page_alloc.c 90.7 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>
42
#include <linux/backing-dev.h>
43
#include <linux/fault-inject.h>
Linus Torvalds's avatar
Linus Torvalds committed
44
45

#include <asm/tlbflush.h>
46
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
47
48
49
50
51
52
#include "internal.h"

/*
 * MCD - HACK: Find somewhere to initialize this EARLY, or make this
 * initializer cleaner
 */
53
nodemask_t node_online_map __read_mostly = { { [0] = 1UL } };
54
EXPORT_SYMBOL(node_online_map);
55
nodemask_t node_possible_map __read_mostly = NODE_MASK_ALL;
56
EXPORT_SYMBOL(node_possible_map);
57
unsigned long totalram_pages __read_mostly;
58
unsigned long totalreserve_pages __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
59
long nr_swap_pages;
60
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
61

62
static void __free_pages_ok(struct page *page, unsigned int order);
63

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

EXPORT_SYMBOL(totalram_pages);

87
static char * const zone_names[MAX_NR_ZONES] = {
88
	 "DMA",
89
#ifdef CONFIG_ZONE_DMA32
90
	 "DMA32",
91
#endif
92
	 "Normal",
93
#ifdef CONFIG_HIGHMEM
94
	 "HighMem"
95
#endif
96
97
};

Linus Torvalds's avatar
Linus Torvalds committed
98
99
int min_free_kbytes = 1024;

100
101
unsigned long __meminitdata nr_kernel_pages;
unsigned long __meminitdata nr_all_pages;
102
static unsigned long __initdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
103

104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
#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];
129
130
131
132
#ifdef CONFIG_MEMORY_HOTPLUG_RESERVE
  unsigned long __initdata node_boundary_start_pfn[MAX_NUMNODES];
  unsigned long __initdata node_boundary_end_pfn[MAX_NUMNODES];
#endif /* CONFIG_MEMORY_HOTPLUG_RESERVE */
133
134
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Nick Piggin's avatar
Nick Piggin committed
135
#ifdef CONFIG_DEBUG_VM
136
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
137
{
138
139
140
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
141

142
143
144
145
146
147
148
149
150
	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;
151
152
153
154
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
Linus Torvalds's avatar
Linus Torvalds committed
155
156
#ifdef CONFIG_HOLES_IN_ZONE
	if (!pfn_valid(page_to_pfn(page)))
157
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
158
159
#endif
	if (zone != page_zone(page))
160
161
162
163
164
165
166
167
168
169
		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
170
		return 1;
171
172
173
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
174
175
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
176
177
178
179
180
181
182
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

183
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
184
{
185
	printk(KERN_EMERG "Bad page state in process '%s'\n"
186
187
188
		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",
189
190
191
		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
192
	dump_stack();
193
194
	page->flags &= ~(1 << PG_lru	|
			1 << PG_private |
Linus Torvalds's avatar
Linus Torvalds committed
195
196
197
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
198
199
			1 << PG_reclaim |
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
200
			1 << PG_swapcache |
201
202
			1 << PG_writeback |
			1 << PG_buddy );
Linus Torvalds's avatar
Linus Torvalds committed
203
204
205
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
206
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
207
208
209
210
211
212
213
214
215
216
217
218
}

/*
 * 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).
 *
219
220
221
 * 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
222
 */
223
224
225
226
227
228

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
229
230
231
232
233
static void prep_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;

234
	set_compound_page_dtor(page, free_compound_page);
235
	page[1].lru.prev = (void *)order;
Linus Torvalds's avatar
Linus Torvalds committed
236
237
238
	for (i = 0; i < nr_pages; i++) {
		struct page *p = page + i;

239
		__SetPageCompound(p);
240
		set_page_private(p, (unsigned long)page);
Linus Torvalds's avatar
Linus Torvalds committed
241
242
243
244
245
246
247
248
	}
}

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

249
	if (unlikely((unsigned long)page[1].lru.prev != order))
250
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
251
252
253
254

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

255
256
257
		if (unlikely(!PageCompound(p) |
				(page_private(p) != (unsigned long)page)))
			bad_page(page);
258
		__ClearPageCompound(p);
Linus Torvalds's avatar
Linus Torvalds committed
259
260
261
	}
}

Nick Piggin's avatar
Nick Piggin committed
262
263
264
265
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

Nick Piggin's avatar
Nick Piggin committed
266
	VM_BUG_ON((gfp_flags & (__GFP_WAIT | __GFP_HIGHMEM)) == __GFP_HIGHMEM);
267
268
269
270
	/*
	 * 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
271
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
272
273
274
275
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

Linus Torvalds's avatar
Linus Torvalds committed
276
277
278
279
280
/*
 * 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.
 */
281
282
static inline unsigned long page_order(struct page *page)
{
283
	return page_private(page);
Linus Torvalds's avatar
Linus Torvalds committed
284
285
}

286
287
static inline void set_page_order(struct page *page, int order)
{
288
	set_page_private(page, order);
289
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
290
291
292
293
}

static inline void rmv_page_order(struct page *page)
{
294
	__ClearPageBuddy(page);
295
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
}

/*
 * 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
313
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
 */
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
332
 * (a) the buddy is not in a hole &&
333
 * (b) the buddy is in the buddy system &&
334
335
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
336
337
338
 *
 * 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
339
 *
340
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
341
 */
342
343
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
344
{
Nick Piggin's avatar
Nick Piggin committed
345
#ifdef CONFIG_HOLES_IN_ZONE
346
	if (!pfn_valid(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
347
348
349
		return 0;
#endif

350
351
352
353
354
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
355
		return 1;
356
	}
357
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
}

/*
 * 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
373
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
374
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
375
376
377
378
379
380
381
382
383
 * 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
384
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
385
386
387
388
389
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;

390
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
391
392
393
394
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
395
396
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
397
398
399
400
401
402
403
404

	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);
405
		if (!page_is_buddy(page, buddy, order))
Linus Torvalds's avatar
Linus Torvalds committed
406
			break;		/* Move the buddy up one level. */
Nick Piggin's avatar
Nick Piggin committed
407

Linus Torvalds's avatar
Linus Torvalds committed
408
409
410
411
		list_del(&buddy->lru);
		area = zone->free_area + order;
		area->nr_free--;
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
412
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
413
414
415
416
417
418
419
420
421
		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++;
}

422
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
423
{
Nick Piggin's avatar
Nick Piggin committed
424
425
426
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
Linus Torvalds's avatar
Linus Torvalds committed
427
428
429
430
431
432
433
434
		(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
435
			1 << PG_writeback |
436
437
			1 << PG_reserved |
			1 << PG_buddy ))))
438
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
439
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
440
		__ClearPageDirty(page);
441
442
443
444
445
446
	/*
	 * 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
447
448
449
450
451
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
452
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
453
454
455
456
457
458
459
 *
 * 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
460
461
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
462
{
Nick Piggin's avatar
Nick Piggin committed
463
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
464
465
	zone->all_unreclaimable = 0;
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
466
467
468
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
469
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
470
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
471
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
472
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
473
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
474
	}
Nick Piggin's avatar
Nick Piggin committed
475
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
476
477
}

Nick Piggin's avatar
Nick Piggin committed
478
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
479
{
480
481
482
	spin_lock(&zone->lock);
	zone->all_unreclaimable = 0;
	zone->pages_scanned = 0;
483
	__free_one_page(page, zone, order);
484
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
485
486
487
488
489
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
490
	int i;
491
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
492
493

	for (i = 0 ; i < (1 << order) ; ++i)
494
		reserved += free_pages_check(page + i);
495
496
497
	if (reserved)
		return;

Nick Piggin's avatar
Nick Piggin committed
498
499
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
Nick Piggin's avatar
Nick Piggin committed
500
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
501
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
502

Nick Piggin's avatar
Nick Piggin committed
503
	local_irq_save(flags);
504
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
505
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
506
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
507
508
}

509
510
511
512
513
514
515
516
/*
 * 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);
517
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
518
		__free_page(page);
519
520
521
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
522
		prefetchw(page);
523
524
525
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
526
527
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
528
529
530
531
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

532
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
533
		__free_pages(page, order);
534
535
536
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551

/*
 * 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
552
static inline void expand(struct zone *zone, struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
553
554
555
556
557
558
559
560
 	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
561
		VM_BUG_ON(bad_range(zone, &page[size]));
Linus Torvalds's avatar
Linus Torvalds committed
562
563
564
565
566
567
568
569
570
		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
571
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
572
{
Nick Piggin's avatar
Nick Piggin committed
573
574
575
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
576
577
		(page->flags & (
			1 << PG_lru	|
Linus Torvalds's avatar
Linus Torvalds committed
578
579
580
581
582
			1 << PG_private	|
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
			1 << PG_reclaim	|
583
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
584
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
585
			1 << PG_writeback |
586
587
			1 << PG_reserved |
			1 << PG_buddy ))))
588
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
589

590
591
592
593
594
595
596
	/*
	 * 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
597
598
599
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error |
			1 << PG_referenced | 1 << PG_arch_1 |
			1 << PG_checked | 1 << PG_mappedtodisk);
600
	set_page_private(page, 0);
601
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
602
603

	arch_alloc_page(page, order);
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
			if (pcp->count) {
689
690
				int to_drain;

691
				local_irq_save(flags);
692
693
694
695
696
697
				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;
698
699
				local_irq_restore(flags);
			}
700
701
702
703
704
		}
	}
}
#endif

Linus Torvalds's avatar
Linus Torvalds committed
705
706
static void __drain_pages(unsigned int cpu)
{
Nick Piggin's avatar
Nick Piggin committed
707
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
708
709
710
711
712
713
	struct zone *zone;
	int i;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;

714
		pset = zone_pcp(zone, cpu);
Linus Torvalds's avatar
Linus Torvalds committed
715
716
717
718
		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
719
			local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
720
721
			free_pages_bulk(zone, pcp->count, &pcp->list, 0);
			pcp->count = 0;
Nick Piggin's avatar
Nick Piggin committed
722
			local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
723
724
725
726
727
728
729
730
		}
	}
}

#ifdef CONFIG_PM

void mark_free_pages(struct zone *zone)
{
731
732
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
733
734
735
736
737
738
739
	int order;
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
740
741
742
743
744
745
746
747
748

	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
749
750
751

	for (order = MAX_ORDER - 1; order >= 0; --order)
		list_for_each(curr, &zone->free_area[order].free_list) {
752
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
753

754
755
756
757
			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
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
784
785

	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;

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

Nick Piggin's avatar
Nick Piggin committed
789
790
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
Nick Piggin's avatar
Nick Piggin committed
791
	arch_free_page(page, 0);
792
793
	kernel_map_pages(page, 1, 0);

794
	pcp = &zone_pcp(zone, get_cpu())->pcp[cold];
Linus Torvalds's avatar
Linus Torvalds committed
795
	local_irq_save(flags);
796
	__count_vm_event(PGFREE);
Linus Torvalds's avatar
Linus Torvalds committed
797
798
	list_add(&page->lru, &pcp->list);
	pcp->count++;
Nick Piggin's avatar
Nick Piggin committed
799
800
801
802
	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
803
804
805
806
807
808
809
810
811
812
813
814
815
816
	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);
}

817
818
819
820
821
822
823
824
825
826
827
828
/*
 * 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
829
830
	VM_BUG_ON(PageCompound(page));
	VM_BUG_ON(!page_count(page));
831
832
	for (i = 1; i < (1 << order); i++)
		set_page_refcounted(page + i);
833
834
}

Linus Torvalds's avatar
Linus Torvalds committed
835
836
837
838
839
/*
 * 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
840
841
static struct page *buffered_rmqueue(struct zonelist *zonelist,
			struct zone *zone, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
842
843
{
	unsigned long flags;
844
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
845
	int cold = !!(gfp_flags & __GFP_COLD);
Nick Piggin's avatar
Nick Piggin committed
846
	int cpu;
Linus Torvalds's avatar
Linus Torvalds committed
847

848
again:
Nick Piggin's avatar
Nick Piggin committed
849
	cpu  = get_cpu();
Nick Piggin's avatar
Nick Piggin committed
850
	if (likely(order == 0)) {
Linus Torvalds's avatar
Linus Torvalds committed
851
852
		struct per_cpu_pages *pcp;

Nick Piggin's avatar
Nick Piggin committed
853
		pcp = &zone_pcp(zone, cpu)->pcp[cold];
Linus Torvalds's avatar
Linus Torvalds committed
854
		local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
855
		if (!pcp->count) {
856
			pcp->count = rmqueue_bulk(zone, 0,
Linus Torvalds's avatar
Linus Torvalds committed
857
						pcp->batch, &pcp->list);
Nick Piggin's avatar
Nick Piggin committed
858
859
			if (unlikely(!pcp->count))
				goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
860
		}
Nick Piggin's avatar
Nick Piggin committed
861
862
863
		page = list_entry(pcp->list.next, struct page, lru);
		list_del(&page->lru);
		pcp->count--;
Rohit Seth's avatar
Rohit Seth committed
864
	} else {
Linus Torvalds's avatar
Linus Torvalds committed
865
866
		spin_lock_irqsave(&zone->lock, flags);
		page = __rmqueue(zone, order);
Nick Piggin's avatar
Nick Piggin committed
867
868
869
		spin_unlock(&zone->lock);
		if (!page)
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
870
871
	}

872
	__count_zone_vm_events(PGALLOC, zone, 1 << order);
873
	zone_statistics(zonelist, zone);
Nick Piggin's avatar
Nick Piggin committed
874
875
	local_irq_restore(flags);
	put_cpu();
Linus Torvalds's avatar
Linus Torvalds committed
876

Nick Piggin's avatar
Nick Piggin committed
877
	VM_BUG_ON(bad_range(zone, page));
Nick Piggin's avatar
Nick Piggin committed
878
	if (prep_new_page(page, order, gfp_flags))
Nick Piggin's avatar
Nick Piggin committed
879
		goto again;
Linus Torvalds's avatar
Linus Torvalds committed
880
	return page;
Nick Piggin's avatar
Nick Piggin committed
881
882
883
884
885

failed:
	local_irq_restore(flags);
	put_cpu();
	return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
886
887
}

Rohit Seth's avatar
Rohit Seth committed
888
#define ALLOC_NO_WATERMARKS	0x01 /* don't check watermarks at all */
889
890
891
892
893
894
#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
895

896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
#ifdef CONFIG_FAIL_PAGE_ALLOC

static struct fail_page_alloc_attr {
	struct fault_attr attr;

	u32 ignore_gfp_highmem;
	u32 ignore_gfp_wait;

#ifdef CONFIG_FAULT_INJECTION_DEBUG_FS

	struct dentry *ignore_gfp_highmem_file;
	struct dentry *ignore_gfp_wait_file;

#endif /* CONFIG_FAULT_INJECTION_DEBUG_FS */

} fail_page_alloc = {
	.attr = FAULT_ATTR_INITIALIZER,
};

static int __init setup_fail_page_alloc(char *str)
{
	return setup_fault_attr(&fail_page_alloc.attr, str);
}
__setup("fail_page_alloc=", setup_fail_page_alloc);

static int should_fail_alloc_page(gfp_t gfp_mask, unsigned int order)
{
	if (gfp_mask & __GFP_NOFAIL)
		return 0;
	if (fail_page_alloc.ignore_gfp_highmem && (gfp_mask & __GFP_HIGHMEM))
		return 0;
	if (fail_page_alloc.ignore_gfp_wait && (gfp_mask & __GFP_WAIT))
		return 0;

	return should_fail(&fail_page_alloc.attr, 1 << order);
}

#ifdef CONFIG_FAULT_INJECTION_DEBUG_FS

static int __init fail_page_alloc_debugfs(void)
{
	mode_t mode = S_IFREG | S_IRUSR | S_IWUSR;
	struct dentry *dir;
	int err;

	err = init_fault_attr_dentries(&fail_page_alloc.attr,
				       "fail_page_alloc");
	if (err)
		return err;
	dir = fail_page_alloc.attr.dentries.dir;

	fail_page_alloc.ignore_gfp_wait_file =
		debugfs_create_bool("ignore-gfp-wait", mode, dir,
				      &fail_page_alloc.ignore_gfp_wait);

	fail_page_alloc.ignore_gfp_highmem_file =
		debugfs_create_bool("ignore-gfp-highmem", mode, dir,
				      &fail_page_alloc.ignore_gfp_highmem);

	if (!fail_page_alloc.ignore_gfp_wait_file ||
			!fail_page_alloc.ignore_gfp_highmem_file) {
		err = -ENOMEM;
		debugfs_remove(fail_page_alloc.ignore_gfp_wait_file);
		debugfs_remove(fail_page_alloc.ignore_gfp_highmem_file);
		cleanup_fault_attr_dentries(&fail_page_alloc.attr);
	}

	return err;
}

late_initcall(fail_page_alloc_debugfs);

#endif /* CONFIG_FAULT_INJECTION_DEBUG_FS */

#else /* CONFIG_FAIL_PAGE_ALLOC */

static inline int should_fail_alloc_page(gfp_t gfp_mask, unsigned int order)
{
	return 0;
}

#endif /* CONFIG_FAIL_PAGE_ALLOC */

Linus Torvalds's avatar
Linus Torvalds committed
979
980
981
982
983
/*
 * 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
984
		      int classzone_idx, int alloc_flags)
Linus Torvalds's avatar
Linus Torvalds committed
985
986
{
	/* free_pages my go negative - that's OK */
987
988
	unsigned long min = mark;
	long free_pages = z->free_pages - (1 << order) + 1;
Linus Torvalds's avatar
Linus Torvalds committed
989
990
	int o;

Rohit Seth's avatar
Rohit Seth committed
991
	if (alloc_flags & ALLOC_HIGH)
Linus Torvalds's avatar
Linus Torvalds committed
992
		min -= min / 2;
Rohit Seth's avatar
Rohit Seth committed
993
	if (alloc_flags & ALLOC_HARDER)
Linus Torvalds's avatar
Linus Torvalds committed
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
		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;
}

1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
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
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
#ifdef CONFIG_NUMA
/*
 * zlc_setup - Setup for "zonelist cache".  Uses cached zone data to
 * skip over zones that are not allowed by the cpuset, or that have
 * been recently (in last second) found to be nearly full.  See further
 * comments in mmzone.h.  Reduces cache footprint of zonelist scans
 * that have to skip over alot of full or unallowed zones.
 *
 * If the zonelist cache is present in the passed in zonelist, then
 * returns a pointer to the allowed node mask (either the current
 * tasks mems_allowed, or node_online_map.)
 *
 * If the zonelist cache is not available for this zonelist, does
 * nothing and returns NULL.
 *
 * If the fullzones BITMAP in the zonelist cache is stale (more than
 * a second since last zap'd) then we zap it out (clear its bits.)
 *
 * We hold off even calling zlc_setup, until after we've checked the
 * first zone in the zonelist, on the theory that most allocations will
 * be satisfied from that first zone, so best to examine that zone as
 * quickly as we can.
 */
static nodemask_t *zlc_setup(struct zonelist *zonelist, int alloc_flags)
{
	struct zonelist_cache *zlc;	/* cached zonelist speedup info */
	nodemask_t *allowednodes;	/* zonelist_cache approximation */

	zlc = zonelist->zlcache_ptr;
	if (!zlc)
		return NULL;

	if (jiffies - zlc->last_full_zap > 1 * HZ) {
		bitmap_zero(zlc->fullzones, MAX_ZONES_PER_ZONELIST);
		zlc->last_full_zap = jiffies;
	}

	allowednodes = !in_interrupt() && (alloc_flags & ALLOC_CPUSET) ?
					&cpuset_current_mems_allowed :
					&node_online_map;
	return allowednodes;
}

/*
 * Given 'z' scanning a zonelist, run a couple of quick checks to see
 * if it is worth looking at further for free memory:
 *  1) Check that the zone isn't thought to be full (doesn't have its
 *     bit set in the zonelist_cache fullzones BITMAP).
 *  2) Check that the zones node (obtained from the zonelist_cache
 *     z_to_n[] mapping) is allowed in the passed in allowednodes mask.
 * Return true (non-zero) if zone is worth looking at further, or
 * else return false (zero) if it is not.
 *
 * This check -ignores- the distinction between various watermarks,
 * such as GFP_HIGH, GFP_ATOMIC, PF_MEMALLOC, ...  If a zone is
 * found to be full for any variation of these watermarks, it will
 * be considered full for up to one second by all requests, unless
 * we are so low on memory on all allowed nodes that we are forced
 * into the second scan of the zonelist.
 *
 * In the second scan we ignore this zonelist cache and exactly
 * apply the watermarks to all zones, even it is slower to do so.
 * We are low on memory in the second scan, and should leave no stone
 * unturned looking for a free page.
 */
static int zlc_zone_worth_trying(struct zonelist *zonelist, struct zone **z,
						nodemask_t *allowednodes)
{
	struct zonelist_cache *zlc;	/* cached zonelist speedup info */
	int i;				/* index of *z in zonelist zones */
	int n;				/* node that zone *z is on */

	zlc = zonelist->zlcache_ptr;
	if (!zlc)
		return 1;

	i = z - zonelist->zones;
	n = zlc->z_to_n[i];

	/* This zone is worth trying if it is allowed but not full */
	return node_isset(n, *allowednodes) && !test_bit(i, zlc->fullzones);
}

/*
 * Given 'z' scanning a zonelist, set the corresponding bit in
 * zlc->fullzones, so that subsequent attempts to allocate a page
 * from that zone don't waste time re-examining it.
 */
static void zlc_mark_zone_full(struct zonelist *zonelist, struct zone **z)
{
	struct zonelist_cache *zlc;	/* cached zonelist speedup info */
	int i;				/* index of *z in zonelist zones */

	zlc = zonelist->zlcache_ptr;
	if (!zlc)
		return;

	i = z - zonelist->zones;

	set_bit(i, zlc->fullzones);
}

#else	/* CONFIG_NUMA */

static nodemask_t *zlc_setup(struct zonelist *zonelist, int alloc_flags)
{
	return NULL;
}

static int zlc_zone_worth_trying(struct zonelist *zonelist, struct zone **z,
				nodemask_t *allowednodes)
{
	return 1;
}

static void zlc_mark_zone_full(struct zonelist *zonelist, struct zone **z)
{
}
#endif	/* CONFIG_NUMA */

Rohit Seth's avatar
Rohit Seth committed
1131
/*
1132
 * get_page_from_freelist goes through the zonelist trying to allocate
Rohit Seth's avatar
Rohit Seth committed
1133
1134
1135
1136
1137
 * 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
1138
{
1139
	struct zone **z;
Rohit Seth's avatar
Rohit Seth committed
1140
	struct page *page = NULL;
1141
	int classzone_idx = zone_idx(zonelist->zones[0]);
1142
	struct zone *zone;
1143
1144
1145
	nodemask_t *allowednodes = NULL;/* zonelist_cache approximation */
	int zlc_active = 0;		/* set if using zonelist_cache */
	int did_zlc_setup = 0;		/* just call zlc_setup() one time */
Rohit Seth's avatar
Rohit Seth committed
1146

1147
zonelist_scan:
Rohit Seth's avatar
Rohit Seth committed
1148
	/*
1149
	 * Scan zonelist, looking for a zone with enough free.
Rohit Seth's avatar
Rohit Seth committed
1150
1151
	 * See also cpuset_zone_allowed() comment in kernel/cpuset.c.
	 */
1152
1153
	z = zonelist->zones;

Rohit Seth's avatar
Rohit Seth committed
1154
	do {
1155
1156
1157
		if (NUMA_BUILD && zlc_active &&
			!zlc_zone_worth_trying(zonelist, z, allowednodes))
				continue;
1158
		zone = *z;
1159
		if (unlikely(NUMA_BUILD && (gfp_mask & __GFP_THISNODE) &&
1160
			zone->zone_pgdat != zonelist->zones[0]->zone_pgdat))
1161
				break;
Rohit Seth's avatar
Rohit Seth committed
1162
		if ((alloc_flags & ALLOC_CPUSET) &&
1163
			!cpuset_zone_allowed(zone, gfp_mask))
1164
				goto try_next_zone;
Rohit Seth's avatar
Rohit Seth committed
1165
1166

		if (!(alloc_flags & ALLOC_NO_WATERMARKS)) {
1167
1168
			unsigned long mark;
			if (alloc_flags & ALLOC_WMARK_MIN)
1169
				mark = zone->pages_min;
1170
			else if (alloc_flags & ALLOC_WMARK_LOW)
1171
				mark = zone->pages_low;
1172
			else
1173
				mark = zone->pages_high;
1174
1175
			if (!zone_watermark_ok(zone, order, mark,
				    classzone_idx, alloc_flags)) {
1176
				if (!zone_reclaim_mode ||
1177
				    !zone_reclaim(zone, gfp_mask, order))
1178
					goto this_zone_full;
1179
			}
Rohit Seth's avatar
Rohit Seth committed
1180
1181
		}

1182
		page = buffered_rmqueue(zonelist, zone, order, gfp_mask);
1183
		if (page)
Rohit Seth's avatar
Rohit Seth committed
1184
			break;
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
this_zone_full:
		if (NUMA_BUILD)
			zlc_mark_zone_full(zonelist, z);
try_next_zone:
		if (NUMA_BUILD && !did_zlc_setup) {
			/* we do zlc_setup after the first zone is tried */
			allowednodes = zlc_setup(zonelist, alloc_flags);
			zlc_active = 1;
			did_zlc_setup = 1;
		}
Rohit Seth's avatar
Rohit Seth committed
1195
	} while (*(++z) != NULL);
1196
1197
1198
1199
1200
1201

	if (unlikely(NUMA_BUILD && page == NULL && zlc_active)) {
		/* Disable zlc cache for second zonelist scan */
		zlc_active = 0;
		goto zonelist_scan;
	}
Rohit Seth's avatar
Rohit Seth committed
1202
	return page;
Martin Hicks's avatar
Martin Hicks committed
1203
1204
}

Linus Torvalds's avatar
Linus Torvalds committed
1205
1206
1207
1208
/*
 * This is the 'heart' of the zoned buddy allocator.
 */
struct page * fastcall
1209
__alloc_pages(gfp_t gfp_mask, unsigned int order,
Linus Torvalds's avatar
Linus Torvalds committed
1210
1211
		struct zonelist *zonelist)
{
Al Viro's avatar
Al Viro committed
1212
	const gfp_t wait = gfp_mask & __GFP_WAIT;
Rohit Seth's avatar
Rohit Seth committed
1213
	struct zone **z;
Linus Torvalds's avatar
Linus Torvalds committed
1214
1215
1216
1217
	struct page *page;
	struct reclaim_state reclaim_state;
	struct task_struct *p = current;
	int do_retry;
Rohit Seth's avatar
Rohit Seth committed
1218
	int alloc_flags;
Linus Torvalds's avatar
Linus Torvalds committed
1219
1220
1221
1222
	int did_some_progress;

	might_sleep_if(wait);

1223
1224
1225
	if (should_fail_alloc_page(gfp_mask, order))
		return NULL;

1226
restart:
Rohit Seth's avatar
Rohit Seth committed
1227
	z = zonelist->zones;  /* the list of zones suitable for gfp_mask */
Linus Torvalds's avatar
Linus Torvalds committed
1228

Rohit Seth's avatar
Rohit Seth committed
1229
	if (unlikely(*z == NULL)) {
Linus Torvalds's avatar
Linus Torvalds committed
1230
1231
1232
		/* Should this ever happen?? */
		return NULL;
	}
1233

Rohit Seth's avatar
Rohit Seth committed
1234
	page = get_page_from_freelist(gfp_mask|__GFP_HARDWALL, order,
1235
				zonelist