page_alloc.c 157 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>
Linus Torvalds's avatar
Linus Torvalds committed
28
29
30
31
32
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
33
#include <linux/ratelimit.h>
34
#include <linux/oom.h>
Linus Torvalds's avatar
Linus Torvalds committed
35
36
37
38
39
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
40
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
41
42
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
43
#include <linux/vmstat.h>
44
#include <linux/mempolicy.h>
45
#include <linux/stop_machine.h>
46
47
#include <linux/sort.h>
#include <linux/pfn.h>
48
#include <linux/backing-dev.h>
49
#include <linux/fault-inject.h>
50
#include <linux/page-isolation.h>
51
#include <linux/page_cgroup.h>
52
#include <linux/debugobjects.h>
53
#include <linux/kmemleak.h>
54
#include <linux/memory.h>
55
#include <linux/compaction.h>
56
#include <trace/events/kmem.h>
57
#include <linux/ftrace_event.h>
58
#include <linux/memcontrol.h>
59
#include <linux/prefetch.h>
60
#include <linux/page-debug-flags.h>
Linus Torvalds's avatar
Linus Torvalds committed
61
62

#include <asm/tlbflush.h>
63
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
64
65
#include "internal.h"

66
67
68
69
70
#ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID
DEFINE_PER_CPU(int, numa_node);
EXPORT_PER_CPU_SYMBOL(numa_node);
#endif

71
72
73
74
75
76
77
78
79
80
81
#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_);
#endif

Linus Torvalds's avatar
Linus Torvalds committed
82
/*
83
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
84
 */
85
86
87
88
89
90
91
92
93
94
95
96
97
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 } },
#endif
	[N_CPU] = { { [0] = 1UL } },
#endif	/* NUMA */
};
EXPORT_SYMBOL(node_states);

98
unsigned long totalram_pages __read_mostly;
99
unsigned long totalreserve_pages __read_mostly;
100
101
102
103
104
105
106
107
/*
 * 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;

108
int percpu_pagelist_fraction;
109
gfp_t gfp_allowed_mask __read_mostly = GFP_BOOT_MASK;
Linus Torvalds's avatar
Linus Torvalds committed
110

111
112
113
114
115
116
117
118
119
#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).
 */
120
121
122
123

static gfp_t saved_gfp_mask;

void pm_restore_gfp_mask(void)
124
125
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
126
127
128
129
	if (saved_gfp_mask) {
		gfp_allowed_mask = saved_gfp_mask;
		saved_gfp_mask = 0;
	}
130
131
}

132
void pm_restrict_gfp_mask(void)
133
134
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
135
136
137
	WARN_ON(saved_gfp_mask);
	saved_gfp_mask = gfp_allowed_mask;
	gfp_allowed_mask &= ~GFP_IOFS;
138
}
139
140
141
142
143
144
145

bool pm_suspended_storage(void)
{
	if ((gfp_allowed_mask & GFP_IOFS) == GFP_IOFS)
		return false;
	return true;
}
146
147
#endif /* CONFIG_PM_SLEEP */

148
149
150
151
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

152
static void __free_pages_ok(struct page *page, unsigned int order);
153

Linus Torvalds's avatar
Linus Torvalds committed
154
155
156
157
158
159
160
/*
 * 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
 *	HIGHMEM allocation will (224M+784M)/256 of ram reserved in ZONE_DMA
161
162
163
 *
 * 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
164
 */
165
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
166
#ifdef CONFIG_ZONE_DMA
167
	 256,
168
#endif
169
#ifdef CONFIG_ZONE_DMA32
170
	 256,
171
#endif
172
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
173
	 32,
174
#endif
Mel Gorman's avatar
Mel Gorman committed
175
	 32,
176
};
Linus Torvalds's avatar
Linus Torvalds committed
177
178
179

EXPORT_SYMBOL(totalram_pages);

180
static char * const zone_names[MAX_NR_ZONES] = {
181
#ifdef CONFIG_ZONE_DMA
182
	 "DMA",
183
#endif
184
#ifdef CONFIG_ZONE_DMA32
185
	 "DMA32",
186
#endif
187
	 "Normal",
188
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
189
	 "HighMem",
190
#endif
Mel Gorman's avatar
Mel Gorman committed
191
	 "Movable",
192
193
};

Linus Torvalds's avatar
Linus Torvalds committed
194
195
int min_free_kbytes = 1024;

196
197
static unsigned long __meminitdata nr_kernel_pages;
static unsigned long __meminitdata nr_all_pages;
198
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
199

Tejun Heo's avatar
Tejun Heo committed
200
201
202
203
204
205
206
207
208
209
210
#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 */
211

Miklos Szeredi's avatar
Miklos Szeredi committed
212
213
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
214
int nr_online_nodes __read_mostly = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
215
EXPORT_SYMBOL(nr_node_ids);
216
EXPORT_SYMBOL(nr_online_nodes);
Miklos Szeredi's avatar
Miklos Szeredi committed
217
218
#endif

219
220
int page_group_by_mobility_disabled __read_mostly;

221
222
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
223
224
225
226

	if (unlikely(page_group_by_mobility_disabled))
		migratetype = MIGRATE_UNMOVABLE;

227
228
229
230
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

231
232
bool oom_killer_disabled __read_mostly;

Nick Piggin's avatar
Nick Piggin committed
233
#ifdef CONFIG_DEBUG_VM
234
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
235
{
236
237
238
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
239

240
241
242
243
244
245
246
247
248
	do {
		seq = zone_span_seqbegin(zone);
		if (pfn >= zone->zone_start_pfn + zone->spanned_pages)
			ret = 1;
		else if (pfn < zone->zone_start_pfn)
			ret = 1;
	} while (zone_span_seqretry(zone, seq));

	return ret;
249
250
251
252
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
253
	if (!pfn_valid_within(page_to_pfn(page)))
254
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
255
	if (zone != page_zone(page))
256
257
258
259
260
261
262
263
264
265
		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
266
		return 1;
267
268
269
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
270
271
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
272
273
274
275
276
277
278
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

279
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
280
{
281
282
283
284
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

285
286
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
287
		reset_page_mapcount(page); /* remove PageBuddy */
288
289
290
		return;
	}

291
292
293
294
295
296
297
298
299
300
	/*
	 * 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) {
301
302
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
303
304
305
306
307
308
309
310
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

311
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
312
		current->comm, page_to_pfn(page));
313
	dump_page(page);
314

315
	print_modules();
Linus Torvalds's avatar
Linus Torvalds committed
316
	dump_stack();
317
out:
318
	/* Leave bad fields for debug, except PageBuddy could make trouble */
319
	reset_page_mapcount(page); /* remove PageBuddy */
320
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
321
322
323
324
325
326
327
328
329
}

/*
 * 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".
 *
330
331
 * 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
332
 *
333
334
335
 * 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
336
 */
337
338
339

static void free_compound_page(struct page *page)
{
340
	__free_pages_ok(page, compound_order(page));
341
342
}

343
void prep_compound_page(struct page *page, unsigned long order)
344
345
346
347
348
349
350
351
352
353
{
	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;
		__SetPageTail(p);
354
		set_page_count(p, 0);
355
356
357
358
		p->first_page = page;
	}
}

359
/* update __split_huge_page_refcount if you change this function */
360
static int destroy_compound_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
361
362
363
{
	int i;
	int nr_pages = 1 << order;
364
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
365

366
367
	if (unlikely(compound_order(page) != order) ||
	    unlikely(!PageHead(page))) {
368
		bad_page(page);
369
370
		bad++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
371

372
	__ClearPageHead(page);
373

374
375
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
Linus Torvalds's avatar
Linus Torvalds committed
376

377
		if (unlikely(!PageTail(p) || (p->first_page != page))) {
378
			bad_page(page);
379
380
			bad++;
		}
381
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
382
	}
383
384

	return bad;
Linus Torvalds's avatar
Linus Torvalds committed
385
386
}

Nick Piggin's avatar
Nick Piggin committed
387
388
389
390
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

391
392
393
394
	/*
	 * clear_highpage() will use KM_USER0, so it's a bug to use __GFP_ZERO
	 * and __GFP_HIGHMEM from hard or soft interrupt context.
	 */
Nick Piggin's avatar
Nick Piggin committed
395
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
396
397
398
399
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
#ifdef CONFIG_DEBUG_PAGEALLOC
unsigned int _debug_guardpage_minorder;

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);

static inline void set_page_guard_flag(struct page *page)
{
	__set_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags);
}

static inline void clear_page_guard_flag(struct page *page)
{
	__clear_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags);
}
#else
static inline void set_page_guard_flag(struct page *page) { }
static inline void clear_page_guard_flag(struct page *page) { }
#endif

431
432
static inline void set_page_order(struct page *page, int order)
{
433
	set_page_private(page, order);
434
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
435
436
437
438
}

static inline void rmv_page_order(struct page *page)
{
439
	__ClearPageBuddy(page);
440
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
}

/*
 * Locate the struct page for both the matching buddy in our
 * pair (buddy1) and the combined O(n+1) page they form (page).
 *
 * 1) Any buddy B1 will have an order O twin B2 which satisfies
 * the following equation:
 *     B2 = B1 ^ (1 << O)
 * For example, if the starting buddy (buddy2) is #8 its order
 * 1 buddy is #10:
 *     B2 = 8 ^ (1 << 1) = 8 ^ 2 = 10
 *
 * 2) Any buddy B will have an order O+1 parent P which
 * satisfies the following equation:
 *     P = B & ~(1 << O)
 *
Andreas Mohr's avatar
Andreas Mohr committed
458
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
459
460
 */
static inline unsigned long
461
__find_buddy_index(unsigned long page_idx, unsigned int order)
Linus Torvalds's avatar
Linus Torvalds committed
462
{
463
	return page_idx ^ (1 << order);
Linus Torvalds's avatar
Linus Torvalds committed
464
465
466
467
468
}

/*
 * 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
469
 * (a) the buddy is not in a hole &&
470
 * (b) the buddy is in the buddy system &&
471
472
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
473
 *
Andrea Arcangeli's avatar
Andrea Arcangeli committed
474
475
 * For recording whether a page is in the buddy system, we set ->_mapcount -2.
 * Setting, clearing, and testing _mapcount -2 is serialized by zone->lock.
Linus Torvalds's avatar
Linus Torvalds committed
476
 *
477
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
478
 */
479
480
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
481
{
482
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
483
484
		return 0;

485
486
487
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

488
489
490
491
492
	if (page_is_guard(buddy) && page_order(buddy) == order) {
		VM_BUG_ON(page_count(buddy) != 0);
		return 1;
	}

493
	if (PageBuddy(buddy) && page_order(buddy) == order) {
494
		VM_BUG_ON(page_count(buddy) != 0);
495
		return 1;
496
	}
497
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
}

/*
 * 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
Andrea Arcangeli's avatar
Andrea Arcangeli committed
513
 * free pages of length of (1 << order) and marked with _mapcount -2. Page's
514
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
515
516
517
518
519
520
521
522
523
 * So when we are allocating or freeing one, we can derive the state of the
 * other.  That is, if we allocate a small block, and both were   
 * free, the remainder of the region must be split into blocks.   
 * If a block is freed, and its buddy is also free, then this
 * triggers coalescing into a block of larger size.            
 *
 * -- wli
 */

Nick Piggin's avatar
Nick Piggin committed
524
static inline void __free_one_page(struct page *page,
525
526
		struct zone *zone, unsigned int order,
		int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
527
528
{
	unsigned long page_idx;
529
	unsigned long combined_idx;
530
	unsigned long uninitialized_var(buddy_idx);
531
	struct page *buddy;
Linus Torvalds's avatar
Linus Torvalds committed
532

533
	if (unlikely(PageCompound(page)))
534
535
		if (unlikely(destroy_compound_page(page, order)))
			return;
Linus Torvalds's avatar
Linus Torvalds committed
536

537
538
	VM_BUG_ON(migratetype == -1);

Linus Torvalds's avatar
Linus Torvalds committed
539
540
	page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);

541
	VM_BUG_ON(page_idx & ((1 << order) - 1));
Nick Piggin's avatar
Nick Piggin committed
542
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
543
544

	while (order < MAX_ORDER-1) {
545
546
		buddy_idx = __find_buddy_index(page_idx, order);
		buddy = page + (buddy_idx - page_idx);
547
		if (!page_is_buddy(page, buddy, order))
548
			break;
549
550
551
552
553
554
555
556
557
558
559
560
561
		/*
		 * 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)) {
			clear_page_guard_flag(buddy);
			set_page_private(page, 0);
			__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
		} else {
			list_del(&buddy->lru);
			zone->free_area[order].nr_free--;
			rmv_page_order(buddy);
		}
562
		combined_idx = buddy_idx & page_idx;
Linus Torvalds's avatar
Linus Torvalds committed
563
564
565
566
567
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
568
569
570
571
572
573
574
575
576

	/*
	 * 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
	 */
577
	if ((order < MAX_ORDER-2) && pfn_valid_within(page_to_pfn(buddy))) {
578
		struct page *higher_page, *higher_buddy;
579
580
581
582
		combined_idx = buddy_idx & page_idx;
		higher_page = page + (combined_idx - page_idx);
		buddy_idx = __find_buddy_index(combined_idx, order + 1);
		higher_buddy = page + (buddy_idx - combined_idx);
583
584
585
586
587
588
589
590
591
		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
592
593
594
	zone->free_area[order].nr_free++;
}

595
596
597
598
599
600
601
602
603
604
605
/*
 * free_page_mlock() -- clean up attempts to free and mlocked() page.
 * Page should not be on lru, so no need to fix that up.
 * free_pages_check() will verify...
 */
static inline void free_page_mlock(struct page *page)
{
	__dec_zone_page_state(page, NR_MLOCK);
	__count_vm_event(UNEVICTABLE_MLOCKFREED);
}

606
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
607
{
Nick Piggin's avatar
Nick Piggin committed
608
609
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
610
		(atomic_read(&page->_count) != 0) |
611
612
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE) |
		(mem_cgroup_bad_page_check(page)))) {
613
		bad_page(page);
614
		return 1;
615
	}
616
617
618
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
619
620
621
}

/*
622
 * Frees a number of pages from the PCP lists
Linus Torvalds's avatar
Linus Torvalds committed
623
 * Assumes all pages on list are in same zone, and of same order.
624
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
625
626
627
628
629
630
631
 *
 * 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.
 */
632
633
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
Linus Torvalds's avatar
Linus Torvalds committed
634
{
635
	int migratetype = 0;
636
	int batch_free = 0;
637
	int to_free = count;
638

Nick Piggin's avatar
Nick Piggin committed
639
	spin_lock(&zone->lock);
640
	zone->all_unreclaimable = 0;
Linus Torvalds's avatar
Linus Torvalds committed
641
	zone->pages_scanned = 0;
642

643
	while (to_free) {
Nick Piggin's avatar
Nick Piggin committed
644
		struct page *page;
645
646
647
		struct list_head *list;

		/*
648
649
650
651
652
		 * 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
653
654
		 */
		do {
655
			batch_free++;
656
657
658
659
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
Nick Piggin's avatar
Nick Piggin committed
660

661
662
663
664
		/* This is the only non-empty list. Free them all. */
		if (batch_free == MIGRATE_PCPTYPES)
			batch_free = to_free;

665
666
667
668
		do {
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
669
670
671
			/* MIGRATE_MOVABLE list may include MIGRATE_RESERVEs */
			__free_one_page(page, zone, 0, page_private(page));
			trace_mm_page_pcpu_drain(page, 0, page_private(page));
672
		} while (--to_free && --batch_free && !list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
673
	}
674
	__mod_zone_page_state(zone, NR_FREE_PAGES, count);
Nick Piggin's avatar
Nick Piggin committed
675
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
676
677
}

678
679
static void free_one_page(struct zone *zone, struct page *page, int order,
				int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
680
{
681
	spin_lock(&zone->lock);
682
	zone->all_unreclaimable = 0;
683
	zone->pages_scanned = 0;
684

685
	__free_one_page(page, zone, order, migratetype);
686
	__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
687
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
688
689
}

690
static bool free_pages_prepare(struct page *page, unsigned int order)
Nick Piggin's avatar
Nick Piggin committed
691
{
Linus Torvalds's avatar
Linus Torvalds committed
692
	int i;
693
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
694

695
	trace_mm_page_free(page, order);
696
697
	kmemcheck_free_shadow(page, order);

Andrea Arcangeli's avatar
Andrea Arcangeli committed
698
699
700
701
	if (PageAnon(page))
		page->mapping = NULL;
	for (i = 0; i < (1 << order); i++)
		bad += free_pages_check(page + i);
702
	if (bad)
703
		return false;
704

705
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
706
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
707
708
709
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
710
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
711
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
712

713
714
715
716
717
718
719
720
721
722
723
	return true;
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
	int wasMlocked = __TestClearPageMlocked(page);

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

Nick Piggin's avatar
Nick Piggin committed
724
	local_irq_save(flags);
725
	if (unlikely(wasMlocked))
726
		free_page_mlock(page);
727
	__count_vm_events(PGFREE, 1 << order);
728
729
	free_one_page(page_zone(page), page, order,
					get_pageblock_migratetype(page));
Nick Piggin's avatar
Nick Piggin committed
730
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
731
732
}

733
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
734
{
735
736
	unsigned int nr_pages = 1 << order;
	unsigned int loop;
737

738
739
740
741
742
743
744
745
	prefetchw(page);
	for (loop = 0; loop < nr_pages; loop++) {
		struct page *p = &page[loop];

		if (loop + 1 < nr_pages)
			prefetchw(p + 1);
		__ClearPageReserved(p);
		set_page_count(p, 0);
746
	}
747
748
749

	set_page_refcounted(page);
	__free_pages(page, order);
750
751
}

Linus Torvalds's avatar
Linus Torvalds committed
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766

/*
 * The order of subdivision here is critical for the IO subsystem.
 * Please do not alter this order without good reasons and regression
 * testing. Specifically, as large blocks of memory are subdivided,
 * the order in which smaller blocks are delivered depends on the order
 * they're subdivided in this function. This is the primary factor
 * influencing the order in which pages are delivered to the IO
 * subsystem according to empirical testing, and this is also justified
 * by considering the behavior of a buddy system containing a single
 * large block of memory acted on by a series of small allocations.
 * This behavior is a critical factor in sglist merging's success.
 *
 * -- wli
 */
Nick Piggin's avatar
Nick Piggin committed
767
static inline void expand(struct zone *zone, struct page *page,
768
769
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
770
771
772
773
774
775
776
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
777
		VM_BUG_ON(bad_range(zone, &page[size]));
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794

#ifdef CONFIG_DEBUG_PAGEALLOC
		if (high < debug_guardpage_minorder()) {
			/*
			 * Mark as guard pages (or page), that will allow to
			 * merge back to allocator when buddy will be freed.
			 * Corresponding page table entries will not be touched,
			 * pages will stay not present in virtual address space
			 */
			INIT_LIST_HEAD(&page[size].lru);
			set_page_guard_flag(&page[size]);
			set_page_private(&page[size], high);
			/* Guard pages are not available for any usage */
			__mod_zone_page_state(zone, NR_FREE_PAGES, -(1 << high));
			continue;
		}
#endif
795
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
796
797
798
799
800
801
802
803
		area->nr_free++;
		set_page_order(&page[size], high);
	}
}

/*
 * This page is about to be returned from the page allocator
 */
804
static inline int check_new_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
805
{
Nick Piggin's avatar
Nick Piggin committed
806
807
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
808
		(atomic_read(&page->_count) != 0)  |
809
810
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP) |
		(mem_cgroup_bad_page_check(page)))) {
811
		bad_page(page);
812
		return 1;
813
	}
814
815
816
817
818
819
820
821
822
823
824
825
	return 0;
}

static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

	for (i = 0; i < (1 << order); i++) {
		struct page *p = page + i;
		if (unlikely(check_new_page(p)))
			return 1;
	}
826

827
	set_page_private(page, 0);
828
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
829
830

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
831
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
832
833
834
835
836
837
838

	if (gfp_flags & __GFP_ZERO)
		prep_zero_page(page, order, gfp_flags);

	if (order && (gfp_flags & __GFP_COMP))
		prep_compound_page(page, order);

839
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
840
841
}

842
843
844
845
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
846
847
static inline
struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
						int migratetype)
{
	unsigned int current_order;
	struct free_area * area;
	struct page *page;

	/* Find a page of the appropriate size in the preferred list */
	for (current_order = order; current_order < MAX_ORDER; ++current_order) {
		area = &(zone->free_area[current_order]);
		if (list_empty(&area->free_list[migratetype]))
			continue;

		page = list_entry(area->free_list[migratetype].next,
							struct page, lru);
		list_del(&page->lru);
		rmv_page_order(page);
		area->nr_free--;
		expand(zone, page, order, current_order, area, migratetype);
		return page;
	}

	return NULL;
}


873
874
875
876
877
/*
 * This array describes the order lists are fallen back to when
 * the free lists for the desirable migrate type are depleted
 */
static int fallbacks[MIGRATE_TYPES][MIGRATE_TYPES-1] = {
878
879
880
881
	[MIGRATE_UNMOVABLE]   = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE,   MIGRATE_RESERVE },
	[MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE,   MIGRATE_MOVABLE,   MIGRATE_RESERVE },
	[MIGRATE_MOVABLE]     = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
	[MIGRATE_RESERVE]     = { MIGRATE_RESERVE,     MIGRATE_RESERVE,   MIGRATE_RESERVE }, /* Never used */
882
883
};

884
885
/*
 * Move the free pages in a range to the free lists of the requested type.
886
 * Note that start_page and end_pages are not aligned on a pageblock
887
888
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
889
890
891
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
892
893
894
{
	struct page *page;
	unsigned long order;
895
	int pages_moved = 0;
896
897
898
899
900
901
902

#ifndef CONFIG_HOLES_IN_ZONE
	/*
	 * page_zone is not safe to call in this context when
	 * CONFIG_HOLES_IN_ZONE is set. This bug check is probably redundant
	 * anyway as we check zone boundaries in move_freepages_block().
	 * Remove at a later date when no bug reports exist related to
Mel Gorman's avatar
Mel Gorman committed
903
	 * grouping pages by mobility
904
905
906
907
908
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
909
910
911
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

912
913
914
915
916
917
918
919
920
921
922
		if (!pfn_valid_within(page_to_pfn(page))) {
			page++;
			continue;
		}

		if (!PageBuddy(page)) {
			page++;
			continue;
		}

		order = page_order(page);
923
924
		list_move(&page->lru,
			  &zone->free_area[order].free_list[migratetype]);
925
		page += 1 << order;
926
		pages_moved += 1 << order;
927
928
	}

929
	return pages_moved;
930
931
}

Adrian Bunk's avatar
Adrian Bunk committed
932
933
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
934
935
936
937
938
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
939
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
940
	start_page = pfn_to_page(start_pfn);
941
942
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
943
944
945
946
947
948
949
950
951
952

	/* Do not cross zone boundaries */
	if (start_pfn < zone->zone_start_pfn)
		start_page = page;
	if (end_pfn >= zone->zone_start_pfn + zone->spanned_pages)
		return 0;

	return move_freepages(zone, start_page, end_page, migratetype);
}

953
954
955
956
957
958
959
960
961
962
963
static void change_pageblock_range(struct page *pageblock_page,
					int start_order, int migratetype)
{
	int nr_pageblocks = 1 << (start_order - pageblock_order);

	while (nr_pageblocks--) {
		set_pageblock_migratetype(pageblock_page, migratetype);
		pageblock_page += pageblock_nr_pages;
	}
}

964
/* Remove an element from the buddy allocator from the fallback list */
965
966
static inline struct page *
__rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
967
968
969
970
971
972
973
974
975
976
977
978
{
	struct free_area * area;
	int current_order;
	struct page *page;
	int migratetype, i;

	/* Find the largest possible block of pages in the other list */
	for (current_order = MAX_ORDER-1; current_order >= order;
						--current_order) {
		for (i = 0; i < MIGRATE_TYPES - 1; i++) {
			migratetype = fallbacks[start_migratetype][i];

979
980
981
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
982

983
984
985
986
987
988
989
990
991
			area = &(zone->free_area[current_order]);
			if (list_empty(&area->free_list[migratetype]))
				continue;

			page = list_entry(area->free_list[migratetype].next,
					struct page, lru);
			area->nr_free--;

			/*
992
			 * If breaking a large block of pages, move all free
993
994
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
Lucas De Marchi's avatar
Lucas De Marchi committed
995
			 * aggressive about taking ownership of free pages
996
			 */
997
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
998
999
					start_migratetype == MIGRATE_RECLAIMABLE ||
					page_group_by_mobility_disabled) {
1000
1001
1002
1003
1004
				unsigned long pages;
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
1005
1006
				if (pages >= (1 << (pageblock_order-1)) ||
						page_group_by_mobility_disabled)
1007
1008
1009
					set_pageblock_migratetype(page,
								start_migratetype);

1010
				migratetype = start_migratetype;
1011
			}
1012
1013
1014
1015
1016

			/* Remove the page from the freelists */
			list_del(&page->lru);
			rmv_page_order(page);

1017
1018
1019
			/* Take ownership for orders >= pageblock_order */
			if (current_order >= pageblock_order)
				change_pageblock_range(page, current_order,
1020
1021
1022
							start_migratetype);

			expand(zone, page, order, current_order, area, migratetype);
1023
1024
1025
1026

			trace_mm_page_alloc_extfrag(page, order, current_order,
				start_migratetype, migratetype);

1027
1028
1029
1030
			return page;
		}
	}

1031
	return NULL;
1032
1033
}

1034
/*
Linus Torvalds's avatar
Linus Torvalds committed
1035
1036
1037
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
1038
1039
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
1040
1041
1042
{
	struct page *page;

1043
retry_reserve:
1044
	page = __rmqueue_smallest(zone, order, migratetype);
1045

1046
	if (unlikely(!page) && migratetype != MIGRATE_RESERVE) {
1047
		page = __rmqueue_fallback(zone, order, migratetype);
1048

1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
		/*
		 * Use MIGRATE_RESERVE rather than fail an allocation. goto
		 * is used because __rmqueue_smallest is an inline function
		 * and we want just one call site
		 */
		if (!page) {
			migratetype = MIGRATE_RESERVE;
			goto retry_reserve;
		}
	}

1060
	trace_mm_page_alloc_zone_locked(page, order, migratetype);
1061
	return page;
Linus Torvalds's avatar
Linus Torvalds committed
1062
1063
1064
1065
1066
1067
1068
1069
}

/* 
 * Obtain a specified number of elements from the buddy allocator, all under
 * a single hold of the lock, for efficiency.  Add them to the supplied list.
 * Returns the number of new pages which were placed at *list.
 */
static int rmqueue_bulk(struct zone *zone, unsigned int order, 
1070
			unsigned long count, struct list_head *list,
1071
			int migratetype, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
1072
1073
1074
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
1075
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
1076
	for (i = 0; i < count; ++i) {
1077
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
1078
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
1079
			break;
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089

		/*
		 * Split buddy pages returned by expand() are received here
		 * in physical page order. The page is added to the callers and
		 * list and the list head then moves forward. From the callers
		 * perspective, the linked list is ordered by page number in
		 * some conditions. This is useful for IO devices that can
		 * merge IO requests if the physical pages are ordered
		 * properly.
		 */
1090
1091
1092
1093
		if (likely(cold == 0))
			list_add(&page->lru, list);
		else
			list_add_tail(&page->lru, list);
1094
		set_page_private(page, migratetype);
1095
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
1096
	}
1097
	__mod_zone_page_state(zone, NR_FREE_PAGES, -(i << order));
Nick Piggin's avatar
Nick Piggin committed
1098
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
1099
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
1100
1101
}

1102
#ifdef CONFIG_NUMA
1103
/*
1104
1105
1106
1107
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
1108
1109
 * Note that this function must be called with the thread pinned to
 * a single processor.
1110
 */
1111
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
1112
1113
{
	unsigned long flags;
1114
	int to_drain;
1115

1116
1117
1118
1119
1120
	local_irq_save(flags);
	if (pcp->count >= pcp->batch)
		to_drain = pcp->batch;
	else
		to_drain = pcp->count;
1121
	free_pcppages_bulk(zone, to_drain, pcp);
1122
1123
	pcp->count -= to_drain;
	local_irq_restore(flags);
1124
1125
1126
}
#endif

1127
1128
1129
1130
1131
1132
1133
1134
/*
 * Drain pages of the indicated processor.
 *
 * The processor must either be the current processor and the
 * thread pinned to the current processor or a processor that
 * is not online.
 */
static void drain_pages(unsigned int cpu)