1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 #include <sys/stdint.h>
21 
22 #include <limits.h>
23 #include <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 #include <sha1.h>
27 #include <zlib.h>
28 #include <ctype.h>
29 
30 #include "got_error.h"
31 #include "got_object.h"
32 #include "got_cancel.h"
33 #include "got_commit_graph.h"
34 #include "got_path.h"
35 
36 #include "got_lib_delta.h"
37 #include "got_lib_inflate.h"
38 #include "got_lib_object.h"
39 #include "got_lib_object_idset.h"
40 
41 struct got_commit_graph_node {
42 	struct got_object_id id;
43 
44 	/* Used only during iteration. */
45 	time_t timestamp;
46 	TAILQ_ENTRY(got_commit_graph_node) entry;
47 };
48 
49 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
50 
51 struct got_commit_graph_branch_tip {
52 	struct got_object_id *commit_id;
53 	struct got_commit_object *commit;
54 	struct got_commit_graph_node *new_node;
55 };
56 
57 struct got_commit_graph {
58 	/* The set of all commits we have traversed. */
59 	struct got_object_idset *node_ids;
60 
61 	int flags;
62 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL		0x01
63 
64 	/*
65 	 * A set of object IDs of known parent commits which we have not yet
66 	 * traversed. Each commit ID in this set represents a branch in commit
67 	 * history: Either the first-parent branch of the head node, or another
68 	 * branch corresponding to a traversed merge commit for which we have
69 	 * not traversed a branch point commit yet.
70 	 *
71 	 * Whenever we add a commit with a matching ID to the graph, we remove
72 	 * its corresponding element from this set, and add new elements for
73 	 * each of that commit's parent commits which were not traversed yet.
74 	 *
75 	 * When API users ask us to fetch more commits, we fetch commits from
76 	 * all currently open branches. This allows API users to process
77 	 * commits in linear order even though the history contains branches.
78 	 */
79 	struct got_object_idset *open_branches;
80 
81 	/* Array of branch tips for fetch_commits_from_open_branches(). */
82 	struct got_commit_graph_branch_tip *tips;
83 	int ntips;
84 
85 	/* Path of tree entry of interest to the API user. */
86 	char *path;
87 
88 	/*
89 	 * Nodes which will be passed to the API user next, sorted by
90 	 * commit timestmap.
91 	 */
92 	struct got_commit_graph_iter_list iter_list;
93 };
94 
95 static const struct got_error *
96 detect_changed_path(int *changed, struct got_commit_object *commit,
97     struct got_object_id *commit_id, const char *path,
98     struct got_repository *repo)
99 {
100 	const struct got_error *err = NULL;
101 	struct got_commit_object *pcommit = NULL;
102 	struct got_tree_object *tree = NULL, *ptree = NULL;
103 	struct got_object_qid *pid;
104 
105 	if (got_path_is_root_dir(path)) {
106 		*changed = 1;
107 		return NULL;
108 	}
109 
110 	*changed = 0;
111 
112 	pid = STAILQ_FIRST(&commit->parent_ids);
113 	if (pid == NULL) {
114 		struct got_object_id *obj_id;
115 		err = got_object_id_by_path(&obj_id, repo, commit_id, path);
116 		if (err) {
117 			if (err->code == GOT_ERR_NO_TREE_ENTRY)
118 				err = NULL;
119 		} else
120 			*changed = 1; /* The path was created in this commit. */
121 		free(obj_id);
122 		return err;
123 	}
124 
125 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
126 	if (err)
127 		return err;
128 
129 	err = got_object_open_as_commit(&pcommit, repo, pid->id);
130 	if (err)
131 		goto done;
132 
133 	err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
134 	if (err)
135 		goto done;
136 
137 	err = got_object_tree_path_changed(changed, tree, ptree, path, repo);
138 done:
139 	if (tree)
140 		got_object_tree_close(tree);
141 	if (ptree)
142 		got_object_tree_close(ptree);
143 	if (pcommit)
144 		got_object_commit_close(pcommit);
145 	return err;
146 }
147 
148 static void
149 add_node_to_iter_list(struct got_commit_graph *graph,
150     struct got_commit_graph_node *node, time_t committer_time)
151 {
152 	struct got_commit_graph_node *n, *next;
153 
154 	node->timestamp = committer_time;
155 
156 	n = TAILQ_FIRST(&graph->iter_list);
157 	while (n) {
158 		next = TAILQ_NEXT(n, entry);
159 		if (next && node->timestamp >= next->timestamp) {
160 			TAILQ_INSERT_BEFORE(next, node, entry);
161 			return;
162 		}
163 		n = next;
164 	}
165 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
166 }
167 
168 static const struct got_error *
169 add_node(struct got_commit_graph_node **new_node,
170     struct got_commit_graph *graph, struct got_object_id *commit_id,
171     struct got_repository *repo)
172 {
173 	const struct got_error *err = NULL;
174 	struct got_commit_graph_node *node;
175 
176 	*new_node = NULL;
177 
178 	node = calloc(1, sizeof(*node));
179 	if (node == NULL)
180 		return got_error_from_errno("calloc");
181 
182 	memcpy(&node->id, commit_id, sizeof(node->id));
183 	err = got_object_idset_add(graph->node_ids, &node->id, NULL);
184 	if (err)
185 		free(node);
186 	else
187 		*new_node = node;
188 	return err;
189 }
190 
191 /*
192  * Ask got-read-pack to traverse first-parent history until a commit is
193  * encountered which modified graph->path, or until the pack file runs
194  * out of relevant commits. This is faster than sending an individual
195  * request for each commit stored in the pack file.
196  */
197 static const struct got_error *
198 packed_first_parent_traversal(int *ncommits_traversed,
199     struct got_commit_graph *graph, struct got_object_id *commit_id,
200     struct got_repository *repo)
201 {
202 	const struct got_error *err = NULL;
203 	struct got_object_id_queue traversed_commits;
204 	struct got_object_qid *qid;
205 
206 	STAILQ_INIT(&traversed_commits);
207 	*ncommits_traversed = 0;
208 
209 	err = got_traverse_packed_commits(&traversed_commits,
210 	    commit_id, graph->path, repo);
211 	if (err)
212 		return err;
213 
214 	/* Add all traversed commits to the graph... */
215 	STAILQ_FOREACH(qid, &traversed_commits, entry) {
216 		struct got_commit_graph_node *node;
217 
218 		if (got_object_idset_contains(graph->open_branches, qid->id))
219 			continue;
220 		if (got_object_idset_contains(graph->node_ids, qid->id))
221 			continue;
222 
223 		(*ncommits_traversed)++;
224 
225 		/* ... except the last commit is the new branch tip. */
226 		if (STAILQ_NEXT(qid, entry) == NULL) {
227 			err = got_object_idset_add(graph->open_branches,
228 			    qid->id, NULL);
229 			break;
230 		}
231 
232 		err = add_node(&node, graph, qid->id, repo);
233 		if (err)
234 			break;
235 	}
236 
237 	got_object_id_queue_free(&traversed_commits);
238 	return err;
239 }
240 
241 static const struct got_error *
242 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
243 {
244 	const struct got_error *err;
245 
246 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
247 	if (err && err->code != GOT_ERR_NO_OBJ)
248 		return err;
249 	return NULL;
250 }
251 
252 static const struct got_error *
253 advance_branch(struct got_commit_graph *graph, struct got_object_id *commit_id,
254     struct got_commit_object *commit, struct got_repository *repo)
255 {
256 	const struct got_error *err;
257 	struct got_object_qid *qid;
258 
259 	err = close_branch(graph, commit_id);
260 	if (err)
261 		return err;
262 
263 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
264 		qid = STAILQ_FIRST(&commit->parent_ids);
265 		if (qid == NULL ||
266 		    got_object_idset_contains(graph->open_branches, qid->id))
267 			return NULL;
268 		/*
269 		 * The root directory always changes by definition, and when
270 		 * logging the root we want to traverse consecutive commits
271 		 * even if they point at the same tree.
272 		 * But if we are looking for a specific path then we can avoid
273 		 * fetching packed commits which did not modify the path and
274 		 * only fetch their IDs. This speeds up 'got blame'.
275 		 */
276 		if (!got_path_is_root_dir(graph->path) &&
277 		    (commit->flags & GOT_COMMIT_FLAG_PACKED)) {
278 			int ncommits = 0;
279 			err = packed_first_parent_traversal(&ncommits,
280 			    graph, qid->id, repo);
281 			if (err || ncommits > 0)
282 				return err;
283 		}
284 		return got_object_idset_add(graph->open_branches,
285 		    qid->id, NULL);
286 	}
287 
288 	/*
289 	 * If we are graphing commits for a specific path, skip branches
290 	 * which do not contribute any content to this path.
291 	 */
292 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
293 		struct got_object_id *merged_id, *prev_id = NULL;
294 		int branches_differ = 0;
295 
296 		err = got_object_id_by_path(&merged_id, repo, commit_id,
297 		    graph->path);
298 		if (err)
299 			return err;
300 
301 		STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
302 			struct got_object_id *id;
303 
304 			if (got_object_idset_contains(graph->open_branches,
305 			    qid->id))
306 				continue;
307 
308 			err = got_object_id_by_path(&id, repo, qid->id,
309 			    graph->path);
310 			if (err) {
311 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
312 					branches_differ = 1;
313 					continue;
314 				}
315 				free(merged_id);
316 				free(prev_id);
317 				return err;
318 			}
319 
320 			if (prev_id) {
321 				if (!branches_differ &&
322 				    got_object_id_cmp(id, prev_id) != 0)
323 					branches_differ = 1;
324 				free(prev_id);
325 			}
326 			prev_id = id;
327 
328 			/*
329 			 * If a branch has created the merged content we can
330 			 * skip any other branches.
331 			 */
332 			if (got_object_id_cmp(merged_id, id) == 0) {
333 				err = got_object_idset_add(graph->open_branches,
334 				    qid->id, NULL);
335 				free(merged_id);
336 				free(id);
337 				return err;
338 			}
339 		}
340 
341 		free(prev_id);
342 		prev_id = NULL;
343 		free(merged_id);
344 		merged_id = NULL;
345 
346 		/*
347 		 * If the path's content is the same on all branches,
348 		 * follow the first parent only.
349 		 */
350 		if (!branches_differ) {
351 			qid = STAILQ_FIRST(&commit->parent_ids);
352 			if (qid == NULL)
353 				return NULL;
354 			if (got_object_idset_contains(graph->open_branches,
355 			    qid->id))
356 				return NULL;
357 			if (got_object_idset_contains(graph->node_ids,
358 			    qid->id))
359 				return NULL; /* parent already traversed */
360 			return got_object_idset_add(graph->open_branches,
361 			    qid->id, NULL);
362 		}
363 	}
364 
365 	STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
366 		if (got_object_idset_contains(graph->open_branches, qid->id))
367 			continue;
368 		if (got_object_idset_contains(graph->node_ids, qid->id))
369 			continue; /* parent already traversed */
370 		err = got_object_idset_add(graph->open_branches, qid->id, NULL);
371 		if (err)
372 			return err;
373 	}
374 
375 	return NULL;
376 }
377 
378 const struct got_error *
379 got_commit_graph_open(struct got_commit_graph **graph,
380     const char *path, int first_parent_traversal)
381 {
382 	const struct got_error *err = NULL;
383 
384 	*graph = calloc(1, sizeof(**graph));
385 	if (*graph == NULL)
386 		return got_error_from_errno("calloc");
387 
388 	TAILQ_INIT(&(*graph)->iter_list);
389 
390 	(*graph)->path = strdup(path);
391 	if ((*graph)->path == NULL) {
392 		err = got_error_from_errno("strdup");
393 		goto done;
394 	}
395 
396 	(*graph)->node_ids = got_object_idset_alloc();
397 	if ((*graph)->node_ids == NULL) {
398 		err = got_error_from_errno("got_object_idset_alloc");
399 		goto done;
400 	}
401 
402 	(*graph)->open_branches = got_object_idset_alloc();
403 	if ((*graph)->open_branches == NULL) {
404 		err = got_error_from_errno("got_object_idset_alloc");
405 		goto done;
406 	}
407 
408 	if (first_parent_traversal)
409 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
410 done:
411 	if (err) {
412 		got_commit_graph_close(*graph);
413 		*graph = NULL;
414 	}
415 	return err;
416 }
417 
418 struct add_branch_tip_arg {
419 	struct got_commit_graph_branch_tip *tips;
420 	int ntips;
421 	struct got_repository *repo;
422 	struct got_commit_graph *graph;
423 };
424 
425 static const struct got_error *
426 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
427 {
428 	const struct got_error *err;
429 	struct add_branch_tip_arg *a = arg;
430 	struct got_commit_graph_node *new_node;
431 	struct got_commit_object *commit;
432 
433 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
434 	if (err)
435 		return err;
436 
437 	err = add_node(&new_node, a->graph, commit_id, a->repo);
438 	if (err)
439 		return err;
440 
441 	a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
442 	a->tips[a->ntips].commit = commit;
443 	a->tips[a->ntips].new_node = new_node;
444 	a->ntips++;
445 
446 	return NULL;
447 }
448 
449 static const struct got_error *
450 fetch_commits_from_open_branches(struct got_commit_graph *graph,
451     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
452 {
453 	const struct got_error *err;
454 	struct add_branch_tip_arg arg;
455 	int i, ntips;
456 
457 	ntips = got_object_idset_num_elements(graph->open_branches);
458 	if (ntips == 0)
459 		return NULL;
460 
461 	/* (Re-)allocate branch tips array if necessary. */
462 	if (graph->ntips < ntips) {
463 		struct got_commit_graph_branch_tip *tips;
464 		tips = recallocarray(graph->tips, graph->ntips, ntips,
465 		    sizeof(*tips));
466 		if (tips == NULL)
467 			return got_error_from_errno("recallocarray");
468 		graph->tips = tips;
469 		graph->ntips = ntips;
470 	}
471 	arg.tips = graph->tips;
472 	arg.ntips = 0; /* add_branch_tip() will increment */
473 	arg.repo = repo;
474 	arg.graph = graph;
475 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
476 	    &arg);
477 	if (err)
478 		goto done;
479 
480 	for (i = 0; i < arg.ntips; i++) {
481 		struct got_object_id *commit_id;
482 		struct got_commit_object *commit;
483 		struct got_commit_graph_node *new_node;
484 		int changed;
485 
486 		if (cancel_cb) {
487 			err = (*cancel_cb)(cancel_arg);
488 			if (err)
489 				break;
490 		}
491 
492 		commit_id = arg.tips[i].commit_id;
493 		commit = arg.tips[i].commit;
494 		new_node = arg.tips[i].new_node;
495 
496 		err = detect_changed_path(&changed, commit, commit_id,
497 		    graph->path, repo);
498 		if (err) {
499 			if (err->code != GOT_ERR_NO_OBJ)
500 				break;
501 			/*
502 			 * History of the path stops here on the current
503 			 * branch. Keep going on other branches.
504 			 */
505 			err = close_branch(graph, commit_id);
506 			if (err)
507 				break;
508 			continue;
509 		}
510 		if (changed)
511 			add_node_to_iter_list(graph, new_node,
512 			    got_object_commit_get_committer_time(commit));
513 		err = advance_branch(graph, commit_id, commit, repo);
514 		if (err)
515 			break;
516 	}
517 done:
518 	for (i = 0; i < arg.ntips; i++)
519 		got_object_commit_close(arg.tips[i].commit);
520 	return err;
521 }
522 
523 void
524 got_commit_graph_close(struct got_commit_graph *graph)
525 {
526 	if (graph->open_branches)
527 		got_object_idset_free(graph->open_branches);
528 	if (graph->node_ids)
529 		got_object_idset_free(graph->node_ids);
530 	free(graph->tips);
531 	free(graph->path);
532 	free(graph);
533 }
534 
535 const struct got_error *
536 got_commit_graph_iter_start(struct got_commit_graph *graph,
537     struct got_object_id *id, struct got_repository *repo,
538     got_cancel_cb cancel_cb, void *cancel_arg)
539 {
540 	const struct got_error *err = NULL;
541 
542 	if (!TAILQ_EMPTY(&graph->iter_list))
543 		return got_error(GOT_ERR_ITER_BUSY);
544 
545 	err = got_object_idset_add(graph->open_branches, id, NULL);
546 	if (err)
547 		return err;
548 
549 	/* Locate first commit which changed graph->path. */
550 	while (TAILQ_EMPTY(&graph->iter_list) &&
551 	    got_object_idset_num_elements(graph->open_branches) > 0) {
552 		err = fetch_commits_from_open_branches(graph, repo,
553 		    cancel_cb, cancel_arg);
554 		if (err)
555 			return err;
556 	}
557 
558 	if (TAILQ_EMPTY(&graph->iter_list)) {
559 		const char *path;
560 		if (got_path_is_root_dir(graph->path))
561 			return got_error_no_obj(id);
562 		path = graph->path;
563 		while (path[0] == '/')
564 			path++;
565 		return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
566 	}
567 
568 	return NULL;
569 }
570 
571 const struct got_error *
572 got_commit_graph_iter_next(struct got_object_id **id,
573     struct got_commit_graph *graph, struct got_repository *repo,
574     got_cancel_cb cancel_cb, void *cancel_arg)
575 {
576 	const struct got_error *err = NULL;
577 	struct got_commit_graph_node *node;
578 
579 	*id = NULL;
580 
581 	node = TAILQ_FIRST(&graph->iter_list);
582 	if (node == NULL) {
583 		/* We are done iterating, or iteration was not started. */
584 		return got_error(GOT_ERR_ITER_COMPLETED);
585 	}
586 
587 	while (TAILQ_NEXT(node, entry) == NULL &&
588 	    got_object_idset_num_elements(graph->open_branches) > 0) {
589 		err = fetch_commits_from_open_branches(graph, repo,
590 		    cancel_cb, cancel_arg);
591 		if (err)
592 			return err;
593 	}
594 
595 	*id = &node->id;
596 	TAILQ_REMOVE(&graph->iter_list, node, entry);
597 	return NULL;
598 }
599 
600 const struct got_error *
601 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
602     struct got_object_id *commit_id, struct got_object_id *commit_id2,
603     int first_parent_traversal,
604     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
605 {
606 	const struct got_error *err = NULL;
607 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
608 	int completed = 0, completed2 = 0;
609 	struct got_object_idset *commit_ids;
610 
611 	*yca_id = NULL;
612 
613 	commit_ids = got_object_idset_alloc();
614 	if (commit_ids == NULL)
615 		return got_error_from_errno("got_object_idset_alloc");
616 
617 	err = got_commit_graph_open(&graph, "/", first_parent_traversal);
618 	if (err)
619 		goto done;
620 
621 	err = got_commit_graph_open(&graph2, "/", first_parent_traversal);
622 	if (err)
623 		goto done;
624 
625 	err = got_commit_graph_iter_start(graph, commit_id, repo,
626 	    cancel_cb, cancel_arg);
627 	if (err)
628 		goto done;
629 
630 	err = got_commit_graph_iter_start(graph2, commit_id2, repo,
631 	    cancel_cb, cancel_arg);
632 	if (err)
633 		goto done;
634 
635 	for (;;) {
636 		struct got_object_id *id = NULL, *id2 = NULL;
637 
638 		if (cancel_cb) {
639 			err = (*cancel_cb)(cancel_arg);
640 			if (err)
641 				break;
642 		}
643 
644 		if (!completed) {
645 			err = got_commit_graph_iter_next(&id, graph, repo,
646 			    cancel_cb, cancel_arg);
647 			if (err) {
648 				if (err->code != GOT_ERR_ITER_COMPLETED)
649 					break;
650 				err = NULL;
651 				completed = 1;
652 			}
653 		}
654 
655 		if (!completed2) {
656 			err = got_commit_graph_iter_next(&id2, graph2, repo,
657 			    cancel_cb, cancel_arg);
658 			if (err) {
659 				if (err->code != GOT_ERR_ITER_COMPLETED)
660 					break;
661 				err = NULL;
662 				completed2 = 1;
663 			}
664 		}
665 
666 		if (id) {
667 			if (got_object_idset_contains(commit_ids, id)) {
668 				*yca_id = got_object_id_dup(id);
669 				if (*yca_id)
670 					break;
671 				err = got_error_from_errno("got_object_id_dup");
672 				break;
673 
674 			}
675 			err = got_object_idset_add(commit_ids, id, NULL);
676 			if (err)
677 				break;
678 		}
679 		if (id2) {
680 			if (got_object_idset_contains(commit_ids, id2)) {
681 				*yca_id = got_object_id_dup(id2);
682 				if (*yca_id)
683 					break;
684 				err = got_error_from_errno("got_object_id_dup");
685 				break;
686 
687 			}
688 			err = got_object_idset_add(commit_ids, id2, NULL);
689 			if (err)
690 				break;
691 		}
692 
693 		if (completed && completed2) {
694 			err = got_error(GOT_ERR_ANCESTRY);
695 			break;
696 		}
697 
698 	}
699 done:
700 	got_object_idset_free(commit_ids);
701 	if (graph)
702 		got_commit_graph_close(graph);
703 	if (graph2)
704 		got_commit_graph_close(graph2);
705 	return err;
706 }