Merge branch 'ma/create-pseudoref-with-null-old-oid'
[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 packing_data *to_pack,
52                                     struct pack_idx_entry **index,
53                                     uint32_t index_nr)
54 {
55         uint32_t i;
56
57         writer.commits = ewah_new();
58         writer.trees = ewah_new();
59         writer.blobs = ewah_new();
60         writer.tags = ewah_new();
61         ALLOC_ARRAY(to_pack->in_pack_pos, to_pack->nr_objects);
62
63         for (i = 0; i < index_nr; ++i) {
64                 struct object_entry *entry = (struct object_entry *)index[i];
65                 enum object_type real_type;
66
67                 oe_set_in_pack_pos(to_pack, entry, i);
68
69                 switch (oe_type(entry)) {
70                 case OBJ_COMMIT:
71                 case OBJ_TREE:
72                 case OBJ_BLOB:
73                 case OBJ_TAG:
74                         real_type = oe_type(entry);
75                         break;
76
77                 default:
78                         real_type = oid_object_info(the_repository,
79                                                     &entry->idx.oid, NULL);
80                         break;
81                 }
82
83                 switch (real_type) {
84                 case OBJ_COMMIT:
85                         ewah_set(writer.commits, i);
86                         break;
87
88                 case OBJ_TREE:
89                         ewah_set(writer.trees, i);
90                         break;
91
92                 case OBJ_BLOB:
93                         ewah_set(writer.blobs, i);
94                         break;
95
96                 case OBJ_TAG:
97                         ewah_set(writer.tags, i);
98                         break;
99
100                 default:
101                         die("Missing type information for %s (%d/%d)",
102                             oid_to_hex(&entry->idx.oid), real_type,
103                             oe_type(entry));
104                 }
105         }
106 }
107
108 /**
109  * Compute the actual bitmaps
110  */
111 static struct object **seen_objects;
112 static unsigned int seen_objects_nr, seen_objects_alloc;
113
114 static inline void push_bitmapped_commit(struct commit *commit, struct ewah_bitmap *reused)
115 {
116         if (writer.selected_nr >= writer.selected_alloc) {
117                 writer.selected_alloc = (writer.selected_alloc + 32) * 2;
118                 REALLOC_ARRAY(writer.selected, writer.selected_alloc);
119         }
120
121         writer.selected[writer.selected_nr].commit = commit;
122         writer.selected[writer.selected_nr].bitmap = reused;
123         writer.selected[writer.selected_nr].flags = 0;
124
125         writer.selected_nr++;
126 }
127
128 static inline void mark_as_seen(struct object *object)
129 {
130         ALLOC_GROW(seen_objects, seen_objects_nr + 1, seen_objects_alloc);
131         seen_objects[seen_objects_nr++] = object;
132 }
133
134 static inline void reset_all_seen(void)
135 {
136         unsigned int i;
137         for (i = 0; i < seen_objects_nr; ++i) {
138                 seen_objects[i]->flags &= ~(SEEN | ADDED | SHOWN);
139         }
140         seen_objects_nr = 0;
141 }
142
143 static uint32_t find_object_pos(const unsigned char *sha1)
144 {
145         struct object_entry *entry = packlist_find(writer.to_pack, sha1, NULL);
146
147         if (!entry) {
148                 die("Failed to write bitmap index. Packfile doesn't have full closure "
149                         "(object %s is missing)", sha1_to_hex(sha1));
150         }
151
152         return oe_in_pack_pos(writer.to_pack, entry);
153 }
154
155 static void show_object(struct object *object, const char *name, void *data)
156 {
157         struct bitmap *base = data;
158         bitmap_set(base, find_object_pos(object->oid.hash));
159         mark_as_seen(object);
160 }
161
162 static void show_commit(struct commit *commit, void *data)
163 {
164         mark_as_seen((struct object *)commit);
165 }
166
167 static int
168 add_to_include_set(struct bitmap *base, struct commit *commit)
169 {
170         khiter_t hash_pos;
171         uint32_t bitmap_pos = find_object_pos(commit->object.oid.hash);
172
173         if (bitmap_get(base, bitmap_pos))
174                 return 0;
175
176         hash_pos = kh_get_sha1(writer.bitmaps, commit->object.oid.hash);
177         if (hash_pos < kh_end(writer.bitmaps)) {
178                 struct bitmapped_commit *bc = kh_value(writer.bitmaps, hash_pos);
179                 bitmap_or_ewah(base, bc->bitmap);
180                 return 0;
181         }
182
183         bitmap_set(base, bitmap_pos);
184         return 1;
185 }
186
187 static int
188 should_include(struct commit *commit, void *_data)
189 {
190         struct bitmap *base = _data;
191
192         if (!add_to_include_set(base, commit)) {
193                 struct commit_list *parent = commit->parents;
194
195                 mark_as_seen((struct object *)commit);
196
197                 while (parent) {
198                         parent->item->object.flags |= SEEN;
199                         mark_as_seen((struct object *)parent->item);
200                         parent = parent->next;
201                 }
202
203                 return 0;
204         }
205
206         return 1;
207 }
208
209 static void compute_xor_offsets(void)
210 {
211         static const int MAX_XOR_OFFSET_SEARCH = 10;
212
213         int i, next = 0;
214
215         while (next < writer.selected_nr) {
216                 struct bitmapped_commit *stored = &writer.selected[next];
217
218                 int best_offset = 0;
219                 struct ewah_bitmap *best_bitmap = stored->bitmap;
220                 struct ewah_bitmap *test_xor;
221
222                 for (i = 1; i <= MAX_XOR_OFFSET_SEARCH; ++i) {
223                         int curr = next - i;
224
225                         if (curr < 0)
226                                 break;
227
228                         test_xor = ewah_pool_new();
229                         ewah_xor(writer.selected[curr].bitmap, stored->bitmap, test_xor);
230
231                         if (test_xor->buffer_size < best_bitmap->buffer_size) {
232                                 if (best_bitmap != stored->bitmap)
233                                         ewah_pool_free(best_bitmap);
234
235                                 best_bitmap = test_xor;
236                                 best_offset = i;
237                         } else {
238                                 ewah_pool_free(test_xor);
239                         }
240                 }
241
242                 stored->xor_offset = best_offset;
243                 stored->write_as = best_bitmap;
244
245                 next++;
246         }
247 }
248
249 void bitmap_writer_build(struct packing_data *to_pack)
250 {
251         static const double REUSE_BITMAP_THRESHOLD = 0.2;
252
253         int i, reuse_after, need_reset;
254         struct bitmap *base = bitmap_new();
255         struct rev_info revs;
256
257         writer.bitmaps = kh_init_sha1();
258         writer.to_pack = to_pack;
259
260         if (writer.show_progress)
261                 writer.progress = start_progress("Building bitmaps", writer.selected_nr);
262
263         init_revisions(&revs, NULL);
264         revs.tag_objects = 1;
265         revs.tree_objects = 1;
266         revs.blob_objects = 1;
267         revs.no_walk = 0;
268
269         revs.include_check = should_include;
270         reset_revision_walk();
271
272         reuse_after = writer.selected_nr * REUSE_BITMAP_THRESHOLD;
273         need_reset = 0;
274
275         for (i = writer.selected_nr - 1; i >= 0; --i) {
276                 struct bitmapped_commit *stored;
277                 struct object *object;
278
279                 khiter_t hash_pos;
280                 int hash_ret;
281
282                 stored = &writer.selected[i];
283                 object = (struct object *)stored->commit;
284
285                 if (stored->bitmap == NULL) {
286                         if (i < writer.selected_nr - 1 &&
287                             (need_reset ||
288                              !in_merge_bases(writer.selected[i + 1].commit,
289                                              stored->commit))) {
290                             bitmap_reset(base);
291                             reset_all_seen();
292                         }
293
294                         add_pending_object(&revs, object, "");
295                         revs.include_check_data = base;
296
297                         if (prepare_revision_walk(&revs))
298                                 die("revision walk setup failed");
299
300                         traverse_commit_list(&revs, show_commit, show_object, base);
301
302                         object_array_clear(&revs.pending);
303
304                         stored->bitmap = bitmap_to_ewah(base);
305                         need_reset = 0;
306                 } else
307                         need_reset = 1;
308
309                 if (i >= reuse_after)
310                         stored->flags |= BITMAP_FLAG_REUSE;
311
312                 hash_pos = kh_put_sha1(writer.bitmaps, object->oid.hash, &hash_ret);
313                 if (hash_ret == 0)
314                         die("Duplicate entry when writing index: %s",
315                             oid_to_hex(&object->oid));
316
317                 kh_value(writer.bitmaps, hash_pos) = stored;
318                 display_progress(writer.progress, writer.selected_nr - i);
319         }
320
321         bitmap_free(base);
322         stop_progress(&writer.progress);
323
324         compute_xor_offsets();
325 }
326
327 /**
328  * Select the commits that will be bitmapped
329  */
330 static inline unsigned int next_commit_index(unsigned int idx)
331 {
332         static const unsigned int MIN_COMMITS = 100;
333         static const unsigned int MAX_COMMITS = 5000;
334
335         static const unsigned int MUST_REGION = 100;
336         static const unsigned int MIN_REGION = 20000;
337
338         unsigned int offset, next;
339
340         if (idx <= MUST_REGION)
341                 return 0;
342
343         if (idx <= MIN_REGION) {
344                 offset = idx - MUST_REGION;
345                 return (offset < MIN_COMMITS) ? offset : MIN_COMMITS;
346         }
347
348         offset = idx - MIN_REGION;
349         next = (offset < MAX_COMMITS) ? offset : MAX_COMMITS;
350
351         return (next > MIN_COMMITS) ? next : MIN_COMMITS;
352 }
353
354 static int date_compare(const void *_a, const void *_b)
355 {
356         struct commit *a = *(struct commit **)_a;
357         struct commit *b = *(struct commit **)_b;
358         return (long)b->date - (long)a->date;
359 }
360
361 void bitmap_writer_reuse_bitmaps(struct packing_data *to_pack)
362 {
363         if (prepare_bitmap_git() < 0)
364                 return;
365
366         writer.reused = kh_init_sha1();
367         rebuild_existing_bitmaps(to_pack, writer.reused, writer.show_progress);
368 }
369
370 static struct ewah_bitmap *find_reused_bitmap(const unsigned char *sha1)
371 {
372         khiter_t hash_pos;
373
374         if (!writer.reused)
375                 return NULL;
376
377         hash_pos = kh_get_sha1(writer.reused, sha1);
378         if (hash_pos >= kh_end(writer.reused))
379                 return NULL;
380
381         return kh_value(writer.reused, hash_pos);
382 }
383
384 void bitmap_writer_select_commits(struct commit **indexed_commits,
385                                   unsigned int indexed_commits_nr,
386                                   int max_bitmaps)
387 {
388         unsigned int i = 0, j, next;
389
390         QSORT(indexed_commits, indexed_commits_nr, date_compare);
391
392         if (writer.show_progress)
393                 writer.progress = start_progress("Selecting bitmap commits", 0);
394
395         if (indexed_commits_nr < 100) {
396                 for (i = 0; i < indexed_commits_nr; ++i)
397                         push_bitmapped_commit(indexed_commits[i], NULL);
398                 return;
399         }
400
401         for (;;) {
402                 struct ewah_bitmap *reused_bitmap = NULL;
403                 struct commit *chosen = NULL;
404
405                 next = next_commit_index(i);
406
407                 if (i + next >= indexed_commits_nr)
408                         break;
409
410                 if (max_bitmaps > 0 && writer.selected_nr >= max_bitmaps) {
411                         writer.selected_nr = max_bitmaps;
412                         break;
413                 }
414
415                 if (next == 0) {
416                         chosen = indexed_commits[i];
417                         reused_bitmap = find_reused_bitmap(chosen->object.oid.hash);
418                 } else {
419                         chosen = indexed_commits[i + next];
420
421                         for (j = 0; j <= next; ++j) {
422                                 struct commit *cm = indexed_commits[i + j];
423
424                                 reused_bitmap = find_reused_bitmap(cm->object.oid.hash);
425                                 if (reused_bitmap || (cm->object.flags & NEEDS_BITMAP) != 0) {
426                                         chosen = cm;
427                                         break;
428                                 }
429
430                                 if (cm->parents && cm->parents->next)
431                                         chosen = cm;
432                         }
433                 }
434
435                 push_bitmapped_commit(chosen, reused_bitmap);
436
437                 i += next + 1;
438                 display_progress(writer.progress, i);
439         }
440
441         stop_progress(&writer.progress);
442 }
443
444
445 static int hashwrite_ewah_helper(void *f, const void *buf, size_t len)
446 {
447         /* hashwrite will die on error */
448         hashwrite(f, buf, len);
449         return len;
450 }
451
452 /**
453  * Write the bitmap index to disk
454  */
455 static inline void dump_bitmap(struct hashfile *f, struct ewah_bitmap *bitmap)
456 {
457         if (ewah_serialize_to(bitmap, hashwrite_ewah_helper, f) < 0)
458                 die("Failed to write bitmap index");
459 }
460
461 static const unsigned char *sha1_access(size_t pos, void *table)
462 {
463         struct pack_idx_entry **index = table;
464         return index[pos]->oid.hash;
465 }
466
467 static void write_selected_commits_v1(struct hashfile *f,
468                                       struct pack_idx_entry **index,
469                                       uint32_t index_nr)
470 {
471         int i;
472
473         for (i = 0; i < writer.selected_nr; ++i) {
474                 struct bitmapped_commit *stored = &writer.selected[i];
475
476                 int commit_pos =
477                         sha1_pos(stored->commit->object.oid.hash, index, index_nr, sha1_access);
478
479                 if (commit_pos < 0)
480                         BUG("trying to write commit not in index");
481
482                 hashwrite_be32(f, commit_pos);
483                 hashwrite_u8(f, stored->xor_offset);
484                 hashwrite_u8(f, stored->flags);
485
486                 dump_bitmap(f, stored->write_as);
487         }
488 }
489
490 static void write_hash_cache(struct hashfile *f,
491                              struct pack_idx_entry **index,
492                              uint32_t index_nr)
493 {
494         uint32_t i;
495
496         for (i = 0; i < index_nr; ++i) {
497                 struct object_entry *entry = (struct object_entry *)index[i];
498                 uint32_t hash_value = htonl(entry->hash);
499                 hashwrite(f, &hash_value, sizeof(hash_value));
500         }
501 }
502
503 void bitmap_writer_set_checksum(unsigned char *sha1)
504 {
505         hashcpy(writer.pack_checksum, sha1);
506 }
507
508 void bitmap_writer_finish(struct pack_idx_entry **index,
509                           uint32_t index_nr,
510                           const char *filename,
511                           uint16_t options)
512 {
513         static uint16_t default_version = 1;
514         static uint16_t flags = BITMAP_OPT_FULL_DAG;
515         struct strbuf tmp_file = STRBUF_INIT;
516         struct hashfile *f;
517
518         struct bitmap_disk_header header;
519
520         int fd = odb_mkstemp(&tmp_file, "pack/tmp_bitmap_XXXXXX");
521
522         f = hashfd(fd, tmp_file.buf);
523
524         memcpy(header.magic, BITMAP_IDX_SIGNATURE, sizeof(BITMAP_IDX_SIGNATURE));
525         header.version = htons(default_version);
526         header.options = htons(flags | options);
527         header.entry_count = htonl(writer.selected_nr);
528         hashcpy(header.checksum, writer.pack_checksum);
529
530         hashwrite(f, &header, sizeof(header));
531         dump_bitmap(f, writer.commits);
532         dump_bitmap(f, writer.trees);
533         dump_bitmap(f, writer.blobs);
534         dump_bitmap(f, writer.tags);
535         write_selected_commits_v1(f, index, index_nr);
536
537         if (options & BITMAP_OPT_HASH_CACHE)
538                 write_hash_cache(f, index, index_nr);
539
540         finalize_hashfile(f, NULL, CSUM_HASH_IN_STREAM | CSUM_FSYNC | CSUM_CLOSE);
541
542         if (adjust_shared_perm(tmp_file.buf))
543                 die_errno("unable to make temporary bitmap file readable");
544
545         if (rename(tmp_file.buf, filename))
546                 die_errno("unable to rename temporary bitmap file to '%s'", filename);
547
548         strbuf_release(&tmp_file);
549 }