page_alloc.c 124 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/memcontrol.h>
Linus Torvalds's avatar
Linus Torvalds committed
48
49

#include <asm/tlbflush.h>
50
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
51
52
53
#include "internal.h"

/*
54
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
55
 */
56
57
58
59
60
61
62
63
64
65
66
67
68
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);

69
unsigned long totalram_pages __read_mostly;
70
unsigned long totalreserve_pages __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
71
long nr_swap_pages;
72
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
73

74
75
76
77
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

78
static void __free_pages_ok(struct page *page, unsigned int order);
79

Linus Torvalds's avatar
Linus Torvalds committed
80
81
82
83
84
85
86
/*
 * 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
87
88
89
 *
 * 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
90
 */
91
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
92
#ifdef CONFIG_ZONE_DMA
93
	 256,
94
#endif
95
#ifdef CONFIG_ZONE_DMA32
96
	 256,
97
#endif
98
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
99
	 32,
100
#endif
Mel Gorman's avatar
Mel Gorman committed
101
	 32,
102
};
Linus Torvalds's avatar
Linus Torvalds committed
103
104
105

EXPORT_SYMBOL(totalram_pages);

106
static char * const zone_names[MAX_NR_ZONES] = {
107
#ifdef CONFIG_ZONE_DMA
108
	 "DMA",
109
#endif
110
#ifdef CONFIG_ZONE_DMA32
111
	 "DMA32",
112
#endif
113
	 "Normal",
114
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
115
	 "HighMem",
116
#endif
Mel Gorman's avatar
Mel Gorman committed
117
	 "Movable",
118
119
};

Linus Torvalds's avatar
Linus Torvalds committed
120
121
int min_free_kbytes = 1024;

122
123
unsigned long __meminitdata nr_kernel_pages;
unsigned long __meminitdata nr_all_pages;
124
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
125

126
127
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
128
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
   * 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

147
148
149
150
  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];
151
#ifdef CONFIG_MEMORY_HOTPLUG_RESERVE
152
153
  static unsigned long __meminitdata node_boundary_start_pfn[MAX_NUMNODES];
  static unsigned long __meminitdata node_boundary_end_pfn[MAX_NUMNODES];
154
#endif /* CONFIG_MEMORY_HOTPLUG_RESERVE */
Mel Gorman's avatar
Mel Gorman committed
155
  unsigned long __initdata required_kernelcore;
156
  static unsigned long __initdata required_movablecore;
157
  unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
158
159
160
161

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
162
163
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
164
165
166
167
168
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
EXPORT_SYMBOL(nr_node_ids);
#endif

169
170
int page_group_by_mobility_disabled __read_mostly;

171
172
173
174
175
176
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

Nick Piggin's avatar
Nick Piggin committed
177
#ifdef CONFIG_DEBUG_VM
178
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
179
{
180
181
182
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
183

184
185
186
187
188
189
190
191
192
	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;
193
194
195
196
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
197
	if (!pfn_valid_within(page_to_pfn(page)))
198
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
199
	if (zone != page_zone(page))
200
201
202
203
204
205
206
207
208
209
		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
210
		return 1;
211
212
213
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
214
215
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
216
217
218
219
220
221
222
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

223
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
224
{
225
	printk(KERN_EMERG "Bad page state in process '%s'\n"
226
227
228
		KERN_EMERG "page:%p flags:0x%0*lx mapping:%p mapcount:%d count:%d\n"
		KERN_EMERG "Trying to fix it up, but a reboot is needed\n"
		KERN_EMERG "Backtrace:\n",
229
230
231
		current->comm, page, (int)(2*sizeof(unsigned long)),
		(unsigned long)page->flags, page->mapping,
		page_mapcount(page), page_count(page));
Linus Torvalds's avatar
Linus Torvalds committed
232
	dump_stack();
233
234
	page->flags &= ~(1 << PG_lru	|
			1 << PG_private |
Linus Torvalds's avatar
Linus Torvalds committed
235
236
237
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
238
239
			1 << PG_reclaim |
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
240
			1 << PG_swapcache |
241
242
			1 << PG_writeback |
			1 << PG_buddy );
Linus Torvalds's avatar
Linus Torvalds committed
243
244
245
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
246
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
247
248
249
250
251
252
253
254
255
256
257
258
}

/*
 * 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).
 *
259
260
261
 * 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
262
 */
263
264
265

static void free_compound_page(struct page *page)
{
266
	__free_pages_ok(page, compound_order(page));
267
268
}

Linus Torvalds's avatar
Linus Torvalds committed
269
270
271
272
273
static void prep_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;

274
	set_compound_page_dtor(page, free_compound_page);
275
	set_compound_order(page, order);
276
	__SetPageHead(page);
277
	for (i = 1; i < nr_pages; i++) {
Linus Torvalds's avatar
Linus Torvalds committed
278
279
		struct page *p = page + i;

280
281
		__SetPageTail(p);
		p->first_page = page;
Linus Torvalds's avatar
Linus Torvalds committed
282
283
284
285
286
287
288
289
	}
}

static void destroy_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;

290
	if (unlikely(compound_order(page) != order))
291
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
292

293
	if (unlikely(!PageHead(page)))
294
			bad_page(page);
295
	__ClearPageHead(page);
296
	for (i = 1; i < nr_pages; i++) {
Linus Torvalds's avatar
Linus Torvalds committed
297
298
		struct page *p = page + i;

299
		if (unlikely(!PageTail(p) |
300
				(p->first_page != page)))
301
			bad_page(page);
302
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
303
304
305
	}
}

Nick Piggin's avatar
Nick Piggin committed
306
307
308
309
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

310
311
312
313
	/*
	 * 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
314
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
315
316
317
318
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

319
320
static inline void set_page_order(struct page *page, int order)
{
321
	set_page_private(page, order);
322
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
323
324
325
326
}

static inline void rmv_page_order(struct page *page)
{
327
	__ClearPageBuddy(page);
328
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
}

/*
 * 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
346
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
 */
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
365
 * (a) the buddy is not in a hole &&
366
 * (b) the buddy is in the buddy system &&
367
368
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
369
370
371
 *
 * 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
372
 *
373
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
374
 */
375
376
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
377
{
378
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
379
380
		return 0;

381
382
383
384
385
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
386
		return 1;
387
	}
388
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
}

/*
 * 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
404
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
405
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
406
407
408
409
410
411
412
413
414
 * 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
415
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
416
417
418
419
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;
420
	int migratetype = get_pageblock_migratetype(page);
Linus Torvalds's avatar
Linus Torvalds committed
421

422
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
423
424
425
426
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
427
428
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
429

430
	__mod_zone_page_state(zone, NR_FREE_PAGES, order_size);
Linus Torvalds's avatar
Linus Torvalds committed
431
432
433
434
435
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
436
		if (!page_is_buddy(page, buddy, order))
Linus Torvalds's avatar
Linus Torvalds committed
437
			break;		/* Move the buddy up one level. */
Nick Piggin's avatar
Nick Piggin committed
438

Linus Torvalds's avatar
Linus Torvalds committed
439
		list_del(&buddy->lru);
440
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
441
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
442
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
443
444
445
446
447
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
448
449
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
450
451
452
	zone->free_area[order].nr_free++;
}

453
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
454
{
Nick Piggin's avatar
Nick Piggin committed
455
456
457
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
Linus Torvalds's avatar
Linus Torvalds committed
458
459
460
461
462
463
464
		(page->flags & (
			1 << PG_lru	|
			1 << PG_private |
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_slab	|
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
465
			1 << PG_writeback |
466
467
			1 << PG_reserved |
			1 << PG_buddy ))))
468
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
469
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
470
		__ClearPageDirty(page);
471
472
473
474
475
476
	/*
	 * For now, we report if PG_reserved was found set, but do not
	 * clear it, and do not free the page.  But we shall soon need
	 * to do more, for when the ZERO_PAGE count wraps negative.
	 */
	return PageReserved(page);
Linus Torvalds's avatar
Linus Torvalds committed
477
478
479
480
481
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
482
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
483
484
485
486
487
488
489
 *
 * 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
490
491
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
492
{
Nick Piggin's avatar
Nick Piggin committed
493
	spin_lock(&zone->lock);
494
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
Linus Torvalds's avatar
Linus Torvalds committed
495
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
496
497
498
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
499
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
500
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
501
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
502
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
503
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
504
	}
Nick Piggin's avatar
Nick Piggin committed
505
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
506
507
}

Nick Piggin's avatar
Nick Piggin committed
508
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
509
{
510
	spin_lock(&zone->lock);
511
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
512
	zone->pages_scanned = 0;
513
	__free_one_page(page, zone, order);
514
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
515
516
517
518
519
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
520
	int i;
521
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
522
523

	for (i = 0 ; i < (1 << order) ; ++i)
524
		reserved += free_pages_check(page + i);
525
526
527
	if (reserved)
		return;

Nick Piggin's avatar
Nick Piggin committed
528
529
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
Nick Piggin's avatar
Nick Piggin committed
530
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
531
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
532

Nick Piggin's avatar
Nick Piggin committed
533
	local_irq_save(flags);
534
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
535
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
536
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
537
538
}

539
540
541
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
Harvey Harrison's avatar
Harvey Harrison committed
542
void __init __free_pages_bootmem(struct page *page, unsigned int order)
543
544
545
546
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
547
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
548
		__free_page(page);
549
550
551
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
552
		prefetchw(page);
553
554
555
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
556
557
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
558
559
560
561
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

562
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
563
		__free_pages(page, order);
564
565
566
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581

/*
 * 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
582
static inline void expand(struct zone *zone, struct page *page,
583
584
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
585
586
587
588
589
590
591
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
592
		VM_BUG_ON(bad_range(zone, &page[size]));
593
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
594
595
596
597
598
599
600
601
		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
602
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
603
{
Nick Piggin's avatar
Nick Piggin committed
604
605
606
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
607
608
		(page->flags & (
			1 << PG_lru	|
Linus Torvalds's avatar
Linus Torvalds committed
609
610
611
612
			1 << PG_private	|
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
613
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
614
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
615
			1 << PG_writeback |
616
617
			1 << PG_reserved |
			1 << PG_buddy ))))
618
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
619

620
621
622
623
624
625
626
	/*
	 * For now, we report if PG_reserved was found set, but do not
	 * clear it, and do not allocate the page: as a safety net.
	 */
	if (PageReserved(page))
		return 1;

627
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error | 1 << PG_readahead |
Linus Torvalds's avatar
Linus Torvalds committed
628
			1 << PG_referenced | 1 << PG_arch_1 |
629
			1 << PG_owner_priv_1 | 1 << PG_mappedtodisk);
630
	set_page_private(page, 0);
631
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
632
633

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
634
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
635
636
637
638
639
640
641

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

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

642
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
643
644
}

645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
/*
 * 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;
}


676
677
678
679
680
/*
 * 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] = {
681
682
683
684
	[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 */
685
686
};

687
688
/*
 * Move the free pages in a range to the free lists of the requested type.
689
 * Note that start_page and end_pages are not aligned on a pageblock
690
691
692
693
694
695
696
697
 * boundary. If alignment is required, use move_freepages_block()
 */
int move_freepages(struct zone *zone,
			struct page *start_page, struct page *end_page,
			int migratetype)
{
	struct page *page;
	unsigned long order;
698
	int pages_moved = 0;
699
700
701
702
703
704
705

#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
706
	 * grouping pages by mobility
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
		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;
727
		pages_moved += 1 << order;
728
729
	}

730
	return pages_moved;
731
732
733
734
735
736
737
738
}

int move_freepages_block(struct zone *zone, struct page *page, int migratetype)
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
739
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
740
	start_page = pfn_to_page(start_pfn);
741
742
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
743
744
745
746
747
748
749
750
751
752

	/* 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);
}

753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
/* 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];

768
769
770
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
771

772
773
774
775
776
777
778
779
780
			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--;

			/*
781
			 * If breaking a large block of pages, move all free
782
783
784
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
785
			 */
786
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
787
788
789
790
791
792
					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 */
793
				if (pages >= (1 << (pageblock_order-1)))
794
795
796
					set_pageblock_migratetype(page,
								start_migratetype);

797
				migratetype = start_migratetype;
798
			}
799
800
801
802
803
804
805

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

806
			if (current_order == pageblock_order)
807
808
809
810
811
812
813
814
				set_pageblock_migratetype(page,
							start_migratetype);

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

815
816
	/* Use MIGRATE_RESERVE rather than fail an allocation */
	return __rmqueue_smallest(zone, order, MIGRATE_RESERVE);
817
818
}

819
/*
Linus Torvalds's avatar
Linus Torvalds committed
820
821
822
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
823
824
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
825
826
827
{
	struct page *page;

828
	page = __rmqueue_smallest(zone, order, migratetype);
829

830
831
	if (unlikely(!page))
		page = __rmqueue_fallback(zone, order, migratetype);
832
833

	return page;
Linus Torvalds's avatar
Linus Torvalds committed
834
835
836
837
838
839
840
841
}

/* 
 * 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, 
842
843
			unsigned long count, struct list_head *list,
			int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
844
845
846
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
847
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
848
	for (i = 0; i < count; ++i) {
849
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
850
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
851
			break;
852
853
854
855
856
857
858
859
860
861

		/*
		 * 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.
		 */
862
863
		list_add(&page->lru, list);
		set_page_private(page, migratetype);
864
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
865
	}
Nick Piggin's avatar
Nick Piggin committed
866
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
867
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
868
869
}

870
#ifdef CONFIG_NUMA
871
/*
872
873
874
875
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
876
877
 * Note that this function must be called with the thread pinned to
 * a single processor.
878
 */
879
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
880
881
{
	unsigned long flags;
882
	int to_drain;
883

884
885
886
887
888
889
890
891
	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);
892
893
894
}
#endif

895
896
897
898
899
900
901
902
/*
 * 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
903
{
Nick Piggin's avatar
Nick Piggin committed
904
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
905
906
907
908
	struct zone *zone;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;
909
		struct per_cpu_pages *pcp;
Linus Torvalds's avatar
Linus Torvalds committed
910

911
912
913
		if (!populated_zone(zone))
			continue;

914
		pset = zone_pcp(zone, cpu);
915
916
917
918
919
920

		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
921
922
923
	}
}

924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
/*
 * 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)
{
	on_each_cpu(drain_local_pages, NULL, 0, 1);
}

940
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
941
942
943

void mark_free_pages(struct zone *zone)
{
944
945
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
946
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
947
948
949
950
951
952
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
953
954
955
956
957
958

	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);

959
960
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
961
		}
Linus Torvalds's avatar
Linus Torvalds committed
962

963
964
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
965
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
966

967
968
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
969
				swsusp_set_page_free(pfn_to_page(pfn + i));
970
		}
971
	}
Linus Torvalds's avatar
Linus Torvalds committed
972
973
	spin_unlock_irqrestore(&zone->lock, flags);
}
974
#endif /* CONFIG_PM */
Linus Torvalds's avatar
Linus Torvalds committed
975
976
977
978

/*
 * Free a 0-order page
 */
Harvey Harrison's avatar
Harvey Harrison committed
979
static void free_hot_cold_page(struct page *page, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
980
981
982
983
984
985
986
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;

	if (PageAnon(page))
		page->mapping = NULL;
987
	if (free_pages_check(page))
988
989
		return;

Nick Piggin's avatar
Nick Piggin committed
990
991
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
992
	VM_BUG_ON(page_get_page_cgroup(page));
Nick Piggin's avatar
Nick Piggin committed
993
	arch_free_page(page, 0);
994
995
	kernel_map_pages(page, 1, 0);

996
	pcp = &zone_pcp(zone, get_cpu())->pcp;
Linus Torvalds's avatar
Linus Torvalds committed
997
	local_irq_save(flags);
998
	__count_vm_event(PGFREE);
999
1000
1001
1002
	if (cold)
		list_add_tail(&page->lru, &pcp->list);
	else
		list_add(&page->lru, &pcp->list);
1003
	set_page_private(page, get_pageblock_migratetype(page));
Linus Torvalds's avatar
Linus Torvalds committed
1004
	pcp->count++;
Nick Piggin's avatar
Nick Piggin committed
1005
1006
1007
1008
	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
1009
1010
1011
1012
	local_irq_restore(flags);
	put_cpu();
}

Harvey Harrison's avatar
Harvey Harrison committed
1013
void free_hot_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
1014
1015
1016
1017
{
	free_hot_cold_page(page, 0);
}
	
Harvey Harrison's avatar
Harvey Harrison committed
1018
void free_cold_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
1019
1020
1021
1022
{
	free_hot_cold_page(page, 1);
}

1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
/*
 * 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
1035
1036
	VM_BUG_ON(PageCompound(page));
	VM_BUG_ON(!page_count(page));
1037
1038
	for (i = 1; i < (1 << order); i++)
		set_page_refcounted(page + i);
1039
1040
}

Linus Torvalds's avatar
Linus Torvalds committed
1041
1042
1043
1044
1045
/*
 * 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.
 */
Nick Piggin's avatar
Nick Piggin committed
1046
1047
static struct page *buffered_rmqueue(struct zonelist *zonelist,
			struct zone *zone, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
1048
1049
{
	unsigned long flags;
1050
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
1051
	int cold = !!(gfp_flags & __GFP_COLD);
Nick Piggin's avatar
Nick Piggin committed
1052
	int cpu;
1053
	int migratetype = allocflags_to_migratetype(gfp_flags);
Linus Torvalds's avatar
Linus Torvalds committed
1054

1055
again:
Nick Piggin's avatar
Nick Piggin committed
1056
	cpu  = get_cpu();
Nick Piggin's avatar
Nick Piggin committed
1057
	if (likely(order == 0)) {
Linus Torvalds's avatar
Linus Torvalds committed
1058
1059
		struct per_cpu_pages *pcp;

1060
		pcp = &zone_pcp(zone, cpu)->pcp;
Linus Torvalds's avatar
Linus Torvalds committed
1061
		local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
1062
		if (!pcp->count) {
1063
			pcp->count = rmqueue_bulk(zone, 0,
1064
					pcp->batch, &pcp->list, migratetype);
Nick Piggin's avatar
Nick Piggin committed
1065
1066
			if (unlikely(!pcp->count))
				goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
1067
		}
1068

1069
		/* Find a page of the appropriate migrate type */
1070
1071
1072
1073
1074
1075
1076
1077
1078
		if (cold) {
			list_for_each_entry_reverse(page, &pcp->list, lru)
				if (page_private(page) == migratetype)
					break;
		} else {
			list_for_each_entry(page, &pcp->list, lru)
				if (page_private(page) == migratetype)
					break;
		}
1079

1080
1081
		/* Allocate more to the pcp list if necessary */
		if (unlikely(&page->lru == &pcp->list)) {
1082
1083
1084
1085
			pcp->count += rmqueue_bulk(zone, 0,
					pcp->batch, &pcp->list, migratetype);
			page = list_entry(pcp->list.next, struct page, lru);
		}