page_alloc.c 154 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/oom.h>
Linus Torvalds's avatar
Linus Torvalds committed
34
35
36
37
38
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
39
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
40
41
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
42
#include <linux/mempolicy.h>
43
#include <linux/stop_machine.h>
44
45
#include <linux/sort.h>
#include <linux/pfn.h>
46
#include <linux/backing-dev.h>
47
#include <linux/fault-inject.h>
48
#include <linux/page-isolation.h>
49
#include <linux/page_cgroup.h>
50
#include <linux/debugobjects.h>
51
#include <linux/kmemleak.h>
52
#include <linux/memory.h>
53
#include <linux/compaction.h>
54
#include <trace/events/kmem.h>
55
#include <linux/ftrace_event.h>
Linus Torvalds's avatar
Linus Torvalds committed
56
57

#include <asm/tlbflush.h>
58
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
59
60
#include "internal.h"

61
62
63
64
65
#ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID
DEFINE_PER_CPU(int, numa_node);
EXPORT_PER_CPU_SYMBOL(numa_node);
#endif

66
67
68
69
70
71
72
73
74
75
76
#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
77
/*
78
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
79
 */
80
81
82
83
84
85
86
87
88
89
90
91
92
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);

93
unsigned long totalram_pages __read_mostly;
94
unsigned long totalreserve_pages __read_mostly;
95
int percpu_pagelist_fraction;
96
gfp_t gfp_allowed_mask __read_mostly = GFP_BOOT_MASK;
Linus Torvalds's avatar
Linus Torvalds committed
97

98
99
100
101
102
103
104
105
106
#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).
 */
107
108
109
110

static gfp_t saved_gfp_mask;

void pm_restore_gfp_mask(void)
111
112
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
113
114
115
116
	if (saved_gfp_mask) {
		gfp_allowed_mask = saved_gfp_mask;
		saved_gfp_mask = 0;
	}
117
118
}

119
void pm_restrict_gfp_mask(void)
120
121
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
122
123
124
	WARN_ON(saved_gfp_mask);
	saved_gfp_mask = gfp_allowed_mask;
	gfp_allowed_mask &= ~GFP_IOFS;
125
126
127
}
#endif /* CONFIG_PM_SLEEP */

128
129
130
131
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

132
static void __free_pages_ok(struct page *page, unsigned int order);
133

Linus Torvalds's avatar
Linus Torvalds committed
134
135
136
137
138
139
140
/*
 * 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
141
142
143
 *
 * 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
144
 */
145
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
146
#ifdef CONFIG_ZONE_DMA
147
	 256,
148
#endif
149
#ifdef CONFIG_ZONE_DMA32
150
	 256,
151
#endif
152
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
153
	 32,
154
#endif
Mel Gorman's avatar
Mel Gorman committed
155
	 32,
156
};
Linus Torvalds's avatar
Linus Torvalds committed
157
158
159

EXPORT_SYMBOL(totalram_pages);

160
static char * const zone_names[MAX_NR_ZONES] = {
161
#ifdef CONFIG_ZONE_DMA
162
	 "DMA",
163
#endif
164
#ifdef CONFIG_ZONE_DMA32
165
	 "DMA32",
166
#endif
167
	 "Normal",
168
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
169
	 "HighMem",
170
#endif
Mel Gorman's avatar
Mel Gorman committed
171
	 "Movable",
172
173
};

Linus Torvalds's avatar
Linus Torvalds committed
174
175
int min_free_kbytes = 1024;

176
177
static unsigned long __meminitdata nr_kernel_pages;
static unsigned long __meminitdata nr_all_pages;
178
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
179

180
181
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
182
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
   * ranges of memory (RAM) that may be registered with add_active_range().
   * Ranges passed to add_active_range() will be merged if possible
   * so the number of times add_active_range() can be called is
   * related to the number of nodes and the number of holes
   */
  #ifdef CONFIG_MAX_ACTIVE_REGIONS
    /* Allow an architecture to set MAX_ACTIVE_REGIONS to save memory */
    #define MAX_ACTIVE_REGIONS CONFIG_MAX_ACTIVE_REGIONS
  #else
    #if MAX_NUMNODES >= 32
      /* If there can be many nodes, allow up to 50 holes per node */
      #define MAX_ACTIVE_REGIONS (MAX_NUMNODES*50)
    #else
      /* By default, allow up to 256 distinct regions */
      #define MAX_ACTIVE_REGIONS 256
    #endif
  #endif

201
202
203
204
  static struct node_active_region __meminitdata early_node_map[MAX_ACTIVE_REGIONS];
  static int __meminitdata nr_nodemap_entries;
  static unsigned long __meminitdata arch_zone_lowest_possible_pfn[MAX_NR_ZONES];
  static unsigned long __meminitdata arch_zone_highest_possible_pfn[MAX_NR_ZONES];
Adrian Bunk's avatar
Adrian Bunk committed
205
  static unsigned long __initdata required_kernelcore;
206
  static unsigned long __initdata required_movablecore;
Adrian Bunk's avatar
Adrian Bunk committed
207
  static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
208
209
210
211

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
212
213
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
214
215
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
216
int nr_online_nodes __read_mostly = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
217
EXPORT_SYMBOL(nr_node_ids);
218
EXPORT_SYMBOL(nr_online_nodes);
Miklos Szeredi's avatar
Miklos Szeredi committed
219
220
#endif

221
222
int page_group_by_mobility_disabled __read_mostly;

223
224
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
225
226
227
228

	if (unlikely(page_group_by_mobility_disabled))
		migratetype = MIGRATE_UNMOVABLE;

229
230
231
232
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

233
234
bool oom_killer_disabled __read_mostly;

Nick Piggin's avatar
Nick Piggin committed
235
#ifdef CONFIG_DEBUG_VM
236
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
237
{
238
239
240
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
241

242
243
244
245
246
247
248
249
250
	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;
251
252
253
254
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
255
	if (!pfn_valid_within(page_to_pfn(page)))
256
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
257
	if (zone != page_zone(page))
258
259
260
261
262
263
264
265
266
267
		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
268
		return 1;
269
270
271
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
272
273
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
274
275
276
277
278
279
280
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

281
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
282
{
283
284
285
286
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

287
288
289
290
291
292
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
		__ClearPageBuddy(page);
		return;
	}

293
294
295
296
297
298
299
300
301
302
	/*
	 * 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) {
303
304
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
305
306
307
308
309
310
311
312
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

313
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
314
		current->comm, page_to_pfn(page));
315
	dump_page(page);
316

Linus Torvalds's avatar
Linus Torvalds committed
317
	dump_stack();
318
out:
319
320
	/* Leave bad fields for debug, except PageBuddy could make trouble */
	__ClearPageBuddy(page);
321
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
322
323
324
325
326
327
328
329
330
331
332
333
}

/*
 * 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".
 *
 * All pages have PG_compound set.  All pages have their ->private pointing at
 * the head page (even the head page has this).
 *
334
335
336
 * 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
337
 */
338
339
340

static void free_compound_page(struct page *page)
{
341
	__free_pages_ok(page, compound_order(page));
342
343
}

344
void prep_compound_page(struct page *page, unsigned long order)
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
{
	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);
		p->first_page = page;
	}
}

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
static inline void set_page_order(struct page *page, int order)
{
402
	set_page_private(page, order);
403
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
404
405
406
407
}

static inline void rmv_page_order(struct page *page)
{
408
	__ClearPageBuddy(page);
409
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
}

/*
 * 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
427
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
 */
static inline struct page *
__page_find_buddy(struct page *page, unsigned long page_idx, unsigned int order)
{
	unsigned long buddy_idx = page_idx ^ (1 << order);

	return page + (buddy_idx - page_idx);
}

static inline unsigned long
__find_combined_index(unsigned long page_idx, unsigned int order)
{
	return (page_idx & ~(1 << order));
}

/*
 * 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
446
 * (a) the buddy is not in a hole &&
447
 * (b) the buddy is in the buddy system &&
448
449
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
450
451
452
 *
 * For recording whether a page is in the buddy system, we use PG_buddy.
 * Setting, clearing, and testing PG_buddy is serialized by zone->lock.
Linus Torvalds's avatar
Linus Torvalds committed
453
 *
454
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
455
 */
456
457
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
458
{
459
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
460
461
		return 0;

462
463
464
465
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
466
		VM_BUG_ON(page_count(buddy) != 0);
467
		return 1;
468
	}
469
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
}

/*
 * 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
485
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
486
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
487
488
489
490
491
492
493
494
495
 * 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
496
static inline void __free_one_page(struct page *page,
497
498
		struct zone *zone, unsigned int order,
		int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
499
500
{
	unsigned long page_idx;
501
502
	unsigned long combined_idx;
	struct page *buddy;
Linus Torvalds's avatar
Linus Torvalds committed
503

504
	if (unlikely(PageCompound(page)))
505
506
		if (unlikely(destroy_compound_page(page, order)))
			return;
Linus Torvalds's avatar
Linus Torvalds committed
507

508
509
	VM_BUG_ON(migratetype == -1);

Linus Torvalds's avatar
Linus Torvalds committed
510
511
	page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);

512
	VM_BUG_ON(page_idx & ((1 << order) - 1));
Nick Piggin's avatar
Nick Piggin committed
513
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
514
515
516

	while (order < MAX_ORDER-1) {
		buddy = __page_find_buddy(page, page_idx, order);
517
		if (!page_is_buddy(page, buddy, order))
518
			break;
Nick Piggin's avatar
Nick Piggin committed
519

520
		/* Our buddy is free, merge with it and move up one order. */
Linus Torvalds's avatar
Linus Torvalds committed
521
		list_del(&buddy->lru);
522
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
523
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
524
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
525
526
527
528
529
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
530
531
532
533
534
535
536
537
538

	/*
	 * 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
	 */
539
	if ((order < MAX_ORDER-2) && pfn_valid_within(page_to_pfn(buddy))) {
540
541
542
543
544
545
546
547
548
549
550
551
552
		struct page *higher_page, *higher_buddy;
		combined_idx = __find_combined_index(page_idx, order);
		higher_page = page + combined_idx - page_idx;
		higher_buddy = __page_find_buddy(higher_page, combined_idx, order + 1);
		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
553
554
555
	zone->free_area[order].nr_free++;
}

556
557
558
559
560
561
562
563
564
565
566
/*
 * 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);
}

567
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
568
{
Nick Piggin's avatar
Nick Piggin committed
569
570
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
571
		(atomic_read(&page->_count) != 0) |
572
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE))) {
573
		bad_page(page);
574
		return 1;
575
	}
576
577
578
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
579
580
581
}

/*
582
 * Frees a number of pages from the PCP lists
Linus Torvalds's avatar
Linus Torvalds committed
583
 * Assumes all pages on list are in same zone, and of same order.
584
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
585
586
587
588
589
590
591
 *
 * 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.
 */
592
593
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
Linus Torvalds's avatar
Linus Torvalds committed
594
{
595
	int migratetype = 0;
596
	int batch_free = 0;
597
	int to_free = count;
598

Nick Piggin's avatar
Nick Piggin committed
599
	spin_lock(&zone->lock);
600
	zone->all_unreclaimable = 0;
Linus Torvalds's avatar
Linus Torvalds committed
601
	zone->pages_scanned = 0;
602

603
	while (to_free) {
Nick Piggin's avatar
Nick Piggin committed
604
		struct page *page;
605
606
607
		struct list_head *list;

		/*
608
609
610
611
612
		 * 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
613
614
		 */
		do {
615
			batch_free++;
616
617
618
619
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
Nick Piggin's avatar
Nick Piggin committed
620

621
622
623
624
		do {
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
625
626
627
			/* 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));
628
		} while (--to_free && --batch_free && !list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
629
	}
630
	__mod_zone_page_state(zone, NR_FREE_PAGES, count);
Nick Piggin's avatar
Nick Piggin committed
631
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
632
633
}

634
635
static void free_one_page(struct zone *zone, struct page *page, int order,
				int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
636
{
637
	spin_lock(&zone->lock);
638
	zone->all_unreclaimable = 0;
639
	zone->pages_scanned = 0;
640

641
	__free_one_page(page, zone, order, migratetype);
642
	__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
643
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
644
645
}

646
static bool free_pages_prepare(struct page *page, unsigned int order)
Nick Piggin's avatar
Nick Piggin committed
647
{
Linus Torvalds's avatar
Linus Torvalds committed
648
	int i;
649
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
650

651
	trace_mm_page_free_direct(page, order);
652
653
	kmemcheck_free_shadow(page, order);

Andrea Arcangeli's avatar
Andrea Arcangeli committed
654
655
656
657
	if (PageAnon(page))
		page->mapping = NULL;
	for (i = 0; i < (1 << order); i++)
		bad += free_pages_check(page + i);
658
	if (bad)
659
		return false;
660

661
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
662
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
663
664
665
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
666
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
667
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
668

669
670
671
672
673
674
675
676
677
678
679
	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
680
	local_irq_save(flags);
681
	if (unlikely(wasMlocked))
682
		free_page_mlock(page);
683
	__count_vm_events(PGFREE, 1 << order);
684
685
	free_one_page(page_zone(page), page, order,
					get_pageblock_migratetype(page));
Nick Piggin's avatar
Nick Piggin committed
686
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
687
688
}

689
690
691
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
692
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
693
694
695
696
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
697
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
698
		__free_page(page);
699
700
701
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
702
		prefetchw(page);
703
704
705
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
706
707
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
708
709
710
711
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

712
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
713
		__free_pages(page, order);
714
715
716
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731

/*
 * 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
732
static inline void expand(struct zone *zone, struct page *page,
733
734
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
735
736
737
738
739
740
741
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
742
		VM_BUG_ON(bad_range(zone, &page[size]));
743
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
744
745
746
747
748
749
750
751
		area->nr_free++;
		set_page_order(&page[size], high);
	}
}

/*
 * This page is about to be returned from the page allocator
 */
752
static inline int check_new_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
753
{
Nick Piggin's avatar
Nick Piggin committed
754
755
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
756
		(atomic_read(&page->_count) != 0)  |
757
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP))) {
758
		bad_page(page);
759
		return 1;
760
	}
761
762
763
764
765
766
767
768
769
770
771
772
	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;
	}
773

774
	set_page_private(page, 0);
775
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
776
777

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
778
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
779
780
781
782
783
784
785

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

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

786
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
787
788
}

789
790
791
792
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
793
794
static inline
struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
						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;
}


820
821
822
823
824
/*
 * 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] = {
825
826
827
828
	[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 */
829
830
};

831
832
/*
 * Move the free pages in a range to the free lists of the requested type.
833
 * Note that start_page and end_pages are not aligned on a pageblock
834
835
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
836
837
838
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
839
840
841
{
	struct page *page;
	unsigned long order;
842
	int pages_moved = 0;
843
844
845
846
847
848
849

#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
850
	 * grouping pages by mobility
851
852
853
854
855
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
856
857
858
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
		if (!pfn_valid_within(page_to_pfn(page))) {
			page++;
			continue;
		}

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

		order = page_order(page);
		list_del(&page->lru);
		list_add(&page->lru,
			&zone->free_area[order].free_list[migratetype]);
		page += 1 << order;
874
		pages_moved += 1 << order;
875
876
	}

877
	return pages_moved;
878
879
}

Adrian Bunk's avatar
Adrian Bunk committed
880
881
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
882
883
884
885
886
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
887
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
888
	start_page = pfn_to_page(start_pfn);
889
890
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
891
892
893
894
895
896
897
898
899
900

	/* 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);
}

901
902
903
904
905
906
907
908
909
910
911
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;
	}
}

912
/* Remove an element from the buddy allocator from the fallback list */
913
914
static inline struct page *
__rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
915
916
917
918
919
920
921
922
923
924
925
926
{
	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];

927
928
929
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
930

931
932
933
934
935
936
937
938
939
			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--;

			/*
940
			 * If breaking a large block of pages, move all free
941
942
943
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
944
			 */
945
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
946
947
					start_migratetype == MIGRATE_RECLAIMABLE ||
					page_group_by_mobility_disabled) {
948
949
950
951
952
				unsigned long pages;
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
953
954
				if (pages >= (1 << (pageblock_order-1)) ||
						page_group_by_mobility_disabled)
955
956
957
					set_pageblock_migratetype(page,
								start_migratetype);

958
				migratetype = start_migratetype;
959
			}
960
961
962
963
964

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

965
966
967
			/* Take ownership for orders >= pageblock_order */
			if (current_order >= pageblock_order)
				change_pageblock_range(page, current_order,
968
969
970
							start_migratetype);

			expand(zone, page, order, current_order, area, migratetype);
971
972
973
974

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

975
976
977
978
			return page;
		}
	}

979
	return NULL;
980
981
}

982
/*
Linus Torvalds's avatar
Linus Torvalds committed
983
984
985
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
986
987
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
988
989
990
{
	struct page *page;

991
retry_reserve:
992
	page = __rmqueue_smallest(zone, order, migratetype);
993

994
	if (unlikely(!page) && migratetype != MIGRATE_RESERVE) {
995
		page = __rmqueue_fallback(zone, order, migratetype);
996

997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
		/*
		 * 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;
		}
	}

1008
	trace_mm_page_alloc_zone_locked(page, order, migratetype);
1009
	return page;
Linus Torvalds's avatar
Linus Torvalds committed
1010
1011
1012
1013
1014
1015
1016
1017
}

/* 
 * 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, 
1018
			unsigned long count, struct list_head *list,
1019
			int migratetype, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
1020
1021
1022
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
1023
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
1024
	for (i = 0; i < count; ++i) {
1025
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
1026
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
1027
			break;
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037

		/*
		 * 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.
		 */
1038
1039
1040
1041
		if (likely(cold == 0))
			list_add(&page->lru, list);
		else
			list_add_tail(&page->lru, list);
1042
		set_page_private(page, migratetype);
1043
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
1044
	}
1045
	__mod_zone_page_state(zone, NR_FREE_PAGES, -(i << order));
Nick Piggin's avatar
Nick Piggin committed
1046
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
1047
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
1048
1049
}

1050
#ifdef CONFIG_NUMA
1051
/*
1052
1053
1054
1055
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
1056
1057
 * Note that this function must be called with the thread pinned to
 * a single processor.
1058
 */
1059
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
1060
1061
{
	unsigned long flags;
1062
	int to_drain;
1063

Christoph Lameter's avatar