page_alloc.c 148 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>
26
#include <linux/kmemcheck.h>
Linus Torvalds's avatar
Linus Torvalds committed
27 28 29 30 31
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
32
#include <linux/oom.h>
Linus Torvalds's avatar
Linus Torvalds committed
33 34 35 36 37
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
38
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
39 40
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
41
#include <linux/mempolicy.h>
42
#include <linux/stop_machine.h>
43 44
#include <linux/sort.h>
#include <linux/pfn.h>
45
#include <linux/backing-dev.h>
46
#include <linux/fault-inject.h>
47
#include <linux/page-isolation.h>
48
#include <linux/page_cgroup.h>
49
#include <linux/debugobjects.h>
50
#include <linux/kmemleak.h>
51
#include <linux/memory.h>
52
#include <linux/compaction.h>
53
#include <trace/events/kmem.h>
54
#include <linux/ftrace_event.h>
Linus Torvalds's avatar
Linus Torvalds committed
55 56

#include <asm/tlbflush.h>
57
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
58 59 60
#include "internal.h"

/*
61
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
62
 */
63 64 65 66 67 68 69 70 71 72 73 74 75
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);

76
unsigned long totalram_pages __read_mostly;
77
unsigned long totalreserve_pages __read_mostly;
78
int percpu_pagelist_fraction;
79
gfp_t gfp_allowed_mask __read_mostly = GFP_BOOT_MASK;
Linus Torvalds's avatar
Linus Torvalds committed
80

81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
#ifdef CONFIG_PM_SLEEP
/*
 * The following functions are used by the suspend/hibernate code to temporarily
 * change gfp_allowed_mask in order to avoid using I/O during memory allocations
 * while devices are suspended.  To avoid races with the suspend/hibernate code,
 * they should always be called with pm_mutex held (gfp_allowed_mask also should
 * only be modified with pm_mutex held, unless the suspend/hibernate code is
 * guaranteed not to run in parallel with that modification).
 */
void set_gfp_allowed_mask(gfp_t mask)
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
	gfp_allowed_mask = mask;
}

gfp_t clear_gfp_allowed_mask(gfp_t mask)
{
	gfp_t ret = gfp_allowed_mask;

	WARN_ON(!mutex_is_locked(&pm_mutex));
	gfp_allowed_mask &= ~mask;
	return ret;
}
#endif /* CONFIG_PM_SLEEP */

106 107 108 109
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

110
static void __free_pages_ok(struct page *page, unsigned int order);
111

Linus Torvalds's avatar
Linus Torvalds committed
112 113 114 115 116 117 118
/*
 * 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
119 120 121
 *
 * 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
122
 */
123
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
124
#ifdef CONFIG_ZONE_DMA
125
	 256,
126
#endif
127
#ifdef CONFIG_ZONE_DMA32
128
	 256,
129
#endif
130
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
131
	 32,
132
#endif
Mel Gorman's avatar
Mel Gorman committed
133
	 32,
134
};
Linus Torvalds's avatar
Linus Torvalds committed
135 136 137

EXPORT_SYMBOL(totalram_pages);

138
static char * const zone_names[MAX_NR_ZONES] = {
139
#ifdef CONFIG_ZONE_DMA
140
	 "DMA",
141
#endif
142
#ifdef CONFIG_ZONE_DMA32
143
	 "DMA32",
144
#endif
145
	 "Normal",
146
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
147
	 "HighMem",
148
#endif
Mel Gorman's avatar
Mel Gorman committed
149
	 "Movable",
150 151
};

Linus Torvalds's avatar
Linus Torvalds committed
152 153
int min_free_kbytes = 1024;

154 155
static unsigned long __meminitdata nr_kernel_pages;
static unsigned long __meminitdata nr_all_pages;
156
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
157

158 159
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
160
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178
   * 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

179 180 181 182
  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
183
  static unsigned long __initdata required_kernelcore;
184
  static unsigned long __initdata required_movablecore;
Adrian Bunk's avatar
Adrian Bunk committed
185
  static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
186 187 188 189

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
190 191
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
192 193
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
194
int nr_online_nodes __read_mostly = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
195
EXPORT_SYMBOL(nr_node_ids);
196
EXPORT_SYMBOL(nr_online_nodes);
Miklos Szeredi's avatar
Miklos Szeredi committed
197 198
#endif

199 200
int page_group_by_mobility_disabled __read_mostly;

201 202
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
203 204 205 206

	if (unlikely(page_group_by_mobility_disabled))
		migratetype = MIGRATE_UNMOVABLE;

207 208 209 210
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

211 212
bool oom_killer_disabled __read_mostly;

Nick Piggin's avatar
Nick Piggin committed
213
#ifdef CONFIG_DEBUG_VM
214
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
215
{
216 217 218
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
219

220 221 222 223 224 225 226 227 228
	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;
229 230 231 232
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
233
	if (!pfn_valid_within(page_to_pfn(page)))
234
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
235
	if (zone != page_zone(page))
236 237 238 239 240 241 242 243 244 245
		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
246
		return 1;
247 248 249
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
250 251
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
252 253 254 255 256 257 258
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

259
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
260
{
261 262 263 264
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

265 266 267 268 269 270
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
		__ClearPageBuddy(page);
		return;
	}

271 272 273 274 275 276 277 278 279 280
	/*
	 * 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) {
281 282
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
283 284 285 286 287 288 289 290
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

291
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
292
		current->comm, page_to_pfn(page));
293
	dump_page(page);
294

Linus Torvalds's avatar
Linus Torvalds committed
295
	dump_stack();
296
out:
297 298
	/* Leave bad fields for debug, except PageBuddy could make trouble */
	__ClearPageBuddy(page);
299
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
300 301 302 303 304 305 306 307 308 309 310 311
}

/*
 * 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).
 *
312 313 314
 * 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
315
 */
316 317 318

static void free_compound_page(struct page *page)
{
319
	__free_pages_ok(page, compound_order(page));
320 321
}

322
void prep_compound_page(struct page *page, unsigned long order)
323 324 325 326 327 328 329 330 331 332 333 334 335 336 337
{
	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;
	}
}

338
static int destroy_compound_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
339 340 341
{
	int i;
	int nr_pages = 1 << order;
342
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
343

344 345
	if (unlikely(compound_order(page) != order) ||
	    unlikely(!PageHead(page))) {
346
		bad_page(page);
347 348
		bad++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
349

350
	__ClearPageHead(page);
351

352 353
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
Linus Torvalds's avatar
Linus Torvalds committed
354

355
		if (unlikely(!PageTail(p) || (p->first_page != page))) {
356
			bad_page(page);
357 358
			bad++;
		}
359
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
360
	}
361 362

	return bad;
Linus Torvalds's avatar
Linus Torvalds committed
363 364
}

Nick Piggin's avatar
Nick Piggin committed
365 366 367 368
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

369 370 371 372
	/*
	 * 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
373
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
374 375 376 377
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

378 379
static inline void set_page_order(struct page *page, int order)
{
380
	set_page_private(page, order);
381
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
382 383 384 385
}

static inline void rmv_page_order(struct page *page)
{
386
	__ClearPageBuddy(page);
387
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404
}

/*
 * 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
405
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423
 */
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
424
 * (a) the buddy is not in a hole &&
425
 * (b) the buddy is in the buddy system &&
426 427
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
428 429 430
 *
 * 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
431
 *
432
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
433
 */
434 435
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
436
{
437
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
438 439
		return 0;

440 441 442 443
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
444
		VM_BUG_ON(page_count(buddy) != 0);
445
		return 1;
446
	}
447
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
}

/*
 * 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
463
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
464
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
465 466 467 468 469 470 471 472 473
 * 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
474
static inline void __free_one_page(struct page *page,
475 476
		struct zone *zone, unsigned int order,
		int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
477 478
{
	unsigned long page_idx;
479 480
	unsigned long combined_idx;
	struct page *buddy;
Linus Torvalds's avatar
Linus Torvalds committed
481

482
	if (unlikely(PageCompound(page)))
483 484
		if (unlikely(destroy_compound_page(page, order)))
			return;
Linus Torvalds's avatar
Linus Torvalds committed
485

486 487
	VM_BUG_ON(migratetype == -1);

Linus Torvalds's avatar
Linus Torvalds committed
488 489
	page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);

490
	VM_BUG_ON(page_idx & ((1 << order) - 1));
Nick Piggin's avatar
Nick Piggin committed
491
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
492 493 494

	while (order < MAX_ORDER-1) {
		buddy = __page_find_buddy(page, page_idx, order);
495
		if (!page_is_buddy(page, buddy, order))
496
			break;
Nick Piggin's avatar
Nick Piggin committed
497

498
		/* Our buddy is free, merge with it and move up one order. */
Linus Torvalds's avatar
Linus Torvalds committed
499
		list_del(&buddy->lru);
500
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
501
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
502
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
503 504 505 506 507
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530

	/*
	 * If this is not the largest possible page, check if the buddy
	 * of the next-highest order is free. If it is, it's possible
	 * that pages are being freed that will coalesce soon. In case,
	 * that is happening, add the free page to the tail of the list
	 * so it's less likely to be used soon and more likely to be merged
	 * as a higher order page
	 */
	if ((order < MAX_ORDER-1) && pfn_valid_within(page_to_pfn(buddy))) {
		struct page *higher_page, *higher_buddy;
		combined_idx = __find_combined_index(page_idx, order);
		higher_page = page + combined_idx - page_idx;
		higher_buddy = __page_find_buddy(higher_page, combined_idx, order + 1);
		if (page_is_buddy(higher_page, higher_buddy, order + 1)) {
			list_add_tail(&page->lru,
				&zone->free_area[order].free_list[migratetype]);
			goto out;
		}
	}

	list_add(&page->lru, &zone->free_area[order].free_list[migratetype]);
out:
Linus Torvalds's avatar
Linus Torvalds committed
531 532 533
	zone->free_area[order].nr_free++;
}

534 535 536 537 538 539 540 541 542 543 544
/*
 * free_page_mlock() -- clean up attempts to free and mlocked() page.
 * Page should not be on lru, so no need to fix that up.
 * free_pages_check() will verify...
 */
static inline void free_page_mlock(struct page *page)
{
	__dec_zone_page_state(page, NR_MLOCK);
	__count_vm_event(UNEVICTABLE_MLOCKFREED);
}

545
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
546
{
Nick Piggin's avatar
Nick Piggin committed
547 548
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
549
		(atomic_read(&page->_count) != 0) |
550
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE))) {
551
		bad_page(page);
552
		return 1;
553
	}
554 555 556
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
557 558 559
}

/*
560
 * Frees a number of pages from the PCP lists
Linus Torvalds's avatar
Linus Torvalds committed
561
 * Assumes all pages on list are in same zone, and of same order.
562
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
563 564 565 566 567 568 569
 *
 * 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.
 */
570 571
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
Linus Torvalds's avatar
Linus Torvalds committed
572
{
573
	int migratetype = 0;
574
	int batch_free = 0;
575

Nick Piggin's avatar
Nick Piggin committed
576
	spin_lock(&zone->lock);
577
	zone->all_unreclaimable = 0;
Linus Torvalds's avatar
Linus Torvalds committed
578
	zone->pages_scanned = 0;
579

580
	__mod_zone_page_state(zone, NR_FREE_PAGES, count);
581
	while (count) {
Nick Piggin's avatar
Nick Piggin committed
582
		struct page *page;
583 584 585
		struct list_head *list;

		/*
586 587 588 589 590
		 * Remove pages from lists in a round-robin fashion. A
		 * batch_free count is maintained that is incremented when an
		 * empty list is encountered.  This is so more pages are freed
		 * off fuller lists instead of spinning excessively around empty
		 * lists
591 592
		 */
		do {
593
			batch_free++;
594 595 596 597
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
Nick Piggin's avatar
Nick Piggin committed
598

599 600 601 602
		do {
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
603 604 605
			/* MIGRATE_MOVABLE list may include MIGRATE_RESERVEs */
			__free_one_page(page, zone, 0, page_private(page));
			trace_mm_page_pcpu_drain(page, 0, page_private(page));
606
		} while (--count && --batch_free && !list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
607
	}
Nick Piggin's avatar
Nick Piggin committed
608
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
609 610
}

611 612
static void free_one_page(struct zone *zone, struct page *page, int order,
				int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
613
{
614
	spin_lock(&zone->lock);
615
	zone->all_unreclaimable = 0;
616
	zone->pages_scanned = 0;
617 618

	__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
619
	__free_one_page(page, zone, order, migratetype);
620
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
621 622 623 624 625
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
626
	int i;
627
	int bad = 0;
628
	int wasMlocked = __TestClearPageMlocked(page);
Linus Torvalds's avatar
Linus Torvalds committed
629

630
	trace_mm_page_free_direct(page, order);
631 632
	kmemcheck_free_shadow(page, order);

Linus Torvalds's avatar
Linus Torvalds committed
633
	for (i = 0 ; i < (1 << order) ; ++i)
634 635
		bad += free_pages_check(page + i);
	if (bad)
636 637
		return;

638
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
639
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
640 641 642
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
643
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
644
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
645

Nick Piggin's avatar
Nick Piggin committed
646
	local_irq_save(flags);
647
	if (unlikely(wasMlocked))
648
		free_page_mlock(page);
649
	__count_vm_events(PGFREE, 1 << order);
650 651
	free_one_page(page_zone(page), page, order,
					get_pageblock_migratetype(page));
Nick Piggin's avatar
Nick Piggin committed
652
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
653 654
}

655 656 657
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
658
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
659 660 661 662
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
663
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
664
		__free_page(page);
665 666 667
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
668
		prefetchw(page);
669 670 671
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
672 673
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
674 675 676 677
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

678
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
679
		__free_pages(page, order);
680 681 682
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
683 684 685 686 687 688 689 690 691 692 693 694 695 696 697

/*
 * 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
698
static inline void expand(struct zone *zone, struct page *page,
699 700
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
701 702 703 704 705 706 707
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
708
		VM_BUG_ON(bad_range(zone, &page[size]));
709
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
710 711 712 713 714 715 716 717
		area->nr_free++;
		set_page_order(&page[size], high);
	}
}

/*
 * This page is about to be returned from the page allocator
 */
718
static inline int check_new_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
719
{
Nick Piggin's avatar
Nick Piggin committed
720 721
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
722
		(atomic_read(&page->_count) != 0)  |
723
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP))) {
724
		bad_page(page);
725
		return 1;
726
	}
727 728 729 730 731 732 733 734 735 736 737 738
	return 0;
}

static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

	for (i = 0; i < (1 << order); i++) {
		struct page *p = page + i;
		if (unlikely(check_new_page(p)))
			return 1;
	}
739

740
	set_page_private(page, 0);
741
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
742 743

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
744
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
745 746 747 748 749 750 751

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

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

752
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
753 754
}

755 756 757 758
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
759 760
static inline
struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
						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--;
		expand(zone, page, order, current_order, area, migratetype);
		return page;
	}

	return NULL;
}


786 787 788 789 790
/*
 * 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] = {
791 792 793 794
	[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 */
795 796
};

797 798
/*
 * Move the free pages in a range to the free lists of the requested type.
799
 * Note that start_page and end_pages are not aligned on a pageblock
800 801
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
802 803 804
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
805 806 807
{
	struct page *page;
	unsigned long order;
808
	int pages_moved = 0;
809 810 811 812 813 814 815

#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
816
	 * grouping pages by mobility
817 818 819 820 821
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
822 823 824
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

825 826 827 828 829 830 831 832 833 834 835 836 837 838 839
		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;
840
		pages_moved += 1 << order;
841 842
	}

843
	return pages_moved;
844 845
}

Adrian Bunk's avatar
Adrian Bunk committed
846 847
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
848 849 850 851 852
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
853
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
854
	start_page = pfn_to_page(start_pfn);
855 856
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
857 858 859 860 861 862 863 864 865 866

	/* 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);
}

867 868 869 870 871 872 873 874 875 876 877
static void change_pageblock_range(struct page *pageblock_page,
					int start_order, int migratetype)
{
	int nr_pageblocks = 1 << (start_order - pageblock_order);

	while (nr_pageblocks--) {
		set_pageblock_migratetype(pageblock_page, migratetype);
		pageblock_page += pageblock_nr_pages;
	}
}

878
/* Remove an element from the buddy allocator from the fallback list */
879 880
static inline struct page *
__rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
881 882 883 884 885 886 887 888 889 890 891 892
{
	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];

893 894 895
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
896

897 898 899 900 901 902 903 904 905
			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--;

			/*
906
			 * If breaking a large block of pages, move all free
907 908 909
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
910
			 */
911
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
912 913
					start_migratetype == MIGRATE_RECLAIMABLE ||
					page_group_by_mobility_disabled) {
914 915 916 917 918
				unsigned long pages;
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
919 920
				if (pages >= (1 << (pageblock_order-1)) ||
						page_group_by_mobility_disabled)
921 922 923
					set_pageblock_migratetype(page,
								start_migratetype);

924
				migratetype = start_migratetype;
925
			}
926 927 928 929 930

			/* Remove the page from the freelists */
			list_del(&page->lru);
			rmv_page_order(page);

931 932 933
			/* Take ownership for orders >= pageblock_order */
			if (current_order >= pageblock_order)
				change_pageblock_range(page, current_order,
934 935 936
							start_migratetype);

			expand(zone, page, order, current_order, area, migratetype);
937 938 939 940

			trace_mm_page_alloc_extfrag(page, order, current_order,
				start_migratetype, migratetype);

941 942 943 944
			return page;
		}
	}

945
	return NULL;
946 947
}

948
/*
Linus Torvalds's avatar
Linus Torvalds committed
949 950 951
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
952 953
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
954 955 956
{
	struct page *page;

957
retry_reserve:
958
	page = __rmqueue_smallest(zone, order, migratetype);
959

960
	if (unlikely(!page) && migratetype != MIGRATE_RESERVE) {
961
		page = __rmqueue_fallback(zone, order, migratetype);
962

963 964 965 966 967 968 969 970 971 972 973
		/*
		 * Use MIGRATE_RESERVE rather than fail an allocation. goto
		 * is used because __rmqueue_smallest is an inline function
		 * and we want just one call site
		 */
		if (!page) {
			migratetype = MIGRATE_RESERVE;
			goto retry_reserve;
		}
	}

974
	trace_mm_page_alloc_zone_locked(page, order, migratetype);
975
	return page;
Linus Torvalds's avatar
Linus Torvalds committed
976 977 978 979 980 981 982 983
}

/* 
 * 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, 
984
			unsigned long count, struct list_head *list,
985
			int migratetype, int cold)
Linus Torvalds's avatar
Linus Torvalds committed
986 987 988
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
989
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
990
	for (i = 0; i < count; ++i) {