page_alloc.c 189 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 *  linux/mm/page_alloc.c
 *
 *  Manages the free list, the system allocates free pages here.
 *  Note that kmalloc() lives in slab.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 *  Swap reorganised 29.12.95, Stephen Tweedie
 *  Support of BIGMEM added by Gerhard Wichert, Siemens AG, July 1999
 *  Reshaped it to be a zoned allocator, Ingo Molnar, Red Hat, 1999
 *  Discontiguous memory support, Kanoj Sarcar, SGI, Nov 1999
 *  Zone balancing, Kanoj Sarcar, SGI, Jan 2000
 *  Per cpu hot/cold page lists, bulk allocation, Martin J. Bligh, Sept 2002
 *          (lots of bits borrowed from Ingo Molnar & Andrew Morton)
 */

#include <linux/stddef.h>
#include <linux/mm.h>
#include <linux/swap.h>
#include <linux/interrupt.h>
#include <linux/pagemap.h>
22
#include <linux/jiffies.h>
Linus Torvalds's avatar
Linus Torvalds committed
23
#include <linux/bootmem.h>
24
#include <linux/memblock.h>
Linus Torvalds's avatar
Linus Torvalds committed
25
#include <linux/compiler.h>
26
#include <linux/kernel.h>
27
#include <linux/kmemcheck.h>
28
#include <linux/kasan.h>
Linus Torvalds's avatar
Linus Torvalds committed
29 30 31 32 33
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
34
#include <linux/ratelimit.h>
35
#include <linux/oom.h>
Linus Torvalds's avatar
Linus Torvalds committed
36 37 38 39 40
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
41
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
42 43
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
44
#include <linux/vmstat.h>
45
#include <linux/mempolicy.h>
46
#include <linux/stop_machine.h>
47 48
#include <linux/sort.h>
#include <linux/pfn.h>
49
#include <linux/backing-dev.h>
50
#include <linux/fault-inject.h>
51
#include <linux/page-isolation.h>
52
#include <linux/page_ext.h>
53
#include <linux/debugobjects.h>
54
#include <linux/kmemleak.h>
55
#include <linux/compaction.h>
56
#include <trace/events/kmem.h>
57
#include <linux/prefetch.h>
58
#include <linux/mm_inline.h>
59
#include <linux/migrate.h>
60
#include <linux/page_ext.h>
61
#include <linux/hugetlb.h>
62
#include <linux/sched/rt.h>
63
#include <linux/page_owner.h>
64
#include <linux/kthread.h>
Linus Torvalds's avatar
Linus Torvalds committed
65

66
#include <asm/sections.h>
Linus Torvalds's avatar
Linus Torvalds committed
67
#include <asm/tlbflush.h>
68
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
69 70
#include "internal.h"

71 72
/* prevent >1 _updater_ of zone percpu pageset ->high and ->batch fields */
static DEFINE_MUTEX(pcp_batch_high_lock);
73
#define MIN_PERCPU_PAGELIST_FRACTION	(8)
74

75 76 77 78 79
#ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID
DEFINE_PER_CPU(int, numa_node);
EXPORT_PER_CPU_SYMBOL(numa_node);
#endif

80 81 82 83 84 85 86 87 88
#ifdef CONFIG_HAVE_MEMORYLESS_NODES
/*
 * N.B., Do NOT reference the '_numa_mem_' per cpu variable directly.
 * It will not be defined when CONFIG_HAVE_MEMORYLESS_NODES is not defined.
 * Use the accessor functions set_numa_mem(), numa_mem_id() and cpu_to_mem()
 * defined in <linux/topology.h>.
 */
DEFINE_PER_CPU(int, _numa_mem_);		/* Kernel "local memory" node */
EXPORT_PER_CPU_SYMBOL(_numa_mem_);
89
int _node_numa_mem_[MAX_NUMNODES];
90 91
#endif

Linus Torvalds's avatar
Linus Torvalds committed
92
/*
93
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
94
 */
95 96 97 98 99 100 101
nodemask_t node_states[NR_NODE_STATES] __read_mostly = {
	[N_POSSIBLE] = NODE_MASK_ALL,
	[N_ONLINE] = { { [0] = 1UL } },
#ifndef CONFIG_NUMA
	[N_NORMAL_MEMORY] = { { [0] = 1UL } },
#ifdef CONFIG_HIGHMEM
	[N_HIGH_MEMORY] = { { [0] = 1UL } },
102 103 104
#endif
#ifdef CONFIG_MOVABLE_NODE
	[N_MEMORY] = { { [0] = 1UL } },
105 106 107 108 109 110
#endif
	[N_CPU] = { { [0] = 1UL } },
#endif	/* NUMA */
};
EXPORT_SYMBOL(node_states);

111 112 113
/* Protect totalram_pages and zone->managed_pages */
static DEFINE_SPINLOCK(managed_page_count_lock);

114
unsigned long totalram_pages __read_mostly;
115
unsigned long totalreserve_pages __read_mostly;
116
unsigned long totalcma_pages __read_mostly;
117 118 119 120 121 122 123 124
/*
 * When calculating the number of globally allowed dirty pages, there
 * is a certain number of per-zone reserves that should not be
 * considered dirtyable memory.  This is the sum of those reserves
 * over all existing zones that contribute dirtyable memory.
 */
unsigned long dirty_balance_reserve __read_mostly;

125
int percpu_pagelist_fraction;
126
gfp_t gfp_allowed_mask __read_mostly = GFP_BOOT_MASK;
Linus Torvalds's avatar
Linus Torvalds committed
127

128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145
/*
 * A cached value of the page's pageblock's migratetype, used when the page is
 * put on a pcplist. Used to avoid the pageblock migratetype lookup when
 * freeing from pcplists in most cases, at the cost of possibly becoming stale.
 * Also the migratetype set in the page does not necessarily match the pcplist
 * index, e.g. page might have MIGRATE_CMA set but be on a pcplist with any
 * other index - this ensures that it will be put on the correct CMA freelist.
 */
static inline int get_pcppage_migratetype(struct page *page)
{
	return page->index;
}

static inline void set_pcppage_migratetype(struct page *page, int migratetype)
{
	page->index = migratetype;
}

146 147 148 149 150 151 152 153 154
#ifdef CONFIG_PM_SLEEP
/*
 * The following functions are used by the suspend/hibernate code to temporarily
 * change gfp_allowed_mask in order to avoid using I/O during memory allocations
 * while devices are suspended.  To avoid races with the suspend/hibernate code,
 * they should always be called with pm_mutex held (gfp_allowed_mask also should
 * only be modified with pm_mutex held, unless the suspend/hibernate code is
 * guaranteed not to run in parallel with that modification).
 */
155 156 157 158

static gfp_t saved_gfp_mask;

void pm_restore_gfp_mask(void)
159 160
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
161 162 163 164
	if (saved_gfp_mask) {
		gfp_allowed_mask = saved_gfp_mask;
		saved_gfp_mask = 0;
	}
165 166
}

167
void pm_restrict_gfp_mask(void)
168 169
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
170 171
	WARN_ON(saved_gfp_mask);
	saved_gfp_mask = gfp_allowed_mask;
172
	gfp_allowed_mask &= ~(__GFP_IO | __GFP_FS);
173
}
174 175 176

bool pm_suspended_storage(void)
{
177
	if ((gfp_allowed_mask & (__GFP_IO | __GFP_FS)) == (__GFP_IO | __GFP_FS))
178 179 180
		return false;
	return true;
}
181 182
#endif /* CONFIG_PM_SLEEP */

183 184 185 186
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

187
static void __free_pages_ok(struct page *page, unsigned int order);
188

Linus Torvalds's avatar
Linus Torvalds committed
189 190 191 192 193 194
/*
 * results with 256, 32 in the lowmem_reserve sysctl:
 *	1G machine -> (16M dma, 800M-16M normal, 1G-800M high)
 *	1G machine -> (16M dma, 784M normal, 224M high)
 *	NORMAL allocation will leave 784M/256 of ram reserved in the ZONE_DMA
 *	HIGHMEM allocation will leave 224M/32 of ram reserved in ZONE_NORMAL
Yaowei Bai's avatar
Yaowei Bai committed
195
 *	HIGHMEM allocation will leave (224M+784M)/256 of ram reserved in ZONE_DMA
196 197 198
 *
 * TBD: should special case ZONE_DMA32 machines here - in those we normally
 * don't need any ZONE_NORMAL reservation
Linus Torvalds's avatar
Linus Torvalds committed
199
 */
200
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
201
#ifdef CONFIG_ZONE_DMA
202
	 256,
203
#endif
204
#ifdef CONFIG_ZONE_DMA32
205
	 256,
206
#endif
207
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
208
	 32,
209
#endif
Mel Gorman's avatar
Mel Gorman committed
210
	 32,
211
};
Linus Torvalds's avatar
Linus Torvalds committed
212 213 214

EXPORT_SYMBOL(totalram_pages);

215
static char * const zone_names[MAX_NR_ZONES] = {
216
#ifdef CONFIG_ZONE_DMA
217
	 "DMA",
218
#endif
219
#ifdef CONFIG_ZONE_DMA32
220
	 "DMA32",
221
#endif
222
	 "Normal",
223
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
224
	 "HighMem",
225
#endif
Mel Gorman's avatar
Mel Gorman committed
226
	 "Movable",
227 228 229
#ifdef CONFIG_ZONE_DEVICE
	 "Device",
#endif
230 231
};

232 233 234 235 236 237 238 239 240
static void free_compound_page(struct page *page);
compound_page_dtor * const compound_page_dtors[] = {
	NULL,
	free_compound_page,
#ifdef CONFIG_HUGETLB_PAGE
	free_huge_page,
#endif
};

Linus Torvalds's avatar
Linus Torvalds committed
241
int min_free_kbytes = 1024;
242
int user_min_free_kbytes = -1;
Linus Torvalds's avatar
Linus Torvalds committed
243

244 245
static unsigned long __meminitdata nr_kernel_pages;
static unsigned long __meminitdata nr_all_pages;
246
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
247

Tejun Heo's avatar
Tejun Heo committed
248 249 250 251 252 253 254 255 256 257 258
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
static unsigned long __meminitdata arch_zone_lowest_possible_pfn[MAX_NR_ZONES];
static unsigned long __meminitdata arch_zone_highest_possible_pfn[MAX_NR_ZONES];
static unsigned long __initdata required_kernelcore;
static unsigned long __initdata required_movablecore;
static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];

/* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
int movable_zone;
EXPORT_SYMBOL(movable_zone);
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
259

Miklos Szeredi's avatar
Miklos Szeredi committed
260 261
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
262
int nr_online_nodes __read_mostly = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
263
EXPORT_SYMBOL(nr_node_ids);
264
EXPORT_SYMBOL(nr_online_nodes);
Miklos Szeredi's avatar
Miklos Szeredi committed
265 266
#endif

267 268
int page_group_by_mobility_disabled __read_mostly;

269 270 271 272 273 274 275
#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
static inline void reset_deferred_meminit(pg_data_t *pgdat)
{
	pgdat->first_deferred_pfn = ULONG_MAX;
}

/* Returns true if the struct page for the pfn is uninitialised */
276
static inline bool __meminit early_page_uninitialised(unsigned long pfn)
277
{
278
	if (pfn >= NODE_DATA(early_pfn_to_nid(pfn))->first_deferred_pfn)
279 280 281 282 283
		return true;

	return false;
}

284 285 286 287 288 289 290 291
static inline bool early_page_nid_uninitialised(unsigned long pfn, int nid)
{
	if (pfn >= NODE_DATA(nid)->first_deferred_pfn)
		return true;

	return false;
}

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
/*
 * Returns false when the remaining initialisation should be deferred until
 * later in the boot cycle when it can be parallelised.
 */
static inline bool update_defer_init(pg_data_t *pgdat,
				unsigned long pfn, unsigned long zone_end,
				unsigned long *nr_initialised)
{
	/* Always populate low zones for address-contrained allocations */
	if (zone_end < pgdat_end_pfn(pgdat))
		return true;

	/* Initialise at least 2G of the highest zone */
	(*nr_initialised)++;
	if (*nr_initialised > (2UL << (30 - PAGE_SHIFT)) &&
	    (pfn & (PAGES_PER_SECTION - 1)) == 0) {
		pgdat->first_deferred_pfn = pfn;
		return false;
	}

	return true;
}
#else
static inline void reset_deferred_meminit(pg_data_t *pgdat)
{
}

static inline bool early_page_uninitialised(unsigned long pfn)
{
	return false;
}

324 325 326 327 328
static inline bool early_page_nid_uninitialised(unsigned long pfn, int nid)
{
	return false;
}

329 330 331 332 333 334 335 336 337
static inline bool update_defer_init(pg_data_t *pgdat,
				unsigned long pfn, unsigned long zone_end,
				unsigned long *nr_initialised)
{
	return true;
}
#endif


338
void set_pageblock_migratetype(struct page *page, int migratetype)
339
{
340 341
	if (unlikely(page_group_by_mobility_disabled &&
		     migratetype < MIGRATE_PCPTYPES))
342 343
		migratetype = MIGRATE_UNMOVABLE;

344 345 346 347
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

Nick Piggin's avatar
Nick Piggin committed
348
#ifdef CONFIG_DEBUG_VM
349
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
350
{
351 352 353
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
354
	unsigned long sp, start_pfn;
355

356 357
	do {
		seq = zone_span_seqbegin(zone);
358 359
		start_pfn = zone->zone_start_pfn;
		sp = zone->spanned_pages;
360
		if (!zone_spans_pfn(zone, pfn))
361 362 363
			ret = 1;
	} while (zone_span_seqretry(zone, seq));

364
	if (ret)
365 366 367
		pr_err("page 0x%lx outside node %d zone %s [ 0x%lx - 0x%lx ]\n",
			pfn, zone_to_nid(zone), zone->name,
			start_pfn, start_pfn + sp);
368

369
	return ret;
370 371 372 373
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
374
	if (!pfn_valid_within(page_to_pfn(page)))
375
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
376
	if (zone != page_zone(page))
377 378 379 380 381 382 383 384 385 386
		return 0;

	return 1;
}
/*
 * Temporary debugging check for pages not lying within a given zone.
 */
static int bad_range(struct zone *zone, struct page *page)
{
	if (page_outside_zone_boundaries(zone, page))
Linus Torvalds's avatar
Linus Torvalds committed
387
		return 1;
388 389 390
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
391 392
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
393 394 395 396 397 398 399
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

400 401
static void bad_page(struct page *page, const char *reason,
		unsigned long bad_flags)
Linus Torvalds's avatar
Linus Torvalds committed
402
{
403 404 405 406
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

407 408
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
409
		page_mapcount_reset(page); /* remove PageBuddy */
410 411 412
		return;
	}

413 414 415 416 417 418 419 420 421 422
	/*
	 * Allow a burst of 60 reports, then keep quiet for that minute;
	 * or allow a steady drip of one report per second.
	 */
	if (nr_shown == 60) {
		if (time_before(jiffies, resume)) {
			nr_unshown++;
			goto out;
		}
		if (nr_unshown) {
423 424
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
425 426 427 428 429 430 431 432
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

433
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
434
		current->comm, page_to_pfn(page));
435
	dump_page_badflags(page, reason, bad_flags);
436

437
	print_modules();
Linus Torvalds's avatar
Linus Torvalds committed
438
	dump_stack();
439
out:
440
	/* Leave bad fields for debug, except PageBuddy could make trouble */
441
	page_mapcount_reset(page); /* remove PageBuddy */
442
	add_taint(TAINT_BAD_PAGE, LOCKDEP_NOW_UNRELIABLE);
Linus Torvalds's avatar
Linus Torvalds committed
443 444 445 446 447
}

/*
 * Higher-order pages are called "compound pages".  They are structured thusly:
 *
448
 * The first PAGE_SIZE page is called the "head page" and have PG_head set.
Linus Torvalds's avatar
Linus Torvalds committed
449
 *
450 451
 * The remaining PAGE_SIZE pages are called "tail pages". PageTail() is encoded
 * in bit 0 of page->compound_head. The rest of bits is pointer to head page.
Linus Torvalds's avatar
Linus Torvalds committed
452
 *
453 454
 * The first tail page's ->compound_dtor holds the offset in array of compound
 * page destructors. See compound_page_dtors.
Linus Torvalds's avatar
Linus Torvalds committed
455
 *
456
 * The first tail page's ->compound_order holds the order of allocation.
457
 * This usage means that zero-order pages may not be compound.
Linus Torvalds's avatar
Linus Torvalds committed
458
 */
459 460 461

static void free_compound_page(struct page *page)
{
462
	__free_pages_ok(page, compound_order(page));
463 464
}

465
void prep_compound_page(struct page *page, unsigned long order)
466 467 468 469
{
	int i;
	int nr_pages = 1 << order;

470
	set_compound_page_dtor(page, COMPOUND_PAGE_DTOR);
471 472 473 474
	set_compound_order(page, order);
	__SetPageHead(page);
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
475
		set_page_count(p, 0);
476
		set_compound_head(p, page);
477 478 479
	}
}

480 481
#ifdef CONFIG_DEBUG_PAGEALLOC
unsigned int _debug_guardpage_minorder;
482
bool _debug_pagealloc_enabled __read_mostly;
483 484
bool _debug_guardpage_enabled __read_mostly;

485 486 487 488 489 490 491 492 493 494 495 496
static int __init early_debug_pagealloc(char *buf)
{
	if (!buf)
		return -EINVAL;

	if (strcmp(buf, "on") == 0)
		_debug_pagealloc_enabled = true;

	return 0;
}
early_param("debug_pagealloc", early_debug_pagealloc);

497 498
static bool need_debug_guardpage(void)
{
499 500 501 502
	/* If we don't use debug_pagealloc, we don't need guard page */
	if (!debug_pagealloc_enabled())
		return false;

503 504 505 506 507
	return true;
}

static void init_debug_guardpage(void)
{
508 509 510
	if (!debug_pagealloc_enabled())
		return;

511 512 513 514 515 516 517
	_debug_guardpage_enabled = true;
}

struct page_ext_operations debug_guardpage_ops = {
	.need = need_debug_guardpage,
	.init = init_debug_guardpage,
};
518 519 520 521 522 523 524 525 526 527 528 529 530 531 532

static int __init debug_guardpage_minorder_setup(char *buf)
{
	unsigned long res;

	if (kstrtoul(buf, 10, &res) < 0 ||  res > MAX_ORDER / 2) {
		printk(KERN_ERR "Bad debug_guardpage_minorder value\n");
		return 0;
	}
	_debug_guardpage_minorder = res;
	printk(KERN_INFO "Setting debug_guardpage_minorder to %lu\n", res);
	return 0;
}
__setup("debug_guardpage_minorder=", debug_guardpage_minorder_setup);

533 534
static inline void set_page_guard(struct zone *zone, struct page *page,
				unsigned int order, int migratetype)
535
{
536 537 538 539 540 541 542 543
	struct page_ext *page_ext;

	if (!debug_guardpage_enabled())
		return;

	page_ext = lookup_page_ext(page);
	__set_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags);

544 545 546 547
	INIT_LIST_HEAD(&page->lru);
	set_page_private(page, order);
	/* Guard pages are not available for any usage */
	__mod_zone_freepage_state(zone, -(1 << order), migratetype);
548 549
}

550 551
static inline void clear_page_guard(struct zone *zone, struct page *page,
				unsigned int order, int migratetype)
552
{
553 554 555 556 557 558 559 560
	struct page_ext *page_ext;

	if (!debug_guardpage_enabled())
		return;

	page_ext = lookup_page_ext(page);
	__clear_bit(PAGE_EXT_DEBUG_GUARD, &page_ext->flags);

561 562 563
	set_page_private(page, 0);
	if (!is_migrate_isolate(migratetype))
		__mod_zone_freepage_state(zone, (1 << order), migratetype);
564 565
}
#else
566
struct page_ext_operations debug_guardpage_ops = { NULL, };
567 568 569 570
static inline void set_page_guard(struct zone *zone, struct page *page,
				unsigned int order, int migratetype) {}
static inline void clear_page_guard(struct zone *zone, struct page *page,
				unsigned int order, int migratetype) {}
571 572
#endif

573
static inline void set_page_order(struct page *page, unsigned int order)
574
{
575
	set_page_private(page, order);
576
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
577 578 579 580
}

static inline void rmv_page_order(struct page *page)
{
581
	__ClearPageBuddy(page);
582
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
583 584 585 586 587
}

/*
 * This function checks whether a page is free && is the buddy
 * we can do coalesce a page and its buddy if
Nick Piggin's avatar
Nick Piggin committed
588
 * (a) the buddy is not in a hole &&
589
 * (b) the buddy is in the buddy system &&
590 591
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
592
 *
593 594 595 596
 * For recording whether a page is in the buddy system, we set ->_mapcount
 * PAGE_BUDDY_MAPCOUNT_VALUE.
 * Setting, clearing, and testing _mapcount PAGE_BUDDY_MAPCOUNT_VALUE is
 * serialized by zone->lock.
Linus Torvalds's avatar
Linus Torvalds committed
597
 *
598
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
599
 */
600
static inline int page_is_buddy(struct page *page, struct page *buddy,
601
							unsigned int order)
Linus Torvalds's avatar
Linus Torvalds committed
602
{
603
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
604 605
		return 0;

606
	if (page_is_guard(buddy) && page_order(buddy) == order) {
607 608 609
		if (page_zone_id(page) != page_zone_id(buddy))
			return 0;

610 611
		VM_BUG_ON_PAGE(page_count(buddy) != 0, buddy);

612 613 614
		return 1;
	}

615
	if (PageBuddy(buddy) && page_order(buddy) == order) {
616 617 618 619 620 621 622 623
		/*
		 * zone check is done late to avoid uselessly
		 * calculating zone/node ids for pages that could
		 * never merge.
		 */
		if (page_zone_id(page) != page_zone_id(buddy))
			return 0;

624 625
		VM_BUG_ON_PAGE(page_count(buddy) != 0, buddy);

626
		return 1;
627
	}
628
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
629 630 631 632 633 634 635 636 637 638 639 640 641 642 643
}

/*
 * Freeing function for a buddy system allocator.
 *
 * The concept of a buddy system is to maintain direct-mapped table
 * (containing bit values) for memory blocks of various "orders".
 * The bottom level table contains the map for the smallest allocatable
 * units of memory (here, pages), and each level above it describes
 * pairs of units from the levels below, hence, "buddies".
 * At a high level, all that happens here is marking the table entry
 * at the bottom level available, and propagating the changes upward
 * as necessary, plus some accounting needed to play nicely with other
 * parts of the VM system.
 * At each level, we keep a list of pages, which are heads of continuous
644 645 646
 * free pages of length of (1 << order) and marked with _mapcount
 * PAGE_BUDDY_MAPCOUNT_VALUE. Page's order is recorded in page_private(page)
 * field.
Linus Torvalds's avatar
Linus Torvalds committed
647
 * So when we are allocating or freeing one, we can derive the state of the
648 649
 * other.  That is, if we allocate a small block, and both were
 * free, the remainder of the region must be split into blocks.
Linus Torvalds's avatar
Linus Torvalds committed
650
 * If a block is freed, and its buddy is also free, then this
651
 * triggers coalescing into a block of larger size.
Linus Torvalds's avatar
Linus Torvalds committed
652
 *
653
 * -- nyc
Linus Torvalds's avatar
Linus Torvalds committed
654 655
 */

Nick Piggin's avatar
Nick Piggin committed
656
static inline void __free_one_page(struct page *page,
657
		unsigned long pfn,
658 659
		struct zone *zone, unsigned int order,
		int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
660 661
{
	unsigned long page_idx;
662
	unsigned long combined_idx;
663
	unsigned long uninitialized_var(buddy_idx);
664
	struct page *buddy;
665
	int max_order = MAX_ORDER;
Linus Torvalds's avatar
Linus Torvalds committed
666

667
	VM_BUG_ON(!zone_is_initialized(zone));
668
	VM_BUG_ON_PAGE(page->flags & PAGE_FLAGS_CHECK_AT_PREP, page);
Linus Torvalds's avatar
Linus Torvalds committed
669

670
	VM_BUG_ON(migratetype == -1);
671 672 673 674 675 676 677 678 679
	if (is_migrate_isolate(migratetype)) {
		/*
		 * We restrict max order of merging to prevent merge
		 * between freepages on isolate pageblock and normal
		 * pageblock. Without this, pageblock isolation
		 * could cause incorrect freepage accounting.
		 */
		max_order = min(MAX_ORDER, pageblock_order + 1);
	} else {
680
		__mod_zone_freepage_state(zone, 1 << order, migratetype);
681
	}
682

683
	page_idx = pfn & ((1 << max_order) - 1);
Linus Torvalds's avatar
Linus Torvalds committed
684

685 686
	VM_BUG_ON_PAGE(page_idx & ((1 << order) - 1), page);
	VM_BUG_ON_PAGE(bad_range(zone, page), page);
Linus Torvalds's avatar
Linus Torvalds committed
687

688
	while (order < max_order - 1) {
689 690
		buddy_idx = __find_buddy_index(page_idx, order);
		buddy = page + (buddy_idx - page_idx);
691
		if (!page_is_buddy(page, buddy, order))
692
			break;
693 694 695 696 697
		/*
		 * Our buddy is free or it is CONFIG_DEBUG_PAGEALLOC guard page,
		 * merge with it and move up one order.
		 */
		if (page_is_guard(buddy)) {
698
			clear_page_guard(zone, buddy, order, migratetype);
699 700 701 702 703
		} else {
			list_del(&buddy->lru);
			zone->free_area[order].nr_free--;
			rmv_page_order(buddy);
		}
704
		combined_idx = buddy_idx & page_idx;
Linus Torvalds's avatar
Linus Torvalds committed
705 706 707 708 709
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
710 711 712 713 714 715 716 717 718

	/*
	 * If this is not the largest possible page, check if the buddy
	 * of the next-highest order is free. If it is, it's possible
	 * that pages are being freed that will coalesce soon. In case,
	 * that is happening, add the free page to the tail of the list
	 * so it's less likely to be used soon and more likely to be merged
	 * as a higher order page
	 */
719
	if ((order < MAX_ORDER-2) && pfn_valid_within(page_to_pfn(buddy))) {
720
		struct page *higher_page, *higher_buddy;
721 722 723
		combined_idx = buddy_idx & page_idx;
		higher_page = page + (combined_idx - page_idx);
		buddy_idx = __find_buddy_index(combined_idx, order + 1);
724
		higher_buddy = higher_page + (buddy_idx - combined_idx);
725 726 727 728 729 730 731 732 733
		if (page_is_buddy(higher_page, higher_buddy, order + 1)) {
			list_add_tail(&page->lru,
				&zone->free_area[order].free_list[migratetype]);
			goto out;
		}
	}

	list_add(&page->lru, &zone->free_area[order].free_list[migratetype]);
out:
Linus Torvalds's avatar
Linus Torvalds committed
734 735 736
	zone->free_area[order].nr_free++;
}

737
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
738
{
739
	const char *bad_reason = NULL;
740 741 742 743 744 745 746 747 748 749 750 751
	unsigned long bad_flags = 0;

	if (unlikely(page_mapcount(page)))
		bad_reason = "nonzero mapcount";
	if (unlikely(page->mapping != NULL))
		bad_reason = "non-NULL mapping";
	if (unlikely(atomic_read(&page->_count) != 0))
		bad_reason = "nonzero _count";
	if (unlikely(page->flags & PAGE_FLAGS_CHECK_AT_FREE)) {
		bad_reason = "PAGE_FLAGS_CHECK_AT_FREE flag(s) set";
		bad_flags = PAGE_FLAGS_CHECK_AT_FREE;
	}
752 753 754 755
#ifdef CONFIG_MEMCG
	if (unlikely(page->mem_cgroup))
		bad_reason = "page still charged to cgroup";
#endif
756 757
	if (unlikely(bad_reason)) {
		bad_page(page, bad_reason, bad_flags);
758
		return 1;
759
	}
760
	page_cpupid_reset_last(page);
761 762 763
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
764 765 766
}

/*
767
 * Frees a number of pages from the PCP lists
Linus Torvalds's avatar
Linus Torvalds committed
768
 * Assumes all pages on list are in same zone, and of same order.
769
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
770 771 772 773 774 775 776
 *
 * If the zone was previously in an "all pages pinned" state then look to
 * see if this freeing clears that state.
 *
 * And clear the zone's pages_scanned counter, to hold off the "all pages are
 * pinned" detection logic.
 */
777 778
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
Linus Torvalds's avatar
Linus Torvalds committed
779
{
780
	int migratetype = 0;
781
	int batch_free = 0;
782
	int to_free = count;
783
	unsigned long nr_scanned;
784

Nick Piggin's avatar
Nick Piggin committed
785
	spin_lock(&zone->lock);
786 787 788
	nr_scanned = zone_page_state(zone, NR_PAGES_SCANNED);
	if (nr_scanned)
		__mod_zone_page_state(zone, NR_PAGES_SCANNED, -nr_scanned);
789

790
	while (to_free) {
Nick Piggin's avatar
Nick Piggin committed
791
		struct page *page;
792 793 794
		struct list_head *list;

		/*
795 796 797 798 799
		 * Remove pages from lists in a round-robin fashion. A
		 * batch_free count is maintained that is incremented when an
		 * empty list is encountered.  This is so more pages are freed
		 * off fuller lists instead of spinning excessively around empty
		 * lists
800 801
		 */
		do {
802
			batch_free++;
803 804 805 806
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
Nick Piggin's avatar
Nick Piggin committed
807

808 809 810 811
		/* This is the only non-empty list. Free them all. */
		if (batch_free == MIGRATE_PCPTYPES)
			batch_free = to_free;

812
		do {
813 814
			int mt;	/* migratetype of the to-be-freed page */

815 816 817
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
818

819
			mt = get_pcppage_migratetype(page);
820 821 822
			/* MIGRATE_ISOLATE page should not go to pcplists */
			VM_BUG_ON_PAGE(is_migrate_isolate(mt), page);
			/* Pageblock could have been isolated meanwhile */
823
			if (unlikely(has_isolate_pageblock(zone)))
824 825
				mt = get_pageblock_migratetype(page);

826
			__free_one_page(page, page_to_pfn(page), zone, 0, mt);
827
			trace_mm_page_pcpu_drain(page, 0, mt);
828
		} while (--to_free && --batch_free && !list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
829
	}
Nick Piggin's avatar
Nick Piggin committed
830
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
831 832
}

833 834
static void free_one_page(struct zone *zone,
				struct page *page, unsigned long pfn,
835
				unsigned int order,
836
				int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
837
{
838
	unsigned long nr_scanned;
839
	spin_lock(&zone->lock);
840 841 842
	nr_scanned = zone_page_state(zone, NR_PAGES_SCANNED);
	if (nr_scanned)
		__mod_zone_page_state(zone, NR_PAGES_SCANNED, -nr_scanned);
843

844 845 846 847
	if (unlikely(has_isolate_pageblock(zone) ||
		is_migrate_isolate(migratetype))) {
		migratetype = get_pfnblock_migratetype(page, pfn);
	}
848
	__free_one_page(page, pfn, zone, order, migratetype);
849
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
850 851
}

852 853
static int free_tail_pages_check(struct page *head_page, struct page *page)
{
854 855 856 857 858 859 860 861 862 863 864 865
	int ret = 1;

	/*
	 * We rely page->lru.next never has bit 0 set, unless the page
	 * is PageTail(). Let's make sure that's true even for poisoned ->lru.
	 */
	BUILD_BUG_ON((unsigned long)LIST_POISON1 & 1);

	if (!IS_ENABLED(CONFIG_DEBUG_VM)) {
		ret = 0;
		goto out;
	}
866 867
	if (unlikely(!PageTail(page))) {
		bad_page(page, "PageTail not set", 0);
868
		goto out;
869
	}
870 871 872
	if (unlikely(compound_head(page) != head_page)) {
		bad_page(page, "compound_head not consistent", 0);
		goto out;
873
	}
874 875 876 877
	ret = 0;
out:
	clear_compound_head(page);
	return ret;
878 879
}

880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901
static void __meminit __init_single_page(struct page *page, unsigned long pfn,
				unsigned long zone, int nid)
{
	set_page_links(page, zone, nid, pfn);
	init_page_count(page);
	page_mapcount_reset(page);
	page_cpupid_reset_last(page);

	INIT_LIST_HEAD(&page->lru);
#ifdef WANT_PAGE_VIRTUAL
	/* The shift won't overflow because ZONE_NORMAL is below 4G. */
	if (!is_highmem_idx(zone))
		set_page_address(page, __va(pfn << PAGE_SHIFT));
#endif
}

static void __meminit __init_single_pfn(unsigned long pfn, unsigned long zone,
					int nid)
{
	return __init_single_page(pfn_to_page(pfn), pfn, zone, nid);
}

902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927
#ifdef CONFIG_DEFERRED_STRUCT_PAGE_INIT
static void init_reserved_page(unsigned long pfn)
{
	pg_data_t *pgdat;
	int nid, zid;

	if (!early_page_uninitialised(pfn))
		return;

	nid = early_pfn_to_nid(pfn);
	pgdat = NODE_DATA(nid);

	for (zid = 0; zid < MAX_NR_ZONES; zid++) {
		struct zone *zone = &pgdat->node_zones[zid];

		if (pfn >= zone->zone_start_pfn && pfn < zone_end_pfn(zone))
			break;
	}
	__init_single_pfn(pfn, zid, nid);
}
#else
static inline void init_reserved_page(unsigned long pfn)
{
}
#endif /* CONFIG_DEFERRED_STRUCT_PAGE_INIT */

928 929 930 931 932 933
/*
 * Initialised pages do not have PageReserved set. This function is
 * called for each range allocated by the bootmem allocator and
 * marks the pages PageReserved. The remaining valid pages are later
 * sent to the buddy page allocator.
 */
934
void __meminit reserve_bootmem_region(unsigned long start, unsigned long end)
935 936 937 938
{
	unsigned long start_pfn = PFN_DOWN(start);
	unsigned long end_pfn = PFN_UP(end);

939 940 941 942 943
	for (; start_pfn < end_pfn; start_pfn++) {
		if (pfn_valid(start_pfn)) {
			struct page *page = pfn_to_page(start_pfn);

			init_reserved_page(start_pfn);
944 945 946 947

			/* Avoid false-positive PageTail() */
			INIT_LIST_HEAD(&page->lru);

948 949 950
			SetPageReserved(page);
		}
	}
951 952
}

953
static bool free_pages_prepare(struct page *page, unsigned int order)
Nick Piggin's avatar
Nick Piggin committed
954
{
955 956
	bool compound = PageCompound(page);
	int i, bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
957

958
	VM_BUG_ON_PAGE(PageTail(page), page);
959
	VM_BUG_ON_PAGE(compound && compound_order(page) != order<