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;
Linus Torvalds's avatar
Linus Torvalds committed
72
long nr_swap_pages;
73
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
74

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

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

Linus Torvalds's avatar
Linus Torvalds committed
81 82 83 84 85 86 87
/*
 * results with 256, 32 in the lowmem_reserve sysctl:
 *	1G machine -> (16M dma, 800M-16M normal, 1G-800M high)
 *	1G machine -> (16M dma, 784M normal, 224M high)
 *	NORMAL allocation will leave 784M/256 of ram reserved in the ZONE_DMA
 *	HIGHMEM allocation will leave 224M/32 of ram reserved in ZONE_NORMAL
 *	HIGHMEM allocation will (224M+784M)/256 of ram reserved in ZONE_DMA
88 89 90
 *
 * TBD: should special case ZONE_DMA32 machines here - in those we normally
 * don't need any ZONE_NORMAL reservation
Linus Torvalds's avatar
Linus Torvalds committed
91
 */
92
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
93
#ifdef CONFIG_ZONE_DMA
94
	 256,
95
#endif
96
#ifdef CONFIG_ZONE_DMA32
97
	 256,
98
#endif
99
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
100
	 32,
101
#endif
Mel Gorman's avatar
Mel Gorman committed
102
	 32,
103
};
Linus Torvalds's avatar
Linus Torvalds committed
104 105 106

EXPORT_SYMBOL(totalram_pages);

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

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

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

127 128
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
129
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147
   * ranges of memory (RAM) that may be registered with add_active_range().
   * Ranges passed to add_active_range() will be merged if possible
   * so the number of times add_active_range() can be called is
   * related to the number of nodes and the number of holes
   */
  #ifdef CONFIG_MAX_ACTIVE_REGIONS
    /* Allow an architecture to set MAX_ACTIVE_REGIONS to save memory */
    #define MAX_ACTIVE_REGIONS CONFIG_MAX_ACTIVE_REGIONS
  #else
    #if MAX_NUMNODES >= 32
      /* If there can be many nodes, allow up to 50 holes per node */
      #define MAX_ACTIVE_REGIONS (MAX_NUMNODES*50)
    #else
      /* By default, allow up to 256 distinct regions */
      #define MAX_ACTIVE_REGIONS 256
    #endif
  #endif

148 149 150 151
  static struct node_active_region __meminitdata early_node_map[MAX_ACTIVE_REGIONS];
  static int __meminitdata nr_nodemap_entries;
  static unsigned long __meminitdata arch_zone_lowest_possible_pfn[MAX_NR_ZONES];
  static unsigned long __meminitdata arch_zone_highest_possible_pfn[MAX_NR_ZONES];
152
#ifdef CONFIG_MEMORY_HOTPLUG_RESERVE
153 154
  static unsigned long __meminitdata node_boundary_start_pfn[MAX_NUMNODES];
  static unsigned long __meminitdata node_boundary_end_pfn[MAX_NUMNODES];
155
#endif /* CONFIG_MEMORY_HOTPLUG_RESERVE */
Adrian Bunk's avatar
Adrian Bunk committed
156
  static unsigned long __initdata required_kernelcore;
157
  static unsigned long __initdata required_movablecore;
Adrian Bunk's avatar
Adrian Bunk committed
158
  static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
159 160 161 162

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

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

170 171
int page_group_by_mobility_disabled __read_mostly;

172 173 174 175 176 177
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

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

185 186 187 188 189 190 191 192 193
	do {
		seq = zone_span_seqbegin(zone);
		if (pfn >= zone->zone_start_pfn + zone->spanned_pages)
			ret = 1;
		else if (pfn < zone->zone_start_pfn)
			ret = 1;
	} while (zone_span_seqretry(zone, seq));

	return ret;
194 195 196 197
}

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

	return 1;
}
/*
 * Temporary debugging check for pages not lying within a given zone.
 */
static int bad_range(struct zone *zone, struct page *page)
{
	if (page_outside_zone_boundaries(zone, page))
Linus Torvalds's avatar
Linus Torvalds committed
211
		return 1;
212 213 214
	if (!page_is_consistent(zone, page))
		return 1;

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

224
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
225
{
226 227
	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",
228 229 230
		current->comm, page, (int)(2*sizeof(unsigned long)),
		(unsigned long)page->flags, page->mapping,
		page_mapcount(page), page_count(page));
231

232 233
	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
234
	dump_stack();
235
	page->flags &= ~PAGE_FLAGS_CLEAR_WHEN_BAD;
Linus Torvalds's avatar
Linus Torvalds committed
236 237 238
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
239
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
240 241 242 243 244 245 246 247 248 249 250 251
}

/*
 * 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).
 *
252 253 254
 * 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
255
 */
256 257 258

static void free_compound_page(struct page *page)
{
259
	__free_pages_ok(page, compound_order(page));
260 261
}

262
void prep_compound_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
263 264 265
{
	int i;
	int nr_pages = 1 << order;
266
	struct page *p = page + 1;
Linus Torvalds's avatar
Linus Torvalds committed
267

268
	set_compound_page_dtor(page, free_compound_page);
269
	set_compound_order(page, order);
270
	__SetPageHead(page);
271 272 273
	for (i = 1; i < nr_pages; i++, p++) {
		if (unlikely((i & (MAX_ORDER_NR_PAGES - 1)) == 0))
			p = pfn_to_page(page_to_pfn(page) + i);
274 275
		__SetPageTail(p);
		p->first_page = page;
Linus Torvalds's avatar
Linus Torvalds committed
276 277 278 279 280 281 282
	}
}

static void destroy_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;
283
	struct page *p = page + 1;
Linus Torvalds's avatar
Linus Torvalds committed
284

285
	if (unlikely(compound_order(page) != order))
286
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
287

288
	if (unlikely(!PageHead(page)))
289
			bad_page(page);
290
	__ClearPageHead(page);
291 292 293
	for (i = 1; i < nr_pages; i++, p++) {
		if (unlikely((i & (MAX_ORDER_NR_PAGES - 1)) == 0))
			p = pfn_to_page(page_to_pfn(page) + i);
Linus Torvalds's avatar
Linus Torvalds committed
294

295
		if (unlikely(!PageTail(p) |
296
				(p->first_page != page)))
297
			bad_page(page);
298
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
299 300 301
	}
}

Nick Piggin's avatar
Nick Piggin committed
302 303 304 305
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

306 307 308 309
	/*
	 * 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
310
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
311 312 313 314
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

315 316
static inline void set_page_order(struct page *page, int order)
{
317
	set_page_private(page, order);
318
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
319 320 321 322
}

static inline void rmv_page_order(struct page *page)
{
323
	__ClearPageBuddy(page);
324
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341
}

/*
 * 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
342
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360
 */
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
361
 * (a) the buddy is not in a hole &&
362
 * (b) the buddy is in the buddy system &&
363 364
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
365 366 367
 *
 * 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
368
 *
369
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
370
 */
371 372
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
373
{
374
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
375 376
		return 0;

377 378 379 380 381
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
382
		return 1;
383
	}
384
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
}

/*
 * 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
400
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
401
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
402 403 404 405 406 407 408 409 410
 * 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
411
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
412 413 414 415
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;
416
	int migratetype = get_pageblock_migratetype(page);
Linus Torvalds's avatar
Linus Torvalds committed
417

418
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
419 420 421 422
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
423 424
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
425

426
	__mod_zone_page_state(zone, NR_FREE_PAGES, order_size);
Linus Torvalds's avatar
Linus Torvalds committed
427 428 429 430 431
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
432
		if (!page_is_buddy(page, buddy, order))
433
			break;
Nick Piggin's avatar
Nick Piggin committed
434

435
		/* Our buddy is free, merge with it and move up one order. */
Linus Torvalds's avatar
Linus Torvalds committed
436
		list_del(&buddy->lru);
437
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
438
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
439
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
440 441 442 443 444
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
445 446
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
447 448 449
	zone->free_area[order].nr_free++;
}

450
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
451
{
452
	free_page_mlock(page);
Nick Piggin's avatar
Nick Piggin committed
453 454 455
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
456
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE)))
457
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
458
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
459
		__ClearPageDirty(page);
460 461
	if (PageSwapBacked(page))
		__ClearPageSwapBacked(page);
462 463 464 465 466 467
	/*
	 * 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
468 469 470 471 472
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
473
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
474 475 476 477 478 479 480
 *
 * 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
481 482
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
483
{
Nick Piggin's avatar
Nick Piggin committed
484
	spin_lock(&zone->lock);
485
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
Linus Torvalds's avatar
Linus Torvalds committed
486
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
487 488 489
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
490
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
491
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
492
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
493
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
494
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
495
	}
Nick Piggin's avatar
Nick Piggin committed
496
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
497 498
}

Nick Piggin's avatar
Nick Piggin committed
499
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
500
{
501
	spin_lock(&zone->lock);
502
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
503
	zone->pages_scanned = 0;
504
	__free_one_page(page, zone, order);
505
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
506 507 508 509 510
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
511
	int i;
512
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
513 514

	for (i = 0 ; i < (1 << order) ; ++i)
515
		reserved += free_pages_check(page + i);
516 517 518
	if (reserved)
		return;

519
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
520
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
521 522 523
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
524
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
525
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
526

Nick Piggin's avatar
Nick Piggin committed
527
	local_irq_save(flags);
528
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
529
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
530
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
531 532
}

533 534 535
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
536
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
537 538 539 540
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
541
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
542
		__free_page(page);
543 544 545
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
546
		prefetchw(page);
547 548 549
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
550 551
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
552 553 554 555
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

556
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
557
		__free_pages(page, order);
558 559 560
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
561 562 563 564 565 566 567 568 569 570 571 572 573 574 575

/*
 * 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
576
static inline void expand(struct zone *zone, struct page *page,
577 578
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
579 580 581 582 583 584 585
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
586
		VM_BUG_ON(bad_range(zone, &page[size]));
587
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
588 589 590 591 592 593 594 595
		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
596
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
597
{
Nick Piggin's avatar
Nick Piggin committed
598 599 600
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
601
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP)))
602
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
603

604 605 606 607 608 609 610
	/*
	 * 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;

611
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error | 1 << PG_reclaim |
Linus Torvalds's avatar
Linus Torvalds committed
612
			1 << PG_referenced | 1 << PG_arch_1 |
613 614 615 616 617
			1 << PG_owner_priv_1 | 1 << PG_mappedtodisk
#ifdef CONFIG_UNEVICTABLE_LRU
			| 1 << PG_mlocked
#endif
			);
618
	set_page_private(page, 0);
619
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
620 621

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
622
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
623 624 625 626 627 628 629

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

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

630
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
631 632
}

633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663
/*
 * 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;
}


664 665 666 667 668
/*
 * 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] = {
669 670 671 672
	[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 */
673 674
};

675 676
/*
 * Move the free pages in a range to the free lists of the requested type.
677
 * Note that start_page and end_pages are not aligned on a pageblock
678 679
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
680 681 682
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
683 684 685
{
	struct page *page;
	unsigned long order;
686
	int pages_moved = 0;
687 688 689 690 691 692 693

#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
694
	 * grouping pages by mobility
695 696 697 698 699
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
700 701 702
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

703 704 705 706 707 708 709 710 711 712 713 714 715 716 717
		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;
718
		pages_moved += 1 << order;
719 720
	}

721
	return pages_moved;
722 723
}

Adrian Bunk's avatar
Adrian Bunk committed
724 725
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
726 727 728 729 730
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
731
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
732
	start_page = pfn_to_page(start_pfn);
733 734
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
735 736 737 738 739 740 741 742 743 744

	/* 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);
}

745 746 747 748 749 750 751 752 753 754 755 756 757 758 759
/* 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];

760 761 762
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
763

764 765 766 767 768 769 770 771 772
			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--;

			/*
773
			 * If breaking a large block of pages, move all free
774 775 776
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
777
			 */
778
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
779 780 781 782 783 784
					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 */
785
				if (pages >= (1 << (pageblock_order-1)))
786 787 788
					set_pageblock_migratetype(page,
								start_migratetype);

789
				migratetype = start_migratetype;
790
			}
791 792 793 794 795 796 797

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

798
			if (current_order == pageblock_order)
799 800 801 802 803 804 805 806
				set_pageblock_migratetype(page,
							start_migratetype);

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

807 808
	/* Use MIGRATE_RESERVE rather than fail an allocation */
	return __rmqueue_smallest(zone, order, MIGRATE_RESERVE);
809 810
}

811
/*
Linus Torvalds's avatar
Linus Torvalds committed
812 813 814
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
815 816
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
817 818 819
{
	struct page *page;

820
	page = __rmqueue_smallest(zone, order, migratetype);
821

822 823
	if (unlikely(!page))
		page = __rmqueue_fallback(zone, order, migratetype);
824 825

	return page;
Linus Torvalds's avatar
Linus Torvalds committed
826 827 828 829 830 831 832 833
}

/* 
 * 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, 
834 835
			unsigned long count, struct list_head *list,
			int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
836 837 838
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
839
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
840
	for (i = 0; i < count; ++i) {
841
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
842
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
843
			break;
844 845 846 847 848 849 850 851 852 853

		/*
		 * 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.
		 */
854 855
		list_add(&page->lru, list);
		set_page_private(page, migratetype);
856
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
857
	}
Nick Piggin's avatar
Nick Piggin committed
858
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
859
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
860 861
}

862
#ifdef CONFIG_NUMA
863
/*
864 865 866 867
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
868 869
 * Note that this function must be called with the thread pinned to
 * a single processor.
870
 */
871
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
872 873
{
	unsigned long flags;
874
	int to_drain;
875

876 877 878 879 880 881 882 883
	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);
884 885 886
}
#endif

887 888 889 890 891 892 893 894
/*
 * 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
895
{
Nick Piggin's avatar
Nick Piggin committed
896
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
897 898 899 900
	struct zone *zone;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;
901
		struct per_cpu_pages *pcp;
Linus Torvalds's avatar
Linus Torvalds committed
902

903 904 905
		if (!populated_zone(zone))
			continue;

906
		pset = zone_pcp(zone, cpu);
907 908 909 910 911 912

		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
913 914 915
	}
}

916 917 918 919 920 921 922 923 924 925 926 927 928
/*
 * 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)
{
929
	on_each_cpu(drain_local_pages, NULL, 1);
930 931
}

932
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
933 934 935

void mark_free_pages(struct zone *zone)
{
936 937
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
938
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
939 940 941 942 943 944
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
945 946 947 948 949 950

	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);

951 952
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
953
		}
Linus Torvalds's avatar
Linus Torvalds committed
954

955 956
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
957
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
958

959 960
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
961
				swsusp_set_page_free(pfn_to_page(pfn + i));
962
		}
963
	}
Linus Torvalds's avatar
Linus Torvalds committed
964 965
	spin_unlock_irqrestore(&zone->lock, flags);
}
966
#endif /* CONFIG_PM */
Linus Torvalds's avatar
Linus Torvalds committed
967 968 969 970

/*
 * Free a 0-order page
 */
Harvey Harrison's avatar
Harvey Harrison committed
971
static void free_hot_cold_page(struct page *page, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
972 973 974 975 976 977 978
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;

	if (PageAnon(page))
		page->mapping = NULL;
979
	if (free_pages_check(page))
980 981
		return;

982
	if (!PageHighMem(page)) {