]>
Commit | Line | Data |
---|---|---|
5d4a6003 | 1 | #include "builtin.h" |
c323ac7d | 2 | #include "cache.h" |
a74db82e | 3 | #include "attr.h" |
c323ac7d | 4 | #include "object.h" |
8e440259 PE |
5 | #include "blob.h" |
6 | #include "commit.h" | |
7 | #include "tag.h" | |
8 | #include "tree.h" | |
c323ac7d | 9 | #include "delta.h" |
a733cb60 | 10 | #include "pack.h" |
c38138cd | 11 | #include "csum-file.h" |
1b0c7174 | 12 | #include "tree-walk.h" |
b5d97e6b JH |
13 | #include "diff.h" |
14 | #include "revision.h" | |
15 | #include "list-objects.h" | |
96a02f8f | 16 | #include "progress.h" |
c323ac7d | 17 | |
231f240b | 18 | static const char pack_usage[] = "\ |
a97773ce BD |
19 | git-pack-objects [{ -q | --progress | --all-progress }] \n\ |
20 | [--max-pack-size=N] [--local] [--incremental] \n\ | |
21 | [--window=N] [--window-memory=N] [--depth=N] \n\ | |
fa736f72 NP |
22 | [--no-reuse-delta] [--no-reuse-object] [--delta-base-offset] \n\ |
23 | [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\ | |
24 | [--stdout | base-name] [<ref-list | <object-list]"; | |
c323ac7d | 25 | |
c323ac7d | 26 | struct object_entry { |
aa7e44bf | 27 | struct pack_idx_entry idx; |
3f9ac8d2 | 28 | unsigned long size; /* uncompressed size */ |
3f9ac8d2 | 29 | struct packed_git *in_pack; /* already in pack */ |
6777a59f | 30 | off_t in_pack_offset; |
a3fbf4df | 31 | struct object_entry *delta; /* delta base object */ |
82e5a82f | 32 | struct object_entry *delta_child; /* deltified objects who bases me */ |
15b4d577 JH |
33 | struct object_entry *delta_sibling; /* other deltified objects who |
34 | * uses the same base as me | |
35 | */ | |
074b2eea | 36 | void *delta_data; /* cached delta (uncompressed) */ |
a3fbf4df | 37 | unsigned long delta_size; /* delta data size (uncompressed) */ |
5a235b5e | 38 | unsigned int hash; /* name hint hash */ |
a3fbf4df NP |
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 | |
a74db82e | 43 | * to be used as the base object to delta |
a3fbf4df NP |
44 | * objects against. |
45 | */ | |
a74db82e | 46 | unsigned char no_try_delta; |
c323ac7d LT |
47 | }; |
48 | ||
3f9ac8d2 | 49 | /* |
82e5a82f | 50 | * Objects we are going to pack are collected in objects array (dynamically |
3f9ac8d2 JH |
51 | * expanded). nr_objects & nr_alloc controls this array. They are stored |
52 | * in the order we see -- typically rev-list --objects order that gives us | |
53 | * nice "minimum seek" order. | |
3f9ac8d2 | 54 | */ |
e4d58311 | 55 | static struct object_entry *objects; |
d01fb92f DH |
56 | static struct object_entry **written_list; |
57 | static uint32_t nr_objects, nr_alloc, nr_result, nr_written; | |
3f9ac8d2 | 58 | |
96f1e58f | 59 | static int non_empty; |
fa736f72 | 60 | static int no_reuse_delta, no_reuse_object; |
96f1e58f DR |
61 | static int local; |
62 | static int incremental; | |
be6b1914 | 63 | static int allow_ofs_delta; |
81a216a5 NP |
64 | static const char *pack_tmp_name, *idx_tmp_name; |
65 | static char tmpname[PATH_MAX]; | |
d01fb92f | 66 | static const char *base_name; |
024701f1 | 67 | static int progress = 1; |
4812a93a | 68 | static int window = 10; |
d01fb92f | 69 | static uint32_t pack_size_limit; |
618e613a | 70 | static int depth = 50; |
df6d6101 | 71 | static int pack_to_stdout; |
8d1d8f83 | 72 | static int num_preferred_base; |
96a02f8f | 73 | static struct progress progress_state; |
960ccca6 DH |
74 | static int pack_compression_level = Z_DEFAULT_COMPRESSION; |
75 | static int pack_compression_seen; | |
c323ac7d | 76 | |
074b2eea MK |
77 | static unsigned long delta_cache_size = 0; |
78 | static unsigned long max_delta_cache_size = 0; | |
e3dfddb3 | 79 | static unsigned long cache_max_small_delta_size = 1000; |
074b2eea | 80 | |
a97773ce BD |
81 | static unsigned long window_memory_usage = 0; |
82 | static unsigned long window_memory_limit = 0; | |
83 | ||
3f9ac8d2 JH |
84 | /* |
85 | * The object names in objects array are hashed with this hashtable, | |
e4d58311 | 86 | * to help looking up the entry by object name. |
3f9ac8d2 JH |
87 | * This hashtable is built after all the objects are seen. |
88 | */ | |
96f1e58f DR |
89 | static int *object_ix; |
90 | static int object_ix_hashsz; | |
3f9ac8d2 JH |
91 | |
92 | /* | |
93 | * Pack index for existing packs give us easy access to the offsets into | |
94 | * corresponding pack file where each object's data starts, but the entries | |
95 | * do not store the size of the compressed representation (uncompressed | |
780e6e73 NP |
96 | * size is easily available by examining the pack entry header). It is |
97 | * also rather expensive to find the sha1 for an object given its offset. | |
98 | * | |
99 | * We build a hashtable of existing packs (pack_revindex), and keep reverse | |
100 | * index here -- pack index file is sorted by object name mapping to offset; | |
101 | * this pack_revindex[].revindex array is a list of offset/index_nr pairs | |
102 | * ordered by offset, so if you know the offset of an object, next offset | |
103 | * is where its packed representation ends and the index_nr can be used to | |
104 | * get the object sha1 from the main index. | |
3f9ac8d2 | 105 | */ |
780e6e73 | 106 | struct revindex_entry { |
6777a59f | 107 | off_t offset; |
780e6e73 NP |
108 | unsigned int nr; |
109 | }; | |
3f9ac8d2 JH |
110 | struct pack_revindex { |
111 | struct packed_git *p; | |
780e6e73 NP |
112 | struct revindex_entry *revindex; |
113 | }; | |
114 | static struct pack_revindex *pack_revindex; | |
96f1e58f | 115 | static int pack_revindex_hashsz; |
3f9ac8d2 JH |
116 | |
117 | /* | |
118 | * stats | |
119 | */ | |
7cadf491 SP |
120 | static uint32_t written, written_delta; |
121 | static uint32_t reused, reused_delta; | |
3f9ac8d2 JH |
122 | |
123 | static int pack_revindex_ix(struct packed_git *p) | |
124 | { | |
2b74cffa | 125 | unsigned long ui = (unsigned long)p; |
3f9ac8d2 JH |
126 | int i; |
127 | ||
128 | ui = ui ^ (ui >> 16); /* defeat structure alignment */ | |
129 | i = (int)(ui % pack_revindex_hashsz); | |
130 | while (pack_revindex[i].p) { | |
131 | if (pack_revindex[i].p == p) | |
132 | return i; | |
133 | if (++i == pack_revindex_hashsz) | |
134 | i = 0; | |
135 | } | |
136 | return -1 - i; | |
137 | } | |
138 | ||
139 | static void prepare_pack_ix(void) | |
140 | { | |
141 | int num; | |
142 | struct packed_git *p; | |
143 | for (num = 0, p = packed_git; p; p = p->next) | |
144 | num++; | |
145 | if (!num) | |
146 | return; | |
147 | pack_revindex_hashsz = num * 11; | |
148 | pack_revindex = xcalloc(sizeof(*pack_revindex), pack_revindex_hashsz); | |
149 | for (p = packed_git; p; p = p->next) { | |
150 | num = pack_revindex_ix(p); | |
151 | num = - 1 - num; | |
152 | pack_revindex[num].p = p; | |
153 | } | |
154 | /* revindex elements are lazily initialized */ | |
155 | } | |
156 | ||
157 | static int cmp_offset(const void *a_, const void *b_) | |
158 | { | |
780e6e73 NP |
159 | const struct revindex_entry *a = a_; |
160 | const struct revindex_entry *b = b_; | |
161 | return (a->offset < b->offset) ? -1 : (a->offset > b->offset) ? 1 : 0; | |
3f9ac8d2 JH |
162 | } |
163 | ||
164 | /* | |
165 | * Ordered list of offsets of objects in the pack. | |
166 | */ | |
167 | static void prepare_pack_revindex(struct pack_revindex *rix) | |
168 | { | |
169 | struct packed_git *p = rix->p; | |
57059091 | 170 | int num_ent = p->num_objects; |
3f9ac8d2 | 171 | int i; |
42873078 | 172 | const char *index = p->index_data; |
3f9ac8d2 | 173 | |
780e6e73 | 174 | rix->revindex = xmalloc(sizeof(*rix->revindex) * (num_ent + 1)); |
c553ca25 NP |
175 | index += 4 * 256; |
176 | ||
177 | if (p->index_version > 1) { | |
178 | const uint32_t *off_32 = | |
179 | (uint32_t *)(index + 8 + p->num_objects * (20 + 4)); | |
180 | const uint32_t *off_64 = off_32 + p->num_objects; | |
181 | for (i = 0; i < num_ent; i++) { | |
182 | uint32_t off = ntohl(*off_32++); | |
183 | if (!(off & 0x80000000)) { | |
184 | rix->revindex[i].offset = off; | |
185 | } else { | |
186 | rix->revindex[i].offset = | |
187 | ((uint64_t)ntohl(*off_64++)) << 32; | |
188 | rix->revindex[i].offset |= | |
189 | ntohl(*off_64++); | |
190 | } | |
191 | rix->revindex[i].nr = i; | |
192 | } | |
193 | } else { | |
194 | for (i = 0; i < num_ent; i++) { | |
195 | uint32_t hl = *((uint32_t *)(index + 24 * i)); | |
196 | rix->revindex[i].offset = ntohl(hl); | |
197 | rix->revindex[i].nr = i; | |
198 | } | |
3f9ac8d2 | 199 | } |
c553ca25 | 200 | |
3f9ac8d2 JH |
201 | /* This knows the pack format -- the 20-byte trailer |
202 | * follows immediately after the last object data. | |
203 | */ | |
780e6e73 NP |
204 | rix->revindex[num_ent].offset = p->pack_size - 20; |
205 | rix->revindex[num_ent].nr = -1; | |
206 | qsort(rix->revindex, num_ent, sizeof(*rix->revindex), cmp_offset); | |
3f9ac8d2 JH |
207 | } |
208 | ||
780e6e73 | 209 | static struct revindex_entry * find_packed_object(struct packed_git *p, |
6777a59f | 210 | off_t ofs) |
3f9ac8d2 JH |
211 | { |
212 | int num; | |
213 | int lo, hi; | |
214 | struct pack_revindex *rix; | |
780e6e73 | 215 | struct revindex_entry *revindex; |
3f9ac8d2 JH |
216 | num = pack_revindex_ix(p); |
217 | if (num < 0) | |
218 | die("internal error: pack revindex uninitialized"); | |
219 | rix = &pack_revindex[num]; | |
220 | if (!rix->revindex) | |
221 | prepare_pack_revindex(rix); | |
222 | revindex = rix->revindex; | |
223 | lo = 0; | |
57059091 | 224 | hi = p->num_objects + 1; |
3f9ac8d2 JH |
225 | do { |
226 | int mi = (lo + hi) / 2; | |
780e6e73 NP |
227 | if (revindex[mi].offset == ofs) { |
228 | return revindex + mi; | |
3f9ac8d2 | 229 | } |
780e6e73 | 230 | else if (ofs < revindex[mi].offset) |
3f9ac8d2 JH |
231 | hi = mi; |
232 | else | |
233 | lo = mi + 1; | |
234 | } while (lo < hi); | |
235 | die("internal error: pack revindex corrupt"); | |
236 | } | |
237 | ||
42873078 NP |
238 | static const unsigned char *find_packed_object_name(struct packed_git *p, |
239 | off_t ofs) | |
780e6e73 NP |
240 | { |
241 | struct revindex_entry *entry = find_packed_object(p, ofs); | |
d72308e0 | 242 | return nth_packed_object_sha1(p, entry->nr); |
780e6e73 NP |
243 | } |
244 | ||
c323ac7d LT |
245 | static void *delta_against(void *buf, unsigned long size, struct object_entry *entry) |
246 | { | |
247 | unsigned long othersize, delta_size; | |
21666f1a | 248 | enum object_type type; |
aa7e44bf | 249 | void *otherbuf = read_sha1_file(entry->delta->idx.sha1, &type, &othersize); |
c323ac7d LT |
250 | void *delta_buf; |
251 | ||
252 | if (!otherbuf) | |
aa7e44bf | 253 | die("unable to read %s", sha1_to_hex(entry->delta->idx.sha1)); |
8ee378a0 | 254 | delta_buf = diff_delta(otherbuf, othersize, |
dcde55bc | 255 | buf, size, &delta_size, 0); |
75c42d8c | 256 | if (!delta_buf || delta_size != entry->delta_size) |
a6080a0a | 257 | die("delta size changed"); |
c323ac7d LT |
258 | free(buf); |
259 | free(otherbuf); | |
260 | return delta_buf; | |
261 | } | |
262 | ||
a733cb60 LT |
263 | /* |
264 | * The per-object header is a pretty dense thing, which is | |
265 | * - first byte: low four bits are "size", then three bits of "type", | |
266 | * and the high bit is "size continues". | |
267 | * - each byte afterwards: low seven bits are size continuation, | |
268 | * with the high bit being "size continues" | |
269 | */ | |
270 | static int encode_header(enum object_type type, unsigned long size, unsigned char *hdr) | |
271 | { | |
01247d87 | 272 | int n = 1; |
a733cb60 LT |
273 | unsigned char c; |
274 | ||
eb32d236 | 275 | if (type < OBJ_COMMIT || type > OBJ_REF_DELTA) |
a733cb60 LT |
276 | die("bad type %d", type); |
277 | ||
01247d87 LT |
278 | c = (type << 4) | (size & 15); |
279 | size >>= 4; | |
280 | while (size) { | |
a733cb60 | 281 | *hdr++ = c | 0x80; |
01247d87 LT |
282 | c = size & 0x7f; |
283 | size >>= 7; | |
284 | n++; | |
a733cb60 LT |
285 | } |
286 | *hdr = c; | |
287 | return n; | |
288 | } | |
289 | ||
780e6e73 NP |
290 | /* |
291 | * we are going to reuse the existing object data as is. make | |
292 | * sure it is not corrupt. | |
293 | */ | |
079afb18 SP |
294 | static int check_pack_inflate(struct packed_git *p, |
295 | struct pack_window **w_curs, | |
6777a59f SP |
296 | off_t offset, |
297 | off_t len, | |
079afb18 SP |
298 | unsigned long expect) |
299 | { | |
300 | z_stream stream; | |
301 | unsigned char fakebuf[4096], *in; | |
302 | int st; | |
303 | ||
304 | memset(&stream, 0, sizeof(stream)); | |
305 | inflateInit(&stream); | |
306 | do { | |
307 | in = use_pack(p, w_curs, offset, &stream.avail_in); | |
308 | stream.next_in = in; | |
309 | stream.next_out = fakebuf; | |
310 | stream.avail_out = sizeof(fakebuf); | |
311 | st = inflate(&stream, Z_FINISH); | |
312 | offset += stream.next_in - in; | |
313 | } while (st == Z_OK || st == Z_BUF_ERROR); | |
314 | inflateEnd(&stream); | |
315 | return (st == Z_STREAM_END && | |
316 | stream.total_out == expect && | |
317 | stream.total_in == len) ? 0 : -1; | |
318 | } | |
319 | ||
91ecbeca NP |
320 | static int check_pack_crc(struct packed_git *p, struct pack_window **w_curs, |
321 | off_t offset, off_t len, unsigned int nr) | |
322 | { | |
323 | const uint32_t *index_crc; | |
324 | uint32_t data_crc = crc32(0, Z_NULL, 0); | |
325 | ||
326 | do { | |
327 | unsigned int avail; | |
328 | void *data = use_pack(p, w_curs, offset, &avail); | |
329 | if (avail > len) | |
330 | avail = len; | |
331 | data_crc = crc32(data_crc, data, avail); | |
332 | offset += avail; | |
333 | len -= avail; | |
334 | } while (len); | |
335 | ||
336 | index_crc = p->index_data; | |
337 | index_crc += 2 + 256 + p->num_objects * (20/4) + nr; | |
338 | ||
339 | return data_crc != ntohl(*index_crc); | |
340 | } | |
341 | ||
079afb18 SP |
342 | static void copy_pack_data(struct sha1file *f, |
343 | struct packed_git *p, | |
344 | struct pack_window **w_curs, | |
6777a59f SP |
345 | off_t offset, |
346 | off_t len) | |
079afb18 SP |
347 | { |
348 | unsigned char *in; | |
349 | unsigned int avail; | |
350 | ||
351 | while (len) { | |
352 | in = use_pack(p, w_curs, offset, &avail); | |
353 | if (avail > len) | |
6777a59f | 354 | avail = (unsigned int)len; |
079afb18 SP |
355 | sha1write(f, in, avail); |
356 | offset += avail; | |
357 | len -= avail; | |
358 | } | |
359 | } | |
360 | ||
d7dd0223 | 361 | static unsigned long write_object(struct sha1file *f, |
17b08f2c DH |
362 | struct object_entry *entry, |
363 | off_t write_offset) | |
c323ac7d LT |
364 | { |
365 | unsigned long size; | |
21666f1a | 366 | enum object_type type; |
3f9ac8d2 | 367 | void *buf; |
a733cb60 | 368 | unsigned char header[10]; |
17b08f2c | 369 | unsigned char dheader[10]; |
6777a59f SP |
370 | unsigned hdrlen; |
371 | off_t datalen; | |
a733cb60 | 372 | enum object_type obj_type; |
ab7cd7bb | 373 | int to_reuse = 0; |
17b08f2c DH |
374 | /* write limit if limited packsize and not first object */ |
375 | unsigned long limit = pack_size_limit && nr_written ? | |
376 | pack_size_limit - write_offset : 0; | |
377 | /* no if no delta */ | |
378 | int usable_delta = !entry->delta ? 0 : | |
379 | /* yes if unlimited packfile */ | |
380 | !pack_size_limit ? 1 : | |
381 | /* no if base written to previous pack */ | |
aa7e44bf | 382 | entry->delta->idx.offset == (off_t)-1 ? 0 : |
17b08f2c DH |
383 | /* otherwise double-check written to this |
384 | * pack, like we do below | |
385 | */ | |
aa7e44bf | 386 | entry->delta->idx.offset ? 1 : 0; |
c323ac7d | 387 | |
78d1e84f NP |
388 | if (!pack_to_stdout) |
389 | crc32_begin(f); | |
390 | ||
a733cb60 | 391 | obj_type = entry->type; |
fa736f72 NP |
392 | if (no_reuse_object) |
393 | to_reuse = 0; /* explicit */ | |
394 | else if (!entry->in_pack) | |
ab7cd7bb | 395 | to_reuse = 0; /* can't reuse what we don't have */ |
780e6e73 | 396 | else if (obj_type == OBJ_REF_DELTA || obj_type == OBJ_OFS_DELTA) |
17b08f2c DH |
397 | /* check_object() decided it for us ... */ |
398 | to_reuse = usable_delta; | |
399 | /* ... but pack split may override that */ | |
ab7cd7bb JH |
400 | else if (obj_type != entry->in_pack_type) |
401 | to_reuse = 0; /* pack has delta which is unusable */ | |
402 | else if (entry->delta) | |
403 | to_reuse = 0; /* we want to pack afresh */ | |
404 | else | |
405 | to_reuse = 1; /* we have it in-pack undeltified, | |
406 | * and we do not need to deltify it. | |
407 | */ | |
408 | ||
df6d6101 | 409 | if (!to_reuse) { |
17b08f2c DH |
410 | z_stream stream; |
411 | unsigned long maxsize; | |
412 | void *out; | |
5476a8ad | 413 | if (!usable_delta) { |
aa7e44bf | 414 | buf = read_sha1_file(entry->idx.sha1, &obj_type, &size); |
5476a8ad | 415 | if (!buf) |
aa7e44bf | 416 | die("unable to read %s", sha1_to_hex(entry->idx.sha1)); |
5476a8ad | 417 | } else if (entry->delta_data) { |
3f9ac8d2 | 418 | size = entry->delta_size; |
5476a8ad NP |
419 | buf = entry->delta_data; |
420 | entry->delta_data = NULL; | |
aa7e44bf | 421 | obj_type = (allow_ofs_delta && entry->delta->idx.offset) ? |
be6b1914 | 422 | OBJ_OFS_DELTA : OBJ_REF_DELTA; |
17b08f2c | 423 | } else { |
aa7e44bf | 424 | buf = read_sha1_file(entry->idx.sha1, &type, &size); |
074b2eea | 425 | if (!buf) |
aa7e44bf | 426 | die("unable to read %s", sha1_to_hex(entry->idx.sha1)); |
5476a8ad NP |
427 | buf = delta_against(buf, size, entry); |
428 | size = entry->delta_size; | |
aa7e44bf | 429 | obj_type = (allow_ofs_delta && entry->delta->idx.offset) ? |
5476a8ad | 430 | OBJ_OFS_DELTA : OBJ_REF_DELTA; |
3f9ac8d2 | 431 | } |
17b08f2c DH |
432 | /* compress the data to store and put compressed length in datalen */ |
433 | memset(&stream, 0, sizeof(stream)); | |
434 | deflateInit(&stream, pack_compression_level); | |
435 | maxsize = deflateBound(&stream, size); | |
436 | out = xmalloc(maxsize); | |
437 | /* Compress it */ | |
438 | stream.next_in = buf; | |
439 | stream.avail_in = size; | |
440 | stream.next_out = out; | |
441 | stream.avail_out = maxsize; | |
442 | while (deflate(&stream, Z_FINISH) == Z_OK) | |
443 | /* nothing */; | |
444 | deflateEnd(&stream); | |
445 | datalen = stream.total_out; | |
446 | deflateEnd(&stream); | |
3f9ac8d2 JH |
447 | /* |
448 | * The object header is a byte of 'type' followed by zero or | |
be6b1914 | 449 | * more bytes of length. |
3f9ac8d2 JH |
450 | */ |
451 | hdrlen = encode_header(obj_type, size, header); | |
3f9ac8d2 | 452 | |
be6b1914 NP |
453 | if (obj_type == OBJ_OFS_DELTA) { |
454 | /* | |
455 | * Deltas with relative base contain an additional | |
456 | * encoding of the relative offset for the delta | |
457 | * base from this object's position in the pack. | |
458 | */ | |
aa7e44bf | 459 | off_t ofs = entry->idx.offset - entry->delta->idx.offset; |
17b08f2c DH |
460 | unsigned pos = sizeof(dheader) - 1; |
461 | dheader[pos] = ofs & 127; | |
be6b1914 | 462 | while (ofs >>= 7) |
17b08f2c DH |
463 | dheader[--pos] = 128 | (--ofs & 127); |
464 | if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) { | |
465 | free(out); | |
466 | free(buf); | |
467 | return 0; | |
468 | } | |
469 | sha1write(f, header, hdrlen); | |
470 | sha1write(f, dheader + pos, sizeof(dheader) - pos); | |
471 | hdrlen += sizeof(dheader) - pos; | |
be6b1914 NP |
472 | } else if (obj_type == OBJ_REF_DELTA) { |
473 | /* | |
474 | * Deltas with a base reference contain | |
475 | * an additional 20 bytes for the base sha1. | |
476 | */ | |
17b08f2c DH |
477 | if (limit && hdrlen + 20 + datalen + 20 >= limit) { |
478 | free(out); | |
479 | free(buf); | |
480 | return 0; | |
481 | } | |
482 | sha1write(f, header, hdrlen); | |
aa7e44bf | 483 | sha1write(f, entry->delta->idx.sha1, 20); |
3f9ac8d2 | 484 | hdrlen += 20; |
17b08f2c DH |
485 | } else { |
486 | if (limit && hdrlen + datalen + 20 >= limit) { | |
487 | free(out); | |
488 | free(buf); | |
489 | return 0; | |
490 | } | |
491 | sha1write(f, header, hdrlen); | |
3f9ac8d2 | 492 | } |
17b08f2c DH |
493 | sha1write(f, out, datalen); |
494 | free(out); | |
3f9ac8d2 | 495 | free(buf); |
c323ac7d | 496 | } |
3f9ac8d2 JH |
497 | else { |
498 | struct packed_git *p = entry->in_pack; | |
03e79c88 | 499 | struct pack_window *w_curs = NULL; |
91ecbeca | 500 | struct revindex_entry *revidx; |
6777a59f | 501 | off_t offset; |
3f9ac8d2 | 502 | |
780e6e73 | 503 | if (entry->delta) { |
aa7e44bf | 504 | obj_type = (allow_ofs_delta && entry->delta->idx.offset) ? |
780e6e73 NP |
505 | OBJ_OFS_DELTA : OBJ_REF_DELTA; |
506 | reused_delta++; | |
507 | } | |
508 | hdrlen = encode_header(obj_type, entry->size, header); | |
91ecbeca NP |
509 | offset = entry->in_pack_offset; |
510 | revidx = find_packed_object(p, offset); | |
511 | datalen = revidx[1].offset - offset; | |
512 | if (!pack_to_stdout && p->index_version > 1 && | |
513 | check_pack_crc(p, &w_curs, offset, datalen, revidx->nr)) | |
aa7e44bf | 514 | die("bad packed object CRC for %s", sha1_to_hex(entry->idx.sha1)); |
91ecbeca NP |
515 | offset += entry->in_pack_header_size; |
516 | datalen -= entry->in_pack_header_size; | |
17b08f2c | 517 | if (obj_type == OBJ_OFS_DELTA) { |
aa7e44bf | 518 | off_t ofs = entry->idx.offset - entry->delta->idx.offset; |
17b08f2c DH |
519 | unsigned pos = sizeof(dheader) - 1; |
520 | dheader[pos] = ofs & 127; | |
521 | while (ofs >>= 7) | |
522 | dheader[--pos] = 128 | (--ofs & 127); | |
523 | if (limit && hdrlen + sizeof(dheader) - pos + datalen + 20 >= limit) | |
524 | return 0; | |
525 | sha1write(f, header, hdrlen); | |
526 | sha1write(f, dheader + pos, sizeof(dheader) - pos); | |
527 | hdrlen += sizeof(dheader) - pos; | |
528 | } else if (obj_type == OBJ_REF_DELTA) { | |
529 | if (limit && hdrlen + 20 + datalen + 20 >= limit) | |
530 | return 0; | |
531 | sha1write(f, header, hdrlen); | |
aa7e44bf | 532 | sha1write(f, entry->delta->idx.sha1, 20); |
17b08f2c DH |
533 | hdrlen += 20; |
534 | } else { | |
535 | if (limit && hdrlen + datalen + 20 >= limit) | |
536 | return 0; | |
537 | sha1write(f, header, hdrlen); | |
538 | } | |
539 | ||
91ecbeca NP |
540 | if (!pack_to_stdout && p->index_version == 1 && |
541 | check_pack_inflate(p, &w_curs, offset, datalen, entry->size)) | |
aa7e44bf | 542 | die("corrupt packed object for %s", sha1_to_hex(entry->idx.sha1)); |
079afb18 | 543 | copy_pack_data(f, p, &w_curs, offset, datalen); |
03e79c88 | 544 | unuse_pack(&w_curs); |
3f9ac8d2 | 545 | reused++; |
a733cb60 | 546 | } |
17b08f2c | 547 | if (usable_delta) |
ab7cd7bb | 548 | written_delta++; |
3f9ac8d2 | 549 | written++; |
78d1e84f | 550 | if (!pack_to_stdout) |
aa7e44bf | 551 | entry->idx.crc32 = crc32_end(f); |
c323ac7d LT |
552 | return hdrlen + datalen; |
553 | } | |
554 | ||
6777a59f | 555 | static off_t write_one(struct sha1file *f, |
9d5ab962 | 556 | struct object_entry *e, |
6777a59f | 557 | off_t offset) |
9d5ab962 | 558 | { |
d7dd0223 NP |
559 | unsigned long size; |
560 | ||
561 | /* offset is non zero if object is written already. */ | |
aa7e44bf | 562 | if (e->idx.offset || e->preferred_base) |
9d5ab962 | 563 | return offset; |
d7dd0223 NP |
564 | |
565 | /* if we are deltified, write out base object first. */ | |
17b08f2c | 566 | if (e->delta) { |
9d5ab962 | 567 | offset = write_one(f, e->delta, offset); |
17b08f2c DH |
568 | if (!offset) |
569 | return 0; | |
570 | } | |
d7dd0223 | 571 | |
aa7e44bf | 572 | e->idx.offset = offset; |
17b08f2c DH |
573 | size = write_object(f, e, offset); |
574 | if (!size) { | |
aa7e44bf | 575 | e->idx.offset = 0; |
17b08f2c DH |
576 | return 0; |
577 | } | |
ebe27b13 | 578 | written_list[nr_written++] = e; |
d7dd0223 NP |
579 | |
580 | /* make sure off_t is sufficiently large not to wrap */ | |
581 | if (offset > offset + size) | |
582 | die("pack too large for current definition of off_t"); | |
583 | return offset + size; | |
9d5ab962 JH |
584 | } |
585 | ||
3082acfa AR |
586 | static int open_object_dir_tmp(const char *path) |
587 | { | |
588 | snprintf(tmpname, sizeof(tmpname), "%s/%s", get_object_directory(), path); | |
7647b17f | 589 | return xmkstemp(tmpname); |
3082acfa AR |
590 | } |
591 | ||
aa7e44bf | 592 | /* forward declaration for write_pack_file */ |
d01fb92f DH |
593 | static int adjust_perm(const char *path, mode_t mode); |
594 | ||
595 | static void write_pack_file(void) | |
c323ac7d | 596 | { |
ebe27b13 | 597 | uint32_t i = 0, j; |
d22b9290 | 598 | struct sha1file *f; |
ebe27b13 | 599 | off_t offset, offset_one, last_obj_offset = 0; |
a733cb60 | 600 | struct pack_header hdr; |
ebe27b13 DH |
601 | int do_progress = progress >> pack_to_stdout; |
602 | uint32_t nr_remaining = nr_result; | |
81a216a5 | 603 | |
13aaf148 NP |
604 | if (do_progress) |
605 | start_progress(&progress_state, "Writing %u objects...", "", nr_result); | |
ebe27b13 | 606 | written_list = xmalloc(nr_objects * sizeof(struct object_entry *)); |
183bdb2c | 607 | |
ebe27b13 | 608 | do { |
aa7e44bf GB |
609 | unsigned char sha1[20]; |
610 | ||
ebe27b13 DH |
611 | if (pack_to_stdout) { |
612 | f = sha1fd(1, "<stdout>"); | |
613 | } else { | |
614 | int fd = open_object_dir_tmp("tmp_pack_XXXXXX"); | |
ebe27b13 DH |
615 | pack_tmp_name = xstrdup(tmpname); |
616 | f = sha1fd(fd, pack_tmp_name); | |
617 | } | |
618 | ||
619 | hdr.hdr_signature = htonl(PACK_SIGNATURE); | |
620 | hdr.hdr_version = htonl(PACK_VERSION); | |
621 | hdr.hdr_entries = htonl(nr_remaining); | |
622 | sha1write(f, &hdr, sizeof(hdr)); | |
623 | offset = sizeof(hdr); | |
624 | nr_written = 0; | |
625 | for (; i < nr_objects; i++) { | |
626 | last_obj_offset = offset; | |
627 | offset_one = write_one(f, objects + i, offset); | |
628 | if (!offset_one) | |
629 | break; | |
630 | offset = offset_one; | |
631 | if (do_progress) | |
632 | display_progress(&progress_state, written); | |
633 | } | |
c553ca25 | 634 | |
ebe27b13 DH |
635 | /* |
636 | * Did we write the wrong # entries in the header? | |
637 | * If so, rewrite it like in fast-import | |
638 | */ | |
639 | if (pack_to_stdout || nr_written == nr_remaining) { | |
aa7e44bf | 640 | sha1close(f, sha1, 1); |
ebe27b13 | 641 | } else { |
aa7e44bf GB |
642 | sha1close(f, sha1, 0); |
643 | fixup_pack_header_footer(f->fd, sha1, pack_tmp_name, nr_written); | |
ebe27b13 DH |
644 | close(f->fd); |
645 | } | |
646 | ||
647 | if (!pack_to_stdout) { | |
d01fb92f DH |
648 | mode_t mode = umask(0); |
649 | ||
650 | umask(mode); | |
651 | mode = 0444 & ~mode; | |
652 | ||
aa7e44bf GB |
653 | idx_tmp_name = write_idx_file(NULL, |
654 | (struct pack_idx_entry **) written_list, nr_written, sha1); | |
d01fb92f | 655 | snprintf(tmpname, sizeof(tmpname), "%s-%s.pack", |
aa7e44bf | 656 | base_name, sha1_to_hex(sha1)); |
d01fb92f DH |
657 | if (adjust_perm(pack_tmp_name, mode)) |
658 | die("unable to make temporary pack file readable: %s", | |
659 | strerror(errno)); | |
660 | if (rename(pack_tmp_name, tmpname)) | |
661 | die("unable to rename temporary pack file: %s", | |
662 | strerror(errno)); | |
663 | snprintf(tmpname, sizeof(tmpname), "%s-%s.idx", | |
aa7e44bf | 664 | base_name, sha1_to_hex(sha1)); |
d01fb92f DH |
665 | if (adjust_perm(idx_tmp_name, mode)) |
666 | die("unable to make temporary index file readable: %s", | |
667 | strerror(errno)); | |
668 | if (rename(idx_tmp_name, tmpname)) | |
669 | die("unable to rename temporary index file: %s", | |
670 | strerror(errno)); | |
aa7e44bf | 671 | puts(sha1_to_hex(sha1)); |
ebe27b13 DH |
672 | } |
673 | ||
674 | /* mark written objects as written to previous pack */ | |
675 | for (j = 0; j < nr_written; j++) { | |
aa7e44bf | 676 | written_list[j]->idx.offset = (off_t)-1; |
ebe27b13 DH |
677 | } |
678 | nr_remaining -= nr_written; | |
679 | } while (nr_remaining && i < nr_objects); | |
680 | ||
681 | free(written_list); | |
183bdb2c | 682 | if (do_progress) |
96a02f8f | 683 | stop_progress(&progress_state); |
67c08ce1 | 684 | if (written != nr_result) |
7cadf491 | 685 | die("wrote %u objects while expecting %u", written, nr_result); |
01c12a23 DH |
686 | /* |
687 | * We have scanned through [0 ... i). Since we have written | |
688 | * the correct number of objects, the remaining [i ... nr_objects) | |
689 | * items must be either already written (due to out-of-order delta base) | |
690 | * or a preferred base. Count those which are neither and complain if any. | |
691 | */ | |
ebe27b13 DH |
692 | for (j = 0; i < nr_objects; i++) { |
693 | struct object_entry *e = objects + i; | |
aa7e44bf | 694 | j += !e->idx.offset && !e->preferred_base; |
d01fb92f | 695 | } |
ebe27b13 DH |
696 | if (j) |
697 | die("wrote %u objects as expected but %u unwritten", written, j); | |
c323ac7d LT |
698 | } |
699 | ||
7a979d99 JH |
700 | static int locate_object_entry_hash(const unsigned char *sha1) |
701 | { | |
702 | int i; | |
703 | unsigned int ui; | |
704 | memcpy(&ui, sha1, sizeof(unsigned int)); | |
705 | i = ui % object_ix_hashsz; | |
706 | while (0 < object_ix[i]) { | |
aa7e44bf | 707 | if (!hashcmp(sha1, objects[object_ix[i] - 1].idx.sha1)) |
7a979d99 JH |
708 | return i; |
709 | if (++i == object_ix_hashsz) | |
710 | i = 0; | |
711 | } | |
712 | return -1 - i; | |
713 | } | |
714 | ||
715 | static struct object_entry *locate_object_entry(const unsigned char *sha1) | |
716 | { | |
717 | int i; | |
718 | ||
719 | if (!object_ix_hashsz) | |
720 | return NULL; | |
721 | ||
722 | i = locate_object_entry_hash(sha1); | |
723 | if (0 <= i) | |
724 | return &objects[object_ix[i]-1]; | |
725 | return NULL; | |
726 | } | |
727 | ||
728 | static void rehash_objects(void) | |
c323ac7d | 729 | { |
7cadf491 | 730 | uint32_t i; |
7a979d99 JH |
731 | struct object_entry *oe; |
732 | ||
733 | object_ix_hashsz = nr_objects * 3; | |
734 | if (object_ix_hashsz < 1024) | |
735 | object_ix_hashsz = 1024; | |
736 | object_ix = xrealloc(object_ix, sizeof(int) * object_ix_hashsz); | |
5379a5c5 | 737 | memset(object_ix, 0, sizeof(int) * object_ix_hashsz); |
7a979d99 | 738 | for (i = 0, oe = objects; i < nr_objects; i++, oe++) { |
aa7e44bf | 739 | int ix = locate_object_entry_hash(oe->idx.sha1); |
7a979d99 JH |
740 | if (0 <= ix) |
741 | continue; | |
742 | ix = -1 - ix; | |
743 | object_ix[ix] = i + 1; | |
744 | } | |
745 | } | |
746 | ||
ce0bd642 | 747 | static unsigned name_hash(const char *name) |
1d6b38cc | 748 | { |
ce0bd642 LT |
749 | unsigned char c; |
750 | unsigned hash = 0; | |
751 | ||
bc32fed5 JH |
752 | if (!name) |
753 | return 0; | |
754 | ||
eeef7135 | 755 | /* |
ce0bd642 LT |
756 | * This effectively just creates a sortable number from the |
757 | * last sixteen non-whitespace characters. Last characters | |
758 | * count "most", so things that end in ".c" sort together. | |
eeef7135 | 759 | */ |
ce0bd642 LT |
760 | while ((c = *name++) != 0) { |
761 | if (isspace(c)) | |
762 | continue; | |
763 | hash = (hash >> 2) + (c << 24); | |
764 | } | |
1d6b38cc JH |
765 | return hash; |
766 | } | |
767 | ||
a74db82e JH |
768 | static void setup_delta_attr_check(struct git_attr_check *check) |
769 | { | |
770 | static struct git_attr *attr_delta; | |
771 | ||
772 | if (!attr_delta) | |
773 | attr_delta = git_attr("delta", 5); | |
774 | ||
775 | check[0].attr = attr_delta; | |
776 | } | |
777 | ||
778 | static int no_try_delta(const char *path) | |
779 | { | |
780 | struct git_attr_check check[1]; | |
781 | ||
782 | setup_delta_attr_check(check); | |
783 | if (git_checkattr(path, ARRAY_SIZE(check), check)) | |
784 | return 0; | |
785 | if (ATTR_FALSE(check->value)) | |
786 | return 1; | |
787 | return 0; | |
788 | } | |
789 | ||
5c49c116 | 790 | static int add_object_entry(const unsigned char *sha1, enum object_type type, |
bc32fed5 | 791 | const char *name, int exclude) |
c323ac7d | 792 | { |
c323ac7d | 793 | struct object_entry *entry; |
29b734e4 | 794 | struct packed_git *p, *found_pack = NULL; |
6777a59f | 795 | off_t found_offset = 0; |
29b734e4 | 796 | int ix; |
bc32fed5 | 797 | unsigned hash = name_hash(name); |
29b734e4 NP |
798 | |
799 | ix = nr_objects ? locate_object_entry_hash(sha1) : -1; | |
800 | if (ix >= 0) { | |
801 | if (exclude) { | |
802 | entry = objects + object_ix[ix] - 1; | |
81a216a5 NP |
803 | if (!entry->preferred_base) |
804 | nr_result--; | |
29b734e4 NP |
805 | entry->preferred_base = 1; |
806 | } | |
807 | return 0; | |
808 | } | |
c323ac7d | 809 | |
5c49c116 NP |
810 | for (p = packed_git; p; p = p->next) { |
811 | off_t offset = find_pack_entry_one(sha1, p); | |
812 | if (offset) { | |
813 | if (!found_pack) { | |
814 | found_offset = offset; | |
815 | found_pack = p; | |
64560374 | 816 | } |
5c49c116 NP |
817 | if (exclude) |
818 | break; | |
819 | if (incremental) | |
820 | return 0; | |
821 | if (local && !p->pack_local) | |
822 | return 0; | |
64560374 LT |
823 | } |
824 | } | |
eb019375 | 825 | |
29b734e4 NP |
826 | if (nr_objects >= nr_alloc) { |
827 | nr_alloc = (nr_alloc + 1024) * 3 / 2; | |
7cadf491 | 828 | objects = xrealloc(objects, nr_alloc * sizeof(*entry)); |
c323ac7d | 829 | } |
29b734e4 NP |
830 | |
831 | entry = objects + nr_objects++; | |
c323ac7d | 832 | memset(entry, 0, sizeof(*entry)); |
aa7e44bf | 833 | hashcpy(entry->idx.sha1, sha1); |
27225f2e | 834 | entry->hash = hash; |
5c49c116 NP |
835 | if (type) |
836 | entry->type = type; | |
29b734e4 NP |
837 | if (exclude) |
838 | entry->preferred_base = 1; | |
81a216a5 NP |
839 | else |
840 | nr_result++; | |
29b734e4 NP |
841 | if (found_pack) { |
842 | entry->in_pack = found_pack; | |
843 | entry->in_pack_offset = found_offset; | |
844 | } | |
7a979d99 JH |
845 | |
846 | if (object_ix_hashsz * 3 <= nr_objects * 4) | |
847 | rehash_objects(); | |
29b734e4 NP |
848 | else |
849 | object_ix[-1 - ix] = nr_objects; | |
7a979d99 | 850 | |
96a02f8f NP |
851 | if (progress) |
852 | display_progress(&progress_state, nr_objects); | |
29b734e4 | 853 | |
a74db82e JH |
854 | if (name && no_try_delta(name)) |
855 | entry->no_try_delta = 1; | |
856 | ||
29b734e4 | 857 | return 1; |
c323ac7d LT |
858 | } |
859 | ||
5379a5c5 JH |
860 | struct pbase_tree_cache { |
861 | unsigned char sha1[20]; | |
862 | int ref; | |
863 | int temporary; | |
864 | void *tree_data; | |
865 | unsigned long tree_size; | |
866 | }; | |
867 | ||
868 | static struct pbase_tree_cache *(pbase_tree_cache[256]); | |
869 | static int pbase_tree_cache_ix(const unsigned char *sha1) | |
870 | { | |
871 | return sha1[0] % ARRAY_SIZE(pbase_tree_cache); | |
872 | } | |
873 | static int pbase_tree_cache_ix_incr(int ix) | |
874 | { | |
875 | return (ix+1) % ARRAY_SIZE(pbase_tree_cache); | |
876 | } | |
877 | ||
878 | static struct pbase_tree { | |
879 | struct pbase_tree *next; | |
880 | /* This is a phony "cache" entry; we are not | |
881 | * going to evict it nor find it through _get() | |
882 | * mechanism -- this is for the toplevel node that | |
883 | * would almost always change with any commit. | |
884 | */ | |
885 | struct pbase_tree_cache pcache; | |
886 | } *pbase_tree; | |
887 | ||
888 | static struct pbase_tree_cache *pbase_tree_get(const unsigned char *sha1) | |
889 | { | |
890 | struct pbase_tree_cache *ent, *nent; | |
891 | void *data; | |
892 | unsigned long size; | |
21666f1a | 893 | enum object_type type; |
5379a5c5 JH |
894 | int neigh; |
895 | int my_ix = pbase_tree_cache_ix(sha1); | |
896 | int available_ix = -1; | |
897 | ||
898 | /* pbase-tree-cache acts as a limited hashtable. | |
899 | * your object will be found at your index or within a few | |
900 | * slots after that slot if it is cached. | |
901 | */ | |
902 | for (neigh = 0; neigh < 8; neigh++) { | |
903 | ent = pbase_tree_cache[my_ix]; | |
a89fccd2 | 904 | if (ent && !hashcmp(ent->sha1, sha1)) { |
5379a5c5 JH |
905 | ent->ref++; |
906 | return ent; | |
907 | } | |
908 | else if (((available_ix < 0) && (!ent || !ent->ref)) || | |
909 | ((0 <= available_ix) && | |
910 | (!ent && pbase_tree_cache[available_ix]))) | |
911 | available_ix = my_ix; | |
912 | if (!ent) | |
913 | break; | |
914 | my_ix = pbase_tree_cache_ix_incr(my_ix); | |
915 | } | |
916 | ||
917 | /* Did not find one. Either we got a bogus request or | |
918 | * we need to read and perhaps cache. | |
919 | */ | |
21666f1a | 920 | data = read_sha1_file(sha1, &type, &size); |
5379a5c5 JH |
921 | if (!data) |
922 | return NULL; | |
21666f1a | 923 | if (type != OBJ_TREE) { |
5379a5c5 JH |
924 | free(data); |
925 | return NULL; | |
926 | } | |
927 | ||
928 | /* We need to either cache or return a throwaway copy */ | |
929 | ||
930 | if (available_ix < 0) | |
931 | ent = NULL; | |
932 | else { | |
933 | ent = pbase_tree_cache[available_ix]; | |
934 | my_ix = available_ix; | |
935 | } | |
936 | ||
937 | if (!ent) { | |
938 | nent = xmalloc(sizeof(*nent)); | |
939 | nent->temporary = (available_ix < 0); | |
940 | } | |
941 | else { | |
942 | /* evict and reuse */ | |
943 | free(ent->tree_data); | |
944 | nent = ent; | |
945 | } | |
e702496e | 946 | hashcpy(nent->sha1, sha1); |
5379a5c5 JH |
947 | nent->tree_data = data; |
948 | nent->tree_size = size; | |
949 | nent->ref = 1; | |
950 | if (!nent->temporary) | |
951 | pbase_tree_cache[my_ix] = nent; | |
952 | return nent; | |
953 | } | |
954 | ||
955 | static void pbase_tree_put(struct pbase_tree_cache *cache) | |
956 | { | |
957 | if (!cache->temporary) { | |
958 | cache->ref--; | |
959 | return; | |
960 | } | |
961 | free(cache->tree_data); | |
962 | free(cache); | |
963 | } | |
964 | ||
965 | static int name_cmp_len(const char *name) | |
966 | { | |
967 | int i; | |
968 | for (i = 0; name[i] && name[i] != '\n' && name[i] != '/'; i++) | |
969 | ; | |
970 | return i; | |
971 | } | |
972 | ||
973 | static void add_pbase_object(struct tree_desc *tree, | |
5379a5c5 | 974 | const char *name, |
ce0bd642 LT |
975 | int cmplen, |
976 | const char *fullname) | |
3f9ac8d2 | 977 | { |
4c068a98 | 978 | struct name_entry entry; |
8a5a8d6c | 979 | int cmp; |
4c068a98 LT |
980 | |
981 | while (tree_entry(tree,&entry)) { | |
1211be6b LT |
982 | if (S_ISGITLINK(entry.mode)) |
983 | continue; | |
8a5a8d6c NP |
984 | cmp = tree_entry_len(entry.path, entry.sha1) != cmplen ? 1 : |
985 | memcmp(name, entry.path, cmplen); | |
986 | if (cmp > 0) | |
7a979d99 | 987 | continue; |
8a5a8d6c NP |
988 | if (cmp < 0) |
989 | return; | |
5379a5c5 | 990 | if (name[cmplen] != '/') { |
5c49c116 NP |
991 | add_object_entry(entry.sha1, |
992 | S_ISDIR(entry.mode) ? OBJ_TREE : OBJ_BLOB, | |
bc32fed5 | 993 | fullname, 1); |
5379a5c5 JH |
994 | return; |
995 | } | |
8a5a8d6c | 996 | if (S_ISDIR(entry.mode)) { |
7a979d99 | 997 | struct tree_desc sub; |
5379a5c5 JH |
998 | struct pbase_tree_cache *tree; |
999 | const char *down = name+cmplen+1; | |
1000 | int downlen = name_cmp_len(down); | |
1001 | ||
4c068a98 | 1002 | tree = pbase_tree_get(entry.sha1); |
5379a5c5 JH |
1003 | if (!tree) |
1004 | return; | |
6fda5e51 | 1005 | init_tree_desc(&sub, tree->tree_data, tree->tree_size); |
5379a5c5 | 1006 | |
ce0bd642 | 1007 | add_pbase_object(&sub, down, downlen, fullname); |
5379a5c5 JH |
1008 | pbase_tree_put(tree); |
1009 | } | |
1010 | } | |
1011 | } | |
1d6b38cc | 1012 | |
5379a5c5 JH |
1013 | static unsigned *done_pbase_paths; |
1014 | static int done_pbase_paths_num; | |
1015 | static int done_pbase_paths_alloc; | |
1016 | static int done_pbase_path_pos(unsigned hash) | |
1017 | { | |
1018 | int lo = 0; | |
1019 | int hi = done_pbase_paths_num; | |
1020 | while (lo < hi) { | |
1021 | int mi = (hi + lo) / 2; | |
1022 | if (done_pbase_paths[mi] == hash) | |
1023 | return mi; | |
1024 | if (done_pbase_paths[mi] < hash) | |
1025 | hi = mi; | |
1026 | else | |
1027 | lo = mi + 1; | |
1028 | } | |
1029 | return -lo-1; | |
1030 | } | |
1031 | ||
1032 | static int check_pbase_path(unsigned hash) | |
1033 | { | |
1034 | int pos = (!done_pbase_paths) ? -1 : done_pbase_path_pos(hash); | |
1035 | if (0 <= pos) | |
1036 | return 1; | |
1037 | pos = -pos - 1; | |
1038 | if (done_pbase_paths_alloc <= done_pbase_paths_num) { | |
1039 | done_pbase_paths_alloc = alloc_nr(done_pbase_paths_alloc); | |
1040 | done_pbase_paths = xrealloc(done_pbase_paths, | |
1041 | done_pbase_paths_alloc * | |
1042 | sizeof(unsigned)); | |
1043 | } | |
1044 | done_pbase_paths_num++; | |
1045 | if (pos < done_pbase_paths_num) | |
1046 | memmove(done_pbase_paths + pos + 1, | |
1047 | done_pbase_paths + pos, | |
1048 | (done_pbase_paths_num - pos - 1) * sizeof(unsigned)); | |
1049 | done_pbase_paths[pos] = hash; | |
1050 | return 0; | |
1051 | } | |
1052 | ||
bc32fed5 | 1053 | static void add_preferred_base_object(const char *name) |
5379a5c5 JH |
1054 | { |
1055 | struct pbase_tree *it; | |
8a5a8d6c | 1056 | int cmplen; |
bc32fed5 | 1057 | unsigned hash = name_hash(name); |
5379a5c5 | 1058 | |
8a5a8d6c | 1059 | if (!num_preferred_base || check_pbase_path(hash)) |
5379a5c5 JH |
1060 | return; |
1061 | ||
8a5a8d6c | 1062 | cmplen = name_cmp_len(name); |
5379a5c5 JH |
1063 | for (it = pbase_tree; it; it = it->next) { |
1064 | if (cmplen == 0) { | |
bc32fed5 | 1065 | add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1); |
5379a5c5 JH |
1066 | } |
1067 | else { | |
1068 | struct tree_desc tree; | |
6fda5e51 | 1069 | init_tree_desc(&tree, it->pcache.tree_data, it->pcache.tree_size); |
ce0bd642 | 1070 | add_pbase_object(&tree, name, cmplen, name); |
7a979d99 | 1071 | } |
3f9ac8d2 | 1072 | } |
3f9ac8d2 JH |
1073 | } |
1074 | ||
7a979d99 | 1075 | static void add_preferred_base(unsigned char *sha1) |
3f9ac8d2 | 1076 | { |
5379a5c5 JH |
1077 | struct pbase_tree *it; |
1078 | void *data; | |
1079 | unsigned long size; | |
1080 | unsigned char tree_sha1[20]; | |
1d6b38cc | 1081 | |
8d1d8f83 JH |
1082 | if (window <= num_preferred_base++) |
1083 | return; | |
1084 | ||
5379a5c5 JH |
1085 | data = read_object_with_reference(sha1, tree_type, &size, tree_sha1); |
1086 | if (!data) | |
7a979d99 | 1087 | return; |
5379a5c5 JH |
1088 | |
1089 | for (it = pbase_tree; it; it = it->next) { | |
a89fccd2 | 1090 | if (!hashcmp(it->pcache.sha1, tree_sha1)) { |
5379a5c5 JH |
1091 | free(data); |
1092 | return; | |
1093 | } | |
1094 | } | |
1095 | ||
1096 | it = xcalloc(1, sizeof(*it)); | |
1097 | it->next = pbase_tree; | |
1098 | pbase_tree = it; | |
1099 | ||
e702496e | 1100 | hashcpy(it->pcache.sha1, tree_sha1); |
5379a5c5 JH |
1101 | it->pcache.tree_data = data; |
1102 | it->pcache.tree_size = size; | |
3f9ac8d2 JH |
1103 | } |
1104 | ||
c323ac7d LT |
1105 | static void check_object(struct object_entry *entry) |
1106 | { | |
5c49c116 | 1107 | if (entry->in_pack) { |
780e6e73 | 1108 | struct packed_git *p = entry->in_pack; |
03e79c88 | 1109 | struct pack_window *w_curs = NULL; |
5c49c116 NP |
1110 | const unsigned char *base_ref = NULL; |
1111 | struct object_entry *base_entry; | |
1112 | unsigned long used, used_0; | |
6777a59f | 1113 | unsigned int avail; |
5c49c116 NP |
1114 | off_t ofs; |
1115 | unsigned char *buf, c; | |
780e6e73 | 1116 | |
6777a59f | 1117 | buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail); |
ab7cd7bb | 1118 | |
5c49c116 | 1119 | /* |
fa736f72 NP |
1120 | * We want in_pack_type even if we do not reuse delta |
1121 | * since non-delta representations could still be reused. | |
ab7cd7bb | 1122 | */ |
6777a59f | 1123 | used = unpack_object_header_gently(buf, avail, |
5c49c116 NP |
1124 | &entry->in_pack_type, |
1125 | &entry->size); | |
ab7cd7bb | 1126 | |
5c49c116 NP |
1127 | /* |
1128 | * Determine if this is a delta and if so whether we can | |
1129 | * reuse it or not. Otherwise let's find out as cheaply as | |
1130 | * possible what the actual type and size for this object is. | |
3f9ac8d2 | 1131 | */ |
5c49c116 NP |
1132 | switch (entry->in_pack_type) { |
1133 | default: | |
1134 | /* Not a delta hence we've already got all we need. */ | |
1135 | entry->type = entry->in_pack_type; | |
1136 | entry->in_pack_header_size = used; | |
1137 | unuse_pack(&w_curs); | |
1138 | return; | |
1139 | case OBJ_REF_DELTA: | |
1140 | if (!no_reuse_delta && !entry->preferred_base) | |
1141 | base_ref = use_pack(p, &w_curs, | |
1142 | entry->in_pack_offset + used, NULL); | |
1143 | entry->in_pack_header_size = used + 20; | |
1144 | break; | |
1145 | case OBJ_OFS_DELTA: | |
1146 | buf = use_pack(p, &w_curs, | |
1147 | entry->in_pack_offset + used, NULL); | |
1148 | used_0 = 0; | |
1149 | c = buf[used_0++]; | |
1150 | ofs = c & 127; | |
1151 | while (c & 128) { | |
1152 | ofs += 1; | |
1153 | if (!ofs || MSB(ofs, 7)) | |
1154 | die("delta base offset overflow in pack for %s", | |
aa7e44bf | 1155 | sha1_to_hex(entry->idx.sha1)); |
5c49c116 NP |
1156 | c = buf[used_0++]; |
1157 | ofs = (ofs << 7) + (c & 127); | |
780e6e73 | 1158 | } |
5c49c116 NP |
1159 | if (ofs >= entry->in_pack_offset) |
1160 | die("delta base offset out of bound for %s", | |
aa7e44bf | 1161 | sha1_to_hex(entry->idx.sha1)); |
5c49c116 NP |
1162 | ofs = entry->in_pack_offset - ofs; |
1163 | if (!no_reuse_delta && !entry->preferred_base) | |
1164 | base_ref = find_packed_object_name(p, ofs); | |
1165 | entry->in_pack_header_size = used + used_0; | |
1166 | break; | |
780e6e73 | 1167 | } |
780e6e73 | 1168 | |
5c49c116 NP |
1169 | if (base_ref && (base_entry = locate_object_entry(base_ref))) { |
1170 | /* | |
1171 | * If base_ref was set above that means we wish to | |
1172 | * reuse delta data, and we even found that base | |
1173 | * in the list of objects we want to pack. Goodie! | |
1174 | * | |
1175 | * Depth value does not matter - find_deltas() will | |
1176 | * never consider reused delta as the base object to | |
1177 | * deltify other objects against, in order to avoid | |
1178 | * circular deltas. | |
3f9ac8d2 | 1179 | */ |
780e6e73 | 1180 | entry->type = entry->in_pack_type; |
5c49c116 | 1181 | entry->delta = base_entry; |
15b4d577 JH |
1182 | entry->delta_sibling = base_entry->delta_child; |
1183 | base_entry->delta_child = entry; | |
5c49c116 NP |
1184 | unuse_pack(&w_curs); |
1185 | return; | |
1186 | } | |
ab7cd7bb | 1187 | |
5c49c116 NP |
1188 | if (entry->type) { |
1189 | /* | |
1190 | * This must be a delta and we already know what the | |
1191 | * final object type is. Let's extract the actual | |
1192 | * object size from the delta header. | |
1193 | */ | |
1194 | entry->size = get_size_from_delta(p, &w_curs, | |
1195 | entry->in_pack_offset + entry->in_pack_header_size); | |
1196 | unuse_pack(&w_curs); | |
3f9ac8d2 JH |
1197 | return; |
1198 | } | |
5c49c116 NP |
1199 | |
1200 | /* | |
1201 | * No choice but to fall back to the recursive delta walk | |
1202 | * with sha1_object_info() to find about the object type | |
1203 | * at this point... | |
1204 | */ | |
1205 | unuse_pack(&w_curs); | |
36e4d74a | 1206 | } |
3f9ac8d2 | 1207 | |
aa7e44bf | 1208 | entry->type = sha1_object_info(entry->idx.sha1, &entry->size); |
21666f1a | 1209 | if (entry->type < 0) |
36e4d74a | 1210 | die("unable to get type of object %s", |
aa7e44bf | 1211 | sha1_to_hex(entry->idx.sha1)); |
3f9ac8d2 JH |
1212 | } |
1213 | ||
5c49c116 NP |
1214 | static int pack_offset_sort(const void *_a, const void *_b) |
1215 | { | |
1216 | const struct object_entry *a = *(struct object_entry **)_a; | |
1217 | const struct object_entry *b = *(struct object_entry **)_b; | |
1218 | ||
1219 | /* avoid filesystem trashing with loose objects */ | |
1220 | if (!a->in_pack && !b->in_pack) | |
aa7e44bf | 1221 | return hashcmp(a->idx.sha1, b->idx.sha1); |
5c49c116 NP |
1222 | |
1223 | if (a->in_pack < b->in_pack) | |
1224 | return -1; | |
1225 | if (a->in_pack > b->in_pack) | |
1226 | return 1; | |
1227 | return a->in_pack_offset < b->in_pack_offset ? -1 : | |
1228 | (a->in_pack_offset > b->in_pack_offset); | |
1229 | } | |
1230 | ||
c323ac7d LT |
1231 | static void get_object_details(void) |
1232 | { | |
7cadf491 | 1233 | uint32_t i; |
5c49c116 NP |
1234 | struct object_entry **sorted_by_offset; |
1235 | ||
1236 | sorted_by_offset = xcalloc(nr_objects, sizeof(struct object_entry *)); | |
1237 | for (i = 0; i < nr_objects; i++) | |
1238 | sorted_by_offset[i] = objects + i; | |
1239 | qsort(sorted_by_offset, nr_objects, sizeof(*sorted_by_offset), pack_offset_sort); | |
c323ac7d | 1240 | |
3f9ac8d2 | 1241 | prepare_pack_ix(); |
5c49c116 NP |
1242 | for (i = 0; i < nr_objects; i++) |
1243 | check_object(sorted_by_offset[i]); | |
1244 | free(sorted_by_offset); | |
c323ac7d LT |
1245 | } |
1246 | ||
9668cf59 | 1247 | static int type_size_sort(const void *_a, const void *_b) |
c323ac7d | 1248 | { |
9668cf59 NP |
1249 | const struct object_entry *a = *(struct object_entry **)_a; |
1250 | const struct object_entry *b = *(struct object_entry **)_b; | |
1251 | ||
c323ac7d LT |
1252 | if (a->type < b->type) |
1253 | return -1; | |
1254 | if (a->type > b->type) | |
1255 | return 1; | |
27225f2e LT |
1256 | if (a->hash < b->hash) |
1257 | return -1; | |
1258 | if (a->hash > b->hash) | |
1259 | return 1; | |
7a979d99 JH |
1260 | if (a->preferred_base < b->preferred_base) |
1261 | return -1; | |
1262 | if (a->preferred_base > b->preferred_base) | |
1263 | return 1; | |
c323ac7d LT |
1264 | if (a->size < b->size) |
1265 | return -1; | |
1266 | if (a->size > b->size) | |
1267 | return 1; | |
adcc7095 | 1268 | return a > b ? -1 : (a < b); /* newest last */ |
c323ac7d LT |
1269 | } |
1270 | ||
1271 | struct unpacked { | |
1272 | struct object_entry *entry; | |
1273 | void *data; | |
f6c7081a | 1274 | struct delta_index *index; |
5a235b5e | 1275 | unsigned depth; |
c323ac7d LT |
1276 | }; |
1277 | ||
d250626c NP |
1278 | static int delta_cacheable(unsigned long src_size, unsigned long trg_size, |
1279 | unsigned long delta_size) | |
074b2eea MK |
1280 | { |
1281 | if (max_delta_cache_size && delta_cache_size + delta_size > max_delta_cache_size) | |
1282 | return 0; | |
1283 | ||
e3dfddb3 MK |
1284 | if (delta_size < cache_max_small_delta_size) |
1285 | return 1; | |
1286 | ||
074b2eea MK |
1287 | /* cache delta, if objects are large enough compared to delta size */ |
1288 | if ((src_size >> 20) + (trg_size >> 21) > (delta_size >> 10)) | |
1289 | return 1; | |
1290 | ||
1291 | return 0; | |
1292 | } | |
1293 | ||
c323ac7d | 1294 | /* |
521a4f4c LT |
1295 | * We search for deltas _backwards_ in a list sorted by type and |
1296 | * by size, so that we see progressively smaller and smaller files. | |
1297 | * That's because we prefer deltas to be from the bigger file | |
1298 | * to the smaller - deletes are potentially cheaper, but perhaps | |
1299 | * more importantly, the bigger file is likely the more recent | |
1300 | * one. | |
c323ac7d | 1301 | */ |
f6c7081a | 1302 | static int try_delta(struct unpacked *trg, struct unpacked *src, |
560b25a8 | 1303 | unsigned max_depth) |
c323ac7d | 1304 | { |
f6c7081a NP |
1305 | struct object_entry *trg_entry = trg->entry; |
1306 | struct object_entry *src_entry = src->entry; | |
560b25a8 | 1307 | unsigned long trg_size, src_size, delta_size, sizediff, max_size, sz; |
c83f032e | 1308 | unsigned ref_depth; |
21666f1a | 1309 | enum object_type type; |
c323ac7d LT |
1310 | void *delta_buf; |
1311 | ||
1312 | /* Don't bother doing diffs between different types */ | |
f6c7081a | 1313 | if (trg_entry->type != src_entry->type) |
c323ac7d LT |
1314 | return -1; |
1315 | ||
7a979d99 JH |
1316 | /* We do not compute delta to *create* objects we are not |
1317 | * going to pack. | |
1318 | */ | |
f6c7081a | 1319 | if (trg_entry->preferred_base) |
75c42d8c | 1320 | return -1; |
7a979d99 | 1321 | |
51d1e83f LT |
1322 | /* |
1323 | * We do not bother to try a delta that we discarded | |
8dbbd14e | 1324 | * on an earlier try, but only when reusing delta data. |
51d1e83f | 1325 | */ |
8dbbd14e | 1326 | if (!no_reuse_delta && trg_entry->in_pack && |
e9195b58 JH |
1327 | trg_entry->in_pack == src_entry->in_pack && |
1328 | trg_entry->in_pack_type != OBJ_REF_DELTA && | |
1329 | trg_entry->in_pack_type != OBJ_OFS_DELTA) | |
51d1e83f LT |
1330 | return 0; |
1331 | ||
898b14ce | 1332 | /* Let's not bust the allowed depth. */ |
5a235b5e | 1333 | if (src->depth >= max_depth) |
d116a45a | 1334 | return 0; |
c323ac7d | 1335 | |
c3b06a69 | 1336 | /* Now some size filtering heuristics. */ |
560b25a8 | 1337 | trg_size = trg_entry->size; |
c83f032e NP |
1338 | if (!trg_entry->delta) { |
1339 | max_size = trg_size/2 - 20; | |
1340 | ref_depth = 1; | |
1341 | } else { | |
1342 | max_size = trg_entry->delta_size; | |
5a235b5e | 1343 | ref_depth = trg->depth; |
c83f032e | 1344 | } |
5a235b5e | 1345 | max_size = max_size * (max_depth - src->depth) / |
c83f032e | 1346 | (max_depth - ref_depth + 1); |
c3b06a69 NP |
1347 | if (max_size == 0) |
1348 | return 0; | |
f6c7081a | 1349 | src_size = src_entry->size; |
560b25a8 | 1350 | sizediff = src_size < trg_size ? trg_size - src_size : 0; |
27225f2e | 1351 | if (sizediff >= max_size) |
f527cb8c | 1352 | return 0; |
a1dab41a BD |
1353 | if (trg_size < src_size / 32) |
1354 | return 0; | |
f6c7081a | 1355 | |
560b25a8 NP |
1356 | /* Load data if not already done */ |
1357 | if (!trg->data) { | |
aa7e44bf | 1358 | trg->data = read_sha1_file(trg_entry->idx.sha1, &type, &sz); |
2e3404c3 JH |
1359 | if (!trg->data) |
1360 | die("object %s cannot be read", | |
1361 | sha1_to_hex(trg_entry->idx.sha1)); | |
560b25a8 NP |
1362 | if (sz != trg_size) |
1363 | die("object %s inconsistent object length (%lu vs %lu)", | |
aa7e44bf | 1364 | sha1_to_hex(trg_entry->idx.sha1), sz, trg_size); |
a97773ce | 1365 | window_memory_usage += sz; |
560b25a8 NP |
1366 | } |
1367 | if (!src->data) { | |
aa7e44bf | 1368 | src->data = read_sha1_file(src_entry->idx.sha1, &type, &sz); |
2e3404c3 JH |
1369 | if (!src->data) |
1370 | die("object %s cannot be read", | |
1371 | sha1_to_hex(src_entry->idx.sha1)); | |
560b25a8 NP |
1372 | if (sz != src_size) |
1373 | die("object %s inconsistent object length (%lu vs %lu)", | |
aa7e44bf | 1374 | sha1_to_hex(src_entry->idx.sha1), sz, src_size); |
a97773ce | 1375 | window_memory_usage += sz; |
560b25a8 NP |
1376 | } |
1377 | if (!src->index) { | |
1378 | src->index = create_delta_index(src->data, src_size); | |
a588d88a MK |
1379 | if (!src->index) { |
1380 | static int warned = 0; | |
1381 | if (!warned++) | |
1382 | warning("suboptimal pack - out of memory"); | |
1383 | return 0; | |
1384 | } | |
a97773ce | 1385 | window_memory_usage += sizeof_delta_index(src->index); |
560b25a8 NP |
1386 | } |
1387 | ||
1388 | delta_buf = create_delta(src->index, trg->data, trg_size, &delta_size, max_size); | |
c323ac7d | 1389 | if (!delta_buf) |
75c42d8c | 1390 | return 0; |
f6c7081a | 1391 | |
074b2eea | 1392 | if (trg_entry->delta_data) { |
848d732c BD |
1393 | /* Prefer only shallower same-sized deltas. */ |
1394 | if (delta_size == trg_entry->delta_size && | |
5a235b5e | 1395 | src->depth + 1 >= trg->depth) { |
848d732c BD |
1396 | free(delta_buf); |
1397 | return 0; | |
1398 | } | |
074b2eea MK |
1399 | delta_cache_size -= trg_entry->delta_size; |
1400 | free(trg_entry->delta_data); | |
5a235b5e | 1401 | trg_entry->delta_data = NULL; |
074b2eea | 1402 | } |
f6c7081a NP |
1403 | trg_entry->delta = src_entry; |
1404 | trg_entry->delta_size = delta_size; | |
5a235b5e | 1405 | trg->depth = src->depth + 1; |
074b2eea | 1406 | |
d250626c | 1407 | if (delta_cacheable(src_size, trg_size, delta_size)) { |
074b2eea MK |
1408 | trg_entry->delta_data = xrealloc(delta_buf, delta_size); |
1409 | delta_cache_size += trg_entry->delta_size; | |
1410 | } else | |
1411 | free(delta_buf); | |
f6c7081a | 1412 | return 1; |
c323ac7d LT |
1413 | } |
1414 | ||
898b14ce | 1415 | static unsigned int check_delta_limit(struct object_entry *me, unsigned int n) |
b2504a0d | 1416 | { |
898b14ce NP |
1417 | struct object_entry *child = me->delta_child; |
1418 | unsigned int m = n; | |
1419 | while (child) { | |
1420 | unsigned int c = check_delta_limit(child, n + 1); | |
1421 | if (m < c) | |
1422 | m = c; | |
1423 | child = child->delta_sibling; | |
1424 | } | |
1425 | return m; | |
b2504a0d NP |
1426 | } |
1427 | ||
a97773ce BD |
1428 | static void free_unpacked(struct unpacked *n) |
1429 | { | |
1430 | window_memory_usage -= sizeof_delta_index(n->index); | |
1431 | free_delta_index(n->index); | |
1432 | n->index = NULL; | |
1433 | if (n->data) { | |
1434 | free(n->data); | |
1435 | n->data = NULL; | |
1436 | window_memory_usage -= n->entry->size; | |
1437 | } | |
1438 | n->entry = NULL; | |
7d7baa5e | 1439 | n->depth = 0; |
a97773ce BD |
1440 | } |
1441 | ||
d116a45a | 1442 | static void find_deltas(struct object_entry **list, int window, int depth) |
c323ac7d | 1443 | { |
a97773ce | 1444 | uint32_t i = nr_objects, idx = 0, count = 0, processed = 0; |
c323ac7d | 1445 | unsigned int array_size = window * sizeof(struct unpacked); |
7cadf491 | 1446 | struct unpacked *array; |
898b14ce | 1447 | int max_depth; |
c323ac7d | 1448 | |
7cadf491 SP |
1449 | if (!nr_objects) |
1450 | return; | |
1451 | array = xmalloc(array_size); | |
c323ac7d | 1452 | memset(array, 0, array_size); |
13aaf148 NP |
1453 | if (progress) |
1454 | start_progress(&progress_state, "Deltifying %u objects...", "", nr_result); | |
21fcd1bd | 1455 | |
7cadf491 SP |
1456 | do { |
1457 | struct object_entry *entry = list[--i]; | |
c323ac7d | 1458 | struct unpacked *n = array + idx; |
c323ac7d LT |
1459 | int j; |
1460 | ||
f0b0af1b JH |
1461 | if (!entry->preferred_base) |
1462 | processed++; | |
1463 | ||
96a02f8f NP |
1464 | if (progress) |
1465 | display_progress(&progress_state, processed); | |
3f9ac8d2 JH |
1466 | |
1467 | if (entry->delta) | |
1468 | /* This happens if we decided to reuse existing | |
ab7cd7bb | 1469 | * delta from a pack. "!no_reuse_delta &&" is implied. |
3f9ac8d2 JH |
1470 | */ |
1471 | continue; | |
1472 | ||
9a8b6a0a JH |
1473 | if (entry->size < 50) |
1474 | continue; | |
a74db82e JH |
1475 | |
1476 | if (entry->no_try_delta) | |
1477 | continue; | |
1478 | ||
a97773ce | 1479 | free_unpacked(n); |
c323ac7d | 1480 | n->entry = entry; |
ab7cd7bb | 1481 | |
a97773ce BD |
1482 | while (window_memory_limit && |
1483 | window_memory_usage > window_memory_limit && | |
1484 | count > 1) { | |
1485 | uint32_t tail = (idx + window - count) % window; | |
1486 | free_unpacked(array + tail); | |
1487 | count--; | |
1488 | } | |
1489 | ||
898b14ce NP |
1490 | /* |
1491 | * If the current object is at pack edge, take the depth the | |
1492 | * objects that depend on the current object into account | |
1493 | * otherwise they would become too deep. | |
1494 | */ | |
1495 | max_depth = depth; | |
1496 | if (entry->delta_child) { | |
1497 | max_depth -= check_delta_limit(entry, 0); | |
1498 | if (max_depth <= 0) | |
1499 | goto next; | |
1500 | } | |
1501 | ||
78817c15 LT |
1502 | j = window; |
1503 | while (--j > 0) { | |
7cadf491 | 1504 | uint32_t other_idx = idx + j; |
c323ac7d | 1505 | struct unpacked *m; |
78817c15 LT |
1506 | if (other_idx >= window) |
1507 | other_idx -= window; | |
c323ac7d LT |
1508 | m = array + other_idx; |
1509 | if (!m->entry) | |
1510 | break; | |
898b14ce | 1511 | if (try_delta(n, m, max_depth) < 0) |
c323ac7d LT |
1512 | break; |
1513 | } | |
898b14ce | 1514 | |
70ca1a3f JH |
1515 | /* if we made n a delta, and if n is already at max |
1516 | * depth, leaving it in the window is pointless. we | |
1517 | * should evict it first. | |
70ca1a3f | 1518 | */ |
5a235b5e | 1519 | if (entry->delta && depth <= n->depth) |
70ca1a3f | 1520 | continue; |
ff45715c | 1521 | |
898b14ce | 1522 | next: |
521a4f4c | 1523 | idx++; |
a97773ce BD |
1524 | if (count + 1 < window) |
1525 | count++; | |
521a4f4c LT |
1526 | if (idx >= window) |
1527 | idx = 0; | |
7cadf491 | 1528 | } while (i > 0); |
adee7bdf | 1529 | |
b2504a0d | 1530 | if (progress) |
96a02f8f | 1531 | stop_progress(&progress_state); |
b2504a0d | 1532 | |
f6c7081a | 1533 | for (i = 0; i < window; ++i) { |
ff45715c | 1534 | free_delta_index(array[i].index); |
adee7bdf | 1535 | free(array[i].data); |
f6c7081a | 1536 | } |
adee7bdf | 1537 | free(array); |
c323ac7d LT |
1538 | } |
1539 | ||
f3123c4a JH |
1540 | static void prepare_pack(int window, int depth) |
1541 | { | |
9668cf59 NP |
1542 | struct object_entry **delta_list; |
1543 | uint32_t i; | |
1544 | ||
3f9ac8d2 | 1545 | get_object_details(); |
9668cf59 NP |
1546 | |
1547 | if (!window || !depth) | |
1548 | return; | |
1549 | ||
1550 | delta_list = xmalloc(nr_objects * sizeof(*delta_list)); | |
1551 | for (i = 0; i < nr_objects; i++) | |
1552 | delta_list[i] = objects + i; | |
1553 | qsort(delta_list, nr_objects, sizeof(*delta_list), type_size_sort); | |
1554 | find_deltas(delta_list, window+1, depth); | |
1555 | free(delta_list); | |
f3123c4a JH |
1556 | } |
1557 | ||
4812a93a JK |
1558 | static int git_pack_config(const char *k, const char *v) |
1559 | { | |
1560 | if(!strcmp(k, "pack.window")) { | |
1561 | window = git_config_int(k, v); | |
1562 | return 0; | |
1563 | } | |
a97773ce BD |
1564 | if (!strcmp(k, "pack.windowmemory")) { |
1565 | window_memory_limit = git_config_ulong(k, v); | |
1566 | return 0; | |
1567 | } | |
1568 | if (!strcmp(k, "pack.depth")) { | |
842aaf93 TT |
1569 | depth = git_config_int(k, v); |
1570 | return 0; | |
1571 | } | |
960ccca6 DH |
1572 | if (!strcmp(k, "pack.compression")) { |
1573 | int level = git_config_int(k, v); | |
1574 | if (level == -1) | |
1575 | level = Z_DEFAULT_COMPRESSION; | |
1576 | else if (level < 0 || level > Z_BEST_COMPRESSION) | |
1577 | die("bad pack compression level %d", level); | |
1578 | pack_compression_level = level; | |
1579 | pack_compression_seen = 1; | |
1580 | return 0; | |
1581 | } | |
074b2eea MK |
1582 | if (!strcmp(k, "pack.deltacachesize")) { |
1583 | max_delta_cache_size = git_config_int(k, v); | |
1584 | return 0; | |
1585 | } | |
e3dfddb3 MK |
1586 | if (!strcmp(k, "pack.deltacachelimit")) { |
1587 | cache_max_small_delta_size = git_config_int(k, v); | |
1588 | return 0; | |
1589 | } | |
4812a93a JK |
1590 | return git_default_config(k, v); |
1591 | } | |
1592 | ||
b5d97e6b | 1593 | static void read_object_list_from_stdin(void) |
c323ac7d | 1594 | { |
b5d97e6b JH |
1595 | char line[40 + 1 + PATH_MAX + 2]; |
1596 | unsigned char sha1[20]; | |
b2504a0d | 1597 | |
da93d12b | 1598 | for (;;) { |
da93d12b LT |
1599 | if (!fgets(line, sizeof(line), stdin)) { |
1600 | if (feof(stdin)) | |
1601 | break; | |
1602 | if (!ferror(stdin)) | |
1603 | die("fgets returned NULL, not EOF, not error!"); | |
687dd75c JH |
1604 | if (errno != EINTR) |
1605 | die("fgets: %s", strerror(errno)); | |
1606 | clearerr(stdin); | |
1607 | continue; | |
da93d12b | 1608 | } |
7a979d99 JH |
1609 | if (line[0] == '-') { |
1610 | if (get_sha1_hex(line+1, sha1)) | |
1611 | die("expected edge sha1, got garbage:\n %s", | |
b5d97e6b | 1612 | line); |
8d1d8f83 | 1613 | add_preferred_base(sha1); |
7a979d99 | 1614 | continue; |
21fcd1bd | 1615 | } |
c323ac7d | 1616 | if (get_sha1_hex(line, sha1)) |
ef07618f | 1617 | die("expected sha1, got garbage:\n %s", line); |
b5d97e6b | 1618 | |
bc32fed5 JH |
1619 | add_preferred_base_object(line+41); |
1620 | add_object_entry(sha1, 0, line+41, 0); | |
c323ac7d | 1621 | } |
b5d97e6b JH |
1622 | } |
1623 | ||
b5d97e6b JH |
1624 | static void show_commit(struct commit *commit) |
1625 | { | |
bc32fed5 | 1626 | add_object_entry(commit->object.sha1, OBJ_COMMIT, NULL, 0); |
b5d97e6b JH |
1627 | } |
1628 | ||
1629 | static void show_object(struct object_array_entry *p) | |
1630 | { | |
bc32fed5 JH |
1631 | add_preferred_base_object(p->name); |
1632 | add_object_entry(p->item->sha1, p->item->type, p->name, 0); | |
b5d97e6b JH |
1633 | } |
1634 | ||
8d1d8f83 JH |
1635 | static void show_edge(struct commit *commit) |
1636 | { | |
1637 | add_preferred_base(commit->object.sha1); | |
1638 | } | |
1639 | ||
1640 | static void get_object_list(int ac, const char **av) | |
b5d97e6b JH |
1641 | { |
1642 | struct rev_info revs; | |
1643 | char line[1000]; | |
b5d97e6b JH |
1644 | int flags = 0; |
1645 | ||
b5d97e6b JH |
1646 | init_revisions(&revs, NULL); |
1647 | save_commit_buffer = 0; | |
1648 | track_object_refs = 0; | |
1649 | setup_revisions(ac, av, &revs, NULL); | |
1650 | ||
b5d97e6b JH |
1651 | while (fgets(line, sizeof(line), stdin) != NULL) { |
1652 | int len = strlen(line); | |
1653 | if (line[len - 1] == '\n') | |
1654 | line[--len] = 0; | |
1655 | if (!len) | |
1656 | break; | |
1657 | if (*line == '-') { | |
1658 | if (!strcmp(line, "--not")) { | |
1659 | flags ^= UNINTERESTING; | |
1660 | continue; | |
1661 | } | |
1662 | die("not a rev '%s'", line); | |
1663 | } | |
1664 | if (handle_revision_arg(line, &revs, flags, 1)) | |
1665 | die("bad revision '%s'", line); | |
1666 | } | |
1667 | ||
1668 | prepare_revision_walk(&revs); | |
8d1d8f83 | 1669 | mark_edges_uninteresting(revs.commits, &revs, show_edge); |
b5d97e6b JH |
1670 | traverse_commit_list(&revs, show_commit, show_object); |
1671 | } | |
1672 | ||
b6b32ccb JH |
1673 | static int adjust_perm(const char *path, mode_t mode) |
1674 | { | |
1675 | if (chmod(path, mode)) | |
1676 | return -1; | |
1677 | return adjust_shared_perm(path); | |
1678 | } | |
1679 | ||
b5d97e6b JH |
1680 | int cmd_pack_objects(int argc, const char **argv, const char *prefix) |
1681 | { | |
b5d97e6b | 1682 | int use_internal_rev_list = 0; |
8d1d8f83 | 1683 | int thin = 0; |
7cadf491 | 1684 | uint32_t i; |
ffa84ffb RD |
1685 | const char **rp_av; |
1686 | int rp_ac_alloc = 64; | |
8d1d8f83 JH |
1687 | int rp_ac; |
1688 | ||
ffa84ffb RD |
1689 | rp_av = xcalloc(rp_ac_alloc, sizeof(*rp_av)); |
1690 | ||
8d1d8f83 JH |
1691 | rp_av[0] = "pack-objects"; |
1692 | rp_av[1] = "--objects"; /* --thin will make it --objects-edge */ | |
1693 | rp_ac = 2; | |
b5d97e6b JH |
1694 | |
1695 | git_config(git_pack_config); | |
960ccca6 DH |
1696 | if (!pack_compression_seen && core_compression_seen) |
1697 | pack_compression_level = core_compression_level; | |
b5d97e6b JH |
1698 | |
1699 | progress = isatty(2); | |
1700 | for (i = 1; i < argc; i++) { | |
1701 | const char *arg = argv[i]; | |
1702 | ||
1703 | if (*arg != '-') | |
1704 | break; | |
1705 | ||
1706 | if (!strcmp("--non-empty", arg)) { | |
1707 | non_empty = 1; | |
1708 | continue; | |
1709 | } | |
1710 | if (!strcmp("--local", arg)) { | |
1711 | local = 1; | |
1712 | continue; | |
1713 | } | |
b5d97e6b JH |
1714 | if (!strcmp("--incremental", arg)) { |
1715 | incremental = 1; | |
1716 | continue; | |
1717 | } | |
960ccca6 DH |
1718 | if (!prefixcmp(arg, "--compression=")) { |
1719 | char *end; | |
1720 | int level = strtoul(arg+14, &end, 0); | |
1721 | if (!arg[14] || *end) | |
1722 | usage(pack_usage); | |
1723 | if (level == -1) | |
1724 | level = Z_DEFAULT_COMPRESSION; | |
1725 | else if (level < 0 || level > Z_BEST_COMPRESSION) | |
1726 | die("bad pack compression level %d", level); | |
1727 | pack_compression_level = level; | |
1728 | continue; | |
1729 | } | |
6b94b1a0 DH |
1730 | if (!prefixcmp(arg, "--max-pack-size=")) { |
1731 | char *end; | |
1732 | pack_size_limit = strtoul(arg+16, &end, 0) * 1024 * 1024; | |
1733 | if (!arg[16] || *end) | |
1734 | usage(pack_usage); | |
1735 | continue; | |
1736 | } | |
599065a3 | 1737 | if (!prefixcmp(arg, "--window=")) { |
b5d97e6b JH |
1738 | char *end; |
1739 | window = strtoul(arg+9, &end, 0); | |
1740 | if (!arg[9] || *end) | |
1741 | usage(pack_usage); | |
1742 | continue; | |
1743 | } | |
a97773ce BD |
1744 | if (!prefixcmp(arg, "--window-memory=")) { |
1745 | if (!git_parse_ulong(arg+16, &window_memory_limit)) | |
1746 | usage(pack_usage); | |
1747 | continue; | |
1748 | } | |
599065a3 | 1749 | if (!prefixcmp(arg, "--depth=")) { |
b5d97e6b JH |
1750 | char *end; |
1751 | depth = strtoul(arg+8, &end, 0); | |
1752 | if (!arg[8] || *end) | |
1753 | usage(pack_usage); | |
1754 | continue; | |
1755 | } | |
1756 | if (!strcmp("--progress", arg)) { | |
1757 | progress = 1; | |
1758 | continue; | |
1759 | } | |
231f240b NP |
1760 | if (!strcmp("--all-progress", arg)) { |
1761 | progress = 2; | |
1762 | continue; | |
1763 | } | |
b5d97e6b JH |
1764 | if (!strcmp("-q", arg)) { |
1765 | progress = 0; | |
1766 | continue; | |
1767 | } | |
1768 | if (!strcmp("--no-reuse-delta", arg)) { | |
1769 | no_reuse_delta = 1; | |
1770 | continue; | |
1771 | } | |
fa736f72 NP |
1772 | if (!strcmp("--no-reuse-object", arg)) { |
1773 | no_reuse_object = no_reuse_delta = 1; | |
1774 | continue; | |
1775 | } | |
be6b1914 | 1776 | if (!strcmp("--delta-base-offset", arg)) { |
780e6e73 | 1777 | allow_ofs_delta = 1; |
be6b1914 NP |
1778 | continue; |
1779 | } | |
b5d97e6b JH |
1780 | if (!strcmp("--stdout", arg)) { |
1781 | pack_to_stdout = 1; | |
1782 | continue; | |
1783 | } | |
1784 | if (!strcmp("--revs", arg)) { | |
1785 | use_internal_rev_list = 1; | |
1786 | continue; | |
1787 | } | |
8d1d8f83 | 1788 | if (!strcmp("--unpacked", arg) || |
599065a3 | 1789 | !prefixcmp(arg, "--unpacked=") || |
63049292 | 1790 | !strcmp("--reflog", arg) || |
8d1d8f83 JH |
1791 | !strcmp("--all", arg)) { |
1792 | use_internal_rev_list = 1; | |
ffa84ffb RD |
1793 | if (rp_ac >= rp_ac_alloc - 1) { |
1794 | rp_ac_alloc = alloc_nr(rp_ac_alloc); | |
1795 | rp_av = xrealloc(rp_av, | |
1796 | rp_ac_alloc * sizeof(*rp_av)); | |
1797 | } | |
8d1d8f83 | 1798 | rp_av[rp_ac++] = arg; |
b5d97e6b JH |
1799 | continue; |
1800 | } | |
8d1d8f83 JH |
1801 | if (!strcmp("--thin", arg)) { |
1802 | use_internal_rev_list = 1; | |
1803 | thin = 1; | |
1804 | rp_av[1] = "--objects-edge"; | |
b5d97e6b | 1805 | continue; |
4ba7d711 NP |
1806 | } |
1807 | if (!prefixcmp(arg, "--index-version=")) { | |
1808 | char *c; | |
aa7e44bf GB |
1809 | pack_idx_default_version = strtoul(arg + 16, &c, 10); |
1810 | if (pack_idx_default_version > 2) | |
4ba7d711 NP |
1811 | die("bad %s", arg); |
1812 | if (*c == ',') | |
aa7e44bf GB |
1813 | pack_idx_off32_limit = strtoul(c+1, &c, 0); |
1814 | if (*c || pack_idx_off32_limit & 0x80000000) | |
4ba7d711 NP |
1815 | die("bad %s", arg); |
1816 | continue; | |
b5d97e6b JH |
1817 | } |
1818 | usage(pack_usage); | |
1819 | } | |
1820 | ||
1821 | /* Traditionally "pack-objects [options] base extra" failed; | |
1822 | * we would however want to take refs parameter that would | |
1823 | * have been given to upstream rev-list ourselves, which means | |
1824 | * we somehow want to say what the base name is. So the | |
1825 | * syntax would be: | |
1826 | * | |
1827 | * pack-objects [options] base <refs...> | |
1828 | * | |
1829 | * in other words, we would treat the first non-option as the | |
1830 | * base_name and send everything else to the internal revision | |
1831 | * walker. | |
1832 | */ | |
1833 | ||
1834 | if (!pack_to_stdout) | |
1835 | base_name = argv[i++]; | |
1836 | ||
1837 | if (pack_to_stdout != !base_name) | |
1838 | usage(pack_usage); | |
1839 | ||
01c12a23 DH |
1840 | if (pack_to_stdout && pack_size_limit) |
1841 | die("--max-pack-size cannot be used to build a pack for transfer."); | |
1842 | ||
8d1d8f83 JH |
1843 | if (!pack_to_stdout && thin) |
1844 | die("--thin cannot be used to build an indexable pack."); | |
b5d97e6b JH |
1845 | |
1846 | prepare_packed_git(); | |
1847 | ||
13aaf148 NP |
1848 | if (progress) |
1849 | start_progress(&progress_state, "Generating pack...", | |
1850 | "Counting objects: ", 0); | |
b5d97e6b JH |
1851 | if (!use_internal_rev_list) |
1852 | read_object_list_from_stdin(); | |
8d1d8f83 JH |
1853 | else { |
1854 | rp_av[rp_ac] = NULL; | |
1855 | get_object_list(rp_ac, rp_av); | |
1856 | } | |
96a02f8f NP |
1857 | if (progress) { |
1858 | stop_progress(&progress_state); | |
7cadf491 | 1859 | fprintf(stderr, "Done counting %u objects.\n", nr_objects); |
96a02f8f NP |
1860 | } |
1861 | ||
f0b0af1b | 1862 | if (non_empty && !nr_result) |
1c4a2912 | 1863 | return 0; |
7a979d99 | 1864 | if (progress && (nr_objects != nr_result)) |
7cadf491 | 1865 | fprintf(stderr, "Result has %u objects.\n", nr_result); |
f7ae6a93 NP |
1866 | if (nr_result) |
1867 | prepare_pack(window, depth); | |
d01fb92f | 1868 | write_pack_file(); |
ab7cd7bb | 1869 | if (progress) |
7cadf491 | 1870 | fprintf(stderr, "Total %u (delta %u), reused %u (delta %u)\n", |
67c08ce1 | 1871 | written, written_delta, reused, reused_delta); |
c323ac7d LT |
1872 | return 0; |
1873 | } |