page_alloc.c 152 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
#include <linux/bootmem.h>
24
#include <linux/memblock.h>
Linus Torvalds's avatar
Linus Torvalds committed
25
#include <linux/compiler.h>
26
#include <linux/kernel.h>
27
#include <linux/kmemcheck.h>
Linus Torvalds's avatar
Linus Torvalds committed
28 29 30 31 32
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
33
#include <linux/oom.h>
Linus Torvalds's avatar
Linus Torvalds committed
34 35 36 37 38
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
39
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
40 41
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
42
#include <linux/mempolicy.h>
43
#include <linux/stop_machine.h>
44 45
#include <linux/sort.h>
#include <linux/pfn.h>
46
#include <linux/backing-dev.h>
47
#include <linux/fault-inject.h>
48
#include <linux/page-isolation.h>
49
#include <linux/page_cgroup.h>
50
#include <linux/debugobjects.h>
51
#include <linux/kmemleak.h>
52
#include <linux/memory.h>
53
#include <linux/compaction.h>
54
#include <trace/events/kmem.h>
55
#include <linux/ftrace_event.h>
Linus Torvalds's avatar
Linus Torvalds committed
56 57

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

61 62 63 64 65
#ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID
DEFINE_PER_CPU(int, numa_node);
EXPORT_PER_CPU_SYMBOL(numa_node);
#endif

66 67 68 69 70 71 72 73 74 75 76
#ifdef CONFIG_HAVE_MEMORYLESS_NODES
/*
 * N.B., Do NOT reference the '_numa_mem_' per cpu variable directly.
 * It will not be defined when CONFIG_HAVE_MEMORYLESS_NODES is not defined.
 * Use the accessor functions set_numa_mem(), numa_mem_id() and cpu_to_mem()
 * defined in <linux/topology.h>.
 */
DEFINE_PER_CPU(int, _numa_mem_);		/* Kernel "local memory" node */
EXPORT_PER_CPU_SYMBOL(_numa_mem_);
#endif

Linus Torvalds's avatar
Linus Torvalds committed
77
/*
78
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
79
 */
80 81 82 83 84 85 86 87 88 89 90 91 92
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);

93
unsigned long totalram_pages __read_mostly;
94
unsigned long totalreserve_pages __read_mostly;
95
int percpu_pagelist_fraction;
96
gfp_t gfp_allowed_mask __read_mostly = GFP_BOOT_MASK;
Linus Torvalds's avatar
Linus Torvalds committed
97

98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
#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 */

123 124 125 126
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

127
static void __free_pages_ok(struct page *page, unsigned int order);
128

Linus Torvalds's avatar
Linus Torvalds committed
129 130 131 132 133 134 135
/*
 * 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
136 137 138
 *
 * 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
139
 */
140
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
141
#ifdef CONFIG_ZONE_DMA
142
	 256,
143
#endif
144
#ifdef CONFIG_ZONE_DMA32
145
	 256,
146
#endif
147
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
148
	 32,
149
#endif
Mel Gorman's avatar
Mel Gorman committed
150
	 32,
151
};
Linus Torvalds's avatar
Linus Torvalds committed
152 153 154

EXPORT_SYMBOL(totalram_pages);

155
static char * const zone_names[MAX_NR_ZONES] = {
156
#ifdef CONFIG_ZONE_DMA
157
	 "DMA",
158
#endif
159
#ifdef CONFIG_ZONE_DMA32
160
	 "DMA32",
161
#endif
162
	 "Normal",
163
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
164
	 "HighMem",
165
#endif
Mel Gorman's avatar
Mel Gorman committed
166
	 "Movable",
167 168
};

Linus Torvalds's avatar
Linus Torvalds committed
169 170
int min_free_kbytes = 1024;

171 172
static unsigned long __meminitdata nr_kernel_pages;
static unsigned long __meminitdata nr_all_pages;
173
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
174

175 176
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
Simon Arlott's avatar
Simon Arlott committed
177
   * MAX_ACTIVE_REGIONS determines the maximum number of distinct
178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
   * 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

196 197 198 199
  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
200
  static unsigned long __initdata required_kernelcore;
201
  static unsigned long __initdata required_movablecore;
Adrian Bunk's avatar
Adrian Bunk committed
202
  static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
203 204 205 206

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
207 208
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
209 210
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
211
int nr_online_nodes __read_mostly = 1;
Miklos Szeredi's avatar
Miklos Szeredi committed
212
EXPORT_SYMBOL(nr_node_ids);
213
EXPORT_SYMBOL(nr_online_nodes);
Miklos Szeredi's avatar
Miklos Szeredi committed
214 215
#endif

216 217
int page_group_by_mobility_disabled __read_mostly;

218 219
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
220 221 222 223

	if (unlikely(page_group_by_mobility_disabled))
		migratetype = MIGRATE_UNMOVABLE;

224 225 226 227
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

228 229
bool oom_killer_disabled __read_mostly;

Nick Piggin's avatar
Nick Piggin committed
230
#ifdef CONFIG_DEBUG_VM
231
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
232
{
233 234 235
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
236

237 238 239 240 241 242 243 244 245
	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;
246 247 248 249
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
250
	if (!pfn_valid_within(page_to_pfn(page)))
251
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
252
	if (zone != page_zone(page))
253 254 255 256 257 258 259 260 261 262
		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
263
		return 1;
264 265 266
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
267 268
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
269 270 271 272 273 274 275
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

276
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
277
{
278 279 280 281
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

282 283 284 285 286 287
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
		__ClearPageBuddy(page);
		return;
	}

288 289 290 291 292 293 294 295 296 297
	/*
	 * 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) {
298 299
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
300 301 302 303 304 305 306 307
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

308
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
309
		current->comm, page_to_pfn(page));
310
	dump_page(page);
311

Linus Torvalds's avatar
Linus Torvalds committed
312
	dump_stack();
313
out:
314 315
	/* Leave bad fields for debug, except PageBuddy could make trouble */
	__ClearPageBuddy(page);
316
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
317 318 319 320 321 322 323 324 325 326 327 328
}

/*
 * 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).
 *
329 330 331
 * 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
332
 */
333 334 335

static void free_compound_page(struct page *page)
{
336
	__free_pages_ok(page, compound_order(page));
337 338
}

339
void prep_compound_page(struct page *page, unsigned long order)
340 341 342 343 344 345 346 347 348 349 350 351 352 353 354
{
	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;
	}
}

355
static int destroy_compound_page(struct page *page, unsigned long order)
Linus Torvalds's avatar
Linus Torvalds committed
356 357 358
{
	int i;
	int nr_pages = 1 << order;
359
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
360

361 362
	if (unlikely(compound_order(page) != order) ||
	    unlikely(!PageHead(page))) {
363
		bad_page(page);
364 365
		bad++;
	}
Linus Torvalds's avatar
Linus Torvalds committed
366

367
	__ClearPageHead(page);
368

369 370
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
Linus Torvalds's avatar
Linus Torvalds committed
371

372
		if (unlikely(!PageTail(p) || (p->first_page != page))) {
373
			bad_page(page);
374 375
			bad++;
		}
376
		__ClearPageTail(p);
Linus Torvalds's avatar
Linus Torvalds committed
377
	}
378 379

	return bad;
Linus Torvalds's avatar
Linus Torvalds committed
380 381
}

Nick Piggin's avatar
Nick Piggin committed
382 383 384 385
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

386 387 388 389
	/*
	 * 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
390
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
391 392 393 394
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

395 396
static inline void set_page_order(struct page *page, int order)
{
397
	set_page_private(page, order);
398
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
399 400 401 402
}

static inline void rmv_page_order(struct page *page)
{
403
	__ClearPageBuddy(page);
404
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421
}

/*
 * 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
422
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440
 */
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
441
 * (a) the buddy is not in a hole &&
442
 * (b) the buddy is in the buddy system &&
443 444
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
445 446 447
 *
 * 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
448
 *
449
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
450
 */
451 452
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
453
{
454
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
455 456
		return 0;

457 458 459 460
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
461
		VM_BUG_ON(page_count(buddy) != 0);
462
		return 1;
463
	}
464
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
465 466 467 468 469 470 471 472 473 474 475 476 477 478 479
}

/*
 * 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
480
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
481
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
482 483 484 485 486 487 488 489 490
 * 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
491
static inline void __free_one_page(struct page *page,
492 493
		struct zone *zone, unsigned int order,
		int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
494 495
{
	unsigned long page_idx;
496 497
	unsigned long combined_idx;
	struct page *buddy;
Linus Torvalds's avatar
Linus Torvalds committed
498

499
	if (unlikely(PageCompound(page)))
500 501
		if (unlikely(destroy_compound_page(page, order)))
			return;
Linus Torvalds's avatar
Linus Torvalds committed
502

503 504
	VM_BUG_ON(migratetype == -1);

Linus Torvalds's avatar
Linus Torvalds committed
505 506
	page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);

507
	VM_BUG_ON(page_idx & ((1 << order) - 1));
Nick Piggin's avatar
Nick Piggin committed
508
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
509 510 511

	while (order < MAX_ORDER-1) {
		buddy = __page_find_buddy(page, page_idx, order);
512
		if (!page_is_buddy(page, buddy, order))
513
			break;
Nick Piggin's avatar
Nick Piggin committed
514

515
		/* Our buddy is free, merge with it and move up one order. */
Linus Torvalds's avatar
Linus Torvalds committed
516
		list_del(&buddy->lru);
517
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
518
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
519
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
520 521 522 523 524
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
525 526 527 528 529 530 531 532 533

	/*
	 * 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
	 */
534
	if ((order < MAX_ORDER-2) && pfn_valid_within(page_to_pfn(buddy))) {
535 536 537 538 539 540 541 542 543 544 545 546 547
		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
548 549 550
	zone->free_area[order].nr_free++;
}

551 552 553 554 555 556 557 558 559 560 561
/*
 * 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);
}

562
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
563
{
Nick Piggin's avatar
Nick Piggin committed
564 565
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
566
		(atomic_read(&page->_count) != 0) |
567
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE))) {
568
		bad_page(page);
569
		return 1;
570
	}
571 572 573
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
574 575 576
}

/*
577
 * Frees a number of pages from the PCP lists
Linus Torvalds's avatar
Linus Torvalds committed
578
 * Assumes all pages on list are in same zone, and of same order.
579
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
580 581 582 583 584 585 586
 *
 * 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.
 */
587 588
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
Linus Torvalds's avatar
Linus Torvalds committed
589
{
590
	int migratetype = 0;
591
	int batch_free = 0;
592
	int to_free = count;
593

Nick Piggin's avatar
Nick Piggin committed
594
	spin_lock(&zone->lock);
595
	zone->all_unreclaimable = 0;
Linus Torvalds's avatar
Linus Torvalds committed
596
	zone->pages_scanned = 0;
597

598
	while (to_free) {
Nick Piggin's avatar
Nick Piggin committed
599
		struct page *page;
600 601 602
		struct list_head *list;

		/*
603 604 605 606 607
		 * 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
608 609
		 */
		do {
610
			batch_free++;
611 612 613 614
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
Nick Piggin's avatar
Nick Piggin committed
615

616 617 618 619
		do {
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
620 621 622
			/* 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));
623
		} while (--to_free && --batch_free && !list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
624
	}
625
	__mod_zone_page_state(zone, NR_FREE_PAGES, count);
Nick Piggin's avatar
Nick Piggin committed
626
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
627 628
}

629 630
static void free_one_page(struct zone *zone, struct page *page, int order,
				int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
631
{
632
	spin_lock(&zone->lock);
633
	zone->all_unreclaimable = 0;
634
	zone->pages_scanned = 0;
635

636
	__free_one_page(page, zone, order, migratetype);
637
	__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
638
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
639 640
}

641
static bool free_pages_prepare(struct page *page, unsigned int order)
Nick Piggin's avatar
Nick Piggin committed
642
{
Linus Torvalds's avatar
Linus Torvalds committed
643
	int i;
644
	int bad = 0;
Linus Torvalds's avatar
Linus Torvalds committed
645

646
	trace_mm_page_free_direct(page, order);
647 648
	kmemcheck_free_shadow(page, order);

649 650 651 652 653 654 655
	for (i = 0; i < (1 << order); i++) {
		struct page *pg = page + i;

		if (PageAnon(pg))
			pg->mapping = NULL;
		bad += free_pages_check(pg);
	}
656
	if (bad)
657
		return false;
658

659
	if (!PageHighMem(page)) {
Nick Piggin's avatar
Nick Piggin committed
660
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
661 662 663
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
Nick Piggin's avatar
Nick Piggin committed
664
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
665
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
666

667 668 669 670 671 672 673 674 675 676 677
	return true;
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
	int wasMlocked = __TestClearPageMlocked(page);

	if (!free_pages_prepare(page, order))
		return;

Nick Piggin's avatar
Nick Piggin committed
678
	local_irq_save(flags);
679
	if (unlikely(wasMlocked))
680
		free_page_mlock(page);
681
	__count_vm_events(PGFREE, 1 << order);
682 683
	free_one_page(page_zone(page), page, order,
					get_pageblock_migratetype(page));
Nick Piggin's avatar
Nick Piggin committed
684
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
685 686
}

687 688 689
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
690
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
691 692 693 694
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
695
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
696
		__free_page(page);
697 698 699
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
700
		prefetchw(page);
701 702 703
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
704 705
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
706 707 708 709
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

710
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
711
		__free_pages(page, order);
712 713 714
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
715 716 717 718 719 720 721 722 723 724 725 726 727 728 729

/*
 * 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
730
static inline void expand(struct zone *zone, struct page *page,
731 732
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
733 734 735 736 737 738 739
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
740
		VM_BUG_ON(bad_range(zone, &page[size]));
741
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
742 743 744 745 746 747 748 749
		area->nr_free++;
		set_page_order(&page[size], high);
	}
}

/*
 * This page is about to be returned from the page allocator
 */
750
static inline int check_new_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
751
{
Nick Piggin's avatar
Nick Piggin committed
752 753
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
754
		(atomic_read(&page->_count) != 0)  |
755
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP))) {
756
		bad_page(page);
757
		return 1;
758
	}
759 760 761 762 763 764 765 766 767 768 769 770
	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;
	}
771

772
	set_page_private(page, 0);
773
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
774 775

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
776
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
777 778 779 780 781 782 783

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

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

784
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
785 786
}

787 788 789 790
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
791 792
static inline
struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817
						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;
}


818 819 820 821 822
/*
 * 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] = {
823 824 825 826
	[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 */
827 828
};

829 830
/*
 * Move the free pages in a range to the free lists of the requested type.
831
 * Note that start_page and end_pages are not aligned on a pageblock
832 833
 * boundary. If alignment is required, use move_freepages_block()
 */
Adrian Bunk's avatar
Adrian Bunk committed
834 835 836
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
837 838 839
{
	struct page *page;
	unsigned long order;
840
	int pages_moved = 0;
841 842 843 844 845 846 847

#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
848
	 * grouping pages by mobility
849 850 851 852 853
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
854 855 856
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

857 858 859 860 861 862 863 864 865 866 867 868 869 870 871
		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;
872
		pages_moved += 1 << order;
873 874
	}

875
	return pages_moved;
876 877
}

Adrian Bunk's avatar
Adrian Bunk committed
878 879
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
880 881 882 883 884
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
885
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
886
	start_page = pfn_to_page(start_pfn);
887 888
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
889 890 891 892 893 894 895 896 897 898

	/* 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);
}

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

910
/* Remove an element from the buddy allocator from the fallback list */
911 912
static inline struct page *
__rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
913 914 915 916 917 918 919 920 921 922 923 924
{
	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];

925 926 927
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
928

929 930 931 932 933 934 935 936 937
			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--;

			/*
938
			 * If breaking a large block of pages, move all free
939 940 941
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
942
			 */
943
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
944 945
					start_migratetype == MIGRATE_RECLAIMABLE ||
					page_group_by_mobility_disabled) {
946 947 948 949 950
				unsigned long pages;
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
951 952
				if (pages >= (1 << (pageblock_order-1)) ||
						page_group_by_mobility_disabled)
953 954 955
					set_pageblock_migratetype(page,
								start_migratetype);

956
				migratetype = start_migratetype;
957
			}
958 959 960 961 962

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

963 964 965
			/* Take ownership for orders >= pageblock_order */
			if (current_order >= pageblock_order)
				change_pageblock_range(page, current_order,
966 967 968
							start_migratetype);

			expand(zone, page, order, current_order, area, migratetype);
969 970 971 972

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

973 974 975 976
			return page;
		}
	}

977
	return NULL;
978 979
}

980
/*
Linus Torvalds's avatar