page_alloc.c 133 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
24
#include <linux/bootmem.h>
#include <linux/compiler.h>
25
#include <linux/kernel.h>
Linus Torvalds's avatar
Linus Torvalds committed
26
27
28
29
30
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
31
#include <linux/oom.h>
Linus Torvalds's avatar
Linus Torvalds committed
32
33
34
35
36
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
37
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
38
39
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
40
#include <linux/mempolicy.h>
41
#include <linux/stop_machine.h>
42
43
#include <linux/sort.h>
#include <linux/pfn.h>
44
#include <linux/backing-dev.h>
45
#include <linux/fault-inject.h>
46
#include <linux/page-isolation.h>
47
#include <linux/page_cgroup.h>
48
#include <linux/debugobjects.h>
49
#include <linux/kmemleak.h>
Linus Torvalds's avatar
Linus Torvalds committed
50
51

#include <asm/tlbflush.h>
52
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
53
54
55
#include "internal.h"

/*
56
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
57
 */
58
59
60
61
62
63
64
65
66
67
68
69
70
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);

71
unsigned long totalram_pages __read_mostly;
72
unsigned long totalreserve_pages __read_mostly;
73
unsigned long highest_memmap_pfn __read_mostly;
74
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
75

76
77
78
79
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

80
static void __free_pages_ok(struct page *page, unsigned int order);
81

Linus Torvalds's avatar
Linus Torvalds committed
82
83
84
85
86
87
88
/*
 * 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
89
90
91
 *
 * 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
92
 */
93
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
94
#ifdef CONFIG_ZONE_DMA
95
	 256,
96
#endif
97
#ifdef CONFIG_ZONE_DMA32
98
	 256,
99
#endif
100
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
101
	 32,
102
#endif
Mel Gorman's avatar
Mel Gorman committed
103
	 32,
104
};
Linus Torvalds's avatar
Linus Torvalds committed
105
106
107

EXPORT_SYMBOL(totalram_pages);

108
static char * const zone_names[MAX_NR_ZONES] = {
109
#ifdef CONFIG_ZONE_DMA
110
	 "DMA",
111
#endif
112
#ifdef CONFIG_ZONE_DMA32
113
	 "DMA32",
114
#endif
115
	 "Normal",
116
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
117
	 "HighMem",
118
#endif
Mel Gorman's avatar
Mel Gorman committed
119
	 "Movable",
120
121
};

Linus Torvalds's avatar
Linus Torvalds committed
122
123
int min_free_kbytes = 1024;

124
125
unsigned long __meminitdata nr_kernel_pages;
unsigned long __meminitdata nr_all_pages;
126
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
127

128
129
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
130
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
   * 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

149
150
151
152
  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
153
  static unsigned long __initdata required_kernelcore;
154
  static unsigned long __initdata required_movablecore;
Adrian Bunk's avatar
Adrian Bunk committed
155
  static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
156
157
158
159

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
160
161
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
162
163
164
165
166
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
EXPORT_SYMBOL(nr_node_ids);
#endif

167
168
int page_group_by_mobility_disabled __read_mostly;

169
170
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
171
172
173
174

	if (unlikely(page_group_by_mobility_disabled))
		migratetype = MIGRATE_UNMOVABLE;

175
176
177
178
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

Nick Piggin's avatar
Nick Piggin committed
179
#ifdef CONFIG_DEBUG_VM
180
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
181
{
182
183
184
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
185

186
187
188
189
190
191
192
193
194
	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;
195
196
197
198
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
199
	if (!pfn_valid_within(page_to_pfn(page)))
200
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
201
	if (zone != page_zone(page))
202
203
204
205
206
207
208
209
210
211
		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
212
		return 1;
213
214
215
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
216
217
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
218
219
220
221
222
223
224
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

225
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
226
{
227
228
229
230
231
232
233
234
235
236
237
238
239
240
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

	/*
	 * 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) {
241
242
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
243
244
245
246
247
248
249
250
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

251
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
252
		current->comm, page_to_pfn(page));
253
	printk(KERN_ALERT
254
255
256
257
		"page:%p flags:%p count:%d mapcount:%d mapping:%p index:%lx\n",
		page, (void *)page->flags, page_count(page),
		page_mapcount(page), page->mapping, page->index);

Linus Torvalds's avatar
Linus Torvalds committed
258
	dump_stack();
259
out:
260
261
	/* Leave bad fields for debug, except PageBuddy could make trouble */
	__ClearPageBuddy(page);
262
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
263
264
265
266
267
268
269
270
271
272
273
274
}

/*
 * 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).
 *
275
276
277
 * 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
278
 */
279
280
281

static void free_compound_page(struct page *page)
{
282
	__free_pages_ok(page, compound_order(page));
283
284
}

285
void prep_compound_page(struct page *page, unsigned long order)
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
{
	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;
	}
}

#ifdef CONFIG_HUGETLBFS
void prep_compound_gigantic_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
303
304
305
{
	int i;
	int nr_pages = 1 << order;
306
	struct page *p = page + 1;
Linus Torvalds's avatar
Linus Torvalds committed
307

308
	set_compound_page_dtor(page, free_compound_page);
309
	set_compound_order(page, order);
310
	__SetPageHead(page);
311
	for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) {
312
313
		__SetPageTail(p);
		p->first_page = page;
Linus Torvalds's avatar
Linus Torvalds committed
314
315
	}
}
316
#endif
Linus Torvalds's avatar
Linus Torvalds committed
317

318
static int destroy_compound_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
319
320
321
{
	int i;
	int nr_pages = 1 << order;
322
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
323

324
325
	if (unlikely(compound_order(page) != order) ||
	    unlikely(!PageHead(page))) {
326
		bad_page(page);
327
328
		bad++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
329

330
	__ClearPageHead(page);
331

332
333
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
Linus Torvalds's avatar
Linus Torvalds committed
334

335
		if (unlikely(!PageTail(p) || (p->first_page != page))) {
336
			bad_page(page);
337
338
			bad++;
		}
339
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
340
	}
341
342

	return bad;
Linus Torvalds's avatar
Linus Torvalds committed
343
344
}

Nick Piggin's avatar
Nick Piggin committed
345
346
347
348
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

349
350
351
352
	/*
	 * 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
353
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
354
355
356
357
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

358
359
static inline void set_page_order(struct page *page, int order)
{
360
	set_page_private(page, order);
361
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
362
363
364
365
}

static inline void rmv_page_order(struct page *page)
{
366
	__ClearPageBuddy(page);
367
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
}

/*
 * 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
385
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
 */
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
404
 * (a) the buddy is not in a hole &&
405
 * (b) the buddy is in the buddy system &&
406
407
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
408
409
410
 *
 * 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
411
 *
412
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
413
 */
414
415
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
416
{
417
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
418
419
		return 0;

420
421
422
423
424
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
425
		return 1;
426
	}
427
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
}

/*
 * 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
443
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
444
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
445
446
447
448
449
450
451
452
453
 * 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
454
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
455
456
457
458
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;
459
	int migratetype = get_pageblock_migratetype(page);
Linus Torvalds's avatar
Linus Torvalds committed
460

461
	if (unlikely(PageCompound(page)))
462
463
		if (unlikely(destroy_compound_page(page, order)))
			return;
Linus Torvalds's avatar
Linus Torvalds committed
464
465
466

	page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);

Nick Piggin's avatar
Nick Piggin committed
467
468
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
469

470
	__mod_zone_page_state(zone, NR_FREE_PAGES, order_size);
Linus Torvalds's avatar
Linus Torvalds committed
471
472
473
474
475
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
476
		if (!page_is_buddy(page, buddy, order))
477
			break;
Nick Piggin's avatar
Nick Piggin committed
478

479
		/* Our buddy is free, merge with it and move up one order. */
Linus Torvalds's avatar
Linus Torvalds committed
480
		list_del(&buddy->lru);
481
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
482
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
483
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
484
485
486
487
488
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
489
490
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
491
492
493
	zone->free_area[order].nr_free++;
}

494
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
495
{
496
	free_page_mlock(page);
Nick Piggin's avatar
Nick Piggin committed
497
498
499
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
500
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE))) {
501
		bad_page(page);
502
		return 1;
503
	}
504
505
506
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
507
508
509
510
511
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
512
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
513
514
515
516
517
518
519
 *
 * 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.
 */
Nick Piggin's avatar
Nick Piggin committed
520
521
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
522
{
Nick Piggin's avatar
Nick Piggin committed
523
	spin_lock(&zone->lock);
524
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
Linus Torvalds's avatar
Linus Torvalds committed
525
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
526
527
528
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
529
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
530
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
531
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
532
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
533
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
534
	}
Nick Piggin's avatar
Nick Piggin committed
535
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
536
537
}

Nick Piggin's avatar
Nick Piggin committed
538
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
539
{
540
	spin_lock(&zone->lock);
541
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
542
	zone->pages_scanned = 0;
543
	__free_one_page(page, zone, order);
544
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
545
546
547
548
549
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
550
	int i;
551
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
552
553

	for (i = 0 ; i < (1 << order) ; ++i)
554
555
		bad += free_pages_check(page + i);
	if (bad)
556
557
		return;

558
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
559
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
560
561
562
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
563
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
564
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
565

Nick Piggin's avatar
Nick Piggin committed
566
	local_irq_save(flags);
567
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
568
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
569
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
570
571
}

572
573
574
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
575
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
576
577
578
579
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
580
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
581
		__free_page(page);
582
583
584
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
585
		prefetchw(page);
586
587
588
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
589
590
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
591
592
593
594
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

595
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
596
		__free_pages(page, order);
597
598
599
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614

/*
 * 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
615
static inline void expand(struct zone *zone, struct page *page,
616
617
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
618
619
620
621
622
623
624
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
625
		VM_BUG_ON(bad_range(zone, &page[size]));
626
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
627
628
629
630
631
632
633
634
		area->nr_free++;
		set_page_order(&page[size], high);
	}
}

/*
 * This page is about to be returned from the page allocator
 */
Nick Piggin's avatar
Nick Piggin committed
635
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
636
{
Nick Piggin's avatar
Nick Piggin committed
637
638
639
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
640
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP))) {
641
		bad_page(page);
642
		return 1;
643
	}
644

645
	set_page_private(page, 0);
646
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
647
648

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
649
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
650
651
652
653
654
655
656

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

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

657
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
658
659
}

660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
static struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
						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--;
		__mod_zone_page_state(zone, NR_FREE_PAGES, - (1UL << order));
		expand(zone, page, order, current_order, area, migratetype);
		return page;
	}

	return NULL;
}


691
692
693
694
695
/*
 * 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] = {
696
697
698
699
	[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 */
700
701
};

702
703
/*
 * Move the free pages in a range to the free lists of the requested type.
704
 * Note that start_page and end_pages are not aligned on a pageblock
705
706
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
707
708
709
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
710
711
712
{
	struct page *page;
	unsigned long order;
713
	int pages_moved = 0;
714
715
716
717
718
719
720

#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
721
	 * grouping pages by mobility
722
723
724
725
726
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
727
728
729
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
		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;
745
		pages_moved += 1 << order;
746
747
	}

748
	return pages_moved;
749
750
}

Adrian Bunk's avatar
Adrian Bunk committed
751
752
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
753
754
755
756
757
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
758
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
759
	start_page = pfn_to_page(start_pfn);
760
761
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
762
763
764
765
766
767
768
769
770
771

	/* 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);
}

772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
/* Remove an element from the buddy allocator from the fallback list */
static struct page *__rmqueue_fallback(struct zone *zone, int order,
						int start_migratetype)
{
	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];

787
788
789
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
790

791
792
793
794
795
796
797
798
799
			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--;

			/*
800
			 * If breaking a large block of pages, move all free
801
802
803
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
804
			 */
805
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
806
807
808
809
810
811
					start_migratetype == MIGRATE_RECLAIMABLE) {
				unsigned long pages;
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
812
				if (pages >= (1 << (pageblock_order-1)))
813
814
815
					set_pageblock_migratetype(page,
								start_migratetype);

816
				migratetype = start_migratetype;
817
			}
818
819
820
821
822
823
824

			/* Remove the page from the freelists */
			list_del(&page->lru);
			rmv_page_order(page);
			__mod_zone_page_state(zone, NR_FREE_PAGES,
							-(1UL << order));

825
			if (current_order == pageblock_order)
826
827
828
829
830
831
832
833
				set_pageblock_migratetype(page,
							start_migratetype);

			expand(zone, page, order, current_order, area, migratetype);
			return page;
		}
	}

834
835
	/* Use MIGRATE_RESERVE rather than fail an allocation */
	return __rmqueue_smallest(zone, order, MIGRATE_RESERVE);
836
837
}

838
/*
Linus Torvalds's avatar
Linus Torvalds committed
839
840
841
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
842
843
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
844
845
846
{
	struct page *page;

847
	page = __rmqueue_smallest(zone, order, migratetype);
848

849
850
	if (unlikely(!page))
		page = __rmqueue_fallback(zone, order, migratetype);
851
852

	return page;
Linus Torvalds's avatar
Linus Torvalds committed
853
854
855
856
857
858
859
860
}

/* 
 * 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, 
861
862
			unsigned long count, struct list_head *list,
			int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
863
864
865
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
866
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
867
	for (i = 0; i < count; ++i) {
868
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
869
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
870
			break;
871
872
873
874
875
876
877
878
879
880

		/*
		 * 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.
		 */
881
882
		list_add(&page->lru, list);
		set_page_private(page, migratetype);
883
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
884
	}
Nick Piggin's avatar
Nick Piggin committed
885
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
886
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
887
888
}

889
#ifdef CONFIG_NUMA
890
/*
891
892
893
894
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
895
896
 * Note that this function must be called with the thread pinned to
 * a single processor.
897
 */
898
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
899
900
{
	unsigned long flags;
901
	int to_drain;
902

903
904
905
906
907
908
909
910
	local_irq_save(flags);
	if (pcp->count >= pcp->batch)
		to_drain = pcp->batch;
	else
		to_drain = pcp->count;
	free_pages_bulk(zone, to_drain, &pcp->list, 0);
	pcp->count -= to_drain;
	local_irq_restore(flags);
911
912
913
}
#endif

914
915
916
917
918
919
920
921
/*
 * 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)
Linus Torvalds's avatar
Linus Torvalds committed
922
{
Nick Piggin's avatar
Nick Piggin committed
923
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
924
925
	struct zone *zone;

926
	for_each_populated_zone(zone) {
Linus Torvalds's avatar
Linus Torvalds committed
927
		struct per_cpu_pageset *pset;
928
		struct per_cpu_pages *pcp;
Linus Torvalds's avatar
Linus Torvalds committed
929

930
		pset = zone_pcp(zone, cpu);
931
932
933
934
935
936

		pcp = &pset->pcp;
		local_irq_save(flags);
		free_pages_bulk(zone, pcp->count, &pcp->list, 0);
		pcp->count = 0;
		local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
937
938
939
	}
}

940
941
942
943
944
945
946
947
948
949
950
951
952
/*
 * Spill all of this CPU's per-cpu pages back into the buddy allocator.
 */
void drain_local_pages(void *arg)
{
	drain_pages(smp_processor_id());
}

/*
 * Spill all the per-cpu pages from all CPUs back into the buddy allocator
 */
void drain_all_pages(void)
{
953
	on_each_cpu(drain_local_pages, NULL, 1);
954
955
}

956
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
957
958
959

void mark_free_pages(struct zone *zone)
{
960
961
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
962
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
963
964
965
966
967
968
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
969
970
971
972
973
974

	max_zone_pfn = zone->zone_start_pfn + zone->spanned_pages;
	for (pfn = zone->zone_start_pfn; pfn < max_zone_pfn; pfn++)
		if (pfn_valid(pfn)) {
			struct page *page = pfn_to_page(pfn);

975
976
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
977
		}
Linus Torvalds's avatar
Linus Torvalds committed
978

979
980
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
981
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
982

983
984
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
985
				swsusp_set_page_free(pfn_to_page(pfn + i));
986
		}
987
	}
Linus Torvalds's avatar
Linus Torvalds committed
988
989
	spin_unlock_irqrestore(&zone->lock, flags);
}
990
#endif /* CONFIG_PM */
Linus Torvalds's avatar
Linus Torvalds committed
991
992
993
994

/*
 * Free a 0-order page
 */
Harvey Harrison's avatar
Harvey Harrison committed
995
static void free_hot_cold_page(struct page *page, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
996
997
998
999
1000
1001
1002
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;

	if (PageAnon(page))
		page->mapping = NULL;
1003
	if (free_pages_check(page))
1004
1005
		return;

1006
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
1007
		debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
1008
1009
		debug_check_no_obj_freed(page_address(page), PAGE_SIZE);
	}
Nick Piggin's avatar
Nick Piggin committed
1010
	arch_free_page(page, 0);
1011
1012
	kernel_map_pages(page, 1, 0);

1013
	pcp = &zone_pcp(zone, get_cpu())->pcp;
Linus Torvalds's avatar
Linus Torvalds committed
1014
	local_irq_save(flags);
1015
	__count_vm_event(PGFREE);
1016
1017
1018
1019
	if (cold)
		list_add_tail(&page->lru, &pcp->list);
	else
		list_add(&page->lru, &pcp->list);
1020
	set_page_private(page, get_pageblock_migratetype(page));
Linus Torvalds's avatar
Linus Torvalds committed
1021
	pcp->count++;
Nick Piggin's avatar
Nick Piggin committed
1022
1023
1024
1025
	if (pcp->count >= pcp->high) {
		free_pages_bulk(zone, pcp->batch, &pcp->list, 0);
		pcp->count -= pcp->batch;
	}
Linus Torvalds's avatar
Linus Torvalds committed
1026
1027
1028
1029
	local_irq_restore(flags);
	put_cpu();
}

Harvey Harrison's avatar
Harvey Harrison committed
1030
void free_hot_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
1031
1032
1033
1034
{
	free_hot_cold_page(page, 0);
}
	
Harvey Harrison's avatar
Harvey Harrison committed
1035
void free_cold_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
1036
1037
1038
1039
{
	free_hot_cold_page(page, 1);
}

1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
/*
 * split_page takes a non-compound higher-order page, and splits it into
 * n (1<<order) sub-pages: page[0..n]
 * Each sub-page must be freed individually.
 *
 * Note: this is probably too low level an operation for use in drivers.
 * Please consult with lkml before using this in your driver.
 */
void split_page(struct page *page, unsigned int order)
{
	int i;

Nick Piggin's avatar
Nick Piggin committed
1052
1053
	VM_BUG_ON(PageCompound(page));
	VM_BUG_ON(!page_count(page));
1054
1055
	for (i = 1; i < (1 << order); i++)
		set_page_refcounted(page + i);
1056
1057
}

Linus Torvalds's avatar
Linus Torvalds committed
1058
1059
1060
1061
1062
/*
 * Really, prep_compound_page() should be called from __rmqueue_bulk().  But
 * we cheat by calling it from here, in the order > 0 path.  Saves a branch
 * or two.
 */
1063
static struct page *buffered_rmqueue(struct zone *preferred_zone,
1064
1065
			struct zone *zone, int order, gfp_t gfp_flags,
			int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
1066
1067
{
	unsigned long flags;
1068
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
1069
	int cold = !!(gfp_flags & __GFP_COLD);
Nick Piggin's avatar
Nick Piggin committed
1070
	int cpu;
Linus Torvalds's avatar
Linus Torvalds committed
1071