page_alloc.c 142 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>
26
#include <linux/kmemcheck.h>
Linus Torvalds's avatar
Linus Torvalds committed
27
28
29
30
31
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
32
#include <linux/oom.h>
Linus Torvalds's avatar
Linus Torvalds committed
33
34
35
36
37
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
38
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
39
40
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
41
#include <linux/mempolicy.h>
42
#include <linux/stop_machine.h>
43
44
#include <linux/sort.h>
#include <linux/pfn.h>
45
#include <linux/backing-dev.h>
46
#include <linux/fault-inject.h>
47
#include <linux/page-isolation.h>
48
#include <linux/page_cgroup.h>
49
#include <linux/debugobjects.h>
50
#include <linux/kmemleak.h>
51
#include <linux/memory.h>
52
#include <trace/events/kmem.h>
Linus Torvalds's avatar
Linus Torvalds committed
53
54

#include <asm/tlbflush.h>
55
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
56
57
58
#include "internal.h"

/*
59
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
60
 */
61
62
63
64
65
66
67
68
69
70
71
72
73
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);

74
unsigned long totalram_pages __read_mostly;
75
unsigned long totalreserve_pages __read_mostly;
76
int percpu_pagelist_fraction;
77
gfp_t gfp_allowed_mask __read_mostly = GFP_BOOT_MASK;
Linus Torvalds's avatar
Linus Torvalds committed
78

79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
#ifdef CONFIG_PM_SLEEP
/*
 * The following functions are used by the suspend/hibernate code to temporarily
 * change gfp_allowed_mask in order to avoid using I/O during memory allocations
 * while devices are suspended.  To avoid races with the suspend/hibernate code,
 * they should always be called with pm_mutex held (gfp_allowed_mask also should
 * only be modified with pm_mutex held, unless the suspend/hibernate code is
 * guaranteed not to run in parallel with that modification).
 */
void set_gfp_allowed_mask(gfp_t mask)
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
	gfp_allowed_mask = mask;
}

gfp_t clear_gfp_allowed_mask(gfp_t mask)
{
	gfp_t ret = gfp_allowed_mask;

	WARN_ON(!mutex_is_locked(&pm_mutex));
	gfp_allowed_mask &= ~mask;
	return ret;
}
#endif /* CONFIG_PM_SLEEP */

104
105
106
107
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

108
static void __free_pages_ok(struct page *page, unsigned int order);
109

Linus Torvalds's avatar
Linus Torvalds committed
110
111
112
113
114
115
116
/*
 * 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
117
118
119
 *
 * 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
120
 */
121
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
122
#ifdef CONFIG_ZONE_DMA
123
	 256,
124
#endif
125
#ifdef CONFIG_ZONE_DMA32
126
	 256,
127
#endif
128
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
129
	 32,
130
#endif
Mel Gorman's avatar
Mel Gorman committed
131
	 32,
132
};
Linus Torvalds's avatar
Linus Torvalds committed
133
134
135

EXPORT_SYMBOL(totalram_pages);

136
static char * const zone_names[MAX_NR_ZONES] = {
137
#ifdef CONFIG_ZONE_DMA
138
	 "DMA",
139
#endif
140
#ifdef CONFIG_ZONE_DMA32
141
	 "DMA32",
142
#endif
143
	 "Normal",
144
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
145
	 "HighMem",
146
#endif
Mel Gorman's avatar
Mel Gorman committed
147
	 "Movable",
148
149
};

Linus Torvalds's avatar
Linus Torvalds committed
150
151
int min_free_kbytes = 1024;

152
153
static unsigned long __meminitdata nr_kernel_pages;
static unsigned long __meminitdata nr_all_pages;
154
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
155

156
157
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
158
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
   * 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

177
178
179
180
  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
181
  static unsigned long __initdata required_kernelcore;
182
  static unsigned long __initdata required_movablecore;
Adrian Bunk's avatar
Adrian Bunk committed
183
  static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
184
185
186
187

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
188
189
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
190
191
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
192
int nr_online_nodes __read_mostly = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
193
EXPORT_SYMBOL(nr_node_ids);
194
EXPORT_SYMBOL(nr_online_nodes);
Miklos Szeredi's avatar
Miklos Szeredi committed
195
196
#endif

197
198
int page_group_by_mobility_disabled __read_mostly;

199
200
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
201
202
203
204

	if (unlikely(page_group_by_mobility_disabled))
		migratetype = MIGRATE_UNMOVABLE;

205
206
207
208
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

209
210
bool oom_killer_disabled __read_mostly;

Nick Piggin's avatar
Nick Piggin committed
211
#ifdef CONFIG_DEBUG_VM
212
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
213
{
214
215
216
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
217

218
219
220
221
222
223
224
225
226
	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;
227
228
229
230
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
231
	if (!pfn_valid_within(page_to_pfn(page)))
232
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
233
	if (zone != page_zone(page))
234
235
236
237
238
239
240
241
242
243
		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
244
		return 1;
245
246
247
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
248
249
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
250
251
252
253
254
255
256
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

257
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
258
{
259
260
261
262
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

263
264
265
266
267
268
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
		__ClearPageBuddy(page);
		return;
	}

269
270
271
272
273
274
275
276
277
278
	/*
	 * 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) {
279
280
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
281
282
283
284
285
286
287
288
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

289
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
290
		current->comm, page_to_pfn(page));
291
	printk(KERN_ALERT
292
293
294
295
		"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
296
	dump_stack();
297
out:
298
299
	/* Leave bad fields for debug, except PageBuddy could make trouble */
	__ClearPageBuddy(page);
300
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
301
302
303
304
305
306
307
308
309
310
311
312
}

/*
 * 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).
 *
313
314
315
 * 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
316
 */
317
318
319

static void free_compound_page(struct page *page)
{
320
	__free_pages_ok(page, compound_order(page));
321
322
}

323
void prep_compound_page(struct page *page, unsigned long order)
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
{
	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;
	}
}

339
static int destroy_compound_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
340
341
342
{
	int i;
	int nr_pages = 1 << order;
343
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
344

345
346
	if (unlikely(compound_order(page) != order) ||
	    unlikely(!PageHead(page))) {
347
		bad_page(page);
348
349
		bad++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
350

351
	__ClearPageHead(page);
352

353
354
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
Linus Torvalds's avatar
Linus Torvalds committed
355

356
		if (unlikely(!PageTail(p) || (p->first_page != page))) {
357
			bad_page(page);
358
359
			bad++;
		}
360
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
361
	}
362
363

	return bad;
Linus Torvalds's avatar
Linus Torvalds committed
364
365
}

Nick Piggin's avatar
Nick Piggin committed
366
367
368
369
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

370
371
372
373
	/*
	 * 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
374
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
375
376
377
378
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

379
380
static inline void set_page_order(struct page *page, int order)
{
381
	set_page_private(page, order);
382
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
383
384
385
386
}

static inline void rmv_page_order(struct page *page)
{
387
	__ClearPageBuddy(page);
388
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
}

/*
 * 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
406
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
 */
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
425
 * (a) the buddy is not in a hole &&
426
 * (b) the buddy is in the buddy system &&
427
428
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
429
430
431
 *
 * 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
432
 *
433
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
434
 */
435
436
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
437
{
438
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
439
440
		return 0;

441
442
443
444
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
445
		VM_BUG_ON(page_count(buddy) != 0);
446
		return 1;
447
	}
448
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
}

/*
 * 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
464
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
465
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
466
467
468
469
470
471
472
473
474
 * 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
475
static inline void __free_one_page(struct page *page,
476
477
		struct zone *zone, unsigned int order,
		int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
478
479
480
{
	unsigned long page_idx;

481
	if (unlikely(PageCompound(page)))
482
483
		if (unlikely(destroy_compound_page(page, order)))
			return;
Linus Torvalds's avatar
Linus Torvalds committed
484

485
486
	VM_BUG_ON(migratetype == -1);

Linus Torvalds's avatar
Linus Torvalds committed
487
488
	page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);

489
	VM_BUG_ON(page_idx & ((1 << order) - 1));
Nick Piggin's avatar
Nick Piggin committed
490
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
491
492
493
494
495
496

	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
497
		if (!page_is_buddy(page, buddy, order))
498
			break;
Nick Piggin's avatar
Nick Piggin committed
499

500
		/* Our buddy is free, merge with it and move up one order. */
Linus Torvalds's avatar
Linus Torvalds committed
501
		list_del(&buddy->lru);
502
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
503
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
504
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
505
506
507
508
509
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
510
511
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
512
513
514
	zone->free_area[order].nr_free++;
}

515
516
517
518
519
520
521
522
523
524
525
/*
 * free_page_mlock() -- clean up attempts to free and mlocked() page.
 * Page should not be on lru, so no need to fix that up.
 * free_pages_check() will verify...
 */
static inline void free_page_mlock(struct page *page)
{
	__dec_zone_page_state(page, NR_MLOCK);
	__count_vm_event(UNEVICTABLE_MLOCKFREED);
}

526
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
527
{
Nick Piggin's avatar
Nick Piggin committed
528
529
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
530
		(atomic_read(&page->_count) != 0) |
531
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE))) {
532
		bad_page(page);
533
		return 1;
534
	}
535
536
537
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
538
539
540
}

/*
541
 * Frees a number of pages from the PCP lists
Linus Torvalds's avatar
Linus Torvalds committed
542
 * Assumes all pages on list are in same zone, and of same order.
543
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
544
545
546
547
548
549
550
 *
 * 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.
 */
551
552
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
Linus Torvalds's avatar
Linus Torvalds committed
553
{
554
	int migratetype = 0;
555
	int batch_free = 0;
556

Nick Piggin's avatar
Nick Piggin committed
557
	spin_lock(&zone->lock);
558
	zone->all_unreclaimable = 0;
Linus Torvalds's avatar
Linus Torvalds committed
559
	zone->pages_scanned = 0;
560

561
	__mod_zone_page_state(zone, NR_FREE_PAGES, count);
562
	while (count) {
Nick Piggin's avatar
Nick Piggin committed
563
		struct page *page;
564
565
566
		struct list_head *list;

		/*
567
568
569
570
571
		 * Remove pages from lists in a round-robin fashion. A
		 * batch_free count is maintained that is incremented when an
		 * empty list is encountered.  This is so more pages are freed
		 * off fuller lists instead of spinning excessively around empty
		 * lists
572
573
		 */
		do {
574
			batch_free++;
575
576
577
578
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
Nick Piggin's avatar
Nick Piggin committed
579

580
581
582
583
		do {
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
584
585
586
			/* MIGRATE_MOVABLE list may include MIGRATE_RESERVEs */
			__free_one_page(page, zone, 0, page_private(page));
			trace_mm_page_pcpu_drain(page, 0, page_private(page));
587
		} while (--count && --batch_free && !list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
588
	}
Nick Piggin's avatar
Nick Piggin committed
589
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
590
591
}

592
593
static void free_one_page(struct zone *zone, struct page *page, int order,
				int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
594
{
595
	spin_lock(&zone->lock);
596
	zone->all_unreclaimable = 0;
597
	zone->pages_scanned = 0;
598
599

	__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
600
	__free_one_page(page, zone, order, migratetype);
601
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
602
603
604
605
606
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
607
	int i;
608
	int bad = 0;
609
	int wasMlocked = __TestClearPageMlocked(page);
Linus Torvalds's avatar
Linus Torvalds committed
610

611
	trace_mm_page_free_direct(page, order);
612
613
	kmemcheck_free_shadow(page, order);

Linus Torvalds's avatar
Linus Torvalds committed
614
	for (i = 0 ; i < (1 << order) ; ++i)
615
616
		bad += free_pages_check(page + i);
	if (bad)
617
618
		return;

619
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
620
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
621
622
623
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
624
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
625
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
626

Nick Piggin's avatar
Nick Piggin committed
627
	local_irq_save(flags);
628
	if (unlikely(wasMlocked))
629
		free_page_mlock(page);
630
	__count_vm_events(PGFREE, 1 << order);
631
632
	free_one_page(page_zone(page), page, order,
					get_pageblock_migratetype(page));
Nick Piggin's avatar
Nick Piggin committed
633
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
634
635
}

636
637
638
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
639
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
640
641
642
643
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
644
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
645
		__free_page(page);
646
647
648
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
649
		prefetchw(page);
650
651
652
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
653
654
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
655
656
657
658
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

659
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
660
		__free_pages(page, order);
661
662
663
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678

/*
 * 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
679
static inline void expand(struct zone *zone, struct page *page,
680
681
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
682
683
684
685
686
687
688
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
689
		VM_BUG_ON(bad_range(zone, &page[size]));
690
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
691
692
693
694
695
696
697
698
		area->nr_free++;
		set_page_order(&page[size], high);
	}
}

/*
 * This page is about to be returned from the page allocator
 */
699
static inline int check_new_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
700
{
Nick Piggin's avatar
Nick Piggin committed
701
702
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
703
		(atomic_read(&page->_count) != 0)  |
704
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP))) {
705
		bad_page(page);
706
		return 1;
707
	}
708
709
710
711
712
713
714
715
716
717
718
719
	return 0;
}

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

	for (i = 0; i < (1 << order); i++) {
		struct page *p = page + i;
		if (unlikely(check_new_page(p)))
			return 1;
	}
720

721
	set_page_private(page, 0);
722
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
723
724

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
725
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
726
727
728
729
730
731
732

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

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

733
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
734
735
}

736
737
738
739
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
740
741
static inline
struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
						int migratetype)
{
	unsigned int current_order;
	struct free_area * area;
	struct page *page;

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

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

	return NULL;
}


767
768
769
770
771
/*
 * 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] = {
772
773
774
775
	[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 */
776
777
};

778
779
/*
 * Move the free pages in a range to the free lists of the requested type.
780
 * Note that start_page and end_pages are not aligned on a pageblock
781
782
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
783
784
785
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
786
787
788
{
	struct page *page;
	unsigned long order;
789
	int pages_moved = 0;
790
791
792
793
794
795
796

#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
797
	 * grouping pages by mobility
798
799
800
801
802
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
803
804
805
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
		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;
821
		pages_moved += 1 << order;
822
823
	}

824
	return pages_moved;
825
826
}

Adrian Bunk's avatar
Adrian Bunk committed
827
828
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
829
830
831
832
833
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
834
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
835
	start_page = pfn_to_page(start_pfn);
836
837
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
838
839
840
841
842
843
844
845
846
847

	/* 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);
}

848
849
850
851
852
853
854
855
856
857
858
static void change_pageblock_range(struct page *pageblock_page,
					int start_order, int migratetype)
{
	int nr_pageblocks = 1 << (start_order - pageblock_order);

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

859
/* Remove an element from the buddy allocator from the fallback list */
860
861
static inline struct page *
__rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
862
863
864
865
866
867
868
869
870
871
872
873
{
	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];

874
875
876
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
877

878
879
880
881
882
883
884
885
886
			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--;

			/*
887
			 * If breaking a large block of pages, move all free
888
889
890
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
891
			 */
892
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
893
894
					start_migratetype == MIGRATE_RECLAIMABLE ||
					page_group_by_mobility_disabled) {
895
896
897
898
899
				unsigned long pages;
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
900
901
				if (pages >= (1 << (pageblock_order-1)) ||
						page_group_by_mobility_disabled)
902
903
904
					set_pageblock_migratetype(page,
								start_migratetype);

905
				migratetype = start_migratetype;
906
			}
907
908
909
910
911

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

912
913
914
			/* Take ownership for orders >= pageblock_order */
			if (current_order >= pageblock_order)
				change_pageblock_range(page, current_order,
915
916
917
							start_migratetype);

			expand(zone, page, order, current_order, area, migratetype);
918
919
920
921

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

922
923
924
925
			return page;
		}
	}

926
	return NULL;
927
928
}

929
/*
Linus Torvalds's avatar
Linus Torvalds committed
930
931
932
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
933
934
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
935
936
937
{
	struct page *page;

938
retry_reserve:
939
	page = __rmqueue_smallest(zone, order, migratetype);
940

941
	if (unlikely(!page) && migratetype != MIGRATE_RESERVE) {
942
		page = __rmqueue_fallback(zone, order, migratetype);
943

944
945
946
947
948
949
950
951
952
953
954
		/*
		 * Use MIGRATE_RESERVE rather than fail an allocation. goto
		 * is used because __rmqueue_smallest is an inline function
		 * and we want just one call site
		 */
		if (!page) {
			migratetype = MIGRATE_RESERVE;
			goto retry_reserve;
		}
	}

955
	trace_mm_page_alloc_zone_locked(page, order, migratetype);
956
	return page;
Linus Torvalds's avatar
Linus Torvalds committed
957
958
959
960
961
962
963
964
}

/* 
 * 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, 
965
			unsigned long count, struct list_head *list,
966
			int migratetype, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
967
968
969
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
970
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
971
	for (i = 0; i < count; ++i) {
972
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
973
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
974
			break;
975
976
977
978
979
980
981
982
983
984

		/*
		 * 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.
		 */
985
986
987
988
		if (likely(cold == 0))
			list_add(&page->lru, list);
		else
			list_add_tail(&page->lru, list);
989
		set_page_private(page, migratetype);
990
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
991
	}
992
	__mod_zone_page_state(zone, NR_FREE_PAGES, -(i << order));
Nick Piggin's avatar
Nick Piggin committed
993
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
994
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
995
996
}

997
#ifdef CONFIG_NUMA
998
/*
999
1000
1001
1002
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
1003
1004
 * Note that this function must be called with the thread pinned to
 * a single processor.
1005
 */
1006
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
1007
1008
{
	unsigned long flags;
1009
	int to_drain;
1010

1011
1012
1013
1014
1015
	local_irq_save(flags);
	if (pcp->count >= pcp->batch)
		to_drain = pcp->batch;
	else
		to_drain = pcp->count;
1016
	free_pcppages_bulk(zone, to_drain, pcp);
1017
1018
	pcp->count -= to_drain;
	local_irq_restore(flags);
1019
1020
1021
}
#endif

1022
1023
1024
1025
1026
1027
1028
1029
/*
 * 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
1030
{
Nick Piggin's avatar
Nick Piggin committed
1031
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
1032
1033
	struct zone *zone;

1034
	for_each_populated_zone(zone) {
Linus Torvalds's avatar
Linus Torvalds committed
1035
		struct per_cpu_pageset *pset;
1036
		struct per_cpu_pages *pcp;
Linus Torvalds's avatar
Linus Torvalds committed
1037

1038
1039
		local_irq_save(flags);
		pset = per_cpu_ptr(zone->pageset, cpu);
1040
1041

		pcp = &pset->pcp;
1042
		free_pcppages_bulk(zone, pcp->count, pcp);
1043
1044
		pcp->count = 0;
		local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
1045
1046
1047
	}
}

1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
/*
 * 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)
{
1061
	on_each_cpu(drain_local_pages, NULL, 1);
1062
1063
}

1064
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
1065
1066
1067

void mark_free_pages(struct zone *zone)
{
1068
1069
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
1070
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
1071