page_alloc.c 126 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>
48
#include <linux/debugobjects.h>
Linus Torvalds's avatar
Linus Torvalds committed
49
50

#include <asm/tlbflush.h>
51
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
52
53
54
#include "internal.h"

/*
55
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
56
 */
57
58
59
60
61
62
63
64
65
66
67
68
69
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);

70
unsigned long totalram_pages __read_mostly;
71
unsigned long totalreserve_pages __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
72
long nr_swap_pages;
73
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
74

75
76
77
78
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

79
static void __free_pages_ok(struct page *page, unsigned int order);
80

Linus Torvalds's avatar
Linus Torvalds committed
81
82
83
84
85
86
87
/*
 * 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
88
89
90
 *
 * 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
91
 */
92
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
93
#ifdef CONFIG_ZONE_DMA
94
	 256,
95
#endif
96
#ifdef CONFIG_ZONE_DMA32
97
	 256,
98
#endif
99
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
100
	 32,
101
#endif
Mel Gorman's avatar
Mel Gorman committed
102
	 32,
103
};
Linus Torvalds's avatar
Linus Torvalds committed
104
105
106

EXPORT_SYMBOL(totalram_pages);

107
static char * const zone_names[MAX_NR_ZONES] = {
108
#ifdef CONFIG_ZONE_DMA
109
	 "DMA",
110
#endif
111
#ifdef CONFIG_ZONE_DMA32
112
	 "DMA32",
113
#endif
114
	 "Normal",
115
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
116
	 "HighMem",
117
#endif
Mel Gorman's avatar
Mel Gorman committed
118
	 "Movable",
119
120
};

Linus Torvalds's avatar
Linus Torvalds committed
121
122
int min_free_kbytes = 1024;

123
124
unsigned long __meminitdata nr_kernel_pages;
unsigned long __meminitdata nr_all_pages;
125
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
126

127
128
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
129
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
   * 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

148
149
150
151
  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];
152
#ifdef CONFIG_MEMORY_HOTPLUG_RESERVE
153
154
  static unsigned long __meminitdata node_boundary_start_pfn[MAX_NUMNODES];
  static unsigned long __meminitdata node_boundary_end_pfn[MAX_NUMNODES];
155
#endif /* CONFIG_MEMORY_HOTPLUG_RESERVE */
Mel Gorman's avatar
Mel Gorman committed
156
  unsigned long __initdata required_kernelcore;
157
  static unsigned long __initdata required_movablecore;
158
  unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
159
160
161
162

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
163
164
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
165
166
167
168
169
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
EXPORT_SYMBOL(nr_node_ids);
#endif

170
171
int page_group_by_mobility_disabled __read_mostly;

172
173
174
175
176
177
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
178
#ifdef CONFIG_DEBUG_VM
179
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
180
{
181
182
183
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
184

185
186
187
188
189
190
191
192
193
	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;
194
195
196
197
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
198
	if (!pfn_valid_within(page_to_pfn(page)))
199
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
200
	if (zone != page_zone(page))
201
202
203
204
205
206
207
208
209
210
		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
211
		return 1;
212
213
214
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
215
216
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
217
218
219
220
221
222
223
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

224
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
225
{
226
227
228
229
	void *pc = page_get_page_cgroup(page);

	printk(KERN_EMERG "Bad page state in process '%s'\n" KERN_EMERG
		"page:%p flags:0x%0*lx mapping:%p mapcount:%d count:%d\n",
230
231
232
		current->comm, page, (int)(2*sizeof(unsigned long)),
		(unsigned long)page->flags, page->mapping,
		page_mapcount(page), page_count(page));
233
234
235
236
237
238
	if (pc) {
		printk(KERN_EMERG "cgroup:%p\n", pc);
		page_reset_bad_cgroup(page);
	}
	printk(KERN_EMERG "Trying to fix it up, but a reboot is needed\n"
		KERN_EMERG "Backtrace:\n");
Linus Torvalds's avatar
Linus Torvalds committed
239
	dump_stack();
240
241
	page->flags &= ~(1 << PG_lru	|
			1 << PG_private |
Linus Torvalds's avatar
Linus Torvalds committed
242
243
244
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
245
246
			1 << PG_reclaim |
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
247
			1 << PG_swapcache |
248
249
			1 << PG_writeback |
			1 << PG_buddy );
Linus Torvalds's avatar
Linus Torvalds committed
250
251
252
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
253
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
254
255
256
257
258
259
260
261
262
263
264
265
}

/*
 * 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).
 *
266
267
268
 * 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
269
 */
270
271
272

static void free_compound_page(struct page *page)
{
273
	__free_pages_ok(page, compound_order(page));
274
275
}

Linus Torvalds's avatar
Linus Torvalds committed
276
277
278
279
280
static void prep_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;

281
	set_compound_page_dtor(page, free_compound_page);
282
	set_compound_order(page, order);
283
	__SetPageHead(page);
284
	for (i = 1; i < nr_pages; i++) {
Linus Torvalds's avatar
Linus Torvalds committed
285
286
		struct page *p = page + i;

287
288
		__SetPageTail(p);
		p->first_page = page;
Linus Torvalds's avatar
Linus Torvalds committed
289
290
291
292
293
294
295
296
	}
}

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

297
	if (unlikely(compound_order(page) != order))
298
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
299

300
	if (unlikely(!PageHead(page)))
301
			bad_page(page);
302
	__ClearPageHead(page);
303
	for (i = 1; i < nr_pages; i++) {
Linus Torvalds's avatar
Linus Torvalds committed
304
305
		struct page *p = page + i;

306
		if (unlikely(!PageTail(p) |
307
				(p->first_page != page)))
308
			bad_page(page);
309
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
310
311
312
	}
}

Nick Piggin's avatar
Nick Piggin committed
313
314
315
316
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

317
318
319
320
	/*
	 * 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
321
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
322
323
324
325
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

326
327
static inline void set_page_order(struct page *page, int order)
{
328
	set_page_private(page, order);
329
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
330
331
332
333
}

static inline void rmv_page_order(struct page *page)
{
334
	__ClearPageBuddy(page);
335
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
}

/*
 * 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
353
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
 */
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
372
 * (a) the buddy is not in a hole &&
373
 * (b) the buddy is in the buddy system &&
374
375
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
376
377
378
 *
 * 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
379
 *
380
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
381
 */
382
383
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
384
{
385
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
386
387
		return 0;

388
389
390
391
392
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
393
		return 1;
394
	}
395
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
}

/*
 * 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
411
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
412
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
413
414
415
416
417
418
419
420
421
 * 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
422
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
423
424
425
426
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;
427
	int migratetype = get_pageblock_migratetype(page);
Linus Torvalds's avatar
Linus Torvalds committed
428

429
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
430
431
432
433
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
434
435
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
436

437
	__mod_zone_page_state(zone, NR_FREE_PAGES, order_size);
Linus Torvalds's avatar
Linus Torvalds committed
438
439
440
441
442
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
443
		if (!page_is_buddy(page, buddy, order))
Linus Torvalds's avatar
Linus Torvalds committed
444
			break;		/* Move the buddy up one level. */
Nick Piggin's avatar
Nick Piggin committed
445

Linus Torvalds's avatar
Linus Torvalds committed
446
		list_del(&buddy->lru);
447
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
448
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
449
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
450
451
452
453
454
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
455
456
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
457
458
459
	zone->free_area[order].nr_free++;
}

460
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
461
{
Nick Piggin's avatar
Nick Piggin committed
462
463
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
464
		(page_get_page_cgroup(page) != NULL) |
Nick Piggin's avatar
Nick Piggin committed
465
		(page_count(page) != 0)  |
Linus Torvalds's avatar
Linus Torvalds committed
466
467
468
469
470
471
472
		(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
473
			1 << PG_writeback |
474
475
			1 << PG_reserved |
			1 << PG_buddy ))))
476
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
477
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
478
		__ClearPageDirty(page);
479
480
481
482
483
484
	/*
	 * 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
485
486
487
488
489
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
490
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
491
492
493
494
495
496
497
 *
 * 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
498
499
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
500
{
Nick Piggin's avatar
Nick Piggin committed
501
	spin_lock(&zone->lock);
502
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
Linus Torvalds's avatar
Linus Torvalds committed
503
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
504
505
506
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
507
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
508
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
509
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
510
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
511
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
512
	}
Nick Piggin's avatar
Nick Piggin committed
513
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
514
515
}

Nick Piggin's avatar
Nick Piggin committed
516
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
517
{
518
	spin_lock(&zone->lock);
519
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
520
	zone->pages_scanned = 0;
521
	__free_one_page(page, zone, order);
522
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
523
524
525
526
527
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
528
	int i;
529
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
530
531

	for (i = 0 ; i < (1 << order) ; ++i)
532
		reserved += free_pages_check(page + i);
533
534
535
	if (reserved)
		return;

536
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
537
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
538
539
540
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
541
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
542
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
543

Nick Piggin's avatar
Nick Piggin committed
544
	local_irq_save(flags);
545
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
546
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
547
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
548
549
}

550
551
552
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
553
void __free_pages_bootmem(struct page *page, unsigned int order)
554
555
556
557
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
558
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
559
		__free_page(page);
560
561
562
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
563
		prefetchw(page);
564
565
566
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
567
568
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
569
570
571
572
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

573
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
574
		__free_pages(page, order);
575
576
577
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592

/*
 * 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
593
static inline void expand(struct zone *zone, struct page *page,
594
595
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
596
597
598
599
600
601
602
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
603
		VM_BUG_ON(bad_range(zone, &page[size]));
604
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
605
606
607
608
609
610
611
612
		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
613
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
614
{
Nick Piggin's avatar
Nick Piggin committed
615
616
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
617
		(page_get_page_cgroup(page) != NULL) |
Nick Piggin's avatar
Nick Piggin committed
618
		(page_count(page) != 0)  |
619
620
		(page->flags & (
			1 << PG_lru	|
Linus Torvalds's avatar
Linus Torvalds committed
621
622
623
624
			1 << PG_private	|
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
625
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
626
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
627
			1 << PG_writeback |
628
629
			1 << PG_reserved |
			1 << PG_buddy ))))
630
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
631

632
633
634
635
636
637
638
	/*
	 * 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;

639
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error | 1 << PG_reclaim |
Linus Torvalds's avatar
Linus Torvalds committed
640
			1 << PG_referenced | 1 << PG_arch_1 |
641
			1 << PG_owner_priv_1 | 1 << PG_mappedtodisk);
642
	set_page_private(page, 0);
643
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
644
645

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
646
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
647
648
649
650
651
652
653

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

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

654
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
655
656
}

657
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
/*
 * 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;
}


688
689
690
691
692
/*
 * 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] = {
693
694
695
696
	[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 */
697
698
};

699
700
/*
 * Move the free pages in a range to the free lists of the requested type.
701
 * Note that start_page and end_pages are not aligned on a pageblock
702
703
704
705
706
707
708
709
 * 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;
710
	int pages_moved = 0;
711
712
713
714
715
716
717

#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
718
	 * grouping pages by mobility
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
	 */
	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;
739
		pages_moved += 1 << order;
740
741
	}

742
	return pages_moved;
743
744
745
746
747
748
749
750
}

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);
751
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
752
	start_page = pfn_to_page(start_pfn);
753
754
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
755
756
757
758
759
760
761
762
763
764

	/* 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);
}

765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
/* 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];

780
781
782
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
783

784
785
786
787
788
789
790
791
792
			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--;

			/*
793
			 * If breaking a large block of pages, move all free
794
795
796
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
797
			 */
798
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
799
800
801
802
803
804
					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 */
805
				if (pages >= (1 << (pageblock_order-1)))
806
807
808
					set_pageblock_migratetype(page,
								start_migratetype);

809
				migratetype = start_migratetype;
810
			}
811
812
813
814
815
816
817

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

818
			if (current_order == pageblock_order)
819
820
821
822
823
824
825
826
				set_pageblock_migratetype(page,
							start_migratetype);

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

827
828
	/* Use MIGRATE_RESERVE rather than fail an allocation */
	return __rmqueue_smallest(zone, order, MIGRATE_RESERVE);
829
830
}

831
/*
Linus Torvalds's avatar
Linus Torvalds committed
832
833
834
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
835
836
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
837
838
839
{
	struct page *page;

840
	page = __rmqueue_smallest(zone, order, migratetype);
841

842
843
	if (unlikely(!page))
		page = __rmqueue_fallback(zone, order, migratetype);
844
845

	return page;
Linus Torvalds's avatar
Linus Torvalds committed
846
847
848
849
850
851
852
853
}

/* 
 * 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, 
854
855
			unsigned long count, struct list_head *list,
			int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
856
857
858
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
859
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
860
	for (i = 0; i < count; ++i) {
861
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
862
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
863
			break;
864
865
866
867
868
869
870
871
872
873

		/*
		 * 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.
		 */
874
875
		list_add(&page->lru, list);
		set_page_private(page, migratetype);
876
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
877
	}
Nick Piggin's avatar
Nick Piggin committed
878
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
879
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
880
881
}

882
#ifdef CONFIG_NUMA
883
/*
884
885
886
887
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
888
889
 * Note that this function must be called with the thread pinned to
 * a single processor.
890
 */
891
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
892
893
{
	unsigned long flags;
894
	int to_drain;
895

896
897
898
899
900
901
902
903
	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);
904
905
906
}
#endif

907
908
909
910
911
912
913
914
/*
 * 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
915
{
Nick Piggin's avatar
Nick Piggin committed
916
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
917
918
919
920
	struct zone *zone;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;
921
		struct per_cpu_pages *pcp;
Linus Torvalds's avatar
Linus Torvalds committed
922

923
924
925
		if (!populated_zone(zone))
			continue;

926
		pset = zone_pcp(zone, cpu);
927
928
929
930
931
932

		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
933
934
935
	}
}

936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
/*
 * 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);
}

952
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
953
954
955

void mark_free_pages(struct zone *zone)
{
956
957
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
958
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
959
960
961
962
963
964
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
965
966
967
968
969
970

	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);

971
972
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
973
		}
Linus Torvalds's avatar
Linus Torvalds committed
974

975
976
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
977
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
978

979
980
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
981
				swsusp_set_page_free(pfn_to_page(pfn + i));
982
		}
983
	}
Linus Torvalds's avatar
Linus Torvalds committed
984
985
	spin_unlock_irqrestore(&zone->lock, flags);
}
986
#endif /* CONFIG_PM */
Linus Torvalds's avatar
Linus Torvalds committed
987
988
989
990

/*
 * Free a 0-order page
 */
Harvey Harrison's avatar
Harvey Harrison committed
991
static void free_hot_cold_page(struct page *page, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
992
993
994
995
996
997
998
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;

	if (PageAnon(page))
		page->mapping = NULL;
999
	if (free_pages_check(page))
1000
1001
		return;

1002
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
1003
		debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
1004
1005
		debug_check_no_obj_freed(page_address(page), PAGE_SIZE);
	}
Nick Piggin's avatar
Nick Piggin committed
1006
	arch_free_page(page, 0);
1007
1008
	kernel_map_pages(page, 1, 0);

1009
	pcp = &zone_pcp(zone, get_cpu())->pcp;
Linus Torvalds's avatar
Linus Torvalds committed
1010
	local_irq_save(flags);
1011
	__count_vm_event(PGFREE);
1012
1013
1014
1015
	if (cold)
		list_add_tail(&page->lru, &pcp->list);
	else
		list_add(&page->lru, &pcp->list);
1016
	set_page_private(page, get_pageblock_migratetype(page));
Linus Torvalds's avatar
Linus Torvalds committed
1017
	pcp->count++;
Nick Piggin's avatar
Nick Piggin committed
1018
1019
1020
1021
	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
1022
1023
1024
1025
	local_irq_restore(flags);
	put_cpu();
}

Harvey Harrison's avatar
Harvey Harrison committed
1026
void free_hot_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
1027
1028
1029
1030
{
	free_hot_cold_page(page, 0);
}
	
Harvey Harrison's avatar
Harvey Harrison committed
1031
void free_cold_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
1032
1033
1034
1035
{
	free_hot_cold_page(page, 1);
}

1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
/*
 * 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
1048
1049
	VM_BUG_ON(PageCompound(page));
	VM_BUG_ON(!page_count(page));
1050
1051
	for (i = 1; i < (1 << order); i++)
		set_page_refcounted(page + i);
1052
1053
}

Linus Torvalds's avatar
Linus Torvalds committed
1054
1055
1056
1057
1058
/*
 * 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.
 */
1059
static struct page *buffered_rmqueue(struct zone *preferred_zone,
Nick Piggin's avatar
Nick Piggin committed
1060
			struct zone *zone, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
1061
1062
{
	unsigned long flags;
1063
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
1064
	int cold = !!(gfp_flags & __GFP_COLD);
Nick Piggin's avatar
Nick Piggin committed
1065
	int cpu;
1066
	int migratetype = allocflags_to_migratetype(gfp_flags);
Linus Torvalds's avatar
Linus Torvalds committed
1067

1068
again:
Nick Piggin's avatar
Nick Piggin committed
1069
	cpu  = get_cpu();
Nick Piggin's avatar
Nick Piggin committed
1070
	if (likely(order == 0)) {
Linus Torvalds's avatar
Linus Torvalds committed
1071
1072
		struct per_cpu_pages *pcp;

1073
		pcp = &zone_pcp(zone, cpu)->pcp;
Linus Torvalds's avatar
Linus Torvalds committed
1074
		local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
1075
		if (!pcp->count) {
1076
			pcp->count = rmqueue_bulk(zone, 0,
1077
					pcp->batch, &pcp->list, migratetype);
Nick Piggin's avatar
Nick Piggin committed