page_owner.c 15.2 KB
Newer Older
1 2 3 4 5 6 7
#include <linux/debugfs.h>
#include <linux/mm.h>
#include <linux/slab.h>
#include <linux/uaccess.h>
#include <linux/bootmem.h>
#include <linux/stacktrace.h>
#include <linux/page_owner.h>
8
#include <linux/jump_label.h>
9
#include <linux/migrate.h>
10
#include <linux/stackdepot.h>
11
#include <linux/seq_file.h>
12

13 14
#include "internal.h"

15 16 17 18 19 20
/*
 * TODO: teach PAGE_OWNER_STACK_DEPTH (__dump_page_owner and save_stack)
 * to use off stack temporal storage
 */
#define PAGE_OWNER_STACK_DEPTH (16)

21 22 23 24 25 26 27
struct page_owner {
	unsigned int order;
	gfp_t gfp_mask;
	int last_migrate_reason;
	depot_stack_handle_t handle;
};

28
static bool page_owner_disabled = true;
29
DEFINE_STATIC_KEY_FALSE(page_owner_inited);
30

31 32
static depot_stack_handle_t dummy_handle;
static depot_stack_handle_t failure_handle;
33
static depot_stack_handle_t early_handle;
34

35 36
static void init_early_allocated_pages(void);

37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56
static int early_page_owner_param(char *buf)
{
	if (!buf)
		return -EINVAL;

	if (strcmp(buf, "on") == 0)
		page_owner_disabled = false;

	return 0;
}
early_param("page_owner", early_page_owner_param);

static bool need_page_owner(void)
{
	if (page_owner_disabled)
		return false;

	return true;
}

57
static __always_inline depot_stack_handle_t create_dummy_stack(void)
58 59 60 61 62 63 64 65 66 67
{
	unsigned long entries[4];
	struct stack_trace dummy;

	dummy.nr_entries = 0;
	dummy.max_entries = ARRAY_SIZE(entries);
	dummy.entries = &entries[0];
	dummy.skip = 0;

	save_stack_trace(&dummy);
68
	return depot_save_stack(&dummy, GFP_KERNEL);
69 70
}

71
static noinline void register_dummy_stack(void)
72
{
73 74
	dummy_handle = create_dummy_stack();
}
75

76 77 78 79
static noinline void register_failure_stack(void)
{
	failure_handle = create_dummy_stack();
}
80

81 82 83
static noinline void register_early_stack(void)
{
	early_handle = create_dummy_stack();
84 85
}

86 87 88 89 90
static void init_page_owner(void)
{
	if (page_owner_disabled)
		return;

91 92
	register_dummy_stack();
	register_failure_stack();
93
	register_early_stack();
94
	static_branch_enable(&page_owner_inited);
95
	init_early_allocated_pages();
96 97 98
}

struct page_ext_operations page_owner_ops = {
99
	.size = sizeof(struct page_owner),
100 101 102 103
	.need = need_page_owner,
	.init = init_page_owner,
};

104 105 106 107 108
static inline struct page_owner *get_page_owner(struct page_ext *page_ext)
{
	return (void *)page_ext + page_owner_ops.offset;
}

109 110 111 112 113 114 115
void __reset_page_owner(struct page *page, unsigned int order)
{
	int i;
	struct page_ext *page_ext;

	for (i = 0; i < (1 << order); i++) {
		page_ext = lookup_page_ext(page + i);
116 117
		if (unlikely(!page_ext))
			continue;
118 119 120 121
		__clear_bit(PAGE_EXT_OWNER, &page_ext->flags);
	}
}

122 123
static inline bool check_recursive_alloc(struct stack_trace *trace,
					unsigned long ip)
124
{
125 126 127 128 129 130 131 132 133
	int i, count;

	if (!trace->nr_entries)
		return false;

	for (i = 0, count = 0; i < trace->nr_entries; i++) {
		if (trace->entries[i] == ip && ++count == 2)
			return true;
	}
134

135 136 137 138 139 140
	return false;
}

static noinline depot_stack_handle_t save_stack(gfp_t flags)
{
	unsigned long entries[PAGE_OWNER_STACK_DEPTH];
141 142
	struct stack_trace trace = {
		.nr_entries = 0,
143 144
		.entries = entries,
		.max_entries = PAGE_OWNER_STACK_DEPTH,
145
		.skip = 2
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
	depot_stack_handle_t handle;

	save_stack_trace(&trace);
	if (trace.nr_entries != 0 &&
	    trace.entries[trace.nr_entries-1] == ULONG_MAX)
		trace.nr_entries--;

	/*
	 * We need to check recursion here because our request to stackdepot
	 * could trigger memory allocation to save new entry. New memory
	 * allocation would reach here and call depot_save_stack() again
	 * if we don't catch it. There is still not enough memory in stackdepot
	 * so it would try to allocate memory again and loop forever.
	 */
	if (check_recursive_alloc(&trace, _RET_IP_))
		return dummy_handle;

	handle = depot_save_stack(&trace, flags);
	if (!handle)
		handle = failure_handle;

	return handle;
}

171 172
static inline void __set_page_owner_handle(struct page_ext *page_ext,
	depot_stack_handle_t handle, unsigned int order, gfp_t gfp_mask)
173
{
174
	struct page_owner *page_owner;
175

176
	page_owner = get_page_owner(page_ext);
177
	page_owner->handle = handle;
178 179 180
	page_owner->order = order;
	page_owner->gfp_mask = gfp_mask;
	page_owner->last_migrate_reason = -1;
181 182 183 184

	__set_bit(PAGE_EXT_OWNER, &page_ext->flags);
}

185 186 187 188 189 190 191 192 193 194 195 196 197
noinline void __set_page_owner(struct page *page, unsigned int order,
					gfp_t gfp_mask)
{
	struct page_ext *page_ext = lookup_page_ext(page);
	depot_stack_handle_t handle;

	if (unlikely(!page_ext))
		return;

	handle = save_stack(gfp_mask);
	__set_page_owner_handle(page_ext, handle, order, gfp_mask);
}

198 199 200
void __set_page_owner_migrate_reason(struct page *page, int reason)
{
	struct page_ext *page_ext = lookup_page_ext(page);
201 202
	struct page_owner *page_owner;

203 204
	if (unlikely(!page_ext))
		return;
205

206 207
	page_owner = get_page_owner(page_ext);
	page_owner->last_migrate_reason = reason;
208 209
}

210
void __split_page_owner(struct page *page, unsigned int order)
211
{
212
	int i;
213
	struct page_ext *page_ext = lookup_page_ext(page);
214
	struct page_owner *page_owner;
215

216
	if (unlikely(!page_ext))
217
		return;
218

219 220
	page_owner = get_page_owner(page_ext);
	page_owner->order = 0;
221 222
	for (i = 1; i < (1 << order); i++)
		__copy_page_owner(page, page + i);
223 224
}

225 226 227 228
void __copy_page_owner(struct page *oldpage, struct page *newpage)
{
	struct page_ext *old_ext = lookup_page_ext(oldpage);
	struct page_ext *new_ext = lookup_page_ext(newpage);
229
	struct page_owner *old_page_owner, *new_page_owner;
230

231 232 233
	if (unlikely(!old_ext || !new_ext))
		return;

234 235 236 237 238 239 240
	old_page_owner = get_page_owner(old_ext);
	new_page_owner = get_page_owner(new_ext);
	new_page_owner->order = old_page_owner->order;
	new_page_owner->gfp_mask = old_page_owner->gfp_mask;
	new_page_owner->last_migrate_reason =
		old_page_owner->last_migrate_reason;
	new_page_owner->handle = old_page_owner->handle;
241 242 243 244 245 246 247 248 249 250 251 252 253

	/*
	 * We don't clear the bit on the oldpage as it's going to be freed
	 * after migration. Until then, the info can be useful in case of
	 * a bug, and the overal stats will be off a bit only temporarily.
	 * Also, migrate_misplaced_transhuge_page() can still fail the
	 * migration and then we want the oldpage to retain the info. But
	 * in that case we also don't need to explicitly clear the info from
	 * the new page, which will be freed.
	 */
	__set_bit(PAGE_EXT_OWNER, &new_ext->flags);
}

254 255 256 257 258
void pagetypeinfo_showmixedcount_print(struct seq_file *m,
				       pg_data_t *pgdat, struct zone *zone)
{
	struct page *page;
	struct page_ext *page_ext;
259
	struct page_owner *page_owner;
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
	unsigned long pfn = zone->zone_start_pfn, block_end_pfn;
	unsigned long end_pfn = pfn + zone->spanned_pages;
	unsigned long count[MIGRATE_TYPES] = { 0, };
	int pageblock_mt, page_mt;
	int i;

	/* Scan block by block. First and last block may be incomplete */
	pfn = zone->zone_start_pfn;

	/*
	 * Walk the zone in pageblock_nr_pages steps. If a page block spans
	 * a zone boundary, it will be double counted between zones. This does
	 * not matter as the mixed block count will still be correct
	 */
	for (; pfn < end_pfn; ) {
		if (!pfn_valid(pfn)) {
			pfn = ALIGN(pfn + 1, MAX_ORDER_NR_PAGES);
			continue;
		}

		block_end_pfn = ALIGN(pfn + 1, pageblock_nr_pages);
		block_end_pfn = min(block_end_pfn, end_pfn);

		page = pfn_to_page(pfn);
		pageblock_mt = get_pageblock_migratetype(page);

		for (; pfn < block_end_pfn; pfn++) {
			if (!pfn_valid_within(pfn))
				continue;

			page = pfn_to_page(pfn);

			if (page_zone(page) != zone)
				continue;

			if (PageBuddy(page)) {
296 297 298 299 300
				unsigned long freepage_order;

				freepage_order = page_order_unsafe(page);
				if (freepage_order < MAX_ORDER)
					pfn += (1UL << freepage_order) - 1;
301 302 303 304 305 306 307 308 309 310 311 312 313
				continue;
			}

			if (PageReserved(page))
				continue;

			page_ext = lookup_page_ext(page);
			if (unlikely(!page_ext))
				continue;

			if (!test_bit(PAGE_EXT_OWNER, &page_ext->flags))
				continue;

314 315 316
			page_owner = get_page_owner(page_ext);
			page_mt = gfpflags_to_migratetype(
					page_owner->gfp_mask);
317 318 319 320 321 322 323 324 325
			if (pageblock_mt != page_mt) {
				if (is_migrate_cma(pageblock_mt))
					count[MIGRATE_MOVABLE]++;
				else
					count[pageblock_mt]++;

				pfn = block_end_pfn;
				break;
			}
326
			pfn += (1UL << page_owner->order) - 1;
327 328 329 330 331 332 333 334 335 336
		}
	}

	/* Print counts */
	seq_printf(m, "Node %d, zone %8s ", pgdat->node_id, zone->name);
	for (i = 0; i < MIGRATE_TYPES; i++)
		seq_printf(m, "%12lu ", count[i]);
	seq_putc(m, '\n');
}

337 338
static ssize_t
print_page_owner(char __user *buf, size_t count, unsigned long pfn,
339
		struct page *page, struct page_owner *page_owner,
340
		depot_stack_handle_t handle)
341 342 343 344
{
	int ret;
	int pageblock_mt, page_mt;
	char *kbuf;
345
	unsigned long entries[PAGE_OWNER_STACK_DEPTH];
346
	struct stack_trace trace = {
347 348 349 350
		.nr_entries = 0,
		.entries = entries,
		.max_entries = PAGE_OWNER_STACK_DEPTH,
		.skip = 0
351
	};
352 353 354 355 356 357

	kbuf = kmalloc(count, GFP_KERNEL);
	if (!kbuf)
		return -ENOMEM;

	ret = snprintf(kbuf, count,
358
			"Page allocated via order %u, mask %#x(%pGg)\n",
359 360
			page_owner->order, page_owner->gfp_mask,
			&page_owner->gfp_mask);
361 362 363 364 365

	if (ret >= count)
		goto err;

	/* Print information relevant to grouping pages by mobility */
366
	pageblock_mt = get_pageblock_migratetype(page);
367
	page_mt  = gfpflags_to_migratetype(page_owner->gfp_mask);
368
	ret += snprintf(kbuf + ret, count - ret,
369
			"PFN %lu type %s Block %lu type %s Flags %#lx(%pGp)\n",
370
			pfn,
371
			migratetype_names[page_mt],
372
			pfn >> pageblock_order,
373 374
			migratetype_names[pageblock_mt],
			page->flags, &page->flags);
375 376 377 378

	if (ret >= count)
		goto err;

379
	depot_fetch_stack(handle, &trace);
380
	ret += snprint_stack_trace(kbuf + ret, count - ret, &trace, 0);
381 382 383
	if (ret >= count)
		goto err;

384
	if (page_owner->last_migrate_reason != -1) {
385 386
		ret += snprintf(kbuf + ret, count - ret,
			"Page has been migrated, last migrate reason: %s\n",
387
			migrate_reason_names[page_owner->last_migrate_reason]);
388 389 390 391
		if (ret >= count)
			goto err;
	}

392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
	ret += snprintf(kbuf + ret, count - ret, "\n");
	if (ret >= count)
		goto err;

	if (copy_to_user(buf, kbuf, ret))
		ret = -EFAULT;

	kfree(kbuf);
	return ret;

err:
	kfree(kbuf);
	return -ENOMEM;
}

407 408 409
void __dump_page_owner(struct page *page)
{
	struct page_ext *page_ext = lookup_page_ext(page);
410
	struct page_owner *page_owner;
411
	unsigned long entries[PAGE_OWNER_STACK_DEPTH];
412
	struct stack_trace trace = {
413 414 415 416
		.nr_entries = 0,
		.entries = entries,
		.max_entries = PAGE_OWNER_STACK_DEPTH,
		.skip = 0
417
	};
418
	depot_stack_handle_t handle;
419 420
	gfp_t gfp_mask;
	int mt;
421

422 423 424 425
	if (unlikely(!page_ext)) {
		pr_alert("There is not page extension available.\n");
		return;
	}
426 427 428

	page_owner = get_page_owner(page_ext);
	gfp_mask = page_owner->gfp_mask;
429
	mt = gfpflags_to_migratetype(gfp_mask);
430

431 432 433 434 435
	if (!test_bit(PAGE_EXT_OWNER, &page_ext->flags)) {
		pr_alert("page_owner info is not active (free page?)\n");
		return;
	}

436
	handle = READ_ONCE(page_owner->handle);
437 438 439 440 441 442
	if (!handle) {
		pr_alert("page_owner info is not active (free page?)\n");
		return;
	}

	depot_fetch_stack(handle, &trace);
Joe Perches's avatar
Joe Perches committed
443
	pr_alert("page allocated via order %u, migratetype %s, gfp_mask %#x(%pGg)\n",
444
		 page_owner->order, migratetype_names[mt], gfp_mask, &gfp_mask);
445 446
	print_stack_trace(&trace, 0);

447
	if (page_owner->last_migrate_reason != -1)
448
		pr_alert("page has been migrated, last migrate reason: %s\n",
449
			migrate_reason_names[page_owner->last_migrate_reason]);
450 451
}

452 453 454 455 456 457
static ssize_t
read_page_owner(struct file *file, char __user *buf, size_t count, loff_t *ppos)
{
	unsigned long pfn;
	struct page *page;
	struct page_ext *page_ext;
458
	struct page_owner *page_owner;
459
	depot_stack_handle_t handle;
460

461
	if (!static_branch_unlikely(&page_owner_inited))
462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497
		return -EINVAL;

	page = NULL;
	pfn = min_low_pfn + *ppos;

	/* Find a valid PFN or the start of a MAX_ORDER_NR_PAGES area */
	while (!pfn_valid(pfn) && (pfn & (MAX_ORDER_NR_PAGES - 1)) != 0)
		pfn++;

	drain_all_pages(NULL);

	/* Find an allocated page */
	for (; pfn < max_pfn; pfn++) {
		/*
		 * If the new page is in a new MAX_ORDER_NR_PAGES area,
		 * validate the area as existing, skip it if not
		 */
		if ((pfn & (MAX_ORDER_NR_PAGES - 1)) == 0 && !pfn_valid(pfn)) {
			pfn += MAX_ORDER_NR_PAGES - 1;
			continue;
		}

		/* Check for holes within a MAX_ORDER area */
		if (!pfn_valid_within(pfn))
			continue;

		page = pfn_to_page(pfn);
		if (PageBuddy(page)) {
			unsigned long freepage_order = page_order_unsafe(page);

			if (freepage_order < MAX_ORDER)
				pfn += (1UL << freepage_order) - 1;
			continue;
		}

		page_ext = lookup_page_ext(page);
498 499
		if (unlikely(!page_ext))
			continue;
500 501

		/*
502 503
		 * Some pages could be missed by concurrent allocation or free,
		 * because we don't hold the zone lock.
504 505 506 507
		 */
		if (!test_bit(PAGE_EXT_OWNER, &page_ext->flags))
			continue;

508 509
		page_owner = get_page_owner(page_ext);

510 511 512 513
		/*
		 * Access to page_ext->handle isn't synchronous so we should
		 * be careful to access it.
		 */
514
		handle = READ_ONCE(page_owner->handle);
515 516 517
		if (!handle)
			continue;

518 519 520
		/* Record the next PFN to read in the file offset */
		*ppos = (pfn - min_low_pfn) + 1;

521
		return print_page_owner(buf, count, pfn, page,
522
				page_owner, handle);
523 524 525 526 527
	}

	return 0;
}

528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560
static void init_pages_in_zone(pg_data_t *pgdat, struct zone *zone)
{
	struct page *page;
	struct page_ext *page_ext;
	unsigned long pfn = zone->zone_start_pfn, block_end_pfn;
	unsigned long end_pfn = pfn + zone->spanned_pages;
	unsigned long count = 0;

	/* Scan block by block. First and last block may be incomplete */
	pfn = zone->zone_start_pfn;

	/*
	 * Walk the zone in pageblock_nr_pages steps. If a page block spans
	 * a zone boundary, it will be double counted between zones. This does
	 * not matter as the mixed block count will still be correct
	 */
	for (; pfn < end_pfn; ) {
		if (!pfn_valid(pfn)) {
			pfn = ALIGN(pfn + 1, MAX_ORDER_NR_PAGES);
			continue;
		}

		block_end_pfn = ALIGN(pfn + 1, pageblock_nr_pages);
		block_end_pfn = min(block_end_pfn, end_pfn);

		page = pfn_to_page(pfn);

		for (; pfn < block_end_pfn; pfn++) {
			if (!pfn_valid_within(pfn))
				continue;

			page = pfn_to_page(pfn);

561 562 563
			if (page_zone(page) != zone)
				continue;

564
			/*
565 566 567 568 569
			 * To avoid having to grab zone->lock, be a little
			 * careful when reading buddy page order. The only
			 * danger is that we skip too much and potentially miss
			 * some early allocated pages, which is better than
			 * heavy lock contention.
570 571
			 */
			if (PageBuddy(page)) {
572 573 574 575
				unsigned long order = page_order_unsafe(page);

				if (order > 0 && order < MAX_ORDER)
					pfn += (1UL << order) - 1;
576 577 578 579 580 581 582
				continue;
			}

			if (PageReserved(page))
				continue;

			page_ext = lookup_page_ext(page);
583 584
			if (unlikely(!page_ext))
				continue;
585

586
			/* Maybe overlapping zone */
587 588 589 590
			if (test_bit(PAGE_EXT_OWNER, &page_ext->flags))
				continue;

			/* Found early allocated page */
591
			__set_page_owner_handle(page_ext, early_handle, 0, 0);
592 593
			count++;
		}
594
		cond_resched();
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
	}

	pr_info("Node %d, zone %8s: page owner found early allocated %lu pages\n",
		pgdat->node_id, zone->name, count);
}

static void init_zones_in_node(pg_data_t *pgdat)
{
	struct zone *zone;
	struct zone *node_zones = pgdat->node_zones;

	for (zone = node_zones; zone - node_zones < MAX_NR_ZONES; ++zone) {
		if (!populated_zone(zone))
			continue;

		init_pages_in_zone(pgdat, zone);
	}
}

static void init_early_allocated_pages(void)
{
	pg_data_t *pgdat;

	drain_all_pages(NULL);
	for_each_online_pgdat(pgdat)
		init_zones_in_node(pgdat);
}

623 624 625 626 627 628 629 630
static const struct file_operations proc_page_owner_operations = {
	.read		= read_page_owner,
};

static int __init pageowner_init(void)
{
	struct dentry *dentry;

631
	if (!static_branch_unlikely(&page_owner_inited)) {
632 633 634 635 636 637 638 639 640 641 642
		pr_info("page_owner is disabled\n");
		return 0;
	}

	dentry = debugfs_create_file("page_owner", S_IRUSR, NULL,
			NULL, &proc_page_owner_operations);
	if (IS_ERR(dentry))
		return PTR_ERR(dentry);

	return 0;
}
643
late_initcall(pageowner_init)