page_alloc.c 125 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/memcontrol.h>
Linus Torvalds's avatar
Linus Torvalds committed
48 49

#include <asm/tlbflush.h>
50
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
51 52 53
#include "internal.h"

/*
54
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
55
 */
56 57 58 59 60 61 62 63 64 65 66 67 68
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);

69
unsigned long totalram_pages __read_mostly;
70
unsigned long totalreserve_pages __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
71
long nr_swap_pages;
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 */
Mel Gorman's avatar
Mel Gorman committed
155
  unsigned long __initdata required_kernelcore;
156
  static unsigned long __initdata required_movablecore;
157
  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
{
225 226 227 228
	void *pc = page_get_page_cgroup(page);

	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",
229 230 231
		current->comm, page, (int)(2*sizeof(unsigned long)),
		(unsigned long)page->flags, page->mapping,
		page_mapcount(page), page_count(page));
232 233 234 235 236 237
	if (pc) {
		printk(KERN_EMERG "cgroup:%p\n", pc);
		page_reset_bad_cgroup(page);
	}
	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
238
	dump_stack();
239 240
	page->flags &= ~(1 << PG_lru	|
			1 << PG_private |
Linus Torvalds's avatar
Linus Torvalds committed
241 242 243
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
244 245
			1 << PG_reclaim |
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
246
			1 << PG_swapcache |
247 248
			1 << PG_writeback |
			1 << PG_buddy );
Linus Torvalds's avatar
Linus Torvalds committed
249 250 251
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
252
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
253 254 255 256 257 258 259 260 261 262 263 264
}

/*
 * 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).
 *
265 266 267
 * 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
268
 */
269 270 271

static void free_compound_page(struct page *page)
{
272
	__free_pages_ok(page, compound_order(page));
273 274
}

Linus Torvalds's avatar
Linus Torvalds committed
275 276 277 278 279
static void prep_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;

280
	set_compound_page_dtor(page, free_compound_page);
281
	set_compound_order(page, order);
282
	__SetPageHead(page);
283
	for (i = 1; i < nr_pages; i++) {
Linus Torvalds's avatar
Linus Torvalds committed
284 285
		struct page *p = page + i;

286 287
		__SetPageTail(p);
		p->first_page = page;
Linus Torvalds's avatar
Linus Torvalds committed
288 289 290 291 292 293 294 295
	}
}

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

296
	if (unlikely(compound_order(page) != order))
297
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
298

299
	if (unlikely(!PageHead(page)))
300
			bad_page(page);
301
	__ClearPageHead(page);
302
	for (i = 1; i < nr_pages; i++) {
Linus Torvalds's avatar
Linus Torvalds committed
303 304
		struct page *p = page + i;

305
		if (unlikely(!PageTail(p) |
306
				(p->first_page != page)))
307
			bad_page(page);
308
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
309 310 311
	}
}

Nick Piggin's avatar
Nick Piggin committed
312 313 314 315
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

316 317 318 319
	/*
	 * 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
320
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
321 322 323 324
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

325 326
static inline void set_page_order(struct page *page, int order)
{
327
	set_page_private(page, order);
328
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
329 330 331 332
}

static inline void rmv_page_order(struct page *page)
{
333
	__ClearPageBuddy(page);
334
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
}

/*
 * 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
352
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370
 */
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
371
 * (a) the buddy is not in a hole &&
372
 * (b) the buddy is in the buddy system &&
373 374
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
375 376 377
 *
 * 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
378
 *
379
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
380
 */
381 382
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
383
{
384
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
385 386
		return 0;

387 388 389 390 391
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
392
		return 1;
393
	}
394
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
}

/*
 * 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
410
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
411
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
412 413 414 415 416 417 418 419 420
 * 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
421
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
422 423 424 425
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;
426
	int migratetype = get_pageblock_migratetype(page);
Linus Torvalds's avatar
Linus Torvalds committed
427

428
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
429 430 431 432
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
433 434
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
435

436
	__mod_zone_page_state(zone, NR_FREE_PAGES, order_size);
Linus Torvalds's avatar
Linus Torvalds committed
437 438 439 440 441
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
442
		if (!page_is_buddy(page, buddy, order))
Linus Torvalds's avatar
Linus Torvalds committed
443
			break;		/* Move the buddy up one level. */
Nick Piggin's avatar
Nick Piggin committed
444

Linus Torvalds's avatar
Linus Torvalds committed
445
		list_del(&buddy->lru);
446
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
447
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
448
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
449 450 451 452 453
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
454 455
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
456 457 458
	zone->free_area[order].nr_free++;
}

459
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
460
{
Nick Piggin's avatar
Nick Piggin committed
461 462
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
463
		(page_get_page_cgroup(page) != NULL) |
Nick Piggin's avatar
Nick Piggin committed
464
		(page_count(page) != 0)  |
Linus Torvalds's avatar
Linus Torvalds committed
465 466 467 468 469 470 471
		(page->flags & (
			1 << PG_lru	|
			1 << PG_private |
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_slab	|
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
472
			1 << PG_writeback |
473 474
			1 << PG_reserved |
			1 << PG_buddy ))))
475
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
476
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
477
		__ClearPageDirty(page);
478 479 480 481 482 483
	/*
	 * 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
484 485 486 487 488
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
489
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
490 491 492 493 494 495 496
 *
 * 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
497 498
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
499
{
Nick Piggin's avatar
Nick Piggin committed
500
	spin_lock(&zone->lock);
501
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
Linus Torvalds's avatar
Linus Torvalds committed
502
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
503 504 505
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
506
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
507
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
508
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
509
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
510
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
511
	}
Nick Piggin's avatar
Nick Piggin committed
512
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
513 514
}

Nick Piggin's avatar
Nick Piggin committed
515
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
516
{
517
	spin_lock(&zone->lock);
518
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
519
	zone->pages_scanned = 0;
520
	__free_one_page(page, zone, order);
521
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
522 523 524 525 526
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
527
	int i;
528
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
529 530

	for (i = 0 ; i < (1 << order) ; ++i)
531
		reserved += free_pages_check(page + i);
532 533 534
	if (reserved)
		return;

Nick Piggin's avatar
Nick Piggin committed
535 536
	if (!PageHighMem(page))
		debug_check_no_locks_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 __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
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
613
		(page_get_page_cgroup(page) != NULL) |
Nick Piggin's avatar
Nick Piggin committed
614
		(page_count(page) != 0)  |
615 616
		(page->flags & (
			1 << PG_lru	|
Linus Torvalds's avatar
Linus Torvalds committed
617 618 619 620
			1 << PG_private	|
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
621
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
622
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
623
			1 << PG_writeback |
624 625
			1 << PG_reserved |
			1 << PG_buddy ))))
626
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
627

628 629 630 631 632 633 634
	/*
	 * 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;

635
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error | 1 << PG_reclaim |
Linus Torvalds's avatar
Linus Torvalds committed
636
			1 << PG_referenced | 1 << PG_arch_1 |
637
			1 << PG_owner_priv_1 | 1 << PG_mappedtodisk);
638
	set_page_private(page, 0);
639
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
640 641

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
642
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
643 644 645 646 647 648 649

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

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

650
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
651 652
}

653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683
/*
 * 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;
}


684 685 686 687 688
/*
 * 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] = {
689 690 691 692
	[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 */
693 694
};

695 696
/*
 * Move the free pages in a range to the free lists of the requested type.
697
 * Note that start_page and end_pages are not aligned on a pageblock
698 699 700 701 702 703 704 705
 * boundary. If alignment is required, use move_freepages_block()
 */
int move_freepages(struct zone *zone,
			struct page *start_page, struct page *end_page,
			int migratetype)
{
	struct page *page;
	unsigned long order;
706
	int pages_moved = 0;
707 708 709 710 711 712 713

#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
714
	 * grouping pages by mobility
715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
		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;
735
		pages_moved += 1 << order;
736 737
	}

738
	return pages_moved;
739 740 741 742 743 744 745 746
}

int move_freepages_block(struct zone *zone, struct page *page, int migratetype)
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
747
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
748
	start_page = pfn_to_page(start_pfn);
749 750
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
751 752 753 754 755 756 757 758 759 760

	/* 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);
}

761 762 763 764 765 766 767 768 769 770 771 772 773 774 775
/* 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];

776 777 778
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
779

780 781 782 783 784 785 786 787 788
			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--;

			/*
789
			 * If breaking a large block of pages, move all free
790 791 792
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
793
			 */
794
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
795 796 797 798 799 800
					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 */
801
				if (pages >= (1 << (pageblock_order-1)))
802 803 804
					set_pageblock_migratetype(page,
								start_migratetype);

805
				migratetype = start_migratetype;
806
			}
807 808 809 810 811 812 813

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

814
			if (current_order == pageblock_order)
815 816 817 818 819 820 821 822
				set_pageblock_migratetype(page,
							start_migratetype);

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

823 824
	/* Use MIGRATE_RESERVE rather than fail an allocation */
	return __rmqueue_smallest(zone, order, MIGRATE_RESERVE);
825 826
}

827
/*
Linus Torvalds's avatar
Linus Torvalds committed
828 829 830
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
831 832
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
833 834 835
{
	struct page *page;

836
	page = __rmqueue_smallest(zone, order, migratetype);
837

838 839
	if (unlikely(!page))
		page = __rmqueue_fallback(zone, order, migratetype);
840 841

	return page;
Linus Torvalds's avatar
Linus Torvalds committed
842 843 844 845 846 847 848 849
}

/* 
 * 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, 
850 851
			unsigned long count, struct list_head *list,
			int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
852 853 854
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
855
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
856
	for (i = 0; i < count; ++i) {
857
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
858
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
859
			break;
860 861 862 863 864 865 866 867 868 869

		/*
		 * 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.
		 */
870 871
		list_add(&page->lru, list);
		set_page_private(page, migratetype);
872
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
873
	}
Nick Piggin's avatar
Nick Piggin committed
874
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
875
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
876 877
}

878
#ifdef CONFIG_NUMA
879
/*
880 881 882 883
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
884 885
 * Note that this function must be called with the thread pinned to
 * a single processor.
886
 */
887
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
888 889
{
	unsigned long flags;
890
	int to_drain;
891

892 893 894 895 896 897 898 899
	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);
900 901 902
}
#endif

903 904 905 906 907 908 909 910
/*
 * 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
911
{
Nick Piggin's avatar
Nick Piggin committed
912
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
913 914 915 916
	struct zone *zone;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;
917
		struct per_cpu_pages *pcp;
Linus Torvalds's avatar
Linus Torvalds committed
918

919 920 921
		if (!populated_zone(zone))
			continue;

922
		pset = zone_pcp(zone, cpu);
923 924 925 926 927 928

		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
929 930 931
	}
}

932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947
/*
 * 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)
{
	on_each_cpu(drain_local_pages, NULL, 0, 1);
}

948
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
949 950 951

void mark_free_pages(struct zone *zone)
{
952 953
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
954
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
955 956 957 958 959 960
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
961 962 963 964 965 966

	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);

967 968
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
969
		}
Linus Torvalds's avatar
Linus Torvalds committed
970

971 972
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
973
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
974

975 976
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
977
				swsusp_set_page_free(pfn_to_page(pfn + i));
978
		}
979
	}
Linus Torvalds's avatar
Linus Torvalds committed
980 981
	spin_unlock_irqrestore(&zone->lock, flags);
}
982
#endif /* CONFIG_PM */
Linus Torvalds's avatar
Linus Torvalds committed
983 984 985 986

/*
 * Free a 0-order page
 */
Harvey Harrison's avatar
Harvey Harrison committed
987
static void free_hot_cold_page(struct page *page, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
988 989 990 991 992 993 994
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;

	if (PageAnon(page))
		page->mapping = NULL;
995
	if (free_pages_check(page))
996 997
		return;

Nick Piggin's avatar
Nick Piggin committed
998 999
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
Nick Piggin's avatar
Nick Piggin committed