2 * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
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.
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.
17 #include <sys/types.h>
19 #include <sys/queue.h>
20 #include <sys/stdint.h>
30 #include "got_error.h"
31 #include "got_object.h"
32 #include "got_cancel.h"
33 #include "got_commit_graph.h"
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"
41 struct got_commit_graph_node {
42 struct got_object_id id;
44 /* Used only during iteration. */
46 TAILQ_ENTRY(got_commit_graph_node) entry;
49 TAILQ_HEAD(got_commit_graph_iter_list, got_commit_graph_node);
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;
57 struct got_commit_graph {
58 /* The set of all commits we have traversed. */
59 struct got_object_idset *node_ids;
62 #define GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL 0x01
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.
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.
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.
79 struct got_object_idset *open_branches;
81 /* Array of branch tips for fetch_commits_from_open_branches(). */
82 struct got_commit_graph_branch_tip *tips;
85 /* Path of tree entry of interest to the API user. */
89 * Nodes which will be passed to the API user next, sorted by
92 struct got_commit_graph_iter_list iter_list;
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)
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;
105 if (got_path_is_root_dir(path)) {
112 pid = STAILQ_FIRST(&commit->parent_ids);
114 struct got_object_id *obj_id;
115 err = got_object_id_by_path(&obj_id, repo, commit, path);
117 if (err->code == GOT_ERR_NO_TREE_ENTRY)
120 *changed = 1; /* The path was created in this commit. */
125 err = got_object_open_as_tree(&tree, repo, commit->tree_id);
129 err = got_object_open_as_commit(&pcommit, repo, pid->id);
133 err = got_object_open_as_tree(&ptree, repo, pcommit->tree_id);
137 err = got_object_tree_path_changed(changed, tree, ptree, path, repo);
140 got_object_tree_close(tree);
142 got_object_tree_close(ptree);
144 got_object_commit_close(pcommit);
149 add_node_to_iter_list(struct got_commit_graph *graph,
150 struct got_commit_graph_node *node, time_t committer_time)
152 struct got_commit_graph_node *n, *next;
154 node->timestamp = committer_time;
156 n = TAILQ_FIRST(&graph->iter_list);
158 next = TAILQ_NEXT(n, entry);
159 if (next && node->timestamp >= next->timestamp) {
160 TAILQ_INSERT_BEFORE(next, node, entry);
165 TAILQ_INSERT_TAIL(&graph->iter_list, node, entry);
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)
173 const struct got_error *err = NULL;
174 struct got_commit_graph_node *node;
178 node = calloc(1, sizeof(*node));
180 return got_error_from_errno("calloc");
182 memcpy(&node->id, commit_id, sizeof(node->id));
183 err = got_object_idset_add(graph->node_ids, &node->id, NULL);
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.
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)
202 const struct got_error *err = NULL;
203 struct got_object_id_queue traversed_commits;
204 struct got_object_qid *qid;
206 STAILQ_INIT(&traversed_commits);
207 *ncommits_traversed = 0;
209 err = got_traverse_packed_commits(&traversed_commits,
210 commit_id, graph->path, repo);
214 /* Add all traversed commits to the graph... */
215 STAILQ_FOREACH(qid, &traversed_commits, entry) {
216 struct got_commit_graph_node *node;
218 if (got_object_idset_contains(graph->open_branches, qid->id))
220 if (got_object_idset_contains(graph->node_ids, qid->id))
223 (*ncommits_traversed)++;
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,
232 err = add_node(&node, graph, qid->id, repo);
237 got_object_id_queue_free(&traversed_commits);
241 static const struct got_error *
242 close_branch(struct got_commit_graph *graph, struct got_object_id *commit_id)
244 const struct got_error *err;
246 err = got_object_idset_remove(NULL, graph->open_branches, commit_id);
247 if (err && err->code != GOT_ERR_NO_OBJ)
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)
256 const struct got_error *err;
257 struct got_object_qid *qid;
259 err = close_branch(graph, commit_id);
263 if (graph->flags & GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL) {
264 qid = STAILQ_FIRST(&commit->parent_ids);
266 got_object_idset_contains(graph->open_branches, qid->id))
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'.
276 if (!got_path_is_root_dir(graph->path) &&
277 (commit->flags & GOT_COMMIT_FLAG_PACKED)) {
279 err = packed_first_parent_traversal(&ncommits,
280 graph, qid->id, repo);
281 if (err || ncommits > 0)
284 return got_object_idset_add(graph->open_branches,
289 * If we are graphing commits for a specific path, skip branches
290 * which do not contribute any content to this path.
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;
296 err = got_object_id_by_path(&merged_id, repo, commit,
301 STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
302 struct got_object_id *id = NULL;
303 struct got_commit_object *pcommit = NULL;
305 if (got_object_idset_contains(graph->open_branches,
309 err = got_object_open_as_commit(&pcommit, repo,
316 err = got_object_id_by_path(&id, repo, pcommit,
318 got_object_commit_close(pcommit);
321 if (err->code == GOT_ERR_NO_TREE_ENTRY) {
331 if (!branches_differ &&
332 got_object_id_cmp(id, prev_id) != 0)
339 * If a branch has created the merged content we can
340 * skip any other branches.
342 if (got_object_id_cmp(merged_id, id) == 0) {
343 err = got_object_idset_add(graph->open_branches,
357 * If the path's content is the same on all branches,
358 * follow the first parent only.
360 if (!branches_differ) {
361 qid = STAILQ_FIRST(&commit->parent_ids);
364 if (got_object_idset_contains(graph->open_branches,
367 if (got_object_idset_contains(graph->node_ids,
369 return NULL; /* parent already traversed */
370 return got_object_idset_add(graph->open_branches,
375 STAILQ_FOREACH(qid, &commit->parent_ids, entry) {
376 if (got_object_idset_contains(graph->open_branches, qid->id))
378 if (got_object_idset_contains(graph->node_ids, qid->id))
379 continue; /* parent already traversed */
380 err = got_object_idset_add(graph->open_branches, qid->id, NULL);
388 const struct got_error *
389 got_commit_graph_open(struct got_commit_graph **graph,
390 const char *path, int first_parent_traversal)
392 const struct got_error *err = NULL;
394 *graph = calloc(1, sizeof(**graph));
396 return got_error_from_errno("calloc");
398 TAILQ_INIT(&(*graph)->iter_list);
400 (*graph)->path = strdup(path);
401 if ((*graph)->path == NULL) {
402 err = got_error_from_errno("strdup");
406 (*graph)->node_ids = got_object_idset_alloc();
407 if ((*graph)->node_ids == NULL) {
408 err = got_error_from_errno("got_object_idset_alloc");
412 (*graph)->open_branches = got_object_idset_alloc();
413 if ((*graph)->open_branches == NULL) {
414 err = got_error_from_errno("got_object_idset_alloc");
418 if (first_parent_traversal)
419 (*graph)->flags |= GOT_COMMIT_GRAPH_FIRST_PARENT_TRAVERSAL;
422 got_commit_graph_close(*graph);
428 struct add_branch_tip_arg {
429 struct got_commit_graph_branch_tip *tips;
431 struct got_repository *repo;
432 struct got_commit_graph *graph;
435 static const struct got_error *
436 add_branch_tip(struct got_object_id *commit_id, void *data, void *arg)
438 const struct got_error *err;
439 struct add_branch_tip_arg *a = arg;
440 struct got_commit_graph_node *new_node;
441 struct got_commit_object *commit;
443 err = got_object_open_as_commit(&commit, a->repo, commit_id);
447 err = add_node(&new_node, a->graph, commit_id, a->repo);
451 a->tips[a->ntips].commit_id = new_node ? &new_node->id : NULL;
452 a->tips[a->ntips].commit = commit;
453 a->tips[a->ntips].new_node = new_node;
459 static const struct got_error *
460 fetch_commits_from_open_branches(struct got_commit_graph *graph,
461 struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
463 const struct got_error *err;
464 struct add_branch_tip_arg arg;
467 ntips = got_object_idset_num_elements(graph->open_branches);
471 /* (Re-)allocate branch tips array if necessary. */
472 if (graph->ntips < ntips) {
473 struct got_commit_graph_branch_tip *tips;
474 tips = recallocarray(graph->tips, graph->ntips, ntips,
477 return got_error_from_errno("recallocarray");
479 graph->ntips = ntips;
481 arg.tips = graph->tips;
482 arg.ntips = 0; /* add_branch_tip() will increment */
485 err = got_object_idset_for_each(graph->open_branches, add_branch_tip,
490 for (i = 0; i < arg.ntips; i++) {
491 struct got_object_id *commit_id;
492 struct got_commit_object *commit;
493 struct got_commit_graph_node *new_node;
497 err = (*cancel_cb)(cancel_arg);
502 commit_id = arg.tips[i].commit_id;
503 commit = arg.tips[i].commit;
504 new_node = arg.tips[i].new_node;
506 err = detect_changed_path(&changed, commit, commit_id,
509 if (err->code != GOT_ERR_NO_OBJ)
512 * History of the path stops here on the current
513 * branch. Keep going on other branches.
515 err = close_branch(graph, commit_id);
521 add_node_to_iter_list(graph, new_node,
522 got_object_commit_get_committer_time(commit));
523 err = advance_branch(graph, commit_id, commit, repo);
528 for (i = 0; i < arg.ntips; i++)
529 got_object_commit_close(arg.tips[i].commit);
534 got_commit_graph_close(struct got_commit_graph *graph)
536 if (graph->open_branches)
537 got_object_idset_free(graph->open_branches);
539 got_object_idset_free(graph->node_ids);
545 const struct got_error *
546 got_commit_graph_iter_start(struct got_commit_graph *graph,
547 struct got_object_id *id, struct got_repository *repo,
548 got_cancel_cb cancel_cb, void *cancel_arg)
550 const struct got_error *err = NULL;
552 if (!TAILQ_EMPTY(&graph->iter_list))
553 return got_error(GOT_ERR_ITER_BUSY);
555 err = got_object_idset_add(graph->open_branches, id, NULL);
559 /* Locate first commit which changed graph->path. */
560 while (TAILQ_EMPTY(&graph->iter_list) &&
561 got_object_idset_num_elements(graph->open_branches) > 0) {
562 err = fetch_commits_from_open_branches(graph, repo,
563 cancel_cb, cancel_arg);
568 if (TAILQ_EMPTY(&graph->iter_list)) {
570 if (got_path_is_root_dir(graph->path))
571 return got_error_no_obj(id);
573 while (path[0] == '/')
575 return got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
581 const struct got_error *
582 got_commit_graph_iter_next(struct got_object_id **id,
583 struct got_commit_graph *graph, struct got_repository *repo,
584 got_cancel_cb cancel_cb, void *cancel_arg)
586 const struct got_error *err = NULL;
587 struct got_commit_graph_node *node;
591 node = TAILQ_FIRST(&graph->iter_list);
593 /* We are done iterating, or iteration was not started. */
594 return got_error(GOT_ERR_ITER_COMPLETED);
597 while (TAILQ_NEXT(node, entry) == NULL &&
598 got_object_idset_num_elements(graph->open_branches) > 0) {
599 err = fetch_commits_from_open_branches(graph, repo,
600 cancel_cb, cancel_arg);
606 TAILQ_REMOVE(&graph->iter_list, node, entry);
610 const struct got_error *
611 got_commit_graph_find_youngest_common_ancestor(struct got_object_id **yca_id,
612 struct got_object_id *commit_id, struct got_object_id *commit_id2,
613 int first_parent_traversal,
614 struct got_repository *repo, got_cancel_cb cancel_cb, void *cancel_arg)
616 const struct got_error *err = NULL;
617 struct got_commit_graph *graph = NULL, *graph2 = NULL;
618 int completed = 0, completed2 = 0;
619 struct got_object_idset *commit_ids;
623 commit_ids = got_object_idset_alloc();
624 if (commit_ids == NULL)
625 return got_error_from_errno("got_object_idset_alloc");
627 err = got_commit_graph_open(&graph, "/", first_parent_traversal);
631 err = got_commit_graph_open(&graph2, "/", first_parent_traversal);
635 err = got_commit_graph_iter_start(graph, commit_id, repo,
636 cancel_cb, cancel_arg);
640 err = got_commit_graph_iter_start(graph2, commit_id2, repo,
641 cancel_cb, cancel_arg);
646 struct got_object_id *id = NULL, *id2 = NULL;
649 err = (*cancel_cb)(cancel_arg);
655 err = got_commit_graph_iter_next(&id, graph, repo,
656 cancel_cb, cancel_arg);
658 if (err->code != GOT_ERR_ITER_COMPLETED)
666 err = got_commit_graph_iter_next(&id2, graph2, repo,
667 cancel_cb, cancel_arg);
669 if (err->code != GOT_ERR_ITER_COMPLETED)
677 if (got_object_idset_contains(commit_ids, id)) {
678 *yca_id = got_object_id_dup(id);
681 err = got_error_from_errno("got_object_id_dup");
685 err = got_object_idset_add(commit_ids, id, NULL);
690 if (got_object_idset_contains(commit_ids, id2)) {
691 *yca_id = got_object_id_dup(id2);
694 err = got_error_from_errno("got_object_id_dup");
698 err = got_object_idset_add(commit_ids, id2, NULL);
703 if (completed && completed2) {
704 err = got_error(GOT_ERR_ANCESTRY);
710 got_object_idset_free(commit_ids);
712 got_commit_graph_close(graph);
714 got_commit_graph_close(graph2);