Merge branch 'sb/z-is-gnutar-ism'
[git] / pack-bitmap-write.c
1 #include "cache.h"
2 #include "commit.h"
3 #include "tag.h"
4 #include "diff.h"
5 #include "revision.h"
6 #include "list-objects.h"
7 #include "progress.h"
8 #include "pack-revindex.h"
9 #include "pack.h"
10 #include "pack-bitmap.h"
11 #include "sha1-lookup.h"
12 #include "pack-objects.h"
13
14 struct bitmapped_commit {
15         struct commit *commit;
16         struct ewah_bitmap *bitmap;
17         struct ewah_bitmap *write_as;
18         int flags;
19         int xor_offset;
20         uint32_t commit_pos;
21 };
22
23 struct bitmap_writer {
24         struct ewah_bitmap *commits;
25         struct ewah_bitmap *trees;
26         struct ewah_bitmap *blobs;
27         struct ewah_bitmap *tags;
28
29         khash_sha1 *bitmaps;
30         khash_sha1 *reused;
31         struct packing_data *to_pack;
32
33         struct bitmapped_commit *selected;
34         unsigned int selected_nr, selected_alloc;
35
36         struct progress *progress;
37         int show_progress;
38         unsigned char pack_checksum[20];
39 };
40
41 static struct bitmap_writer writer;
42
43 void bitmap_writer_show_progress(int show)
44 {
45         writer.show_progress = show;
46 }
47
48 /**
49  * Build the initial type index for the packfile
50  */
51 void bitmap_writer_build_type_index(struct pack_idx_entry **index,
52                                     uint32_t index_nr)
53 {
54         uint32_t i;
55
56         writer.commits = ewah_new();
57         writer.trees = ewah_new();
58         writer.blobs = ewah_new();
59         writer.tags = ewah_new();
60
61         for (i = 0; i < index_nr; ++i) {
62                 struct object_entry *entry = (struct object_entry *)index[i];
63                 enum object_type real_type;
64
65                 entry->in_pack_pos = i;
66
67                 switch (entry->type) {
68                 case OBJ_COMMIT:
69                 case OBJ_TREE:
70                 case OBJ_BLOB:
71                 case OBJ_TAG:
72                         real_type = entry->type;
73                         break;
74
75                 default:
76                         real_type = sha1_object_info(entry->idx.sha1, NULL);
77                         break;
78                 }
79
80                 switch (real_type) {
81                 case OBJ_COMMIT:
82                         ewah_set(writer.commits, i);
83                         break;
84
85                 case OBJ_TREE:
86                         ewah_set(writer.trees, i);
87                         break;
88
89                 case OBJ_BLOB:
90                         ewah_set(writer.blobs, i);
91                         break;
92
93                 case OBJ_TAG:
94                         ewah_set(writer.tags, i);
95                         break;
96
97                 default:
98                         die("Missing type information for %s (%d/%d)",
99                             sha1_to_hex(entry->idx.sha1), real_type, entry->type);
100                 }
101         }
102 }
103
104 /**
105  * Compute the actual bitmaps
106  */
107 static struct object **seen_objects;
108 static unsigned int seen_objects_nr, seen_objects_alloc;
109
110 static inline void push_bitmapped_commit(struct commit *commit, struct ewah_bitmap *reused)
111 {
112         if (writer.selected_nr >= writer.selected_alloc) {
113                 writer.selected_alloc = (writer.selected_alloc + 32) * 2;
114                 REALLOC_ARRAY(writer.selected, writer.selected_alloc);
115         }
116
117         writer.selected[writer.selected_nr].commit = commit;
118         writer.selected[writer.selected_nr].bitmap = reused;
119         writer.selected[writer.selected_nr].flags = 0;
120
121         writer.selected_nr++;
122 }
123
124 static inline void mark_as_seen(struct object *object)
125 {
126         ALLOC_GROW(seen_objects, seen_objects_nr + 1, seen_objects_alloc);
127         seen_objects[seen_objects_nr++] = object;
128 }
129
130 static inline void reset_all_seen(void)
131 {
132         unsigned int i;
133         for (i = 0; i < seen_objects_nr; ++i) {
134                 seen_objects[i]->flags &= ~(SEEN | ADDED | SHOWN);
135         }
136         seen_objects_nr = 0;
137 }
138
139 static uint32_t find_object_pos(const unsigned char *sha1)
140 {
141         struct object_entry *entry = packlist_find(writer.to_pack, sha1, NULL);
142
143         if (!entry) {
144                 die("Failed to write bitmap index. Packfile doesn't have full closure "
145                         "(object %s is missing)", sha1_to_hex(sha1));
146         }
147
148         return entry->in_pack_pos;
149 }
150
151 static void show_object(struct object *object, const char *name, void *data)
152 {
153         struct bitmap *base = data;
154         bitmap_set(base, find_object_pos(object->oid.hash));
155         mark_as_seen(object);
156 }
157
158 static void show_commit(struct commit *commit, void *data)
159 {
160         mark_as_seen((struct object *)commit);
161 }
162
163 static int
164 add_to_include_set(struct bitmap *base, struct commit *commit)
165 {
166         khiter_t hash_pos;
167         uint32_t bitmap_pos = find_object_pos(commit->object.oid.hash);
168
169         if (bitmap_get(base, bitmap_pos))
170                 return 0;
171
172         hash_pos = kh_get_sha1(writer.bitmaps, commit->object.oid.hash);
173         if (hash_pos < kh_end(writer.bitmaps)) {
174                 struct bitmapped_commit *bc = kh_value(writer.bitmaps, hash_pos);
175                 bitmap_or_ewah(base, bc->bitmap);
176                 return 0;
177         }
178
179         bitmap_set(base, bitmap_pos);
180         return 1;
181 }
182
183 static int
184 should_include(struct commit *commit, void *_data)
185 {
186         struct bitmap *base = _data;
187
188         if (!add_to_include_set(base, commit)) {
189                 struct commit_list *parent = commit->parents;
190
191                 mark_as_seen((struct object *)commit);
192
193                 while (parent) {
194                         parent->item->object.flags |= SEEN;
195                         mark_as_seen((struct object *)parent->item);
196                         parent = parent->next;
197                 }
198
199                 return 0;
200         }
201
202         return 1;
203 }
204
205 static void compute_xor_offsets(void)
206 {
207         static const int MAX_XOR_OFFSET_SEARCH = 10;
208
209         int i, next = 0;
210
211         while (next < writer.selected_nr) {
212                 struct bitmapped_commit *stored = &writer.selected[next];
213
214                 int best_offset = 0;
215                 struct ewah_bitmap *best_bitmap = stored->bitmap;
216                 struct ewah_bitmap *test_xor;
217
218                 for (i = 1; i <= MAX_XOR_OFFSET_SEARCH; ++i) {
219                         int curr = next - i;
220
221                         if (curr < 0)
222                                 break;
223
224                         test_xor = ewah_pool_new();
225                         ewah_xor(writer.selected[curr].bitmap, stored->bitmap, test_xor);
226
227                         if (test_xor->buffer_size < best_bitmap->buffer_size) {
228                                 if (best_bitmap != stored->bitmap)
229                                         ewah_pool_free(best_bitmap);
230
231                                 best_bitmap = test_xor;
232                                 best_offset = i;
233                         } else {
234                                 ewah_pool_free(test_xor);
235                         }
236                 }
237
238                 stored->xor_offset = best_offset;
239                 stored->write_as = best_bitmap;
240
241                 next++;
242         }
243 }
244
245 void bitmap_writer_build(struct packing_data *to_pack)
246 {
247         static const double REUSE_BITMAP_THRESHOLD = 0.2;
248
249         int i, reuse_after, need_reset;
250         struct bitmap *base = bitmap_new();
251         struct rev_info revs;
252
253         writer.bitmaps = kh_init_sha1();
254         writer.to_pack = to_pack;
255
256         if (writer.show_progress)
257                 writer.progress = start_progress("Building bitmaps", writer.selected_nr);
258
259         init_revisions(&revs, NULL);
260         revs.tag_objects = 1;
261         revs.tree_objects = 1;
262         revs.blob_objects = 1;
263         revs.no_walk = 0;
264
265         revs.include_check = should_include;
266         reset_revision_walk();
267
268         reuse_after = writer.selected_nr * REUSE_BITMAP_THRESHOLD;
269         need_reset = 0;
270
271         for (i = writer.selected_nr - 1; i >= 0; --i) {
272                 struct bitmapped_commit *stored;
273                 struct object *object;
274
275                 khiter_t hash_pos;
276                 int hash_ret;
277
278                 stored = &writer.selected[i];
279                 object = (struct object *)stored->commit;
280
281                 if (stored->bitmap == NULL) {
282                         if (i < writer.selected_nr - 1 &&
283                             (need_reset ||
284                              !in_merge_bases(writer.selected[i + 1].commit,
285                                              stored->commit))) {
286                             bitmap_reset(base);
287                             reset_all_seen();
288                         }
289
290                         add_pending_object(&revs, object, "");
291                         revs.include_check_data = base;
292
293                         if (prepare_revision_walk(&revs))
294                                 die("revision walk setup failed");
295
296                         traverse_commit_list(&revs, show_commit, show_object, base);
297
298                         revs.pending.nr = 0;
299                         revs.pending.alloc = 0;
300                         revs.pending.objects = NULL;
301
302                         stored->bitmap = bitmap_to_ewah(base);
303                         need_reset = 0;
304                 } else
305                         need_reset = 1;
306
307                 if (i >= reuse_after)
308                         stored->flags |= BITMAP_FLAG_REUSE;
309
310                 hash_pos = kh_put_sha1(writer.bitmaps, object->oid.hash, &hash_ret);
311                 if (hash_ret == 0)
312                         die("Duplicate entry when writing index: %s",
313                             oid_to_hex(&object->oid));
314
315                 kh_value(writer.bitmaps, hash_pos) = stored;
316                 display_progress(writer.progress, writer.selected_nr - i);
317         }
318
319         bitmap_free(base);
320         stop_progress(&writer.progress);
321
322         compute_xor_offsets();
323 }
324
325 /**
326  * Select the commits that will be bitmapped
327  */
328 static inline unsigned int next_commit_index(unsigned int idx)
329 {
330         static const unsigned int MIN_COMMITS = 100;
331         static const unsigned int MAX_COMMITS = 5000;
332
333         static const unsigned int MUST_REGION = 100;
334         static const unsigned int MIN_REGION = 20000;
335
336         unsigned int offset, next;
337
338         if (idx <= MUST_REGION)
339                 return 0;
340
341         if (idx <= MIN_REGION) {
342                 offset = idx - MUST_REGION;
343                 return (offset < MIN_COMMITS) ? offset : MIN_COMMITS;
344         }
345
346         offset = idx - MIN_REGION;
347         next = (offset < MAX_COMMITS) ? offset : MAX_COMMITS;
348
349         return (next > MIN_COMMITS) ? next : MIN_COMMITS;
350 }
351
352 static int date_compare(const void *_a, const void *_b)
353 {
354         struct commit *a = *(struct commit **)_a;
355         struct commit *b = *(struct commit **)_b;
356         return (long)b->date - (long)a->date;
357 }
358
359 void bitmap_writer_reuse_bitmaps(struct packing_data *to_pack)
360 {
361         if (prepare_bitmap_git() < 0)
362                 return;
363
364         writer.reused = kh_init_sha1();
365         rebuild_existing_bitmaps(to_pack, writer.reused, writer.show_progress);
366 }
367
368 static struct ewah_bitmap *find_reused_bitmap(const unsigned char *sha1)
369 {
370         khiter_t hash_pos;
371
372         if (!writer.reused)
373                 return NULL;
374
375         hash_pos = kh_get_sha1(writer.reused, sha1);
376         if (hash_pos >= kh_end(writer.reused))
377                 return NULL;
378
379         return kh_value(writer.reused, hash_pos);
380 }
381
382 void bitmap_writer_select_commits(struct commit **indexed_commits,
383                                   unsigned int indexed_commits_nr,
384                                   int max_bitmaps)
385 {
386         unsigned int i = 0, j, next;
387
388         qsort(indexed_commits, indexed_commits_nr, sizeof(indexed_commits[0]),
389               date_compare);
390
391         if (writer.show_progress)
392                 writer.progress = start_progress("Selecting bitmap commits", 0);
393
394         if (indexed_commits_nr < 100) {
395                 for (i = 0; i < indexed_commits_nr; ++i)
396                         push_bitmapped_commit(indexed_commits[i], NULL);
397                 return;
398         }
399
400         for (;;) {
401                 struct ewah_bitmap *reused_bitmap = NULL;
402                 struct commit *chosen = NULL;
403
404                 next = next_commit_index(i);
405
406                 if (i + next >= indexed_commits_nr)
407                         break;
408
409                 if (max_bitmaps > 0 && writer.selected_nr >= max_bitmaps) {
410                         writer.selected_nr = max_bitmaps;
411                         break;
412                 }
413
414                 if (next == 0) {
415                         chosen = indexed_commits[i];
416                         reused_bitmap = find_reused_bitmap(chosen->object.oid.hash);
417                 } else {
418                         chosen = indexed_commits[i + next];
419
420                         for (j = 0; j <= next; ++j) {
421                                 struct commit *cm = indexed_commits[i + j];
422
423                                 reused_bitmap = find_reused_bitmap(cm->object.oid.hash);
424                                 if (reused_bitmap || (cm->object.flags & NEEDS_BITMAP) != 0) {
425                                         chosen = cm;
426                                         break;
427                                 }
428
429                                 if (cm->parents && cm->parents->next)
430                                         chosen = cm;
431                         }
432                 }
433
434                 push_bitmapped_commit(chosen, reused_bitmap);
435
436                 i += next + 1;
437                 display_progress(writer.progress, i);
438         }
439
440         stop_progress(&writer.progress);
441 }
442
443
444 static int sha1write_ewah_helper(void *f, const void *buf, size_t len)
445 {
446         /* sha1write will die on error */
447         sha1write(f, buf, len);
448         return len;
449 }
450
451 /**
452  * Write the bitmap index to disk
453  */
454 static inline void dump_bitmap(struct sha1file *f, struct ewah_bitmap *bitmap)
455 {
456         if (ewah_serialize_to(bitmap, sha1write_ewah_helper, f) < 0)
457                 die("Failed to write bitmap index");
458 }
459
460 static const unsigned char *sha1_access(size_t pos, void *table)
461 {
462         struct pack_idx_entry **index = table;
463         return index[pos]->sha1;
464 }
465
466 static void write_selected_commits_v1(struct sha1file *f,
467                                       struct pack_idx_entry **index,
468                                       uint32_t index_nr)
469 {
470         int i;
471
472         for (i = 0; i < writer.selected_nr; ++i) {
473                 struct bitmapped_commit *stored = &writer.selected[i];
474
475                 int commit_pos =
476                         sha1_pos(stored->commit->object.oid.hash, index, index_nr, sha1_access);
477
478                 if (commit_pos < 0)
479                         die("BUG: trying to write commit not in index");
480
481                 sha1write_be32(f, commit_pos);
482                 sha1write_u8(f, stored->xor_offset);
483                 sha1write_u8(f, stored->flags);
484
485                 dump_bitmap(f, stored->write_as);
486         }
487 }
488
489 static void write_hash_cache(struct sha1file *f,
490                              struct pack_idx_entry **index,
491                              uint32_t index_nr)
492 {
493         uint32_t i;
494
495         for (i = 0; i < index_nr; ++i) {
496                 struct object_entry *entry = (struct object_entry *)index[i];
497                 uint32_t hash_value = htonl(entry->hash);
498                 sha1write(f, &hash_value, sizeof(hash_value));
499         }
500 }
501
502 void bitmap_writer_set_checksum(unsigned char *sha1)
503 {
504         hashcpy(writer.pack_checksum, sha1);
505 }
506
507 void bitmap_writer_finish(struct pack_idx_entry **index,
508                           uint32_t index_nr,
509                           const char *filename,
510                           uint16_t options)
511 {
512         static char tmp_file[PATH_MAX];
513         static uint16_t default_version = 1;
514         static uint16_t flags = BITMAP_OPT_FULL_DAG;
515         struct sha1file *f;
516
517         struct bitmap_disk_header header;
518
519         int fd = odb_mkstemp(tmp_file, sizeof(tmp_file), "pack/tmp_bitmap_XXXXXX");
520
521         if (fd < 0)
522                 die_errno("unable to create '%s'", tmp_file);
523         f = sha1fd(fd, tmp_file);
524
525         memcpy(header.magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE));
526         header.version = htons(default_version);
527         header.options = htons(flags | options);
528         header.entry_count = htonl(writer.selected_nr);
529         hashcpy(header.checksum, writer.pack_checksum);
530
531         sha1write(f, &header, sizeof(header));
532         dump_bitmap(f, writer.commits);
533         dump_bitmap(f, writer.trees);
534         dump_bitmap(f, writer.blobs);
535         dump_bitmap(f, writer.tags);
536         write_selected_commits_v1(f, index, index_nr);
537
538         if (options & BITMAP_OPT_HASH_CACHE)
539                 write_hash_cache(f, index, index_nr);
540
541         sha1close(f, NULL, CSUM_FSYNC);
542
543         if (adjust_shared_perm(tmp_file))
544                 die_errno("unable to make temporary bitmap file readable");
545
546         if (rename(tmp_file, filename))
547                 die_errno("unable to rename temporary bitmap file to '%s'", filename);
548 }