page_alloc.c 187 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>
Linus Torvalds's avatar
Linus Torvalds committed
64

65
#include <asm/sections.h>
Linus Torvalds's avatar
Linus Torvalds committed
66
#include <asm/tlbflush.h>
67
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
68 69
#include "internal.h"

70 71
/* prevent >1 _updater_ of zone percpu pageset ->high and ->batch fields */
static DEFINE_MUTEX(pcp_batch_high_lock);
72
#define MIN_PERCPU_PAGELIST_FRACTION	(8)
73

74 75 76 77 78
#ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID
DEFINE_PER_CPU(int, numa_node);
EXPORT_PER_CPU_SYMBOL(numa_node);
#endif

79 80 81 82 83 84 85 86 87
#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_);
88
int _node_numa_mem_[MAX_NUMNODES];
89 90
#endif

Linus Torvalds's avatar
Linus Torvalds committed
91
/*
92
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
93
 */
94 95 96 97 98 99 100
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 } },
101 102 103
#endif
#ifdef CONFIG_MOVABLE_NODE
	[N_MEMORY] = { { [0] = 1UL } },
104 105 106 107 108 109
#endif
	[N_CPU] = { { [0] = 1UL } },
#endif	/* NUMA */
};
EXPORT_SYMBOL(node_states);

110 111 112
/* Protect totalram_pages and zone->managed_pages */
static DEFINE_SPINLOCK(managed_page_count_lock);

113
unsigned long totalram_pages __read_mostly;
114
unsigned long totalreserve_pages __read_mostly;
115
unsigned long totalcma_pages __read_mostly;
116 117 118 119 120 121 122 123
/*
 * 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;

124
int percpu_pagelist_fraction;
125
gfp_t gfp_allowed_mask __read_mostly = GFP_BOOT_MASK;
Linus Torvalds's avatar
Linus Torvalds committed
126

127 128 129 130 131 132 133 134 135
#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).
 */
136 137 138 139

static gfp_t saved_gfp_mask;

void pm_restore_gfp_mask(void)
140 141
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
142 143 144 145
	if (saved_gfp_mask) {
		gfp_allowed_mask = saved_gfp_mask;
		saved_gfp_mask = 0;
	}
146 147
}

148
void pm_restrict_gfp_mask(void)
149 150
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
151 152 153
	WARN_ON(saved_gfp_mask);
	saved_gfp_mask = gfp_allowed_mask;
	gfp_allowed_mask &= ~GFP_IOFS;
154
}
155 156 157 158 159 160 161

bool pm_suspended_storage(void)
{
	if ((gfp_allowed_mask & GFP_IOFS) == GFP_IOFS)
		return false;
	return true;
}
162 163
#endif /* CONFIG_PM_SLEEP */

164 165 166 167
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

168
static void __free_pages_ok(struct page *page, unsigned int order);
169

Linus Torvalds's avatar
Linus Torvalds committed
170 171 172 173 174 175
/*
 * 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
176
 *	HIGHMEM allocation will leave (224M+784M)/256 of ram reserved in ZONE_DMA
177 178 179
 *
 * 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
180
 */
181
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
182
#ifdef CONFIG_ZONE_DMA
183
	 256,
184
#endif
185
#ifdef CONFIG_ZONE_DMA32
186
	 256,
187
#endif
188
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
189
	 32,
190
#endif
Mel Gorman's avatar
Mel Gorman committed
191
	 32,
192
};
Linus Torvalds's avatar
Linus Torvalds committed
193 194 195

EXPORT_SYMBOL(totalram_pages);

196
static char * const zone_names[MAX_NR_ZONES] = {
197
#ifdef CONFIG_ZONE_DMA
198
	 "DMA",
199
#endif
200
#ifdef CONFIG_ZONE_DMA32
201
	 "DMA32",
202
#endif
203
	 "Normal",
204
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
205
	 "HighMem",
206
#endif
Mel Gorman's avatar
Mel Gorman committed
207
	 "Movable",
208 209
};

Linus Torvalds's avatar
Linus Torvalds committed
210
int min_free_kbytes = 1024;
211
int user_min_free_kbytes = -1;
Linus Torvalds's avatar
Linus Torvalds committed
212

213 214
static unsigned long __meminitdata nr_kernel_pages;
static unsigned long __meminitdata nr_all_pages;
215
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
216

Tejun Heo's avatar
Tejun Heo committed
217 218 219 220 221 222 223 224 225 226 227
#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 */
228

Miklos Szeredi's avatar
Miklos Szeredi committed
229 230
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
231
int nr_online_nodes __read_mostly = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
232
EXPORT_SYMBOL(nr_node_ids);
233
EXPORT_SYMBOL(nr_online_nodes);
Miklos Szeredi's avatar
Miklos Szeredi committed
234 235
#endif

236 237
int page_group_by_mobility_disabled __read_mostly;

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
#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 */
static inline bool __defermem_init early_page_uninitialised(unsigned long pfn)
{
	int nid = early_pfn_to_nid(pfn);

	if (pfn >= NODE_DATA(nid)->first_deferred_pfn)
		return true;

	return false;
}

/*
 * 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;
}

static inline bool update_defer_init(pg_data_t *pgdat,
				unsigned long pfn, unsigned long zone_end,
				unsigned long *nr_initialised)
{
	return true;
}
#endif


296
void set_pageblock_migratetype(struct page *page, int migratetype)
297
{
298 299
	if (unlikely(page_group_by_mobility_disabled &&
		     migratetype < MIGRATE_PCPTYPES))
300 301
		migratetype = MIGRATE_UNMOVABLE;

302 303 304 305
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

Nick Piggin's avatar
Nick Piggin committed
306
#ifdef CONFIG_DEBUG_VM
307
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
308
{
309 310 311
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
312
	unsigned long sp, start_pfn;
313

314 315
	do {
		seq = zone_span_seqbegin(zone);
316 317
		start_pfn = zone->zone_start_pfn;
		sp = zone->spanned_pages;
318
		if (!zone_spans_pfn(zone, pfn))
319 320 321
			ret = 1;
	} while (zone_span_seqretry(zone, seq));

322
	if (ret)
323 324 325
		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);
326

327
	return ret;
328 329 330 331
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
332
	if (!pfn_valid_within(page_to_pfn(page)))
333
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
334
	if (zone != page_zone(page))
335 336 337 338 339 340 341 342 343 344
		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
345
		return 1;
346 347 348
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
349 350
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
351 352 353 354 355 356 357
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

358 359
static void bad_page(struct page *page, const char *reason,
		unsigned long bad_flags)
Linus Torvalds's avatar
Linus Torvalds committed
360
{
361 362 363 364
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

365 366
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
367
		page_mapcount_reset(page); /* remove PageBuddy */
368 369 370
		return;
	}

371 372 373 374 375 376 377 378 379 380
	/*
	 * 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) {
381 382
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
383 384 385 386 387 388 389 390
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

391
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
392
		current->comm, page_to_pfn(page));
393
	dump_page_badflags(page, reason, bad_flags);
394

395
	print_modules();
Linus Torvalds's avatar
Linus Torvalds committed
396
	dump_stack();
397
out:
398
	/* Leave bad fields for debug, except PageBuddy could make trouble */
399
	page_mapcount_reset(page); /* remove PageBuddy */
400
	add_taint(TAINT_BAD_PAGE, LOCKDEP_NOW_UNRELIABLE);
Linus Torvalds's avatar
Linus Torvalds committed
401 402 403 404 405 406 407 408 409
}

/*
 * Higher-order pages are called "compound pages".  They are structured thusly:
 *
 * The first PAGE_SIZE page is called the "head page".
 *
 * The remaining PAGE_SIZE pages are called "tail pages".
 *
410 411
 * All pages have PG_compound set.  All tail pages have their ->first_page
 * pointing at the head page.
Linus Torvalds's avatar
Linus Torvalds committed
412
 *
413 414 415
 * The first tail page's ->lru.next holds the address of the compound page's
 * put_page() function.  Its ->lru.prev holds the order of allocation.
 * This usage means that zero-order pages may not be compound.
Linus Torvalds's avatar
Linus Torvalds committed
416
 */
417 418 419

static void free_compound_page(struct page *page)
{
420
	__free_pages_ok(page, compound_order(page));
421 422
}

423
void prep_compound_page(struct page *page, unsigned long order)
424 425 426 427 428 429 430 431 432
{
	int i;
	int nr_pages = 1 << order;

	set_compound_page_dtor(page, free_compound_page);
	set_compound_order(page, order);
	__SetPageHead(page);
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
433
		set_page_count(p, 0);
434
		p->first_page = page;
David Rientjes's avatar
David Rientjes committed
435 436 437
		/* Make sure p->first_page is always valid for PageTail() */
		smp_wmb();
		__SetPageTail(p);
438 439 440
	}
}

441 442
#ifdef CONFIG_DEBUG_PAGEALLOC
unsigned int _debug_guardpage_minorder;
443
bool _debug_pagealloc_enabled __read_mostly;
444 445
bool _debug_guardpage_enabled __read_mostly;

446 447 448 449 450 451 452 453 454 455 456 457
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);

458 459
static bool need_debug_guardpage(void)
{
460 461 462 463
	/* If we don't use debug_pagealloc, we don't need guard page */
	if (!debug_pagealloc_enabled())
		return false;

464 465 466 467 468
	return true;
}

static void init_debug_guardpage(void)
{
469 470 471
	if (!debug_pagealloc_enabled())
		return;

472 473 474 475 476 477 478
	_debug_guardpage_enabled = true;
}

struct page_ext_operations debug_guardpage_ops = {
	.need = need_debug_guardpage,
	.init = init_debug_guardpage,
};
479 480 481 482 483 484 485 486 487 488 489 490 491 492 493

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);

494 495
static inline void set_page_guard(struct zone *zone, struct page *page,
				unsigned int order, int migratetype)
496
{
497 498 499 500 501 502 503 504
	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);

505 506 507 508
	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);
509 510
}

511 512
static inline void clear_page_guard(struct zone *zone, struct page *page,
				unsigned int order, int migratetype)
513
{
514 515 516 517 518 519 520 521
	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);

522 523 524
	set_page_private(page, 0);
	if (!is_migrate_isolate(migratetype))
		__mod_zone_freepage_state(zone, (1 << order), migratetype);
525 526
}
#else
527
struct page_ext_operations debug_guardpage_ops = { NULL, };
528 529 530 531
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) {}
532 533
#endif

534
static inline void set_page_order(struct page *page, unsigned int order)
535
{
536
	set_page_private(page, order);
537
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
538 539 540 541
}

static inline void rmv_page_order(struct page *page)
{
542
	__ClearPageBuddy(page);
543
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
544 545 546 547 548
}

/*
 * 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
549
 * (a) the buddy is not in a hole &&
550
 * (b) the buddy is in the buddy system &&
551 552
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
553
 *
554 555 556 557
 * 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
558
 *
559
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
560
 */
561
static inline int page_is_buddy(struct page *page, struct page *buddy,
562
							unsigned int order)
Linus Torvalds's avatar
Linus Torvalds committed
563
{
564
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
565 566
		return 0;

567
	if (page_is_guard(buddy) && page_order(buddy) == order) {
568 569 570
		if (page_zone_id(page) != page_zone_id(buddy))
			return 0;

571 572
		VM_BUG_ON_PAGE(page_count(buddy) != 0, buddy);

573 574 575
		return 1;
	}

576
	if (PageBuddy(buddy) && page_order(buddy) == order) {
577 578 579 580 581 582 583 584
		/*
		 * 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;

585 586
		VM_BUG_ON_PAGE(page_count(buddy) != 0, buddy);

587
		return 1;
588
	}
589
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
}

/*
 * 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
605 606 607
 * 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
608
 * So when we are allocating or freeing one, we can derive the state of the
609 610
 * 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
611
 * If a block is freed, and its buddy is also free, then this
612
 * triggers coalescing into a block of larger size.
Linus Torvalds's avatar
Linus Torvalds committed
613
 *
614
 * -- nyc
Linus Torvalds's avatar
Linus Torvalds committed
615 616
 */

Nick Piggin's avatar
Nick Piggin committed
617
static inline void __free_one_page(struct page *page,
618
		unsigned long pfn,
619 620
		struct zone *zone, unsigned int order,
		int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
621 622
{
	unsigned long page_idx;
623
	unsigned long combined_idx;
624
	unsigned long uninitialized_var(buddy_idx);
625
	struct page *buddy;
626
	int max_order = MAX_ORDER;
Linus Torvalds's avatar
Linus Torvalds committed
627

628
	VM_BUG_ON(!zone_is_initialized(zone));
629
	VM_BUG_ON_PAGE(page->flags & PAGE_FLAGS_CHECK_AT_PREP, page);
Linus Torvalds's avatar
Linus Torvalds committed
630

631
	VM_BUG_ON(migratetype == -1);
632 633 634 635 636 637 638 639 640
	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 {
641
		__mod_zone_freepage_state(zone, 1 << order, migratetype);
642
	}
643

644
	page_idx = pfn & ((1 << max_order) - 1);
Linus Torvalds's avatar
Linus Torvalds committed
645

646 647
	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
648

649
	while (order < max_order - 1) {
650 651
		buddy_idx = __find_buddy_index(page_idx, order);
		buddy = page + (buddy_idx - page_idx);
652
		if (!page_is_buddy(page, buddy, order))
653
			break;
654 655 656 657 658
		/*
		 * 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)) {
659
			clear_page_guard(zone, buddy, order, migratetype);
660 661 662 663 664
		} else {
			list_del(&buddy->lru);
			zone->free_area[order].nr_free--;
			rmv_page_order(buddy);
		}
665
		combined_idx = buddy_idx & page_idx;
Linus Torvalds's avatar
Linus Torvalds committed
666 667 668 669 670
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
671 672 673 674 675 676 677 678 679

	/*
	 * 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
	 */
680
	if ((order < MAX_ORDER-2) && pfn_valid_within(page_to_pfn(buddy))) {
681
		struct page *higher_page, *higher_buddy;
682 683 684
		combined_idx = buddy_idx & page_idx;
		higher_page = page + (combined_idx - page_idx);
		buddy_idx = __find_buddy_index(combined_idx, order + 1);
685
		higher_buddy = higher_page + (buddy_idx - combined_idx);
686 687 688 689 690 691 692 693 694
		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
695 696 697
	zone->free_area[order].nr_free++;
}

698
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
699
{
700
	const char *bad_reason = NULL;
701 702 703 704 705 706 707 708 709 710 711 712
	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;
	}
713 714 715 716
#ifdef CONFIG_MEMCG
	if (unlikely(page->mem_cgroup))
		bad_reason = "page still charged to cgroup";
#endif
717 718
	if (unlikely(bad_reason)) {
		bad_page(page, bad_reason, bad_flags);
719
		return 1;
720
	}
721
	page_cpupid_reset_last(page);
722 723 724
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
725 726 727
}

/*
728
 * Frees a number of pages from the PCP lists
Linus Torvalds's avatar
Linus Torvalds committed
729
 * Assumes all pages on list are in same zone, and of same order.
730
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
731 732 733 734 735 736 737
 *
 * 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.
 */
738 739
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
Linus Torvalds's avatar
Linus Torvalds committed
740
{
741
	int migratetype = 0;
742
	int batch_free = 0;
743
	int to_free = count;
744
	unsigned long nr_scanned;
745

Nick Piggin's avatar
Nick Piggin committed
746
	spin_lock(&zone->lock);
747 748 749
	nr_scanned = zone_page_state(zone, NR_PAGES_SCANNED);
	if (nr_scanned)
		__mod_zone_page_state(zone, NR_PAGES_SCANNED, -nr_scanned);
750

751
	while (to_free) {
Nick Piggin's avatar
Nick Piggin committed
752
		struct page *page;
753 754 755
		struct list_head *list;

		/*
756 757 758 759 760
		 * 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
761 762
		 */
		do {
763
			batch_free++;
764 765 766 767
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
Nick Piggin's avatar
Nick Piggin committed
768

769 770 771 772
		/* This is the only non-empty list. Free them all. */
		if (batch_free == MIGRATE_PCPTYPES)
			batch_free = to_free;

773
		do {
774 775
			int mt;	/* migratetype of the to-be-freed page */

776 777 778
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
779
			mt = get_freepage_migratetype(page);
780
			if (unlikely(has_isolate_pageblock(zone)))
781 782
				mt = get_pageblock_migratetype(page);

783
			/* MIGRATE_MOVABLE list may include MIGRATE_RESERVEs */
784
			__free_one_page(page, page_to_pfn(page), zone, 0, mt);
785
			trace_mm_page_pcpu_drain(page, 0, mt);
786
		} while (--to_free && --batch_free && !list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
787
	}
Nick Piggin's avatar
Nick Piggin committed
788
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
789 790
}

791 792
static void free_one_page(struct zone *zone,
				struct page *page, unsigned long pfn,
793
				unsigned int order,
794
				int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
795
{
796
	unsigned long nr_scanned;
797
	spin_lock(&zone->lock);
798 799 800
	nr_scanned = zone_page_state(zone, NR_PAGES_SCANNED);
	if (nr_scanned)
		__mod_zone_page_state(zone, NR_PAGES_SCANNED, -nr_scanned);
801

802 803 804 805
	if (unlikely(has_isolate_pageblock(zone) ||
		is_migrate_isolate(migratetype))) {
		migratetype = get_pfnblock_migratetype(page, pfn);
	}
806
	__free_one_page(page, pfn, zone, order, migratetype);
807
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
808 809
}

810 811 812 813 814 815 816 817 818 819 820 821 822 823 824
static int free_tail_pages_check(struct page *head_page, struct page *page)
{
	if (!IS_ENABLED(CONFIG_DEBUG_VM))
		return 0;
	if (unlikely(!PageTail(page))) {
		bad_page(page, "PageTail not set", 0);
		return 1;
	}
	if (unlikely(page->first_page != head_page)) {
		bad_page(page, "first_page not consistent", 0);
		return 1;
	}
	return 0;
}

825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868
static void __meminit __init_single_page(struct page *page, unsigned long pfn,
				unsigned long zone, int nid)
{
	struct zone *z = &NODE_DATA(nid)->node_zones[zone];

	set_page_links(page, zone, nid, pfn);
	mminit_verify_page_links(page, zone, nid, pfn);
	init_page_count(page);
	page_mapcount_reset(page);
	page_cpupid_reset_last(page);

	/*
	 * Mark the block movable so that blocks are reserved for
	 * movable at startup. This will force kernel allocations
	 * to reserve their blocks rather than leaking throughout
	 * the address space during boot when many long-lived
	 * kernel allocations are made. Later some blocks near
	 * the start are marked MIGRATE_RESERVE by
	 * setup_zone_migrate_reserve()
	 *
	 * bitmap is created for zone's valid pfn range. but memmap
	 * can be created for invalid pages (for alignment)
	 * check here not to call set_pageblock_migratetype() against
	 * pfn out of zone.
	 */
	if ((z->zone_start_pfn <= pfn)
	    && (pfn < zone_end_pfn(z))
	    && !(pfn & (pageblock_nr_pages - 1)))
		set_pageblock_migratetype(page, MIGRATE_MOVABLE);

	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);
}

869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884
/*
 * 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.
 */
void reserve_bootmem_region(unsigned long start, unsigned long end)
{
	unsigned long start_pfn = PFN_DOWN(start);
	unsigned long end_pfn = PFN_UP(end);

	for (; start_pfn < end_pfn; start_pfn++)
		if (pfn_valid(start_pfn))
			SetPageReserved(pfn_to_page(start_pfn));
}

885
static bool free_pages_prepare(struct page *page, unsigned int order)
Nick Piggin's avatar
Nick Piggin committed
886
{
887 888
	bool compound = PageCompound(page);
	int i, bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
889

890
	VM_BUG_ON_PAGE(PageTail(page), page);
891
	VM_BUG_ON_PAGE(compound && compound_order(page) != order, page);
892

893
	trace_mm_page_free(page, order);
894
	kmemcheck_free_shadow(page, order);
895
	kasan_free_pages(page, order);
896

Andrea Arcangeli's avatar
Andrea Arcangeli committed
897 898
	if (PageAnon(page))
		page->mapping = NULL;
899 900 901 902
	bad += free_pages_check(page);
	for (i = 1; i < (1 << order); i++) {
		if (compound)
			bad += free_tail_pages_check(page, page + i);
Andrea Arcangeli's avatar
Andrea Arcangeli committed
903
		bad += free_pages_check(page + i);
904
	}
905
	if (bad)
906
		return false;
907

908 909
	reset_page_owner(page, order);

910
	if (!PageHighMem(page)) {
911 912
		debug_check_no_locks_freed(page_address(page),
					   PAGE_SIZE << order);
913 914 915
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
916
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
917
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
918

919 920 921 922 923 924
	return true;
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
925
	int migratetype;
926
	unsigned long pfn = page_to_pfn(page);
927 928 929 930

	if (!free_pages_prepare(page, order))
		return;

931
	migratetype = get_pfnblock_migratetype(page, pfn);
Nick Piggin's avatar
Nick Piggin committed
932
	local_irq_save(flags);
933
	__count_vm_events(PGFREE, 1 << order);
934
	set_freepage_migratetype(page, migratetype);
935
	free_one_page(page_zone(page), page, pfn, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
936
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
937 938
}

939 940
static void __defer_init __free_pages_boot_core(struct page *page,
					unsigned long pfn, unsigned int order)
941
{
942
	unsigned int nr_pages = 1 << order;
943
	struct page *p = page;
944
	unsigned int loop;
945

946 947 948
	prefetchw(p);
	for (loop = 0; loop < (nr_pages - 1); loop++, p++) {
		prefetchw(p + 1);
949 950
		__ClearPageReserved(p);
		set_page_count(p, 0);