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 <sha2.h>
28 #include <zlib.h>
29 #include <ctype.h>
30 
31 #include "got_error.h"
32 #include "got_object.h"
33 #include "got_cancel.h"
34 #include "got_commit_graph.h"
35 #include "got_path.h"
36 
37 #include "got_lib_delta.h"
38 #include "got_lib_inflate.h"
39 #include "got_lib_object.h"
40 #include "got_lib_object_idset.h"
41 #include "got_lib_object_qid.h"
42 
43 #ifndef nitems
44 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
45 #endif
46 
47 struct got_commit_graph_node {
48 	struct got_object_id id;
49 
50 	/* Used for topological sorting. */
51 	struct got_commit_graph_node *parents[2];
52 	struct got_commit_graph_node **more_parents;
53 	int nparents;
54 	int indegree;
55 
56 	/* Used only during iteration. */
57 	time_t timestamp;
58 	TAILQ_ENTRY(got_commit_graph_node) entry;
59 };
60 
61 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
62 
63 struct got_commit_graph_branch_tip {
64 	struct got_object_id *commit_id;
65 	struct got_commit_object *commit;
66 	struct got_commit_graph_node *new_node;
67 };
68 
69 struct got_commit_graph {
70 	/* The set of all commits we have traversed. */
71 	struct got_object_idset *node_ids;
72 
73 	int flags;
74 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL		0x01
75 #define GOT_COMMIT_GRAPH_TOPOSORT			0x02
76 
77 	/*
78 	 * A set of object IDs of known parent commits which we have not yet
79 	 * traversed. Each commit ID in this set represents a branch in commit
80 	 * history: Either the first-parent branch of the head node, or another
81 	 * branch corresponding to a traversed merge commit for which we have
82 	 * not traversed a branch point commit yet.
83 	 *
84 	 * Whenever we add a commit with a matching ID to the graph, we remove
85 	 * its corresponding element from this set, and add new elements for
86 	 * each of that commit's parent commits which were not traversed yet.
87 	 *
88 	 * When API users ask us to fetch more commits, we fetch commits from
89 	 * all currently open branches. This allows API users to process
90 	 * commits in linear order even though the history contains branches.
91 	 */
92 	struct got_object_idset *open_branches;
93 
94 	/* Array of branch tips for fetch_commits_from_open_branches(). */
95 	struct got_commit_graph_branch_tip *tips;
96 	int ntips;
97 
98 	/* Path of tree entry of interest to the API user. */
99 	char *path;
100 
101 	/*
102 	 * Nodes which will be passed to the API user next, sorted by
103 	 * commit timestamp. Sorted in topological order only if topological
104 	 * sorting was requested.
105 	 */
106 	struct got_commit_graph_iter_list iter_list;
107 };
108 
109 static const struct got_error *
110 detect_changed_path(int *changed, struct got_commit_object *commit,
111     struct got_object_id *commit_id, const char *path,
112     struct got_repository *repo)
113 {
114 	const struct got_error *err = NULL;
115 	struct got_commit_object *pcommit = NULL;
116 	struct got_tree_object *tree = NULL, *ptree = NULL;
117 	struct got_object_qid *pid;
118 
119 	if (got_path_is_root_dir(path)) {
120 		*changed = 1;
121 		return NULL;
122 	}
123 
124 	*changed = 0;
125 
126 	pid = STAILQ_FIRST(&commit->parent_ids);
127 	if (pid == NULL) {
128 		struct got_object_id *obj_id;
129 		err = got_object_id_by_path(&obj_id, repo, commit, path);
130 		if (err) {
131 			if (err->code == GOT_ERR_NO_TREE_ENTRY)
132 				err = NULL;
133 		} else
134 			*changed = 1; /* The path was created in this commit. */
135 		free(obj_id);
136 		return err;
137 	}
138 
139 	err = got_object_open_as_tree(&tree, repo, commit->tree_id);
140 	if (err)
141 		return err;
142 
143 	err = got_object_open_as_commit(&pcommit, repo, &pid->id);
144 	if (err)
145 		goto done;
146 
147 	err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
148 	if (err)
149 		goto done;
150 
151 	err = got_object_tree_path_changed(changed, tree, ptree, path, repo);
152 done:
153 	if (tree)
154 		got_object_tree_close(tree);
155 	if (ptree)
156 		got_object_tree_close(ptree);
157 	if (pcommit)
158 		got_object_commit_close(pcommit);
159 	return err;
160 }
161 
162 static void
163 add_node_to_iter_list(struct got_commit_graph *graph,
164     struct got_commit_graph_node *node, time_t committer_time)
165 {
166 	struct got_commit_graph_node *n, *next;
167 
168 	node->timestamp = committer_time;
169 
170 	n = TAILQ_FIRST(&graph->iter_list);
171 	while (n) {
172 		next = TAILQ_NEXT(n, entry);
173 		if (next && node->timestamp >= next->timestamp) {
174 			TAILQ_INSERT_BEFORE(next, node, entry);
175 			return;
176 		}
177 		n = next;
178 	}
179 	TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
180 }
181 
182 static const struct got_error *
183 add_node(struct got_commit_graph_node **new_node,
184     struct got_commit_graph *graph, struct got_object_id *commit_id,
185     struct got_repository *repo)
186 {
187 	const struct got_error *err = NULL;
188 	struct got_commit_graph_node *node;
189 
190 	*new_node = NULL;
191 
192 	node = calloc(1, sizeof(*node));
193 	if (node == NULL)
194 		return got_error_from_errno("calloc");
195 
196 	memcpy(&node->id, commit_id, sizeof(node->id));
197 	node->nparents = -1;
198 	err = got_object_idset_add(graph->node_ids, &node->id, node);
199 	if (err)
200 		free(node);
201 	else
202 		*new_node = node;
203 	return err;
204 }
205 
206 /*
207  * Ask got-read-pack to traverse first-parent history until a commit is
208  * encountered which modified graph->path, or until the pack file runs
209  * out of relevant commits. This is faster than sending an individual
210  * request for each commit stored in the pack file.
211  */
212 static const struct got_error *
213 packed_first_parent_traversal(int *ncommits_traversed,
214     struct got_commit_graph *graph, struct got_object_id *commit_id,
215     struct got_repository *repo)
216 {
217 	const struct got_error *err = NULL;
218 	struct got_object_id_queue traversed_commits;
219 	struct got_object_qid *qid;
220 
221 	STAILQ_INIT(&traversed_commits);
222 	*ncommits_traversed = 0;
223 
224 	err = got_traverse_packed_commits(&traversed_commits,
225 	    commit_id, graph->path, repo);
226 	if (err)
227 		return err;
228 
229 	/* Add all traversed commits to the graph... */
230 	STAILQ_FOREACH(qid, &traversed_commits, entry) {
231 		if (got_object_idset_contains(graph->open_branches, &qid->id))
232 			continue;
233 		if (got_object_idset_contains(graph->node_ids, &qid->id))
234 			continue;
235 
236 		(*ncommits_traversed)++;
237 
238 		/* ... except the last commit is the new branch tip. */
239 		if (STAILQ_NEXT(qid, entry) == NULL) {
240 			err = got_object_idset_add(graph->open_branches,
241 			    &qid->id, NULL);
242 			break;
243 		}
244 
245 		err = got_object_idset_add(graph->node_ids, &qid->id, NULL);
246 		if (err)
247 			break;
248 	}
249 
250 	got_object_id_queue_free(&traversed_commits);
251 	return err;
252 }
253 
254 static const struct got_error *
255 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
256 {
257 	const struct got_error *err;
258 
259 	err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
260 	if (err && err->code != GOT_ERR_NO_OBJ)
261 		return err;
262 	return NULL;
263 }
264 
265 static const struct got_error *
266 advance_branch(struct got_commit_graph *graph, struct got_object_id *commit_id,
267     struct got_commit_object *commit, struct got_repository *repo)
268 {
269 	const struct got_error *err;
270 	struct got_object_qid *qid;
271 	struct got_object_id *merged_id = NULL;
272 
273 	err = close_branch(graph, commit_id);
274 	if (err)
275 		return err;
276 
277 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
278 		qid = STAILQ_FIRST(&commit->parent_ids);
279 		if (qid == NULL ||
280 		    got_object_idset_contains(graph->open_branches, &qid->id))
281 			return NULL;
282 		/*
283 		 * The root directory always changes by definition, and when
284 		 * logging the root we want to traverse consecutive commits
285 		 * even if they point at the same tree.
286 		 * But if we are looking for a specific path then we can avoid
287 		 * fetching packed commits which did not modify the path and
288 		 * only fetch their IDs. This speeds up 'got blame'.
289 		 */
290 		if (!got_path_is_root_dir(graph->path) &&
291 		    (commit->flags & GOT_COMMIT_FLAG_PACKED)) {
292 			int ncommits = 0;
293 			err = packed_first_parent_traversal(&ncommits,
294 			    graph, &qid->id, repo);
295 			if (err || ncommits > 0)
296 				return err;
297 		}
298 		return got_object_idset_add(graph->open_branches,
299 		    &qid->id, NULL);
300 	}
301 
302 	/*
303 	 * If we are graphing commits for a specific path, skip branches
304 	 * which do not contribute any content to this path.
305 	 */
306 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path)) {
307 		err = got_object_id_by_path(&merged_id, repo, commit, graph->path);
308 		if (err && err->code != GOT_ERR_NO_TREE_ENTRY)
309 			return err;
310 		/* The requested path does not exist in this merge commit. */
311 	}
312 	if (commit->nparents > 1 && !got_path_is_root_dir(graph->path) &&
313 	    merged_id != NULL) {
314 		struct got_object_id *prev_id = NULL;
315 		int branches_differ = 0;
316 
317 
318 		STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
319 			struct got_object_id *id = NULL;
320 			struct got_commit_object *pcommit = NULL;
321 
322 			if (got_object_idset_contains(graph->open_branches,
323 			    &qid->id))
324 				continue;
325 
326 			err = got_object_open_as_commit(&pcommit, repo,
327 			    &qid->id);
328 			if (err) {
329 				free(merged_id);
330 				free(prev_id);
331 				return err;
332 			}
333 			err = got_object_id_by_path(&id, repo, pcommit,
334 			    graph->path);
335 			got_object_commit_close(pcommit);
336 			pcommit = NULL;
337 			if (err) {
338 				if (err->code == GOT_ERR_NO_TREE_ENTRY) {
339 					branches_differ = 1;
340 					continue;
341 				}
342 				free(merged_id);
343 				free(prev_id);
344 				return err;
345 			}
346 
347 			if (prev_id) {
348 				if (!branches_differ &&
349 				    got_object_id_cmp(id, prev_id) != 0)
350 					branches_differ = 1;
351 				free(prev_id);
352 			}
353 			prev_id = id;
354 
355 			/*
356 			 * If a branch has created the merged content we can
357 			 * skip any other branches.
358 			 */
359 			if (got_object_id_cmp(merged_id, id) == 0) {
360 				err = got_object_idset_add(graph->open_branches,
361 				    &qid->id, NULL);
362 				free(merged_id);
363 				free(id);
364 				return err;
365 			}
366 		}
367 
368 		free(prev_id);
369 		prev_id = NULL;
370 		free(merged_id);
371 		merged_id = NULL;
372 
373 		/*
374 		 * If the path's content is the same on all branches,
375 		 * follow the first parent only.
376 		 */
377 		if (!branches_differ) {
378 			qid = STAILQ_FIRST(&commit->parent_ids);
379 			if (qid == NULL)
380 				return NULL;
381 			if (got_object_idset_contains(graph->open_branches,
382 			    &qid->id))
383 				return NULL;
384 			if (got_object_idset_contains(graph->node_ids,
385 			    &qid->id))
386 				return NULL; /* parent already traversed */
387 			return got_object_idset_add(graph->open_branches,
388 			    &qid->id, NULL);
389 		}
390 	}
391 
392 	STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
393 		if (got_object_idset_contains(graph->open_branches, &qid->id))
394 			continue;
395 		if (got_object_idset_contains(graph->node_ids, &qid->id))
396 			continue; /* parent already traversed */
397 		err = got_object_idset_add(graph->open_branches, &qid->id,
398 		    NULL);
399 		if (err)
400 			return err;
401 	}
402 
403 	return NULL;
404 }
405 
406 const struct got_error *
407 got_commit_graph_open(struct got_commit_graph **graph,
408     const char *path, int first_parent_traversal)
409 {
410 	const struct got_error *err = NULL;
411 
412 	*graph = calloc(1, sizeof(**graph));
413 	if (*graph == NULL)
414 		return got_error_from_errno("calloc");
415 
416 	TAILQ_INIT(&(*graph)->iter_list);
417 
418 	(*graph)->path = strdup(path);
419 	if ((*graph)->path == NULL) {
420 		err = got_error_from_errno("strdup");
421 		goto done;
422 	}
423 
424 	(*graph)->node_ids = got_object_idset_alloc();
425 	if ((*graph)->node_ids == NULL) {
426 		err = got_error_from_errno("got_object_idset_alloc");
427 		goto done;
428 	}
429 
430 	(*graph)->open_branches = got_object_idset_alloc();
431 	if ((*graph)->open_branches == NULL) {
432 		err = got_error_from_errno("got_object_idset_alloc");
433 		goto done;
434 	}
435 
436 	if (first_parent_traversal)
437 		(*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
438 done:
439 	if (err) {
440 		got_commit_graph_close(*graph);
441 		*graph = NULL;
442 	}
443 	return err;
444 }
445 
446 struct add_branch_tip_arg {
447 	struct got_commit_graph_branch_tip *tips;
448 	int ntips;
449 	struct got_repository *repo;
450 	struct got_commit_graph *graph;
451 };
452 
453 static const struct got_error *
454 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
455 {
456 	const struct got_error *err;
457 	struct add_branch_tip_arg *a = arg;
458 	struct got_commit_graph_node *new_node;
459 	struct got_commit_object *commit;
460 
461 	err = got_object_open_as_commit(&commit, a->repo, commit_id);
462 	if (err)
463 		return err;
464 
465 	err = add_node(&new_node, a->graph, commit_id, a->repo);
466 	if (err) {
467 		got_object_commit_close(commit);
468 		return err;
469 	}
470 
471 	a->tips[a->ntips].commit_id = &new_node->id;
472 	a->tips[a->ntips].commit = commit;
473 	a->tips[a->ntips].new_node = new_node;
474 	a->ntips++;
475 
476 	return NULL;
477 }
478 
479 static const struct got_error *
480 fetch_commits_from_open_branches(struct got_commit_graph *graph,
481     struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
482 {
483 	const struct got_error *err;
484 	struct add_branch_tip_arg arg;
485 	int i, ntips;
486 
487 	ntips = got_object_idset_num_elements(graph->open_branches);
488 	if (ntips == 0)
489 		return NULL;
490 
491 	/* (Re-)allocate branch tips array if necessary. */
492 	if (graph->ntips < ntips) {
493 		struct got_commit_graph_branch_tip *tips;
494 		tips = recallocarray(graph->tips, graph->ntips, ntips,
495 		    sizeof(*tips));
496 		if (tips == NULL)
497 			return got_error_from_errno("recallocarray");
498 		graph->tips = tips;
499 		graph->ntips = ntips;
500 	}
501 	arg.tips = graph->tips;
502 	arg.ntips = 0; /* add_branch_tip() will increment */
503 	arg.repo = repo;
504 	arg.graph = graph;
505 	err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
506 	    &arg);
507 	if (err)
508 		goto done;
509 
510 	for (i = 0; i < arg.ntips; i++) {
511 		struct got_object_id *commit_id;
512 		struct got_commit_object *commit;
513 		struct got_commit_graph_node *new_node;
514 		int changed;
515 
516 		if (cancel_cb) {
517 			err = (*cancel_cb)(cancel_arg);
518 			if (err)
519 				break;
520 		}
521 
522 		commit_id = arg.tips[i].commit_id;
523 		commit = arg.tips[i].commit;
524 		new_node = arg.tips[i].new_node;
525 
526 		err = detect_changed_path(&changed, commit, commit_id,
527 		    graph->path, repo);
528 		if (err) {
529 			if (err->code != GOT_ERR_NO_OBJ)
530 				break;
531 			/*
532 			 * History of the path stops here on the current
533 			 * branch. Keep going on other branches.
534 			 */
535 			err = close_branch(graph, commit_id);
536 			if (err)
537 				break;
538 			continue;
539 		}
540 		if (changed) {
541 			add_node_to_iter_list(graph, new_node,
542 			    got_object_commit_get_committer_time(commit));
543 			arg.tips[i].new_node = NULL;
544 		}
545 		err = advance_branch(graph, commit_id, commit, repo);
546 		if (err)
547 			break;
548 	}
549 done:
550 	for (i = 0; i < arg.ntips; i++) {
551 		got_object_commit_close(arg.tips[i].commit);
552 		free(arg.tips[i].new_node);
553 	}
554 	return err;
555 }
556 
557 void
558 got_commit_graph_close(struct got_commit_graph *graph)
559 {
560 	struct got_commit_graph_node *node;
561 
562 	while ((node = TAILQ_FIRST(&graph->iter_list))) {
563 		TAILQ_REMOVE(&graph->iter_list, node, entry);
564 		free(node->more_parents);
565 		free(node);
566 	}
567 
568 	if (graph->open_branches)
569 		got_object_idset_free(graph->open_branches);
570 	if (graph->node_ids)
571 		got_object_idset_free(graph->node_ids);
572 	free(graph->tips);
573 	free(graph->path);
574 	free(graph);
575 }
576 
577 static const struct got_error *
578 remove_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
579 {
580 	struct got_object_idset *open_branches = arg;
581 
582 	return got_object_idset_remove(NULL, open_branches, commit_id);
583 }
584 
585 const struct got_error *
586 got_commit_graph_bfsort(struct got_commit_graph *graph,
587     struct got_object_id *id, struct got_repository *repo,
588     got_cancel_cb cancel_cb, void *cancel_arg)
589 {
590 	const struct got_error *err = NULL;
591 	struct got_commit_graph_node *node;
592 
593 	graph->flags &= ~GOT_COMMIT_GRAPH_TOPOSORT;
594 
595 	/* Clear left-over state from previous iteration attempts. */
596 	while ((node = TAILQ_FIRST(&graph->iter_list)))
597 		TAILQ_REMOVE(&graph->iter_list, node, entry);
598 	err = got_object_idset_for_each(graph->open_branches,
599 	    remove_branch_tip, graph->open_branches);
600 	if (err)
601 		return err;
602 
603 	err = got_object_idset_add(graph->open_branches, id, NULL);
604 	if (err)
605 		return err;
606 
607 	/* Locate first commit which changed graph->path. */
608 	while (TAILQ_EMPTY(&graph->iter_list) &&
609 	    got_object_idset_num_elements(graph->open_branches) > 0) {
610 		err = fetch_commits_from_open_branches(graph, repo,
611 		    cancel_cb, cancel_arg);
612 		if (err)
613 			return err;
614 	}
615 
616 	if (TAILQ_EMPTY(&graph->iter_list)) {
617 		const char *path;
618 		if (got_path_is_root_dir(graph->path))
619 			return got_error_no_obj(id);
620 		path = graph->path;
621 		while (path[0] == '/')
622 			path++;
623 		return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
624 	}
625 
626 	return NULL;
627 }
628 
629 const struct got_error *
630 got_commit_graph_iter_next(struct got_object_id *id,
631     struct got_commit_graph *graph, struct got_repository *repo,
632     got_cancel_cb cancel_cb, void *cancel_arg)
633 {
634 	const struct got_error *err = NULL;
635 	struct got_commit_graph_node *node, *pnode;
636 	int i;
637 
638 	node = TAILQ_FIRST(&graph->iter_list);
639 	if (node == NULL) {
640 		/* We are done iterating, or iteration was not started. */
641 		return got_error(GOT_ERR_ITER_COMPLETED);
642 	}
643 
644 	if (graph->flags & GOT_COMMIT_GRAPH_TOPOSORT) {
645 		/* At least one node with in-degree zero must exist. */
646 		while (node->indegree != 0)
647 			node = TAILQ_NEXT(node, entry);
648 	} else {
649 		while (TAILQ_NEXT(node, entry) == NULL &&
650 		    got_object_idset_num_elements(graph->open_branches) > 0) {
651 			err = fetch_commits_from_open_branches(graph, repo,
652 			    cancel_cb, cancel_arg);
653 			if (err)
654 				return err;
655 		}
656 	}
657 
658 	memcpy(id, &node->id, sizeof(*id));
659 
660 	TAILQ_REMOVE(&graph->iter_list, node, entry);
661 	if (graph->flags & GOT_COMMIT_GRAPH_TOPOSORT) {
662 		/* When visiting a commit decrement in-degree of all parents. */
663 		for (i = 0; i < node->nparents; i++) {
664 			if (i < nitems(node->parents))
665 				pnode = node->parents[i];
666 			else
667 				pnode = node->more_parents[i];
668 			pnode->indegree--;
669 		}
670 	}
671 	free(node);
672 	return NULL;
673 }
674 
675 static const struct got_error *
676 find_yca_add_id(struct got_object_id **yca_id, struct got_commit_graph *graph,
677     struct got_object_idset *commit_ids, struct got_repository *repo,
678     got_cancel_cb cancel_cb, void *cancel_arg)
679 {
680 	const struct got_error *err = NULL;
681 	struct got_object_id id;
682 
683 	err = got_commit_graph_iter_next(&id, graph, repo, cancel_cb,
684 	    cancel_arg);
685 	if (err)
686 		return err;
687 
688 	if (got_object_idset_contains(commit_ids, &id)) {
689 		*yca_id = got_object_id_dup(&id);
690 		if (*yca_id == NULL)
691 			err = got_error_from_errno("got_object_id_dup");
692 		return err;
693 	}
694 
695 	return got_object_idset_add(commit_ids, &id, NULL);
696 }
697 
698 /*
699  * Sets *yca_id to the youngest common ancestor of commit_id and
700  * commit_id2. Returns got_error(GOT_ERR_ANCESTRY) if they have no
701  * common ancestors.
702  *
703  * If first_parent_traversal is nonzero, only linear history is considered.
704  * If toposort is set then sort commits in topological order before
705  * traversing them.
706  */
707 const struct got_error *
708 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
709     struct got_object_id *commit_id, struct got_object_id *commit_id2,
710     int first_parent_traversal, int toposort, struct got_repository *repo,
711     got_cancel_cb cancel_cb, void *cancel_arg)
712 {
713 	const struct got_error *err = NULL;
714 	struct got_commit_graph *graph = NULL, *graph2 = NULL;
715 	int completed = 0, completed2 = 0;
716 	struct got_object_idset *commit_ids;
717 
718 	*yca_id = NULL;
719 
720 	commit_ids = got_object_idset_alloc();
721 	if (commit_ids == NULL)
722 		return got_error_from_errno("got_object_idset_alloc");
723 
724 	err = got_commit_graph_open(&graph, "/", first_parent_traversal);
725 	if (err)
726 		goto done;
727 
728 	err = got_commit_graph_open(&graph2, "/", first_parent_traversal);
729 	if (err)
730 		goto done;
731 
732 	if (toposort) {
733 		err = got_commit_graph_toposort(graph, commit_id, repo,
734 		    cancel_cb, cancel_arg);
735 		if (err)
736 			goto done;
737 
738 		err = got_commit_graph_toposort(graph2, commit_id2, repo,
739 		    cancel_cb, cancel_arg);
740 		if (err)
741 			goto done;
742 	} else {
743 		err = got_commit_graph_bfsort(graph, commit_id, repo,
744 		    cancel_cb, cancel_arg);
745 		if (err)
746 			goto done;
747 
748 		err = got_commit_graph_bfsort(graph2, commit_id2, repo,
749 		    cancel_cb, cancel_arg);
750 		if (err)
751 			goto done;
752 	}
753 
754 	for (;;) {
755 		if (cancel_cb) {
756 			err = (*cancel_cb)(cancel_arg);
757 			if (err)
758 				break;
759 		}
760 
761 		if (!completed) {
762 			err = find_yca_add_id(yca_id, graph, commit_ids, repo,
763 			    cancel_cb, cancel_arg);
764 			if (err) {
765 				if (err->code != GOT_ERR_ITER_COMPLETED)
766 					break;
767 				err = NULL;
768 				completed = 1;
769 			}
770 			if (*yca_id)
771 				break;
772 		}
773 
774 		if (!completed2) {
775 			err = find_yca_add_id(yca_id, graph2, commit_ids, repo,
776 			    cancel_cb, cancel_arg);
777 			if (err) {
778 				if (err->code != GOT_ERR_ITER_COMPLETED)
779 					break;
780 				err = NULL;
781 				completed2 = 1;
782 			}
783 			if (*yca_id)
784 				break;
785 		}
786 
787 		if (completed && completed2) {
788 			err = got_error(GOT_ERR_ANCESTRY);
789 			break;
790 		}
791 	}
792 done:
793 	got_object_idset_free(commit_ids);
794 	if (graph)
795 		got_commit_graph_close(graph);
796 	if (graph2)
797 		got_commit_graph_close(graph2);
798 	return err;
799 }
800 
801 /*
802  * Sort the graph for traversal in topological order.
803  *
804  * This implementation is based on the description of topological sorting
805  * of git commits by Derrick Stolee at
806  * https://github.blog/2022-08-30-gits-database-internals-ii-commit-history-queries/#topological-sorting
807  * which reads as follows:
808  *
809  * The basic algorithm for topological sorting is Kahn’s algorithm which
810  * follows two big steps:
811  * 1. Walk all reachable commits, counting the number of times a commit appears
812  * as a parent of another commit. Call these numbers the in-degree of the
813  * commit, referencing the number of incoming edges.
814  * 2. Walk the reachable commits, but only visit a commit if its in-degree
815  * value is zero. When visiting a commit, decrement the in-degree value of
816  * each parent.
817  * 
818  * This algorithm works because at least one of our starting points will
819  * have in-degree zero, and then decrementing the in-degree value is similar
820  * to deleting the commit from the graph, always having at least one commit
821  * with in-degree zero.
822  */
823 const struct got_error *
824 got_commit_graph_toposort(struct got_commit_graph *graph,
825     struct got_object_id *id, struct got_repository *repo,
826     got_cancel_cb cancel_cb, void *cancel_arg)
827 {
828 	const struct got_error *err = NULL;
829 	struct got_commit_graph_node *node = NULL, *pnode = NULL;
830 	struct got_commit_object *commit = NULL;
831 	struct got_object_id_queue commits;
832 	const struct got_object_id_queue *parent_ids;
833 	struct got_object_qid *qid = NULL, *pid;
834 	int i;
835 
836 	STAILQ_INIT(&commits);
837 
838 	if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL)
839 		return got_commit_graph_bfsort(graph, id, repo,
840 		    cancel_cb, cancel_arg);
841 
842 	/* Clear left-over state from previous iteration attempts. */
843 	while ((node = TAILQ_FIRST(&graph->iter_list)))
844 		TAILQ_REMOVE(&graph->iter_list, node, entry);
845 	err = got_object_idset_for_each(graph->open_branches,
846 	    remove_branch_tip, graph->open_branches);
847 	if (err)
848 		return err;
849 
850 	graph->flags |= GOT_COMMIT_GRAPH_TOPOSORT;
851 
852 	/*
853 	 * Sorting the commit graph in topological order requires visiting
854 	 * every reachable commit. This is very expensive but there are
855 	 * ways to speed this up significantly in the future:
856 	 * 1) Run this loop in got-read-pack if possible.
857 	 * 2) Use Git's commit-graph file to compute the result incrementally.
858 	 *    See the blog post linked above for details.
859 	 */
860 	err = got_object_qid_alloc_partial(&qid);
861 	if (err)
862 		return err;
863 	memcpy(&qid->id, id, sizeof(qid->id));
864 	STAILQ_INSERT_TAIL(&commits, qid, entry);
865 	while (!STAILQ_EMPTY(&commits)) {
866 		if (cancel_cb) {
867 			err = (*cancel_cb)(cancel_arg);
868 			if (err)
869 				break;
870 		}
871 
872 		qid = STAILQ_FIRST(&commits);
873 		STAILQ_REMOVE_HEAD(&commits, entry);
874 		err = got_object_open_as_commit(&commit, repo, &qid->id);
875 		if (err)
876 			break;
877 
878 		node = got_object_idset_get(graph->node_ids, &qid->id);
879 		if (node == NULL) {
880 			err = add_node(&node, graph, id, repo);
881 			if (err)
882 				break;
883 			TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
884 		}
885 
886 		got_object_qid_free(qid);
887 		qid = NULL;
888 
889 		if (node->timestamp != 0) /* already traversed once */
890 			continue;
891 
892 		if (node->nparents == -1) {
893 			node->nparents = got_object_commit_get_nparents(commit);
894 			if (node->nparents > nitems(node->parents)) {
895 				node->more_parents = calloc(node->nparents,
896 				    sizeof(*node->more_parents));
897 				if (node->more_parents == NULL) {
898 					err = got_error_from_errno("calloc");
899 					break;
900 				}
901 			}
902 
903 		}
904 
905 		node->timestamp = got_object_commit_get_committer_time(commit);
906 		parent_ids = got_object_commit_get_parent_ids(commit);
907 		i = 0;
908 		STAILQ_FOREACH(pid, parent_ids, entry) {
909 			if (cancel_cb) {
910 				err = (*cancel_cb)(cancel_arg);
911 				if (err)
912 					goto done;
913 			}
914 
915 			/*
916 			 * Increment the in-degree counter every time a given
917 			 * commit appears as the parent of another commit.
918 			 */
919 			pnode = got_object_idset_get(graph->node_ids, &pid->id);
920 			if (pnode == NULL) {
921 				err = add_node(&pnode, graph, &pid->id, repo);
922 				if (err)
923 					goto done;
924 				TAILQ_INSERT_TAIL(&graph->iter_list, pnode,
925 				    entry);
926 			}
927 			pnode->indegree++;
928 
929 			/*
930 			 * Cache parent pointers on the node to make future
931 			 * in-degree updates easier.
932 			 */
933 			if (node->nparents <= nitems(node->parents)) {
934 				node->parents[i] = pnode;
935 			} else {
936 				node->more_parents[i] = pnode;
937 				if (i < nitems(node->parents))
938 					node->parents[i] = pnode;
939 			}
940 			i++;
941 
942 			/* Keep traversing through all parent commits. */
943 			err = got_object_qid_alloc_partial(&qid);
944 			if (err)
945 				goto done;
946 			memcpy(&qid->id, &pid->id, sizeof(qid->id));
947 			STAILQ_INSERT_TAIL(&commits, qid, entry);
948 			qid = NULL;
949 		}
950 
951 		got_object_commit_close(commit);
952 		commit = NULL;
953 	}
954 done:
955 	if (commit)
956 		got_object_commit_close(commit);
957 	got_object_qid_free(qid);
958 	got_object_id_queue_free(&commits);
959 	return err;
960 }