page_alloc.c 81.2 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
22
23
/*
 *  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>
#include <linux/bootmem.h>
#include <linux/compiler.h>
24
#include <linux/kernel.h>
Linus Torvalds's avatar
Linus Torvalds committed
25
26
27
28
29
30
31
32
33
34
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
35
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
36
37
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
38
#include <linux/mempolicy.h>
39
#include <linux/stop_machine.h>
40
41
#include <linux/sort.h>
#include <linux/pfn.h>
Linus Torvalds's avatar
Linus Torvalds committed
42
43

#include <asm/tlbflush.h>
44
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
45
46
47
48
49
50
#include "internal.h"

/*
 * MCD - HACK: Find somewhere to initialize this EARLY, or make this
 * initializer cleaner
 */
51
nodemask_t node_online_map __read_mostly = { { [0] = 1UL } };
52
EXPORT_SYMBOL(node_online_map);
53
nodemask_t node_possible_map __read_mostly = NODE_MASK_ALL;
54
EXPORT_SYMBOL(node_possible_map);
55
unsigned long totalram_pages __read_mostly;
56
unsigned long totalreserve_pages __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
57
long nr_swap_pages;
58
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
59

60
static void __free_pages_ok(struct page *page, unsigned int order);
61

Linus Torvalds's avatar
Linus Torvalds committed
62
63
64
65
66
67
68
/*
 * 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
69
70
71
 *
 * 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
72
 */
73
74
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
	 256,
75
#ifdef CONFIG_ZONE_DMA32
76
	 256,
77
#endif
78
#ifdef CONFIG_HIGHMEM
79
	 32
80
#endif
81
};
Linus Torvalds's avatar
Linus Torvalds committed
82
83
84
85
86
87
88

EXPORT_SYMBOL(totalram_pages);

/*
 * Used by page_zone() to look up the address of the struct zone whose
 * id is encoded in the upper bits of page->flags
 */
89
struct zone *zone_table[1 << ZONETABLE_SHIFT] __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
90
91
EXPORT_SYMBOL(zone_table);

92
93
static char *zone_names[MAX_NR_ZONES] = {
	 "DMA",
94
#ifdef CONFIG_ZONE_DMA32
95
	 "DMA32",
96
#endif
97
	 "Normal",
98
#ifdef CONFIG_HIGHMEM
99
	 "HighMem"
100
#endif
101
102
};

Linus Torvalds's avatar
Linus Torvalds committed
103
104
int min_free_kbytes = 1024;

105
106
unsigned long __meminitdata nr_kernel_pages;
unsigned long __meminitdata nr_all_pages;
107
static unsigned long __initdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
108

109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
   * MAX_ACTIVE_REGIONS determines the maxmimum number of distinct
   * 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

  struct node_active_region __initdata early_node_map[MAX_ACTIVE_REGIONS];
  int __initdata nr_nodemap_entries;
  unsigned long __initdata arch_zone_lowest_possible_pfn[MAX_NR_ZONES];
  unsigned long __initdata arch_zone_highest_possible_pfn[MAX_NR_ZONES];
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Nick Piggin's avatar
Nick Piggin committed
136
#ifdef CONFIG_DEBUG_VM
137
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
138
{
139
140
141
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
142

143
144
145
146
147
148
149
150
151
	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;
152
153
154
155
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
Linus Torvalds's avatar
Linus Torvalds committed
156
157
#ifdef CONFIG_HOLES_IN_ZONE
	if (!pfn_valid(page_to_pfn(page)))
158
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
159
160
#endif
	if (zone != page_zone(page))
161
162
163
164
165
166
167
168
169
170
		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
171
		return 1;
172
173
174
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
175
176
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
177
178
179
180
181
182
183
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

184
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
185
{
186
	printk(KERN_EMERG "Bad page state in process '%s'\n"
187
188
189
		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",
190
191
192
		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
193
	dump_stack();
194
195
	page->flags &= ~(1 << PG_lru	|
			1 << PG_private |
Linus Torvalds's avatar
Linus Torvalds committed
196
197
198
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
199
200
			1 << PG_reclaim |
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
201
			1 << PG_swapcache |
202
203
			1 << PG_writeback |
			1 << PG_buddy );
Linus Torvalds's avatar
Linus Torvalds committed
204
205
206
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
207
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
208
209
210
211
212
213
214
215
216
217
218
219
}

/*
 * 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).
 *
220
221
222
 * 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
223
 */
224
225
226
227
228
229

static void free_compound_page(struct page *page)
{
	__free_pages_ok(page, (unsigned long)page[1].lru.prev);
}

Linus Torvalds's avatar
Linus Torvalds committed
230
231
232
233
234
static void prep_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;

235
	page[1].lru.next = (void *)free_compound_page;	/* set dtor */
236
	page[1].lru.prev = (void *)order;
Linus Torvalds's avatar
Linus Torvalds committed
237
238
239
	for (i = 0; i < nr_pages; i++) {
		struct page *p = page + i;

240
		__SetPageCompound(p);
241
		set_page_private(p, (unsigned long)page);
Linus Torvalds's avatar
Linus Torvalds committed
242
243
244
245
246
247
248
249
	}
}

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

250
	if (unlikely((unsigned long)page[1].lru.prev != order))
251
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
252
253
254
255

	for (i = 0; i < nr_pages; i++) {
		struct page *p = page + i;

256
257
258
		if (unlikely(!PageCompound(p) |
				(page_private(p) != (unsigned long)page)))
			bad_page(page);
259
		__ClearPageCompound(p);
Linus Torvalds's avatar
Linus Torvalds committed
260
261
262
	}
}

Nick Piggin's avatar
Nick Piggin committed
263
264
265
266
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

Nick Piggin's avatar
Nick Piggin committed
267
	VM_BUG_ON((gfp_flags & (__GFP_WAIT | __GFP_HIGHMEM)) == __GFP_HIGHMEM);
268
269
270
271
	/*
	 * 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
272
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
273
274
275
276
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

Linus Torvalds's avatar
Linus Torvalds committed
277
278
279
280
281
/*
 * function for dealing with page's order in buddy system.
 * zone->lock is already acquired when we use these.
 * So, we don't need atomic page->flags operations here.
 */
282
283
static inline unsigned long page_order(struct page *page)
{
284
	return page_private(page);
Linus Torvalds's avatar
Linus Torvalds committed
285
286
}

287
288
static inline void set_page_order(struct page *page, int order)
{
289
	set_page_private(page, order);
290
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
291
292
293
294
}

static inline void rmv_page_order(struct page *page)
{
295
	__ClearPageBuddy(page);
296
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
}

/*
 * 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
314
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
 */
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
333
 * (a) the buddy is not in a hole &&
334
 * (b) the buddy is in the buddy system &&
335
336
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
337
338
339
 *
 * 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
340
 *
341
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
342
 */
343
344
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
345
{
Nick Piggin's avatar
Nick Piggin committed
346
#ifdef CONFIG_HOLES_IN_ZONE
347
	if (!pfn_valid(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
348
349
350
		return 0;
#endif

351
352
353
354
355
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
356
		return 1;
357
	}
358
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
}

/*
 * 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
374
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
375
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
376
377
378
379
380
381
382
383
384
 * 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
385
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
386
387
388
389
390
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;

391
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
392
393
394
395
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
396
397
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
398
399
400
401
402
403
404
405

	zone->free_pages += order_size;
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct free_area *area;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
406
		if (!page_is_buddy(page, buddy, order))
Linus Torvalds's avatar
Linus Torvalds committed
407
			break;		/* Move the buddy up one level. */
Nick Piggin's avatar
Nick Piggin committed
408

Linus Torvalds's avatar
Linus Torvalds committed
409
410
411
412
		list_del(&buddy->lru);
		area = zone->free_area + order;
		area->nr_free--;
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
413
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
414
415
416
417
418
419
420
421
422
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
	list_add(&page->lru, &zone->free_area[order].free_list);
	zone->free_area[order].nr_free++;
}

423
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
424
{
Nick Piggin's avatar
Nick Piggin committed
425
426
427
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
Linus Torvalds's avatar
Linus Torvalds committed
428
429
430
431
432
433
434
435
		(page->flags & (
			1 << PG_lru	|
			1 << PG_private |
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_reclaim	|
			1 << PG_slab	|
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
436
			1 << PG_writeback |
437
438
			1 << PG_reserved |
			1 << PG_buddy ))))
439
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
440
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
441
		__ClearPageDirty(page);
442
443
444
445
446
447
	/*
	 * 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
448
449
450
451
452
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
453
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
454
455
456
457
458
459
460
 *
 * 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
461
462
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
463
{
Nick Piggin's avatar
Nick Piggin committed
464
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
465
466
	zone->all_unreclaimable = 0;
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
467
468
469
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
470
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
471
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
472
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
473
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
474
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
475
	}
Nick Piggin's avatar
Nick Piggin committed
476
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
477
478
}

Nick Piggin's avatar
Nick Piggin committed
479
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
480
{
481
482
483
484
485
	spin_lock(&zone->lock);
	zone->all_unreclaimable = 0;
	zone->pages_scanned = 0;
	__free_one_page(page, zone ,order);
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
486
487
488
489
490
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
491
	int i;
492
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
493
494

	arch_free_page(page, order);
495
	if (!PageHighMem(page))
496
497
		debug_check_no_locks_freed(page_address(page),
					   PAGE_SIZE<<order);
Linus Torvalds's avatar
Linus Torvalds committed
498
499

	for (i = 0 ; i < (1 << order) ; ++i)
500
		reserved += free_pages_check(page + i);
501
502
503
	if (reserved)
		return;

Nick Piggin's avatar
Nick Piggin committed
504
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
505
	local_irq_save(flags);
506
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
507
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
508
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
509
510
}

511
512
513
514
515
516
517
518
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
void fastcall __init __free_pages_bootmem(struct page *page, unsigned int order)
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
519
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
520
		__free_page(page);
521
522
523
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
524
		prefetchw(page);
525
526
527
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
528
529
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
530
531
532
533
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

534
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
535
		__free_pages(page, order);
536
537
538
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553

/*
 * 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
554
static inline void expand(struct zone *zone, struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
555
556
557
558
559
560
561
562
 	int low, int high, struct free_area *area)
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
563
		VM_BUG_ON(bad_range(zone, &page[size]));
Linus Torvalds's avatar
Linus Torvalds committed
564
565
566
567
568
569
570
571
572
		list_add(&page[size].lru, &area->free_list);
		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
573
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
574
{
Nick Piggin's avatar
Nick Piggin committed
575
576
577
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
578
579
		(page->flags & (
			1 << PG_lru	|
Linus Torvalds's avatar
Linus Torvalds committed
580
581
582
583
584
			1 << PG_private	|
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
			1 << PG_reclaim	|
585
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
586
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
587
			1 << PG_writeback |
588
589
			1 << PG_reserved |
			1 << PG_buddy ))))
590
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
591

592
593
594
595
596
597
598
	/*
	 * 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;

Linus Torvalds's avatar
Linus Torvalds committed
599
600
601
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error |
			1 << PG_referenced | 1 << PG_arch_1 |
			1 << PG_checked | 1 << PG_mappedtodisk);
602
	set_page_private(page, 0);
603
	set_page_refcounted(page);
Linus Torvalds's avatar
Linus Torvalds committed
604
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
605
606
607
608
609
610
611

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

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

612
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
}

/* 
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
static struct page *__rmqueue(struct zone *zone, unsigned int order)
{
	struct free_area * area;
	unsigned int current_order;
	struct page *page;

	for (current_order = order; current_order < MAX_ORDER; ++current_order) {
		area = zone->free_area + current_order;
		if (list_empty(&area->free_list))
			continue;

		page = list_entry(area->free_list.next, struct page, lru);
		list_del(&page->lru);
		rmv_page_order(page);
		area->nr_free--;
		zone->free_pages -= 1UL << order;
Nick Piggin's avatar
Nick Piggin committed
635
636
		expand(zone, page, order, current_order, area);
		return page;
Linus Torvalds's avatar
Linus Torvalds committed
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
	}

	return NULL;
}

/* 
 * 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, 
			unsigned long count, struct list_head *list)
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
652
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
653
	for (i = 0; i < count; ++i) {
Nick Piggin's avatar
Nick Piggin committed
654
655
		struct page *page = __rmqueue(zone, order);
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
656
657
658
			break;
		list_add_tail(&page->lru, list);
	}
Nick Piggin's avatar
Nick Piggin committed
659
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
660
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
661
662
}

663
#ifdef CONFIG_NUMA
664
665
/*
 * Called from the slab reaper to drain pagesets on a particular node that
666
 * belongs to the currently executing processor.
667
668
 * Note that this function must be called with the thread pinned to
 * a single processor.
669
670
 */
void drain_node_pages(int nodeid)
671
{
672
673
	int i;
	enum zone_type z;
674
675
	unsigned long flags;

676
677
	for (z = 0; z < MAX_NR_ZONES; z++) {
		struct zone *zone = NODE_DATA(nodeid)->node_zones + z;
678
679
		struct per_cpu_pageset *pset;

680
681
682
		if (!populated_zone(zone))
			continue;

Nick Piggin's avatar
Nick Piggin committed
683
		pset = zone_pcp(zone, smp_processor_id());
684
685
686
687
		for (i = 0; i < ARRAY_SIZE(pset->pcp); i++) {
			struct per_cpu_pages *pcp;

			pcp = &pset->pcp[i];
688
689
690
691
692
693
			if (pcp->count) {
				local_irq_save(flags);
				free_pages_bulk(zone, pcp->count, &pcp->list, 0);
				pcp->count = 0;
				local_irq_restore(flags);
			}
694
695
696
697
698
		}
	}
}
#endif

Linus Torvalds's avatar
Linus Torvalds committed
699
700
701
#if defined(CONFIG_PM) || defined(CONFIG_HOTPLUG_CPU)
static void __drain_pages(unsigned int cpu)
{
Nick Piggin's avatar
Nick Piggin committed
702
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
703
704
705
706
707
708
	struct zone *zone;
	int i;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;

709
		pset = zone_pcp(zone, cpu);
Linus Torvalds's avatar
Linus Torvalds committed
710
711
712
713
		for (i = 0; i < ARRAY_SIZE(pset->pcp); i++) {
			struct per_cpu_pages *pcp;

			pcp = &pset->pcp[i];
Nick Piggin's avatar
Nick Piggin committed
714
			local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
715
716
			free_pages_bulk(zone, pcp->count, &pcp->list, 0);
			pcp->count = 0;
Nick Piggin's avatar
Nick Piggin committed
717
			local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
718
719
720
721
722
723
724
725
726
		}
	}
}
#endif /* CONFIG_PM || CONFIG_HOTPLUG_CPU */

#ifdef CONFIG_PM

void mark_free_pages(struct zone *zone)
{
727
728
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
729
730
731
732
733
734
735
	int order;
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
736
737
738
739
740
741
742
743
744

	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);

			if (!PageNosave(page))
				ClearPageNosaveFree(page);
		}
Linus Torvalds's avatar
Linus Torvalds committed
745
746
747

	for (order = MAX_ORDER - 1; order >= 0; --order)
		list_for_each(curr, &zone->free_area[order].free_list) {
748
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
749

750
751
752
753
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
				SetPageNosaveFree(pfn_to_page(pfn + i));
		}
Linus Torvalds's avatar
Linus Torvalds committed
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783

	spin_unlock_irqrestore(&zone->lock, flags);
}

/*
 * Spill all of this CPU's per-cpu pages back into the buddy allocator.
 */
void drain_local_pages(void)
{
	unsigned long flags;

	local_irq_save(flags);	
	__drain_pages(smp_processor_id());
	local_irq_restore(flags);	
}
#endif /* CONFIG_PM */

/*
 * Free a 0-order page
 */
static void fastcall free_hot_cold_page(struct page *page, int cold)
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;

	arch_free_page(page, 0);

	if (PageAnon(page))
		page->mapping = NULL;
784
	if (free_pages_check(page))
785
786
787
788
		return;

	kernel_map_pages(page, 1, 0);

789
	pcp = &zone_pcp(zone, get_cpu())->pcp[cold];
Linus Torvalds's avatar
Linus Torvalds committed
790
	local_irq_save(flags);
791
	__count_vm_event(PGFREE);
Linus Torvalds's avatar
Linus Torvalds committed
792
793
	list_add(&page->lru, &pcp->list);
	pcp->count++;
Nick Piggin's avatar
Nick Piggin committed
794
795
796
797
	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
798
799
800
801
802
803
804
805
806
807
808
809
810
811
	local_irq_restore(flags);
	put_cpu();
}

void fastcall free_hot_page(struct page *page)
{
	free_hot_cold_page(page, 0);
}
	
void fastcall free_cold_page(struct page *page)
{
	free_hot_cold_page(page, 1);
}

812
813
814
815
816
817
818
819
820
821
822
823
/*
 * 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
824
825
	VM_BUG_ON(PageCompound(page));
	VM_BUG_ON(!page_count(page));
826
827
	for (i = 1; i < (1 << order); i++)
		set_page_refcounted(page + i);
828
829
}

Linus Torvalds's avatar
Linus Torvalds committed
830
831
832
833
834
/*
 * 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
835
836
static struct page *buffered_rmqueue(struct zonelist *zonelist,
			struct zone *zone, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
837
838
{
	unsigned long flags;
839
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
840
	int cold = !!(gfp_flags & __GFP_COLD);
Nick Piggin's avatar
Nick Piggin committed
841
	int cpu;
Linus Torvalds's avatar
Linus Torvalds committed
842

843
again:
Nick Piggin's avatar
Nick Piggin committed
844
	cpu  = get_cpu();
Nick Piggin's avatar
Nick Piggin committed
845
	if (likely(order == 0)) {
Linus Torvalds's avatar
Linus Torvalds committed
846
847
		struct per_cpu_pages *pcp;

Nick Piggin's avatar
Nick Piggin committed
848
		pcp = &zone_pcp(zone, cpu)->pcp[cold];
Linus Torvalds's avatar
Linus Torvalds committed
849
		local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
850
		if (!pcp->count) {
Linus Torvalds's avatar
Linus Torvalds committed
851
852
			pcp->count += rmqueue_bulk(zone, 0,
						pcp->batch, &pcp->list);
Nick Piggin's avatar
Nick Piggin committed
853
854
			if (unlikely(!pcp->count))
				goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
855
		}
Nick Piggin's avatar
Nick Piggin committed
856
857
858
		page = list_entry(pcp->list.next, struct page, lru);
		list_del(&page->lru);
		pcp->count--;
Rohit Seth's avatar
Rohit Seth committed
859
	} else {
Linus Torvalds's avatar
Linus Torvalds committed
860
861
		spin_lock_irqsave(&zone->lock, flags);
		page = __rmqueue(zone, order);
Nick Piggin's avatar
Nick Piggin committed
862
863
864
		spin_unlock(&zone->lock);
		if (!page)
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
865
866
	}

867
	__count_zone_vm_events(PGALLOC, zone, 1 << order);
868
	zone_statistics(zonelist, zone);
Nick Piggin's avatar
Nick Piggin committed
869
870
	local_irq_restore(flags);
	put_cpu();
Linus Torvalds's avatar
Linus Torvalds committed
871

Nick Piggin's avatar
Nick Piggin committed
872
	VM_BUG_ON(bad_range(zone, page));
Nick Piggin's avatar
Nick Piggin committed
873
	if (prep_new_page(page, order, gfp_flags))
Nick Piggin's avatar
Nick Piggin committed
874
		goto again;
Linus Torvalds's avatar
Linus Torvalds committed
875
	return page;
Nick Piggin's avatar
Nick Piggin committed
876
877
878
879
880

failed:
	local_irq_restore(flags);
	put_cpu();
	return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
881
882
}

Rohit Seth's avatar
Rohit Seth committed
883
#define ALLOC_NO_WATERMARKS	0x01 /* don't check watermarks at all */
884
885
886
887
888
889
#define ALLOC_WMARK_MIN		0x02 /* use pages_min watermark */
#define ALLOC_WMARK_LOW		0x04 /* use pages_low watermark */
#define ALLOC_WMARK_HIGH	0x08 /* use pages_high watermark */
#define ALLOC_HARDER		0x10 /* try to alloc harder */
#define ALLOC_HIGH		0x20 /* __GFP_HIGH set */
#define ALLOC_CPUSET		0x40 /* check for correct cpuset */
Rohit Seth's avatar
Rohit Seth committed
890

Linus Torvalds's avatar
Linus Torvalds committed
891
892
893
894
895
/*
 * Return 1 if free pages are above 'mark'. This takes into account the order
 * of the allocation.
 */
int zone_watermark_ok(struct zone *z, int order, unsigned long mark,
Rohit Seth's avatar
Rohit Seth committed
896
		      int classzone_idx, int alloc_flags)
Linus Torvalds's avatar
Linus Torvalds committed
897
898
899
900
901
{
	/* free_pages my go negative - that's OK */
	long min = mark, free_pages = z->free_pages - (1 << order) + 1;
	int o;

Rohit Seth's avatar
Rohit Seth committed
902
	if (alloc_flags & ALLOC_HIGH)
Linus Torvalds's avatar
Linus Torvalds committed
903
		min -= min / 2;
Rohit Seth's avatar
Rohit Seth committed
904
	if (alloc_flags & ALLOC_HARDER)
Linus Torvalds's avatar
Linus Torvalds committed
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
		min -= min / 4;

	if (free_pages <= min + z->lowmem_reserve[classzone_idx])
		return 0;
	for (o = 0; o < order; o++) {
		/* At the next order, this order's pages become unavailable */
		free_pages -= z->free_area[o].nr_free << o;

		/* Require fewer higher order pages to be free */
		min >>= 1;

		if (free_pages <= min)
			return 0;
	}
	return 1;
}

Rohit Seth's avatar
Rohit Seth committed
922
923
924
925
926
927
928
/*
 * get_page_from_freeliest goes through the zonelist trying to allocate
 * a page.
 */
static struct page *
get_page_from_freelist(gfp_t gfp_mask, unsigned int order,
		struct zonelist *zonelist, int alloc_flags)
Martin Hicks's avatar
Martin Hicks committed
929
{
Rohit Seth's avatar
Rohit Seth committed
930
931
932
	struct zone **z = zonelist->zones;
	struct page *page = NULL;
	int classzone_idx = zone_idx(*z);
933
	struct zone *zone;
Rohit Seth's avatar
Rohit Seth committed
934
935
936
937
938
939

	/*
	 * Go through the zonelist once, looking for a zone with enough free.
	 * See also cpuset_zone_allowed() comment in kernel/cpuset.c.
	 */
	do {
940
		zone = *z;
941
		if (unlikely((gfp_mask & __GFP_THISNODE) &&
942
			zone->zone_pgdat != zonelist->zones[0]->zone_pgdat))
943
				break;
Rohit Seth's avatar
Rohit Seth committed
944
		if ((alloc_flags & ALLOC_CPUSET) &&
945
				!cpuset_zone_allowed(zone, gfp_mask))
Rohit Seth's avatar
Rohit Seth committed
946
947
948
			continue;

		if (!(alloc_flags & ALLOC_NO_WATERMARKS)) {
949
950
			unsigned long mark;
			if (alloc_flags & ALLOC_WMARK_MIN)
951
				mark = zone->pages_min;
952
			else if (alloc_flags & ALLOC_WMARK_LOW)
953
				mark = zone->pages_low;
954
			else
955
956
				mark = zone->pages_high;
			if (!zone_watermark_ok(zone , order, mark,
Rohit Seth's avatar
Rohit Seth committed
957
				    classzone_idx, alloc_flags))
958
				if (!zone_reclaim_mode ||
959
				    !zone_reclaim(zone, gfp_mask, order))
960
					continue;
Rohit Seth's avatar
Rohit Seth committed
961
962
		}

963
		page = buffered_rmqueue(zonelist, zone, order, gfp_mask);
Rohit Seth's avatar
Rohit Seth committed
964
965
966
967
968
		if (page) {
			break;
		}
	} while (*(++z) != NULL);
	return page;
Martin Hicks's avatar
Martin Hicks committed
969
970
}

Linus Torvalds's avatar
Linus Torvalds committed
971
972
973
974
/*
 * This is the 'heart' of the zoned buddy allocator.
 */
struct page * fastcall
975
__alloc_pages(gfp_t gfp_mask, unsigned int order,
Linus Torvalds's avatar
Linus Torvalds committed
976
977
		struct zonelist *zonelist)
{
Al Viro's avatar
Al Viro committed
978
	const gfp_t wait = gfp_mask & __GFP_WAIT;
Rohit Seth's avatar
Rohit Seth committed
979
	struct zone **z;
Linus Torvalds's avatar
Linus Torvalds committed
980
981
982
983
	struct page *page;
	struct reclaim_state reclaim_state;
	struct task_struct *p = current;
	int do_retry;
Rohit Seth's avatar
Rohit Seth committed
984
	int alloc_flags;
Linus Torvalds's avatar
Linus Torvalds committed
985
986
987
988
	int did_some_progress;

	might_sleep_if(wait);

989
restart:
Rohit Seth's avatar
Rohit Seth committed
990
	z = zonelist->zones;  /* the list of zones suitable for gfp_mask */
Linus Torvalds's avatar
Linus Torvalds committed
991

Rohit Seth's avatar
Rohit Seth committed
992
	if (unlikely(*z == NULL)) {
Linus Torvalds's avatar
Linus Torvalds committed
993
994
995
		/* Should this ever happen?? */
		return NULL;
	}
996

Rohit Seth's avatar
Rohit Seth committed
997
	page = get_page_from_freelist(gfp_mask|__GFP_HARDWALL, order,
998
				zonelist, ALLOC_WMARK_LOW|ALLOC_CPUSET);
Rohit Seth's avatar
Rohit Seth committed
999
1000
	if (page)
		goto got_pg;
Linus Torvalds's avatar
Linus Torvalds committed
1001

1002
	do {
1003
		wakeup_kswapd(*z, order);
1004
	} while (*(++z));
Linus Torvalds's avatar
Linus Torvalds committed
1005

1006
	/*
Rohit Seth's avatar
Rohit Seth committed
1007
1008
1009
1010
1011
1012
	 * OK, we're below the kswapd watermark and have kicked background
	 * reclaim. Now things get more complex, so set up alloc_flags according
	 * to how we want to proceed.
	 *
	 * The caller may dip into page reserves a bit more if the caller
	 * cannot run direct reclaim, or if the caller has realtime scheduling
Paul Jackson's avatar
Paul Jackson committed
1013
1014
	 * policy or is asking for __GFP_HIGH memory.  GFP_ATOMIC requests will
	 * set both ALLOC_HARDER (!wait) and ALLOC_HIGH (__GFP_HIGH).
1015
	 */
1016
	alloc_flags = ALLOC_WMARK_MIN;
Rohit Seth's avatar
Rohit Seth committed
1017
1018
1019
1020
	if ((unlikely(rt_task(p)) && !in_interrupt()) || !wait)
		alloc_flags |= ALLOC_HARDER;
	if (gfp_mask & __GFP_HIGH)
		alloc_flags |= ALLOC_HIGH;
1021
1022
	if (wait)
		alloc_flags |= ALLOC_CPUSET;
Linus Torvalds's avatar
Linus Torvalds committed
1023
1024
1025

	/*
	 * Go through the zonelist again. Let __GFP_HIGH and allocations
Rohit Seth's avatar
Rohit Seth committed
1026
	 * coming from realtime tasks go deeper into reserves.
Linus Torvalds's avatar
Linus Torvalds committed
1027
1028
1029
	 *
	 * This is the last chance, in general, before the goto nopage.
	 * Ignore cpuset if GFP_ATOMIC (!wait) rather than fail alloc.
1030
	 * See also cpuset_zone_allowed() comment in kernel/cpuset.c.
Linus Torvalds's avatar
Linus Torvalds committed
1031
	 */
Rohit Seth's avatar
Rohit Seth committed
1032
1033
1034
	page = get_page_from_freelist(gfp_mask, order, zonelist, alloc_flags);
	if (page)
		goto got_pg;
Linus Torvalds's avatar
Linus Torvalds committed
1035
1036

	/* This allocation should allow future memory freeing. */
1037
1038
1039
1040

	if (((p->flags & PF_MEMALLOC) || unlikely(test_thread_flag(TIF_MEMDIE)))
			&& !in_interrupt()) {
		if (!(gfp_mask & __GFP_NOMEMALLOC)) {
Kirill Korotaev's avatar
Kirill Korotaev committed
1041
nofail_alloc:
1042
			/* go through the zonelist yet again, ignoring mins */
Rohit Seth's avatar
Rohit Seth committed
1043
			page = get_page_from_freelist(gfp_mask, order,
1044
				zonelist, ALLOC_NO_WATERMARKS);
Rohit Seth's avatar
Rohit Seth committed
1045
1046
			if (page)
				goto got_pg;
Kirill Korotaev's avatar
Kirill Korotaev committed
1047
1048
1049
1050
			if (gfp_mask & __GFP_NOFAIL) {
				blk_congestion_wait(WRITE, HZ/50);
				goto nofail_alloc;
			}
Linus Torvalds's avatar
Linus Torvalds committed
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
		}
		goto nopage;
	}

	/* Atomic allocations - we can't balance anything */
	if (!wait)
		goto nopage;

rebalance:
	cond_resched();

	/* We now go into synchronous reclaim */
1063
	cpuset_memory_pressure_bump();
Linus Torvalds's avatar
Linus Torvalds committed
1064
1065
1066
1067
	p->flags |= PF_MEMALLOC;
	reclaim_state.reclaimed_slab = 0;
	p->reclaim_state = &reclaim_state;

Rohit Seth's avatar
Rohit Seth committed
1068
	did_some_progress = try_to_free_pages(zonelist->zones, gfp_mask);
Linus Torvalds's avatar
Linus Torvalds committed
1069
1070
1071
1072
1073
1074
1075

	p->reclaim_state = NULL;
	p->flags &= ~PF_MEMALLOC;

	cond_resched();

	if (likely(did_some_progress)) {
Rohit Seth's avatar
Rohit Seth committed
1076
1077
1078
1079
		page = get_page_from_freelist(gfp_mask, order,
						zonelist, alloc_flags);
		if (page)
			goto got_pg;
Linus Torvalds's avatar
Linus Torvalds committed
1080
1081
1082
1083
1084
1085
1086
	} else if ((gfp_mask & __GFP_FS) && !(gfp_mask & __GFP_NORETRY)) {
		/*
		 * Go through the zonelist yet one more time, keep
		 * very high watermark here, this is only to catch
		 * a parallel oom killing, we must fail if we're still
		 * under heavy pressure.
		 */
Rohit Seth's avatar
Rohit Seth committed
1087
		page = get_page_from_freelist(gfp_mask|__GFP_HARDWALL, order,
1088
				zonelist, ALLOC_WMARK_HIGH|ALLOC_CPUSET);
Rohit Seth's avatar
Rohit Seth committed
1089
1090
		if (page)
			goto got_pg;
Linus Torvalds's avatar
Linus Torvalds committed
1091

1092
		out_of_memory(zonelist, gfp_mask, order);
Linus Torvalds's avatar
Linus Torvalds committed
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
		goto restart;
	}

	/*
	 * Don't let big-order allocations loop unless the caller explicitly
	 * requests that.  Wait for some write requests to complete then retry.
	 *
	 * In this implementation, __GFP_REPEAT means __GFP_NOFAIL for order
	 * <= 3, but that may not be true in other implementations.
	 */
	do_retry = 0;
	if (!(gfp_mask & __GFP_NORETRY)) {
		if ((order <= 3) || (gfp_mask & __GFP_REPEAT))
			do_retry = 1;
		if (gfp_mask & __GFP_NOFAIL)
			do_retry = 1;
	}
	if (do_retry) {
		blk_congestion_wait(WRITE, HZ/50);
		goto rebalance;
	}

nopage:
	if (!(gfp_mask & __GFP_NOWARN) && printk_ratelimit()) {
		printk(KERN_WARNING "%s: page allocation failure."
			" order:%d, mode:0x%x\n",
			p->comm, order, gfp_mask);
		dump_stack();
Janet Morgan's avatar
Janet Morgan committed
1121
		show_mem();
Linus Torvalds's avatar
Linus Torvalds committed
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
	}
got_pg:
	return page;
}

EXPORT_SYMBOL(__alloc_pages);

/*
 * Common helper functions.
 */
1132
fastcall unsigned long __get_free_pages(gfp_t gfp_mask, unsigned int order)
Linus Torvalds's avatar
Linus Torvalds committed
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
{
	struct page * page;
	page = alloc_pages(gfp_mask, order);
	if (!page)
		return 0;
	return (unsigned long) page_address(page);
}

EXPORT_SYMBOL(__get_free_pages);

1143
fastcall unsigned long get_zeroed_page(gfp_t gfp_mask)
Linus Torvalds's avatar
Linus Torvalds committed
1144
1145
1146
1147
1148
1149
1150
{
	struct page * page;

	/*
	 * get_zeroed_page() returns a 32-bit address, which cannot represent
	 * a highmem page
	 */
Nick Piggin's avatar
Nick Piggin committed
1151
	VM_BUG_ON((gfp_mask & __GFP_HIGHMEM) != 0);
Linus Torvalds's avatar
Linus Torvalds committed
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170

	page = alloc_pages(gfp_mask | __GFP_ZERO, 0);
	if (page)
		return (unsigned long) page_address(page);
	return 0;
}

EXPORT_SYMBOL(get_zeroed_page);

void __pagevec_free(struct pagevec *pvec)
{
	int i = pagevec_count(pvec);

	while (--i >= 0)
		free_hot_cold_page(pvec->pages[i], pvec->cold);
}

fastcall void __free_pages(struct page *page, unsigned int order)
{
Nick Piggin's avatar
Nick Piggin committed
1171
	if (put_page_testzero(page)) {
Linus Torvalds's avatar
Linus Torvalds committed
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
		if (order == 0)
			free_hot_page(page);
		else
			__free_pages_ok(page, order);
	}
}

EXPORT_SYMBOL(__free_pages);

fastcall void free_pages(unsigned long addr, unsigned int order)
{
	if (addr != 0) {
Nick Piggin's avatar
Nick Piggin committed
1184
		VM_BUG_ON(!virt_addr_valid((void *)addr));
Linus Torvalds's avatar
Linus Torvalds committed
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205