page_alloc.c 129 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>
49
#include <linux/kmemleak.h>
Linus Torvalds's avatar
Linus Torvalds committed
50 51

#include <asm/tlbflush.h>
52
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
53 54 55
#include "internal.h"

/*
56
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
57
 */
58 59 60 61 62 63 64 65 66 67 68 69 70
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);

71
unsigned long totalram_pages __read_mostly;
72
unsigned long totalreserve_pages __read_mostly;
73
unsigned long highest_memmap_pfn __read_mostly;
74
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
75

76 77 78 79
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

80
static void __free_pages_ok(struct page *page, unsigned int order);
81

Linus Torvalds's avatar
Linus Torvalds committed
82 83 84 85 86 87 88
/*
 * 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
89 90 91
 *
 * 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
92
 */
93
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
94
#ifdef CONFIG_ZONE_DMA
95
	 256,
96
#endif
97
#ifdef CONFIG_ZONE_DMA32
98
	 256,
99
#endif
100
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
101
	 32,
102
#endif
Mel Gorman's avatar
Mel Gorman committed
103
	 32,
104
};
Linus Torvalds's avatar
Linus Torvalds committed
105 106 107

EXPORT_SYMBOL(totalram_pages);

108
static char * const zone_names[MAX_NR_ZONES] = {
109
#ifdef CONFIG_ZONE_DMA
110
	 "DMA",
111
#endif
112
#ifdef CONFIG_ZONE_DMA32
113
	 "DMA32",
114
#endif
115
	 "Normal",
116
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
117
	 "HighMem",
118
#endif
Mel Gorman's avatar
Mel Gorman committed
119
	 "Movable",
120 121
};

Linus Torvalds's avatar
Linus Torvalds committed
122 123
int min_free_kbytes = 1024;

124 125
unsigned long __meminitdata nr_kernel_pages;
unsigned long __meminitdata nr_all_pages;
126
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
127

128 129
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
130
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
   * 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

149 150 151 152
  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];
Adrian Bunk's avatar
Adrian Bunk committed
153
  static unsigned long __initdata required_kernelcore;
154
  static unsigned long __initdata required_movablecore;
Adrian Bunk's avatar
Adrian Bunk committed
155
  static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
156 157 158 159

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
160 161
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
162 163 164 165 166
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
EXPORT_SYMBOL(nr_node_ids);
#endif

167 168
int page_group_by_mobility_disabled __read_mostly;

169 170 171 172 173 174
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
175
#ifdef CONFIG_DEBUG_VM
176
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
177
{
178 179 180
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
181

182 183 184 185 186 187 188 189 190
	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;
191 192 193 194
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
195
	if (!pfn_valid_within(page_to_pfn(page)))
196
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
197
	if (zone != page_zone(page))
198 199 200 201 202 203 204 205 206 207
		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
208
		return 1;
209 210 211
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
212 213
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
214 215 216 217 218 219 220
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

221
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
222
{
223 224 225 226 227 228 229 230 231 232 233 234 235 236
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

	/*
	 * Allow a burst of 60 reports, then keep quiet for that minute;
	 * or allow a steady drip of one report per second.
	 */
	if (nr_shown == 60) {
		if (time_before(jiffies, resume)) {
			nr_unshown++;
			goto out;
		}
		if (nr_unshown) {
237 238
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
239 240 241 242 243 244 245 246
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

247
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
248
		current->comm, page_to_pfn(page));
249
	printk(KERN_ALERT
250 251 252 253
		"page:%p flags:%p count:%d mapcount:%d mapping:%p index:%lx\n",
		page, (void *)page->flags, page_count(page),
		page_mapcount(page), page->mapping, page->index);

Linus Torvalds's avatar
Linus Torvalds committed
254
	dump_stack();
255
out:
256 257
	/* Leave bad fields for debug, except PageBuddy could make trouble */
	__ClearPageBuddy(page);
258
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
259 260 261 262 263 264 265 266 267 268 269 270
}

/*
 * 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).
 *
271 272 273
 * 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
274
 */
275 276 277

static void free_compound_page(struct page *page)
{
278
	__free_pages_ok(page, compound_order(page));
279 280
}

281
void prep_compound_page(struct page *page, unsigned long order)
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298
{
	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
299 300 301
{
	int i;
	int nr_pages = 1 << order;
302
	struct page *p = page + 1;
Linus Torvalds's avatar
Linus Torvalds committed
303

304
	set_compound_page_dtor(page, free_compound_page);
305
	set_compound_order(page, order);
306
	__SetPageHead(page);
307
	for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) {
308 309
		__SetPageTail(p);
		p->first_page = page;
Linus Torvalds's avatar
Linus Torvalds committed
310 311
	}
}
312
#endif
Linus Torvalds's avatar
Linus Torvalds committed
313

314
static int destroy_compound_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
315 316 317
{
	int i;
	int nr_pages = 1 << order;
318
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
319

320 321
	if (unlikely(compound_order(page) != order) ||
	    unlikely(!PageHead(page))) {
322
		bad_page(page);
323 324
		bad++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
325

326
	__ClearPageHead(page);
327

328 329
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
Linus Torvalds's avatar
Linus Torvalds committed
330

331
		if (unlikely(!PageTail(p) || (p->first_page != page))) {
332
			bad_page(page);
333 334
			bad++;
		}
335
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
336
	}
337 338

	return bad;
Linus Torvalds's avatar
Linus Torvalds committed
339 340
}

Nick Piggin's avatar
Nick Piggin committed
341 342 343 344
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

345 346 347 348
	/*
	 * 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
349
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
350 351 352 353
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

354 355
static inline void set_page_order(struct page *page, int order)
{
356
	set_page_private(page, order);
357
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
358 359 360 361
}

static inline void rmv_page_order(struct page *page)
{
362
	__ClearPageBuddy(page);
363
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380
}

/*
 * 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
381
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
 */
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
400
 * (a) the buddy is not in a hole &&
401
 * (b) the buddy is in the buddy system &&
402 403
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
404 405 406
 *
 * 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
407
 *
408
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
409
 */
410 411
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
412
{
413
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
414 415
		return 0;

416 417 418 419 420
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
421
		return 1;
422
	}
423
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
424 425 426 427 428 429 430 431 432 433 434 435 436 437 438
}

/*
 * 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
439
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
440
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
441 442 443 444 445 446 447 448 449
 * 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
450
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
451 452 453 454
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;
455
	int migratetype = get_pageblock_migratetype(page);
Linus Torvalds's avatar
Linus Torvalds committed
456

457
	if (unlikely(PageCompound(page)))
458 459
		if (unlikely(destroy_compound_page(page, order)))
			return;
Linus Torvalds's avatar
Linus Torvalds committed
460 461 462

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

Nick Piggin's avatar
Nick Piggin committed
463 464
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
465

466
	__mod_zone_page_state(zone, NR_FREE_PAGES, order_size);
Linus Torvalds's avatar
Linus Torvalds committed
467 468 469 470 471
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
472
		if (!page_is_buddy(page, buddy, order))
473
			break;
Nick Piggin's avatar
Nick Piggin committed
474

475
		/* Our buddy is free, merge with it and move up one order. */
Linus Torvalds's avatar
Linus Torvalds committed
476
		list_del(&buddy->lru);
477
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
478
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
479
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
480 481 482 483 484
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
485 486
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
487 488 489
	zone->free_area[order].nr_free++;
}

490
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
491
{
492
	free_page_mlock(page);
Nick Piggin's avatar
Nick Piggin committed
493 494 495
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
496
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE))) {
497
		bad_page(page);
498
		return 1;
499
	}
500 501 502
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
503 504 505 506 507
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
508
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
509 510 511 512 513 514 515
 *
 * 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
516 517
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
518
{
Nick Piggin's avatar
Nick Piggin committed
519
	spin_lock(&zone->lock);
520
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
Linus Torvalds's avatar
Linus Torvalds committed
521
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
522 523 524
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
525
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
526
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
527
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
528
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
529
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
530
	}
Nick Piggin's avatar
Nick Piggin committed
531
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
532 533
}

Nick Piggin's avatar
Nick Piggin committed
534
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
535
{
536
	spin_lock(&zone->lock);
537
	zone_clear_flag(zone, ZONE_ALL_UNRECLAIMABLE);
538
	zone->pages_scanned = 0;
539
	__free_one_page(page, zone, order);
540
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
541 542 543 544 545
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
546
	int i;
547
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
548 549

	for (i = 0 ; i < (1 << order) ; ++i)
550 551
		bad += free_pages_check(page + i);
	if (bad)
552 553
		return;

554
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
555
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
556 557 558
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
559
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
560
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
561

Nick Piggin's avatar
Nick Piggin committed
562
	local_irq_save(flags);
563
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
564
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
565
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
566 567
}

568 569 570
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
571
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
572 573 574 575
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
576
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
577
		__free_page(page);
578 579 580
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
581
		prefetchw(page);
582 583 584
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
585 586
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
587 588 589 590
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

591
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
592
		__free_pages(page, order);
593 594 595
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
596 597 598 599 600 601 602 603 604 605 606 607 608 609 610

/*
 * 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
611
static inline void expand(struct zone *zone, struct page *page,
612 613
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
614 615 616 617 618 619 620
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
621
		VM_BUG_ON(bad_range(zone, &page[size]));
622
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
623 624 625 626 627 628 629 630
		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
631
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
632
{
Nick Piggin's avatar
Nick Piggin committed
633 634 635
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
636
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP))) {
637
		bad_page(page);
638
		return 1;
639
	}
640

641
	set_page_private(page, 0);
642
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
643 644

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
645
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
646 647 648 649 650 651 652

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

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

653
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
654 655
}

656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686
/*
 * 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;
}


687 688 689 690 691
/*
 * 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] = {
692 693 694 695
	[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 */
696 697
};

698 699
/*
 * Move the free pages in a range to the free lists of the requested type.
700
 * Note that start_page and end_pages are not aligned on a pageblock
701 702
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
703 704 705
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
706 707 708
{
	struct page *page;
	unsigned long order;
709
	int pages_moved = 0;
710 711 712 713 714 715 716

#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
717
	 * grouping pages by mobility
718 719 720 721 722
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
723 724 725
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

726 727 728 729 730 731 732 733 734 735 736 737 738 739 740
		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;
741
		pages_moved += 1 << order;
742 743
	}

744
	return pages_moved;
745 746
}

Adrian Bunk's avatar
Adrian Bunk committed
747 748
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
749 750 751 752 753
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
754
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
755
	start_page = pfn_to_page(start_pfn);
756 757
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
758 759 760 761 762 763 764 765 766 767

	/* 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);
}

768 769 770 771 772 773 774 775 776 777 778 779 780 781 782
/* 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];

783 784 785
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
786

787 788 789 790 791 792 793 794 795
			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--;

			/*
796
			 * If breaking a large block of pages, move all free
797 798 799
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
800
			 */
801
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
802 803 804 805 806 807
					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 */
808
				if (pages >= (1 << (pageblock_order-1)))
809 810 811
					set_pageblock_migratetype(page,
								start_migratetype);

812
				migratetype = start_migratetype;
813
			}
814 815 816 817 818 819 820

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

821
			if (current_order == pageblock_order)
822 823 824 825 826 827 828 829
				set_pageblock_migratetype(page,
							start_migratetype);

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

830 831
	/* Use MIGRATE_RESERVE rather than fail an allocation */
	return __rmqueue_smallest(zone, order, MIGRATE_RESERVE);
832 833
}

834
/*
Linus Torvalds's avatar
Linus Torvalds committed
835 836 837
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
838 839
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
840 841 842
{
	struct page *page;

843
	page = __rmqueue_smallest(zone, order, migratetype);
844

845 846
	if (unlikely(!page))
		page = __rmqueue_fallback(zone, order, migratetype);
847 848

	return page;
Linus Torvalds's avatar
Linus Torvalds committed
849 850 851 852 853 854 855 856
}

/* 
 * 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, 
857 858
			unsigned long count, struct list_head *list,
			int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
859 860 861
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
862
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
863
	for (i = 0; i < count; ++i) {
864
		struct page *page = __rmqueue(zone, order, migratetype);
Nick Piggin's avatar
Nick Piggin committed
865
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
866
			break;
867 868 869 870 871 872 873 874 875 876

		/*
		 * 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.
		 */
877 878
		list_add(&page->lru, list);
		set_page_private(page, migratetype);
879
		list = &page->lru;
Linus Torvalds's avatar
Linus Torvalds committed
880
	}
Nick Piggin's avatar
Nick Piggin committed
881
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
882
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
883 884
}

885
#ifdef CONFIG_NUMA
886
/*
887 888 889 890
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
891 892
 * Note that this function must be called with the thread pinned to
 * a single processor.
893
 */
894
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
895 896
{
	unsigned long flags;
897
	int to_drain;
898

899 900 901 902 903 904 905 906
	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);
907 908 909
}
#endif

910 911 912 913 914 915 916 917
/*
 * 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
918
{
Nick Piggin's avatar
Nick Piggin committed
919
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
920 921
	struct zone *zone;

922
	for_each_populated_zone(zone) {
Linus Torvalds's avatar
Linus Torvalds committed
923
		struct per_cpu_pageset *pset;
924
		struct per_cpu_pages *pcp;
Linus Torvalds's avatar
Linus Torvalds committed
925

926
		pset = zone_pcp(zone, cpu);
927 928 929 930 931 932

		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
933 934 935
	}
}

936 937 938 939 940 941 942 943 944 945 946 947 948
/*
 * 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)
{
949
	on_each_cpu(drain_local_pages, NULL, 1);
950 951
}

952
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
953 954 955

void mark_free_pages(struct zone *zone)
{
956 957
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
958
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
959 960 961 962 963 964
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
965 966 967 968 969 970

	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);

971 972
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
973
		}
Linus Torvalds's avatar
Linus Torvalds committed
974

975 976
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
977
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
978

979 980
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
981
				swsusp_set_page_free(pfn_to_page(pfn + i));
982
		}
983
	}
Linus Torvalds's avatar
Linus Torvalds committed
</