bisect.c 15.6 KB
Newer Older
1 2 3 4
#include "cache.h"
#include "commit.h"
#include "diff.h"
#include "revision.h"
5 6
#include "refs.h"
#include "list-objects.h"
7
#include "quote.h"
8
#include "sha1-lookup.h"
9
#include "run-command.h"
10 11
#include "bisect.h"

12 13 14 15
struct sha1_array {
	unsigned char (*sha1)[20];
	int sha1_nr;
	int sha1_alloc;
16
	int sorted;
17 18
};

19
static struct sha1_array good_revs;
20
static struct sha1_array skipped_revs;
21

22 23
static const unsigned char *current_bad_sha1;

24 25 26 27 28 29
struct argv_array {
	const char **argv;
	int argv_nr;
	int argv_alloc;
};

30 31 32 33
static const char *argv_diff_tree[] = {"diff-tree", "--pretty", NULL, NULL};
static const char *argv_checkout[] = {"checkout", "-q", NULL, "--", NULL};
static const char *argv_show_branch[] = {"show-branch", NULL, NULL};

34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
/* bits #0-15 in revision.h */

#define COUNTED		(1u<<16)

/*
 * This is a truly stupid algorithm, but it's only
 * used for bisection, and we just don't care enough.
 *
 * We care just barely enough to avoid recursing for
 * non-merge entries.
 */
static int count_distance(struct commit_list *entry)
{
	int nr = 0;

	while (entry) {
		struct commit *commit = entry->item;
		struct commit_list *p;

		if (commit->object.flags & (UNINTERESTING | COUNTED))
			break;
		if (!(commit->object.flags & TREESAME))
			nr++;
		commit->object.flags |= COUNTED;
		p = commit->parents;
		entry = p;
		if (p) {
			p = p->next;
			while (p) {
				nr += count_distance(p);
				p = p->next;
			}
		}
	}

	return nr;
}

static void clear_distance(struct commit_list *list)
{
	while (list) {
		struct commit *commit = list->item;
		commit->object.flags &= ~COUNTED;
		list = list->next;
	}
}

#define DEBUG_BISECT 0

static inline int weight(struct commit_list *elem)
{
	return *((int*)(elem->item->util));
}

static inline void weight_set(struct commit_list *elem, int weight)
{
	*((int*)(elem->item->util)) = weight;
}

static int count_interesting_parents(struct commit *commit)
{
	struct commit_list *p;
	int count;

	for (count = 0, p = commit->parents; p; p = p->next) {
		if (p->item->object.flags & UNINTERESTING)
			continue;
		count++;
	}
	return count;
}

static inline int halfway(struct commit_list *p, int nr)
{
	/*
	 * Don't short-cut something we are not going to return!
	 */
	if (p->item->object.flags & TREESAME)
		return 0;
	if (DEBUG_BISECT)
		return 0;
	/*
	 * 2 and 3 are halfway of 5.
	 * 3 is halfway of 6 but 2 and 4 are not.
	 */
	switch (2 * weight(p) - nr) {
	case -1: case 0: case 1:
		return 1;
	default:
		return 0;
	}
}

#if !DEBUG_BISECT
#define show_list(a,b,c,d) do { ; } while (0)
#else
static void show_list(const char *debug, int counted, int nr,
		      struct commit_list *list)
{
	struct commit_list *p;

	fprintf(stderr, "%s (%d/%d)\n", debug, counted, nr);

	for (p = list; p; p = p->next) {
		struct commit_list *pp;
		struct commit *commit = p->item;
		unsigned flags = commit->object.flags;
		enum object_type type;
		unsigned long size;
		char *buf = read_sha1_file(commit->object.sha1, &type, &size);
		char *ep, *sp;

		fprintf(stderr, "%c%c%c ",
			(flags & TREESAME) ? ' ' : 'T',
			(flags & UNINTERESTING) ? 'U' : ' ',
			(flags & COUNTED) ? 'C' : ' ');
		if (commit->util)
			fprintf(stderr, "%3d", weight(p));
		else
			fprintf(stderr, "---");
		fprintf(stderr, " %.*s", 8, sha1_to_hex(commit->object.sha1));
		for (pp = commit->parents; pp; pp = pp->next)
			fprintf(stderr, " %.*s", 8,
				sha1_to_hex(pp->item->object.sha1));

		sp = strstr(buf, "\n\n");
		if (sp) {
			sp += 2;
			for (ep = sp; *ep && *ep != '\n'; ep++)
				;
			fprintf(stderr, " %.*s", (int)(ep - sp), sp);
		}
		fprintf(stderr, "\n");
	}
}
#endif /* DEBUG_BISECT */

static struct commit_list *best_bisection(struct commit_list *list, int nr)
{
	struct commit_list *p, *best;
	int best_distance = -1;

	best = list;
	for (p = list; p; p = p->next) {
		int distance;
		unsigned flags = p->item->object.flags;

		if (flags & TREESAME)
			continue;
		distance = weight(p);
		if (nr - distance < distance)
			distance = nr - distance;
		if (distance > best_distance) {
			best = p;
			best_distance = distance;
		}
	}

	return best;
}

struct commit_dist {
	struct commit *commit;
	int distance;
};

static int compare_commit_dist(const void *a_, const void *b_)
{
	struct commit_dist *a, *b;

	a = (struct commit_dist *)a_;
	b = (struct commit_dist *)b_;
	if (a->distance != b->distance)
		return b->distance - a->distance; /* desc sort */
	return hashcmp(a->commit->object.sha1, b->commit->object.sha1);
}

static struct commit_list *best_bisection_sorted(struct commit_list *list, int nr)
{
	struct commit_list *p;
	struct commit_dist *array = xcalloc(nr, sizeof(*array));
	int cnt, i;

	for (p = list, cnt = 0; p; p = p->next) {
		int distance;
		unsigned flags = p->item->object.flags;

		if (flags & TREESAME)
			continue;
		distance = weight(p);
		if (nr - distance < distance)
			distance = nr - distance;
		array[cnt].commit = p->item;
		array[cnt].distance = distance;
		cnt++;
	}
	qsort(array, cnt, sizeof(*array), compare_commit_dist);
	for (p = list, i = 0; i < cnt; i++) {
		struct name_decoration *r = xmalloc(sizeof(*r) + 100);
		struct object *obj = &(array[i].commit->object);

		sprintf(r->name, "dist=%d", array[i].distance);
		r->next = add_decoration(&name_decoration, obj, r);
		p->item = array[i].commit;
		p = p->next;
	}
	if (p)
		p->next = NULL;
	free(array);
	return list;
}

/*
 * zero or positive weight is the number of interesting commits it can
 * reach, including itself.  Especially, weight = 0 means it does not
 * reach any tree-changing commits (e.g. just above uninteresting one
 * but traversal is with pathspec).
 *
 * weight = -1 means it has one parent and its distance is yet to
 * be computed.
 *
 * weight = -2 means it has more than one parent and its distance is
 * unknown.  After running count_distance() first, they will get zero
 * or positive distance.
 */
static struct commit_list *do_find_bisection(struct commit_list *list,
					     int nr, int *weights,
					     int find_all)
{
	int n, counted;
	struct commit_list *p;

	counted = 0;

	for (n = 0, p = list; p; p = p->next) {
		struct commit *commit = p->item;
		unsigned flags = commit->object.flags;

		p->item->util = &weights[n++];
		switch (count_interesting_parents(commit)) {
		case 0:
			if (!(flags & TREESAME)) {
				weight_set(p, 1);
				counted++;
				show_list("bisection 2 count one",
					  counted, nr, list);
			}
			/*
			 * otherwise, it is known not to reach any
			 * tree-changing commit and gets weight 0.
			 */
			break;
		case 1:
			weight_set(p, -1);
			break;
		default:
			weight_set(p, -2);
			break;
		}
	}

	show_list("bisection 2 initialize", counted, nr, list);

	/*
	 * If you have only one parent in the resulting set
	 * then you can reach one commit more than that parent
	 * can reach.  So we do not have to run the expensive
	 * count_distance() for single strand of pearls.
	 *
	 * However, if you have more than one parents, you cannot
	 * just add their distance and one for yourself, since
	 * they usually reach the same ancestor and you would
	 * end up counting them twice that way.
	 *
	 * So we will first count distance of merges the usual
	 * way, and then fill the blanks using cheaper algorithm.
	 */
	for (p = list; p; p = p->next) {
		if (p->item->object.flags & UNINTERESTING)
			continue;
		if (weight(p) != -2)
			continue;
		weight_set(p, count_distance(p));
		clear_distance(list);

		/* Does it happen to be at exactly half-way? */
		if (!find_all && halfway(p, nr))
			return p;
		counted++;
	}

	show_list("bisection 2 count_distance", counted, nr, list);

	while (counted < nr) {
		for (p = list; p; p = p->next) {
			struct commit_list *q;
			unsigned flags = p->item->object.flags;

			if (0 <= weight(p))
				continue;
			for (q = p->item->parents; q; q = q->next) {
				if (q->item->object.flags & UNINTERESTING)
					continue;
				if (0 <= weight(q))
					break;
			}
			if (!q)
				continue;

			/*
			 * weight for p is unknown but q is known.
			 * add one for p itself if p is to be counted,
			 * otherwise inherit it from q directly.
			 */
			if (!(flags & TREESAME)) {
				weight_set(p, weight(q)+1);
				counted++;
				show_list("bisection 2 count one",
					  counted, nr, list);
			}
			else
				weight_set(p, weight(q));

			/* Does it happen to be at exactly half-way? */
			if (!find_all && halfway(p, nr))
				return p;
		}
	}

	show_list("bisection 2 counted all", counted, nr, list);

	if (!find_all)
		return best_bisection(list, nr);
	else
		return best_bisection_sorted(list, nr);
}

struct commit_list *find_bisection(struct commit_list *list,
					  int *reaches, int *all,
					  int find_all)
{
	int nr, on_list;
	struct commit_list *p, *best, *next, *last;
	int *weights;

	show_list("bisection 2 entry", 0, 0, list);

	/*
	 * Count the number of total and tree-changing items on the
	 * list, while reversing the list.
	 */
	for (nr = on_list = 0, last = NULL, p = list;
	     p;
	     p = next) {
		unsigned flags = p->item->object.flags;

		next = p->next;
		if (flags & UNINTERESTING)
			continue;
		p->next = last;
		last = p;
		if (!(flags & TREESAME))
			nr++;
		on_list++;
	}
	list = last;
	show_list("bisection 2 sorted", 0, nr, list);

	*all = nr;
	weights = xcalloc(on_list, sizeof(*weights));

	/* Do the real work of finding bisection commit. */
	best = do_find_bisection(list, nr, weights, find_all);
	if (best) {
		if (!find_all)
			best->next = NULL;
		*reaches = weight(best);
	}
	free(weights);
	return best;
}

416
static void argv_array_push(struct argv_array *array, const char *string)
417
{
418 419 420
	ALLOC_GROW(array->argv, array->argv_nr + 1, array->argv_alloc);
	array->argv[array->argv_nr++] = string;
}
421

422 423 424 425 426
static void argv_array_push_sha1(struct argv_array *array,
				 const unsigned char *sha1,
				 const char *format)
{
	struct strbuf buf = STRBUF_INIT;
427
	strbuf_addf(&buf, format, sha1_to_hex(sha1));
428
	argv_array_push(array, strbuf_detach(&buf, NULL));
429 430
}

431 432 433 434 435 436 437
static void sha1_array_push(struct sha1_array *array,
			    const unsigned char *sha1)
{
	ALLOC_GROW(array->sha1, array->sha1_nr + 1, array->sha1_alloc);
	hashcpy(array->sha1[array->sha1_nr++], sha1);
}

438 439 440 441
static int register_ref(const char *refname, const unsigned char *sha1,
			int flags, void *cb_data)
{
	if (!strcmp(refname, "bad")) {
442
		current_bad_sha1 = sha1;
443
	} else if (!prefixcmp(refname, "good-")) {
444
		sha1_array_push(&good_revs, sha1);
445
	} else if (!prefixcmp(refname, "skip-")) {
446
		sha1_array_push(&skipped_revs, sha1);
447 448 449 450 451 452 453 454 455 456
	}

	return 0;
}

static int read_bisect_refs(void)
{
	return for_each_ref_in("refs/bisect/", register_ref, NULL);
}

457
void read_bisect_paths(struct argv_array *array)
458 459 460 461 462 463 464 465 466 467 468 469 470 471
{
	struct strbuf str = STRBUF_INIT;
	const char *filename = git_path("BISECT_NAMES");
	FILE *fp = fopen(filename, "r");

	if (!fp)
		die("Could not open file '%s': %s", filename, strerror(errno));

	while (strbuf_getline(&str, fp, '\n') != EOF) {
		char *quoted;
		int res;

		strbuf_trim(&str);
		quoted = strbuf_detach(&str, NULL);
472 473
		res = sq_dequote_to_argv(quoted, &array->argv,
					 &array->argv_nr, &array->argv_alloc);
474 475 476 477 478 479 480 481 482
		if (res)
			die("Badly quoted content in file '%s': %s",
			    filename, quoted);
	}

	strbuf_release(&str);
	fclose(fp);
}

483
static int array_cmp(const void *a, const void *b)
484 485 486 487
{
	return hashcmp(a, b);
}

488
static void sort_sha1_array(struct sha1_array *array)
489
{
490
	qsort(array->sha1, array->sha1_nr, sizeof(*array->sha1), array_cmp);
491 492

	array->sorted = 1;
493 494
}

495
static const unsigned char *sha1_access(size_t index, void *table)
496
{
497 498
	unsigned char (*array)[20] = table;
	return array[index];
499 500
}

501 502
static int lookup_sha1_array(struct sha1_array *array,
			     const unsigned char *sha1)
503
{
504 505 506
	if (!array->sorted)
		sort_sha1_array(array);

507
	return sha1_pos(sha1, array->sha1, array->sha1_nr, sha1_access);
508 509 510 511 512 513 514 515 516 517
}

struct commit_list *filter_skipped(struct commit_list *list,
				   struct commit_list **tried,
				   int show_all)
{
	struct commit_list *filtered = NULL, **f = &filtered;

	*tried = NULL;

518
	if (!skipped_revs.sha1_nr)
519 520 521 522 523
		return list;

	while (list) {
		struct commit_list *next = list->next;
		list->next = NULL;
524 525
		if (0 <= lookup_sha1_array(&skipped_revs,
					   list->item->object.sha1)) {
526 527 528 529 530 531 532 533 534 535 536 537 538 539 540
			/* Move current to tried list */
			*tried = list;
			tried = &list->next;
		} else {
			if (!show_all)
				return list;
			/* Move current to filtered list */
			*f = list;
			f = &list->next;
		}
		list = next;
	}

	return filtered;
}
541

542
static void bisect_rev_setup(struct rev_info *revs, const char *prefix)
543
{
544
	struct argv_array rev_argv = { NULL, 0, 0 };
545 546
	int i;

547 548 549
	init_revisions(revs, prefix);
	revs->abbrev = 0;
	revs->commit_format = CMIT_FMT_UNSPECIFIED;
550

551 552 553
	/* rev_argv.argv[0] will be ignored by setup_revisions */
	argv_array_push(&rev_argv, xstrdup("bisect_rev_setup"));
	argv_array_push_sha1(&rev_argv, current_bad_sha1, "%s");
554
	for (i = 0; i < good_revs.sha1_nr; i++)
555 556 557 558
		argv_array_push_sha1(&rev_argv, good_revs.sha1[i], "^%s");
	argv_array_push(&rev_argv, xstrdup("--"));
	read_bisect_paths(&rev_argv);
	argv_array_push(&rev_argv, NULL);
559

560
	setup_revisions(rev_argv.argv_nr, rev_argv.argv, revs, NULL);
561 562 563
	revs->limited = 1;
}

564
static void bisect_common(struct rev_info *revs, int *reaches, int *all)
565 566 567 568 569 570 571
{
	if (prepare_revision_walk(revs))
		die("revision walk setup failed");
	if (revs->tree_objects)
		mark_edges_uninteresting(revs->commits, revs, NULL);

	revs->commits = find_bisection(revs->commits, reaches, all,
572
				       !!skipped_revs.sha1_nr);
573 574
}

575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635
static void exit_if_skipped_commits(struct commit_list *tried,
				    const unsigned char *bad)
{
	if (!tried)
		return;

	printf("There are only 'skip'ped commits left to test.\n"
	       "The first bad commit could be any of:\n");
	print_commit_list(tried, "%s\n", "%s\n");
	if (bad)
		printf("%s\n", sha1_to_hex(bad));
	printf("We cannot bisect more!\n");
	exit(2);
}

static void mark_expected_rev(char *bisect_rev_hex)
{
	int len = strlen(bisect_rev_hex);
	const char *filename = git_path("BISECT_EXPECTED_REV");
	int fd = open(filename, O_CREAT | O_TRUNC | O_WRONLY, 0600);

	if (fd < 0)
		die("could not create file '%s': %s",
		    filename, strerror(errno));

	bisect_rev_hex[len] = '\n';
	write_or_die(fd, bisect_rev_hex, len + 1);
	bisect_rev_hex[len] = '\0';

	if (close(fd) < 0)
		die("closing file %s: %s", filename, strerror(errno));
}

static int bisect_checkout(char *bisect_rev_hex)
{
	int res;

	mark_expected_rev(bisect_rev_hex);

	argv_checkout[2] = bisect_rev_hex;
	res = run_command_v_opt(argv_checkout, RUN_GIT_CMD);
	if (res)
		exit(res);

	argv_show_branch[1] = bisect_rev_hex;
	return run_command_v_opt(argv_show_branch, RUN_GIT_CMD);
}

/*
 * We use the convention that exiting with an exit code 10 means that
 * the bisection process finished successfully.
 * In this case the calling shell script should exit 0.
 */
int bisect_next_exit(const char *prefix)
{
	struct rev_info revs;
	struct commit_list *tried;
	int reaches = 0, all = 0, nr;
	const unsigned char *bisect_rev;
	char bisect_rev_hex[41];

636 637 638 639 640 641
	if (read_bisect_refs())
		die("reading bisect refs failed");

	bisect_rev_setup(&revs, prefix);

	bisect_common(&revs, &reaches, &all);
642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675

	revs.commits = filter_skipped(revs.commits, &tried, 0);

	if (!revs.commits) {
		/*
		 * We should exit here only if the "bad"
		 * commit is also a "skip" commit.
		 */
		exit_if_skipped_commits(tried, NULL);

		printf("%s was both good and bad\n",
		       sha1_to_hex(current_bad_sha1));
		exit(1);
	}

	bisect_rev = revs.commits->item->object.sha1;
	memcpy(bisect_rev_hex, sha1_to_hex(bisect_rev), 41);

	if (!hashcmp(bisect_rev, current_bad_sha1)) {
		exit_if_skipped_commits(tried, current_bad_sha1);
		printf("%s is first bad commit\n", bisect_rev_hex);
		argv_diff_tree[2] = bisect_rev_hex;
		run_command_v_opt(argv_diff_tree, RUN_GIT_CMD);
		/* This means the bisection process succeeded. */
		exit(10);
	}

	nr = all - reaches - 1;
	printf("Bisecting: %d revisions left to test after this "
	       "(roughly %d steps)\n", nr, estimate_bisect_steps(all));

	return bisect_checkout(bisect_rev_hex);
}