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