page_alloc.c 130 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
/*
 *  linux/mm/page_alloc.c
 *
 *  Manages the free list, the system allocates free pages here.
 *  Note that kmalloc() lives in slab.c
 *
 *  Copyright (C) 1991, 1992, 1993, 1994  Linus Torvalds
 *  Swap reorganised 29.12.95, Stephen Tweedie
 *  Support of BIGMEM added by Gerhard Wichert, Siemens AG, July 1999
 *  Reshaped it to be a zoned allocator, Ingo Molnar, Red Hat, 1999
 *  Discontiguous memory support, Kanoj Sarcar, SGI, Nov 1999
 *  Zone balancing, Kanoj Sarcar, SGI, Jan 2000
 *  Per cpu hot/cold page lists, bulk allocation, Martin J. Bligh, Sept 2002
 *          (lots of bits borrowed from Ingo Molnar & Andrew Morton)
 */

#include <linux/stddef.h>
#include <linux/mm.h>
#include <linux/swap.h>
#include <linux/interrupt.h>
#include <linux/pagemap.h>
22
#include <linux/jiffies.h>
Linus Torvalds's avatar
Linus Torvalds committed
23 24
#include <linux/bootmem.h>
#include <linux/compiler.h>
25
#include <linux/kernel.h>
Linus Torvalds's avatar
Linus Torvalds committed
26 27 28 29 30
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
31
#include <linux/oom.h>
Linus Torvalds's avatar
Linus Torvalds committed
32 33 34 35 36
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
37
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
38 39
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
40
#include <linux/mempolicy.h>
41
#include <linux/stop_machine.h>
42 43
#include <linux/sort.h>
#include <linux/pfn.h>
44
#include <linux/backing-dev.h>
45
#include <linux/fault-inject.h>
46
#include <linux/page-isolation.h>
47
#include <linux/page_cgroup.h>
48
#include <linux/debugobjects.h>
Linus Torvalds's avatar
Linus Torvalds committed
49 50

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

/*
55
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
56
 */
57 58 59 60 61 62 63 64 65 66 67 68 69
nodemask_t node_states[NR_NODE_STATES] __read_mostly = {
	[N_POSSIBLE] = NODE_MASK_ALL,
	[N_ONLINE] = { { [0] = 1UL } },
#ifndef CONFIG_NUMA
	[N_NORMAL_MEMORY] = { { [0] = 1UL } },
#ifdef CONFIG_HIGHMEM
	[N_HIGH_MEMORY] = { { [0] = 1UL } },
#endif
	[N_CPU] = { { [0] = 1UL } },
#endif	/* NUMA */
};
EXPORT_SYMBOL(node_states);

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

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

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

Linus Torvalds's avatar
Linus Torvalds committed
80 81 82 83 84 85 86
/*
 * results with 256, 32 in the lowmem_reserve sysctl:
 *	1G machine -> (16M dma, 800M-16M normal, 1G-800M high)
 *	1G machine -> (16M dma, 784M normal, 224M high)
 *	NORMAL allocation will leave 784M/256 of ram reserved in the ZONE_DMA
 *	HIGHMEM allocation will leave 224M/32 of ram reserved in ZONE_NORMAL
 *	HIGHMEM allocation will (224M+784M)/256 of ram reserved in ZONE_DMA
87 88 89
 *
 * TBD: should special case ZONE_DMA32 machines here - in those we normally
 * don't need any ZONE_NORMAL reservation
Linus Torvalds's avatar
Linus Torvalds committed
90
 */
91
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
92
#ifdef CONFIG_ZONE_DMA
93
	 256,
94
#endif
95
#ifdef CONFIG_ZONE_DMA32
96
	 256,
97
#endif
98
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
99
	 32,
100
#endif
Mel Gorman's avatar
Mel Gorman committed
101
	 32,
102
};
Linus Torvalds's avatar
Linus Torvalds committed
103 104 105

EXPORT_SYMBOL(totalram_pages);

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

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

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

126 127
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
128
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146
   * ranges of memory (RAM) that may be registered with add_active_range().
   * Ranges passed to add_active_range() will be merged if possible
   * so the number of times add_active_range() can be called is
   * related to the number of nodes and the number of holes
   */
  #ifdef CONFIG_MAX_ACTIVE_REGIONS
    /* Allow an architecture to set MAX_ACTIVE_REGIONS to save memory */
    #define MAX_ACTIVE_REGIONS CONFIG_MAX_ACTIVE_REGIONS
  #else
    #if MAX_NUMNODES >= 32
      /* If there can be many nodes, allow up to 50 holes per node */
      #define MAX_ACTIVE_REGIONS (MAX_NUMNODES*50)
    #else
      /* By default, allow up to 256 distinct regions */
      #define MAX_ACTIVE_REGIONS 256
    #endif
  #endif

147 148 149 150
  static struct node_active_region __meminitdata early_node_map[MAX_ACTIVE_REGIONS];
  static int __meminitdata nr_nodemap_entries;
  static unsigned long __meminitdata arch_zone_lowest_possible_pfn[MAX_NR_ZONES];
  static unsigned long __meminitdata arch_zone_highest_possible_pfn[MAX_NR_ZONES];
151
#ifdef CONFIG_MEMORY_HOTPLUG_RESERVE
152 153
  static unsigned long __meminitdata node_boundary_start_pfn[MAX_NUMNODES];
  static unsigned long __meminitdata node_boundary_end_pfn[MAX_NUMNODES];
154
#endif /* CONFIG_MEMORY_HOTPLUG_RESERVE */
Adrian Bunk's avatar
Adrian Bunk committed
155
  static unsigned long __initdata required_kernelcore;
156
  static unsigned long __initdata required_movablecore;
Adrian Bunk's avatar
Adrian Bunk committed
157
  static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
158 159 160 161

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

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

169 170
int page_group_by_mobility_disabled __read_mostly;

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

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

184 185 186 187 188 189 190 191 192
	do {
		seq = zone_span_seqbegin(zone);
		if (pfn >= zone->zone_start_pfn + zone->spanned_pages)
			ret = 1;
		else if (pfn < zone->zone_start_pfn)
			ret = 1;
	} while (zone_span_seqretry(zone, seq));

	return ret;
193 194 195 196
}

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

	return 1;
}
/*
 * Temporary debugging check for pages not lying within a given zone.
 */
static int bad_range(struct zone *zone, struct page *page)
{
	if (page_outside_zone_boundaries(zone, page))
Linus Torvalds's avatar
Linus Torvalds committed
210
		return 1;
211 212 213
	if (!page_is_consistent(zone, page))
		return 1;

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

223
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
224
{
225 226
	printk(KERN_EMERG "Bad page state in process '%s'\n" KERN_EMERG
		"page:%p flags:0x%0*lx mapping:%p mapcount:%d count:%d\n",
227 228 229
		current->comm, page, (int)(2*sizeof(unsigned long)),
		(unsigned long)page->flags, page->mapping,
		page_mapcount(page), page_count(page));
230

231 232
	printk(KERN_EMERG "Trying to fix it up, but a reboot is needed\n"
		KERN_EMERG "Backtrace:\n");
Linus Torvalds's avatar
Linus Torvalds committed
233
	dump_stack();
234
	page->flags &= ~PAGE_FLAGS_CLEAR_WHEN_BAD;
Linus Torvalds's avatar
Linus Torvalds committed
235 236 237
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
238
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
239 240 241 242 243 244 245 246 247 248 249 250
}

/*
 * 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).
 *
251 252 253
 * 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
254
 */
255 256 257

static void free_compound_page(struct page *page)
{
258
	__free_pages_ok(page, compound_order(page));
259 260
}

261
void prep_compound_page(struct page *page, unsigned long order)
262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278
{
	int i;
	int nr_pages = 1 << order;

	set_compound_page_dtor(page, free_compound_page);
	set_compound_order(page, order);
	__SetPageHead(page);
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;

		__SetPageTail(p);
		p->first_page = page;
	}
}

#ifdef CONFIG_HUGETLBFS
void prep_compound_gigantic_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
279 280 281
{
	int i;
	int nr_pages = 1 << order;
282
	struct page *p = page + 1;
Linus Torvalds's avatar
Linus Torvalds committed
283

284
	set_compound_page_dtor(page, free_compound_page);
285
	set_compound_order(page, order);
286
	__SetPageHead(page);
287
	for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) {
288 289
		__SetPageTail(p);
		p->first_page = page;
Linus Torvalds's avatar
Linus Torvalds committed
290 291
	}
}
292
#endif
Linus Torvalds's avatar
Linus Torvalds committed
293 294 295 296 297 298

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

299
	if (unlikely(compound_order(page) != order))
300
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
301

302
	if (unlikely(!PageHead(page)))
303
			bad_page(page);
304
	__ClearPageHead(page);
305 306
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
Linus Torvalds's avatar
Linus Torvalds committed
307

308
		if (unlikely(!PageTail(p) |
309
				(p->first_page != page)))
310
			bad_page(page);
311
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
312 313 314
	}
}

Nick Piggin's avatar
Nick Piggin committed
315 316 317 318
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

319 320 321 322
	/*
	 * 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
323
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
324 325 326 327
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

328 329
static inline void set_page_order(struct page *page, int order)
{
330
	set_page_private(page, order);
331
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
332 333 334 335
}

static inline void rmv_page_order(struct page *page)
{
336
	__ClearPageBuddy(page);
337
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
}

/*
 * 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
355
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373
 */
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
374
 * (a) the buddy is not in a hole &&
375
 * (b) the buddy is in the buddy system &&
376 377
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
378 379 380
 *
 * 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
381
 *
382
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
383
 */
384 385
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
386
{
387
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
388 389
		return 0;

390 391 392 393 394
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
395
		return 1;
396
	}
397
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
398 399 400 401 402 403 404 405 406 407 408 409 410 411 412
}

/*
 * 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
413
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
414
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
415 416 417 418 419 420 421 422 423
 * 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
424
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
425 426 427 428
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;
429
	int migratetype = get_pageblock_migratetype(page);
Linus Torvalds's avatar
Linus Torvalds committed
430

431
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
432 433 434 435
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
436 437
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
438

439
	__mod_zone_page_state(zone, NR_FREE_PAGES, order_size);
Linus Torvalds's avatar
Linus Torvalds committed
440 441 442 443 444
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
445
		if (!page_is_buddy(page, buddy, order))
446
			break;
Nick Piggin's avatar
Nick Piggin committed
447

448
		/* Our buddy is free, merge with it and move up one order. */
Linus Torvalds's avatar
Linus Torvalds committed
449
		list_del(&buddy->lru);
450
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
451
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
452
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
453 454 455 456 457
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
458 459
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
460 461 462
	zone->free_area[order].nr_free++;
}

463
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
464
{
465
	free_page_mlock(page);
Nick Piggin's avatar
Nick Piggin committed
466 467 468
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
469
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE)))
470
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
471
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
472
		__ClearPageDirty(page);
473 474
	if (PageSwapBacked(page))
		__ClearPageSwapBacked(page);
475 476 477 478 479 480
	/*
	 * 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
481 482 483 484 485
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
486
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
487 488 489 490 491 492 493
 *
 * 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
494 495
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
496
{
Nick Piggin's avatar
Nick Piggin committed
497
	spin_lock(&zone->lock);
498
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
Linus Torvalds's avatar
Linus Torvalds committed
499
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
500 501 502
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
503
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
504
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
505
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
506
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
507
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
508
	}
Nick Piggin's avatar
Nick Piggin committed
509
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
510 511
}

Nick Piggin's avatar
Nick Piggin committed
512
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
513
{
514
	spin_lock(&zone->lock);
515
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
516
	zone->pages_scanned = 0;
517
	__free_one_page(page, zone, order);
518
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
519 520 521 522 523
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
524
	int i;
525
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
526 527

	for (i = 0 ; i < (1 << order) ; ++i)
528
		reserved += free_pages_check(page + i);
529 530 531
	if (reserved)
		return;

532
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
533
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
534 535 536
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
537
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
538
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
539

Nick Piggin's avatar
Nick Piggin committed
540
	local_irq_save(flags);
541
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
542
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
543
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
544 545
}

546 547 548
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
549
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
550 551 552 553
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
554
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
555
		__free_page(page);
556 557 558
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
559
		prefetchw(page);
560 561 562
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
563 564
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
565 566 567 568
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

569
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
570
		__free_pages(page, order);
571 572 573
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
574 575 576 577 578 579 580 581 582 583 584 585 586 587 588

/*
 * 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
589
static inline void expand(struct zone *zone, struct page *page,
590 591
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
592 593 594 595 596 597 598
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
599
		VM_BUG_ON(bad_range(zone, &page[size]));
600
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
601 602 603 604 605 606 607 608
		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
609
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
610
{
Nick Piggin's avatar
Nick Piggin committed
611 612 613
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
614
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP)))
615
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
616

617 618 619 620 621 622 623
	/*
	 * 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;

624
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error | 1 << PG_reclaim |
Linus Torvalds's avatar
Linus Torvalds committed
625
			1 << PG_referenced | 1 << PG_arch_1 |
626 627 628 629 630
			1 << PG_owner_priv_1 | 1 << PG_mappedtodisk
#ifdef CONFIG_UNEVICTABLE_LRU
			| 1 << PG_mlocked
#endif
			);
631
	set_page_private(page, 0);
632
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
633 634

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
635
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
636 637 638 639 640 641 642

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

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

643
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
644 645
}

646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
static struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
						int migratetype)
{
	unsigned int current_order;
	struct free_area * area;
	struct page *page;

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

		page = list_entry(area->free_list[migratetype].next,
							struct page, lru);
		list_del(&page->lru);
		rmv_page_order(page);
		area->nr_free--;
		__mod_zone_page_state(zone, NR_FREE_PAGES, - (1UL << order));
		expand(zone, page, order, current_order, area, migratetype);
		return page;
	}

	return NULL;
}


677 678 679 680 681
/*
 * This array describes the order lists are fallen back to when
 * the free lists for the desirable migrate type are depleted
 */
static int fallbacks[MIGRATE_TYPES][MIGRATE_TYPES-1] = {
682 683 684 685
	[MIGRATE_UNMOVABLE]   = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE,   MIGRATE_RESERVE },
	[MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE,   MIGRATE_MOVABLE,   MIGRATE_RESERVE },
	[MIGRATE_MOVABLE]     = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
	[MIGRATE_RESERVE]     = { MIGRATE_RESERVE,     MIGRATE_RESERVE,   MIGRATE_RESERVE }, /* Never used */
686 687
};

688 689
/*
 * Move the free pages in a range to the free lists of the requested type.
690
 * Note that start_page and end_pages are not aligned on a pageblock
691 692
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
693 694 695
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
696 697 698
{
	struct page *page;
	unsigned long order;
699
	int pages_moved = 0;
700 701 702 703 704 705 706

#ifndef CONFIG_HOLES_IN_ZONE
	/*
	 * page_zone is not safe to call in this context when
	 * CONFIG_HOLES_IN_ZONE is set. This bug check is probably redundant
	 * anyway as we check zone boundaries in move_freepages_block().
	 * Remove at a later date when no bug reports exist related to
Mel Gorman's avatar
Mel Gorman committed
707
	 * grouping pages by mobility
708 709 710 711 712
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
713 714 715
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

716 717 718 719 720 721 722 723 724 725 726 727 728 729 730
		if (!pfn_valid_within(page_to_pfn(page))) {
			page++;
			continue;
		}

		if (!PageBuddy(page)) {
			page++;
			continue;
		}

		order = page_order(page);
		list_del(&page->lru);
		list_add(&page->lru,
			&zone->free_area[order].free_list[migratetype]);
		page += 1 << order;
731
		pages_moved += 1 << order;
732 733
	}

734
	return pages_moved;
735 736
}

Adrian Bunk's avatar
Adrian Bunk committed
737 738
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
739 740 741 742 743
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
744
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
745
	start_page = pfn_to_page(start_pfn);
746 747
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
748 749 750 751 752 753 754 755 756 757

	/* Do not cross zone boundaries */
	if (start_pfn < zone->zone_start_pfn)
		start_page = page;
	if (end_pfn >= zone->zone_start_pfn + zone->spanned_pages)
		return 0;

	return move_freepages(zone, start_page, end_page, migratetype);
}

758 759 760 761 762 763 764 765 766 767 768 769 770 771 772
/* Remove an element from the buddy allocator from the fallback list */
static struct page *__rmqueue_fallback(struct zone *zone, int order,
						int start_migratetype)
{
	struct free_area * area;
	int current_order;
	struct page *page;
	int migratetype, i;

	/* Find the largest possible block of pages in the other list */
	for (current_order = MAX_ORDER-1; current_order >= order;
						--current_order) {
		for (i = 0; i < MIGRATE_TYPES - 1; i++) {
			migratetype = fallbacks[start_migratetype][i];

773 774 775
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
776

777 778 779 780 781 782 783 784 785
			area = &(zone->free_area[current_order]);
			if (list_empty(&area->free_list[migratetype]))
				continue;

			page = list_entry(area->free_list[migratetype].next,
					struct page, lru);
			area->nr_free--;

			/*
786
			 * If breaking a large block of pages, move all free
787 788 789
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
790
			 */
791
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
792 793 794 795 796 797
					start_migratetype == MIGRATE_RECLAIMABLE) {
				unsigned long pages;
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
798
				if (pages >= (1 << (pageblock_order-1)))
799 800 801
					set_pageblock_migratetype(page,
								start_migratetype);

802
				migratetype = start_migratetype;
803
			}
804 805 806 807 808 809 810

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

811
			if (current_order == pageblock_order)
812 813 814 815 816 817 818 819
				set_pageblock_migratetype(page,
							start_migratetype);

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

820 821
	/* Use MIGRATE_RESERVE rather than fail an allocation */
	return __rmqueue_smallest(zone, order, MIGRATE_RESERVE);
822 823
}

824
/*
Linus Torvalds's avatar
Linus Torvalds committed
825 826 827
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
828 829
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
830 831 832
{
	struct page *page;

833
	page = __rmqueue_smallest(zone, order, migratetype);
834

835 836
	if (unlikely(!page))
		page = __rmqueue_fallback(zone, order, migratetype);
837 838

	return page;
Linus Torvalds's avatar
Linus Torvalds committed
839 840 841 842 843 844 845 846
}

/* 
 * 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, 
847 848
			unsigned long count, struct list_head *list,
			int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
849 850 851
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
852
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
853
	for (i = 0; i < count; ++i) {
854
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
855
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
856
			break;
857 858 859 860 861 862 863 864 865 866

		/*
		 * Split buddy pages returned by expand() are received here
		 * in physical page order. The page is added to the callers and
		 * list and the list head then moves forward. From the callers
		 * perspective, the linked list is ordered by page number in
		 * some conditions. This is useful for IO devices that can
		 * merge IO requests if the physical pages are ordered
		 * properly.
		 */
867 868
		list_add(&page->lru, list);
		set_page_private(page, migratetype);
869
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
870
	}
Nick Piggin's avatar
Nick Piggin committed
871
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
872
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
873 874
}

875
#ifdef CONFIG_NUMA
876
/*
877 878 879 880
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
881 882
 * Note that this function must be called with the thread pinned to
 * a single processor.
883
 */
884
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
885 886
{
	unsigned long flags;
887
	int to_drain;
888

889 890 891 892 893 894 895 896
	local_irq_save(flags);
	if (pcp->count >= pcp->batch)
		to_drain = pcp->batch;
	else
		to_drain = pcp->count;
	free_pages_bulk(zone, to_drain, &pcp->list, 0);
	pcp->count -= to_drain;
	local_irq_restore(flags);
897 898 899
}
#endif

900 901 902 903 904 905 906 907
/*
 * Drain pages of the indicated processor.
 *
 * The processor must either be the current processor and the
 * thread pinned to the current processor or a processor that
 * is not online.
 */
static void drain_pages(unsigned int cpu)
Linus Torvalds's avatar
Linus Torvalds committed
908
{
Nick Piggin's avatar
Nick Piggin committed
909
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
910 911 912 913
	struct zone *zone;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;
914
		struct per_cpu_pages *pcp;
Linus Torvalds's avatar
Linus Torvalds committed
915

916 917 918
		if (!populated_zone(zone))
			continue;

919
		pset = zone_pcp(zone, cpu);
920 921 922 923 924 925

		pcp = &pset->pcp;
		local_irq_save(flags);
		free_pages_bulk(zone, pcp->count, &pcp->list, 0);
		pcp->count = 0;
		local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
926 927 928
	}
}

929 930 931 932 933 934 935 936 937 938 939 940 941
/*
 * Spill all of this CPU's per-cpu pages back into the buddy allocator.
 */
void drain_local_pages(void *arg)
{
	drain_pages(smp_processor_id());
}

/*
 * Spill all the per-cpu pages from all CPUs back into the buddy allocator
 */
void drain_all_pages(void)
{
942
	on_each_cpu(drain_local_pages, NULL, 1);
943 944
}

945
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
946 947 948

void mark_free_pages(struct zone *zone)
{
949 950
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
951
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
952 953 954 955 956 957
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
958 959 960 961 962 963

	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);

964 965
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
966
		}
Linus Torvalds's avatar
Linus Torvalds committed
967

968 969
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
970
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
971

972 973
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
974
				swsusp_set_page_free(pfn_to_page(pfn + i));
975
		}
976
	}
Linus Torvalds's avatar
Linus Torvalds committed
977 978
	spin_unlock_irqrestore(&zone->lock, flags);
}
979
#endif /* CONFIG_PM */
Linus Torvalds's avatar
Linus Torvalds committed
980 981 982 983

/*
 * Free a 0-order page
 */
Harvey Harrison's avatar
Harvey Harrison committed
984
static void free_hot_cold_page(struct page *page, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
985 986 987 988 989 990 991
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;

	if (PageAnon(page))
		page->mapping = NULL;