller - deletes are potentially cheaper, but perhaps
1290 * more importantly, the bigger file is likely the more recent
1291 * one.
1292 */
1293 static int try_delta(struct unpacked *trg, struct unpacked *src,
1294 unsigned max_depth)
1295 {
1296 struct object_entry *trg_entry = trg->entry;
1297 struct object_entry *src_entry = src->entry;
1298 unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz;
1299 enum object_type type;
1300 void *delta_buf;
1302 /* Don't bother doing diffs between different types */
1303 if (trg_entry->type != src_entry->type)
1304 return -1;
1306 /* We do not compute delta to *create* objects we are not
1307 * going to pack.
1308 */
1309 if (trg_entry->preferred_base)
1310 return -1;
1312 /*
1313 * We do not bother to try a delta that we discarded
1314 * on an earlier try, but only when reusing delta data.
1315 */
1316 if (!no_reuse_delta && trg_entry->in_pack &&
1317 trg_entry->in_pack == src_entry->in_pack &&
1318 trg_entry->in_pack_type != OBJ_REF_DELTA &&
1319 trg_entry->in_pack_type != OBJ_OFS_DELTA)
1320 return 0;
1322 /* Let's not bust the allowed depth. */
1323 if (src_entry->depth >= max_depth)
1324 return 0;
1326 /* Now some size filtering heuristics. */
1327 trg_size = trg_entry->size;
1328 max_size = trg_size/2 - 20;
1329 max_size = max_size * (max_depth - src_entry->depth) / max_depth;
1330 if (max_size == 0)
1331 return 0;
1332 if (trg_entry->delta && trg_entry->delta_size <= max_size)
1333 max_size = trg_entry->delta_size-1;
1334 src_size = src_entry->size;
1335 sizediff = src_size < trg_size ? trg_size - src_size : 0;
1336 if (sizediff >= max_size)
1337 return 0;
1339 /* Load data if not already done */
1340 if (!trg->data) {
1341 trg->data = read_sha1_file(trg_entry->sha1, &type, &sz);
1342 if (sz != trg_size)
1343 die("object %s inconsistent object length (%lu vs %lu)",
1344 sha1_to_hex(trg_entry->sha1), sz, trg_size);
1345 }
1346 if (!src->data) {
1347 src->data = read_sha1_file(src_entry->sha1, &type, &sz);
1348 if (sz != src_size)
1349 die("object %s inconsistent object length (%lu vs %lu)",
1350 sha1_to_hex(src_entry->sha1), sz, src_size);
1351 }
1352 if (!src->index) {
1353 src->index = create_delta_index(src->data, src_size);
1354 if (!src->index)
1355 die("out of memory");
1356 }
1358 delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size);
1359 if (!delta_buf)
1360 return 0;
1362 trg_entry->delta = src_entry;
1363 trg_entry->delta_size = delta_size;
1364 trg_entry->depth = src_entry->depth + 1;
1365 free(delta_buf);
1366 return 1;
1367 }
1369 static unsigned int check_delta_limit(struct object_entry *me, unsigned int n)
1370 {
1371 struct object_entry *child = me->delta_child;
1372 unsigned int m = n;
1373 while (child) {
1374 unsigned int c = check_delta_limit(child, n + 1);
1375 if (m < c)
1376 m = c;
1377 child = child->delta_sibling;
1378 }
1379 return m;
1380 }
1382 static void find_deltas(struct object_entry **list, int window, int depth)
1383 {
1384 uint32_t i = nr_objects, idx = 0, processed = 0;
1385 unsigned int array_size = window * sizeof(struct unpacked);
1386 struct unpacked *array;
1387 int max_depth;
1389 if (!nr_objects)
1390 return;
1391 array = xmalloc(array_size);
1392 memset(array, 0, array_size);
1393 if (progress)
1394 start_progress(&progress_state, "Deltifying %u objects...", "", nr_result);
1396 do {
1397 struct object_entry *entry = list[--i];
1398 struct unpacked *n = array + idx;
1399 int j;
1401 if (!entry->preferred_base)
1402 processed++;
1404 if (progress)
1405 display_progress(&progress_state, processed);
1407 if (entry->delta)
1408 /* This happens if we decided to reuse existing
1409 * delta from a pack. "!no_reuse_delta &&" is implied.
1410 */
1411 continue;
1413 if (entry->size < 50)
1414 continue;
1415 free_delta_index(n->index);
1416 n->index = NULL;
1417 free(n->data);
1418 n->data = NULL;
1419 n->entry = entry;
1421 /*
1422 * If the current object is at pack edge, take the depth the
1423 * objects that depend on the current object into account
1424 * otherwise they would become too deep.
1425 */
1426 max_depth = depth;
1427 if (entry->delta_child) {
1428 max_depth -= check_delta_limit(entry, 0);
1429 if (max_depth <= 0)
1430 goto next;
1431 }
1433 j = window;
1434 while (--j > 0) {
1435 uint32_t other_idx = idx + j;
1436 struct unpacked *m;
1437 if (other_idx >= window)
1438 other_idx -= window;
1439 m = array + other_idx;
1440 if (!m->entry)
1441 break;
1442 if (try_delta(n, m, max_depth) < 0)
1443 break;
1444 }
1446 /* if we made n a delta, and if n is already at max
1447 * depth, leaving it in the window is pointless. we
1448 * should evict it first.
1449 */
1450 if (entry->delta && depth <= entry->depth)
1451 continue;
1453 next:
1454 idx++;
1455 if (idx >= window)
1456 idx = 0;
1457 } while (i > 0);
1459 if (progress)
1460 stop_progress(&progress_state);
1462 for (i = 0; i < window; ++i) {
1463 free_delta_index(array[i].index);
1464 free(array[i].data);
1465 }
1466 free(array);
1467 }
1469 static void prepare_pack(int window, int depth)
1470 {
1471 struct object_entry **delta_list;
1472 uint32_t i;
1474 get_object_details();
1476 if (!window || !depth)
1477 return;
1479 delta_list = xmalloc(nr_objects * sizeof(*delta_list));
1480 for (i = 0; i < nr_objects; i++)
1481 delta_list[i] = objects + i;
1482 qsort(delta_list, nr_objects, sizeof(*delta_list), type_size_sort);
1483 find_deltas(delta_list, window+1, depth);
1484 free(delta_list);
1485 }
1487 static int git_pack_config(const char *k, const char *v)
1488 {
1489 if(!strcmp(k, "pack.window")) {
1490 window = git_config_int(k, v);
1491 return 0;
1492 }
1493 if(!strcmp(k, "pack.depth")) {
1494 depth = git_config_int(k, v);
1495 return 0;
1496 }
1497 return git_default_config(k, v);
1498 }
1500 static void read_object_list_from_stdin(void)
1501 {
1502 char line[40 + 1 + PATH_MAX + 2];
1503 unsigned char sha1[20];
1504 unsigned hash;
1506 for (;;) {
1507 if (!fgets(line, sizeof(line), stdin)) {
1508 if (feof(stdin))
1509 break;
1510 if (!ferror(stdin))
1511 die("fgets returned NULL, not EOF, not error!");
1512 if (errno != EINTR)
1513 die("fgets: %s", strerror(errno));
1514 clearerr(stdin);
1515 continue;
1516 }
1517 if (line[0] == '-') {
1518 if (get_sha1_hex(line+1, sha1))
1519 die("expected edge sha1, got garbage:\n %s",
1520 line);
1521 add_preferred_base(sha1);
1522 continue;
1523 }
1524 if (get_sha1_hex(line, sha1))
1525 die("expected sha1, got garbage:\n %s", line);
1527 hash = name_hash(line+41);
1528 add_preferred_base_object(line+41, hash);
1529 add_object_entry(sha1, 0, hash, 0);
1530 }
1531 }
1533 static void show_commit(struct commit *commit)
1534 {
1535 add_object_entry(commit->object.sha1, OBJ_COMMIT, 0, 0);
1536 }
1538 static void show_object(struct object_array_entry *p)
1539 {
1540 unsigned hash = name_hash(p->name);
1541 add_preferred_base_object(p->name, hash);
1542 add_object_entry(p->item->sha1, p->item->type, hash, 0);
1543 }
1545 static void show_edge(struct commit *commit)
1546 {
1547 add_preferred_base(commit->object.sha1);
1548 }
1550 static void get_object_list(int ac, const char **av)
1551 {
1552 struct rev_info revs;
1553 char line[1000];
1554 int flags = 0;
1556 init_revisions(&revs, NULL);
1557 save_commit_buffer = 0;
1558 track_object_refs = 0;
1559 setup_revisions(ac, av, &revs, NULL);
1561 while (fgets(line, sizeof(line), stdin) != NULL) {
1562 int len = strlen(line);
1563 if (line[len - 1] == '\n')
1564 line[--len] = 0;
1565 if (!len)
1566 break;
1567 if (*line == '-') {
1568 if (!strcmp(line, "--not")) {
1569 flags ^= UNINTERESTING;
1570 continue;
1571 }
1572 die("not a rev '%s'", line);
1573 }
1574 if (handle_revision_arg(line, &revs, flags, 1))
1575 die("bad revision '%s'", line);
1576 }
1578 prepare_revision_walk(&revs);
1579 mark_edges_uninteresting(revs.commits, &revs, show_edge);
1580 traverse_commit_list(&revs, show_commit, show_object);
1581 }
1583 static int adjust_perm(const char *path, mode_t mode)
1584 {
1585 if (chmod(path, mode))
1586 return -1;
1587 return adjust_shared_perm(path);
1588 }
1590 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
1591 {
1592 int use_internal_rev_list = 0;
1593 int thin = 0;
1594 uint32_t i;
1595 off_t last_obj_offset;
1596 const char *base_name = NULL;
1597 const char **rp_av;
1598 int rp_ac_alloc = 64;
1599 int rp_ac;
1601 rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av));
1603 rp_av[0] = "pack-objects";
1604 rp_av[1] = "--objects"; /* --thin will make it --objects-edge */
1605 rp_ac = 2;
1607 git_config(git_pack_config);
1609 progress = isatty(2);
1610 for (i = 1; i < argc; i++) {
1611 const char *arg = argv[i];
1613 if (*arg != '-')
1614 break;
1616 if (!strcmp("--non-empty", arg)) {
1617 non_empty = 1;
1618 continue;
1619 }
1620 if (!strcmp("--local", arg)) {
1621 local = 1;
1622 continue;
1623 }
1624 if (!strcmp("--incremental", arg)) {
1625 incremental = 1;
1626 continue;
1627 }
1628 if (!prefixcmp(arg, "--window=")) {
1629 char *end;
1630 window = strtoul(arg+9, &end, 0);
1631 if (!arg[9] || *end)
1632 usage(pack_usage);
1633 continue;
1634 }
1635 if (!prefixcmp(arg, "--depth=")) {
1636 char *end;
1637 depth = strtoul(arg+8, &end, 0);
1638 if (!arg[8] || *end)
1639 usage(pack_usage);
1640 continue;
1641 }
1642 if (!strcmp("--progress", arg)) {
1643 progress = 1;
1644 continue;
1645 }
1646 if (!strcmp("--all-progress", arg)) {
1647 progress = 2;
1648 continue;
1649 }
1650 if (!strcmp("-q", arg)) {
1651 progress = 0;
1652 continue;
1653 }
1654 if (!strcmp("--no-reuse-delta", arg)) {
1655 no_reuse_delta = 1;
1656 continue;
1657 }
1658 if (!strcmp("--delta-base-offset", arg)) {
1659 allow_ofs_delta = 1;
1660 continue;
1661 }
1662 if (!strcmp("--stdout", arg)) {
1663 pack_to_stdout = 1;
1664 continue;
1665 }
1666 if (!strcmp("--revs", arg)) {
1667 use_internal_rev_list = 1;
1668 continue;
1669 }
1670 if (!strcmp("--unpacked", arg) ||
1671 !prefixcmp(arg, "--unpacked=") ||
1672 !strcmp("--reflog", arg) ||
1673 !strcmp("--all", arg)) {
1674 use_internal_rev_list = 1;
1675 if (rp_ac >= rp_ac_alloc - 1) {
1676 rp_ac_alloc = alloc_nr(rp_ac_alloc);
1677 rp_av = xrealloc(rp_av,
1678 rp_ac_alloc * sizeof(*rp_av));
1679 }
1680 rp_av[rp_ac++] = arg;
1681 continue;
1682 }
1683 if (!strcmp("--thin", arg)) {
1684 use_internal_rev_list = 1;
1685 thin = 1;
1686 rp_av[1] = "--objects-edge";
1687 continue;
1688 }
1689 if (!prefixcmp(arg, "--index-version=")) {
1690 char *c;
1691 index_default_version = strtoul(arg + 16, &c, 10);
1692 if (index_default_version > 2)
1693 die("bad %s", arg);
1694 if (*c == ',')
1695 index_off32_limit = strtoul(c+1, &c, 0);
1696 if (*c || index_off32_limit & 0x80000000)
1697 die("bad %s", arg);
1698 continue;
1699 }
1700 usage(pack_usage);
1701 }
1703 /* Traditionally "pack-objects [options] base extra" failed;
1704 * we would however want to take refs parameter that would
1705 * have been given to upstream rev-list ourselves, which means
1706 * we somehow want to say what the base name is. So the
1707 * syntax would be:
1708 *
1709 * pack-objects [options] base <refs...>
1710 *
1711 * in other words, we would treat the first non-option as the
1712 * base_name and send everything else to the internal revision
1713 * walker.
1714 */
1716 if (!pack_to_stdout)
1717 base_name = argv[i++];
1719 if (pack_to_stdout != !base_name)
1720 usage(pack_usage);
1722 if (!pack_to_stdout && thin)
1723 die("--thin cannot be used to build an indexable pack.");
1725 prepare_packed_git();
1727 if (progress)
1728 start_progress(&progress_state, "Generating pack...",
1729 "Counting objects: ", 0);
1730 if (!use_internal_rev_list)
1731 read_object_list_from_stdin();
1732 else {
1733 rp_av[rp_ac] = NULL;
1734 get_object_list(rp_ac, rp_av);
1735 }
1736 if (progress) {
1737 stop_progress(&progress_state);
1738 fprintf(stderr, "Done counting %u objects.\n", nr_objects);
1739 }
1741 if (non_empty && !nr_result)
1742 return 0;
1743 if (progress && (nr_objects != nr_result))
1744 fprintf(stderr, "Result has %u objects.\n", nr_result);
1745 if (nr_result)
1746 prepare_pack(window, depth);
1747 last_obj_offset = write_pack_file();
1748 if (!pack_to_stdout) {
1749 unsigned char object_list_sha1[20];
1750 mode_t mode = umask(0);
1752 umask(mode);
1753 mode = 0444 & ~mode;
1755 write_index_file(last_obj_offset, object_list_sha1);
1756 snprintf(tmpname, sizeof(tmpname), "%s-%s.pack",
1757 base_name, sha1_to_hex(object_list_sha1));
1758 if (adjust_perm(pack_tmp_name, mode))
1759 die("unable to make temporary pack file readable: %s",
1760 strerror(errno));
1761 if (rename(pack_tmp_name, tmpname))
1762 die("unable to rename temporary pack file: %s",
1763 strerror(errno));
1764 snprintf(tmpname, sizeof(tmpname), "%s-%s.idx",
1765 base_name, sha1_to_hex(object_list_sha1));
1766 if (adjust_perm(idx_tmp_name, mode))
1767 die("unable to make temporary index file readable: %s",
1768 strerror(errno));
1769 if (rename(idx_tmp_name, tmpname))
1770 die("unable to rename temporary index file: %s",
1771 strerror(errno));
1772 puts(sha1_to_hex(object_list_sha1));
1773 }
1774 if (progress)
1775 fprintf(stderr, "Total %u (delta %u), reused %u (delta %u)\n",
1776 written, written_delta, reused, reused_delta);
1777 return 0;
1778 }