page_alloc.c 121 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 22 23
/*
 *  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>
#include <linux/bootmem.h>
#include <linux/compiler.h>
24
#include <linux/kernel.h>
Linus Torvalds's avatar
Linus Torvalds committed
25 26 27 28 29 30 31 32 33 34
#include <linux/module.h>
#include <linux/suspend.h>
#include <linux/pagevec.h>
#include <linux/blkdev.h>
#include <linux/slab.h>
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
35
#include <linux/memory_hotplug.h>
Linus Torvalds's avatar
Linus Torvalds committed
36 37
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
38
#include <linux/mempolicy.h>
39
#include <linux/stop_machine.h>
40 41
#include <linux/sort.h>
#include <linux/pfn.h>
42
#include <linux/backing-dev.h>
43
#include <linux/fault-inject.h>
Linus Torvalds's avatar
Linus Torvalds committed
44 45

#include <asm/tlbflush.h>
46
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
47 48 49
#include "internal.h"

/*
50
 * Array of node states.
Linus Torvalds's avatar
Linus Torvalds committed
51
 */
52 53 54 55 56 57 58 59 60 61 62 63 64
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);

65
unsigned long totalram_pages __read_mostly;
66
unsigned long totalreserve_pages __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
67
long nr_swap_pages;
68
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
69

70 71 72 73
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

74
static void __free_pages_ok(struct page *page, unsigned int order);
75

Linus Torvalds's avatar
Linus Torvalds committed
76 77 78 79 80 81 82
/*
 * 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
83 84 85
 *
 * 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
86
 */
87
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
88
#ifdef CONFIG_ZONE_DMA
89
	 256,
90
#endif
91
#ifdef CONFIG_ZONE_DMA32
92
	 256,
93
#endif
94
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
95
	 32,
96
#endif
Mel Gorman's avatar
Mel Gorman committed
97
	 32,
98
};
Linus Torvalds's avatar
Linus Torvalds committed
99 100 101

EXPORT_SYMBOL(totalram_pages);

102
static char * const zone_names[MAX_NR_ZONES] = {
103
#ifdef CONFIG_ZONE_DMA
104
	 "DMA",
105
#endif
106
#ifdef CONFIG_ZONE_DMA32
107
	 "DMA32",
108
#endif
109
	 "Normal",
110
#ifdef CONFIG_HIGHMEM
Mel Gorman's avatar
Mel Gorman committed
111
	 "HighMem",
112
#endif
Mel Gorman's avatar
Mel Gorman committed
113
	 "Movable",
114 115
};

Linus Torvalds's avatar
Linus Torvalds committed
116 117
int min_free_kbytes = 1024;

118 119
unsigned long __meminitdata nr_kernel_pages;
unsigned long __meminitdata nr_all_pages;
120
static unsigned long __meminitdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
121

122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142
#ifdef CONFIG_ARCH_POPULATES_NODE_MAP
  /*
   * MAX_ACTIVE_REGIONS determines the maxmimum number of distinct
   * 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

143 144 145 146
  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];
147
#ifdef CONFIG_MEMORY_HOTPLUG_RESERVE
148 149
  static unsigned long __meminitdata node_boundary_start_pfn[MAX_NUMNODES];
  static unsigned long __meminitdata node_boundary_end_pfn[MAX_NUMNODES];
150
#endif /* CONFIG_MEMORY_HOTPLUG_RESERVE */
Mel Gorman's avatar
Mel Gorman committed
151
  unsigned long __initdata required_kernelcore;
152
  static unsigned long __initdata required_movablecore;
153
  unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];
Mel Gorman's avatar
Mel Gorman committed
154 155 156 157

  /* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
  int movable_zone;
  EXPORT_SYMBOL(movable_zone);
158 159
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Miklos Szeredi's avatar
Miklos Szeredi committed
160 161 162 163 164
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
EXPORT_SYMBOL(nr_node_ids);
#endif

165 166
int page_group_by_mobility_disabled __read_mostly;

167 168 169 170 171 172
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
173
#ifdef CONFIG_DEBUG_VM
174
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
175
{
176 177 178
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
179

180 181 182 183 184 185 186 187 188
	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;
189 190 191 192
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
193
	if (!pfn_valid_within(page_to_pfn(page)))
194
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
195
	if (zone != page_zone(page))
196 197 198 199 200 201 202 203 204 205
		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
206
		return 1;
207 208 209
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
210 211
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
212 213 214 215 216 217 218
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

219
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
220
{
221
	printk(KERN_EMERG "Bad page state in process '%s'\n"
222 223 224
		KERN_EMERG "page:%p flags:0x%0*lx mapping:%p mapcount:%d count:%d\n"
		KERN_EMERG "Trying to fix it up, but a reboot is needed\n"
		KERN_EMERG "Backtrace:\n",
225 226 227
		current->comm, page, (int)(2*sizeof(unsigned long)),
		(unsigned long)page->flags, page->mapping,
		page_mapcount(page), page_count(page));
Linus Torvalds's avatar
Linus Torvalds committed
228
	dump_stack();
229 230
	page->flags &= ~(1 << PG_lru	|
			1 << PG_private |
Linus Torvalds's avatar
Linus Torvalds committed
231 232 233
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
234 235
			1 << PG_reclaim |
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
236
			1 << PG_swapcache |
237 238
			1 << PG_writeback |
			1 << PG_buddy );
Linus Torvalds's avatar
Linus Torvalds committed
239 240 241
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
242
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
243 244 245 246 247 248 249 250 251 252 253 254
}

/*
 * 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).
 *
255 256 257
 * 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
258
 */
259 260 261

static void free_compound_page(struct page *page)
{
262
	__free_pages_ok(page, compound_order(page));
263 264
}

Linus Torvalds's avatar
Linus Torvalds committed
265 266 267 268 269
static void prep_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;

270
	set_compound_page_dtor(page, free_compound_page);
271
	set_compound_order(page, order);
272
	__SetPageHead(page);
273
	for (i = 1; i < nr_pages; i++) {
Linus Torvalds's avatar
Linus Torvalds committed
274 275
		struct page *p = page + i;

276 277
		__SetPageTail(p);
		p->first_page = page;
Linus Torvalds's avatar
Linus Torvalds committed
278 279 280 281 282 283 284 285
	}
}

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

286
	if (unlikely(compound_order(page) != order))
287
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
288

289
	if (unlikely(!PageHead(page)))
290
			bad_page(page);
291
	__ClearPageHead(page);
292
	for (i = 1; i < nr_pages; i++) {
Linus Torvalds's avatar
Linus Torvalds committed
293 294
		struct page *p = page + i;

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

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

Nick Piggin's avatar
Nick Piggin committed
306
	VM_BUG_ON((gfp_flags & (__GFP_WAIT | __GFP_HIGHMEM)) == __GFP_HIGHMEM);
307 308 309 310
	/*
	 * 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
311
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
312 313 314 315
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

316 317
static inline void set_page_order(struct page *page, int order)
{
318
	set_page_private(page, order);
319
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
320 321 322 323
}

static inline void rmv_page_order(struct page *page)
{
324
	__ClearPageBuddy(page);
325
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342
}

/*
 * 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
343
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361
 */
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
362
 * (a) the buddy is not in a hole &&
363
 * (b) the buddy is in the buddy system &&
364 365
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
366 367 368
 *
 * 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
369
 *
370
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
371
 */
372 373
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
374
{
375
	if (!pfn_valid_within(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
376 377
		return 0;

378 379 380 381 382
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
383
		return 1;
384
	}
385
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
386 387 388 389 390 391 392 393 394 395 396 397 398 399 400
}

/*
 * 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
401
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
402
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
403 404 405 406 407 408 409 410 411
 * 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
412
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
413 414 415 416
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;
417
	int migratetype = get_pageblock_migratetype(page);
Linus Torvalds's avatar
Linus Torvalds committed
418

419
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
420 421 422 423
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
424 425
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
426

427
	__mod_zone_page_state(zone, NR_FREE_PAGES, order_size);
Linus Torvalds's avatar
Linus Torvalds committed
428 429 430 431 432
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
433
		if (!page_is_buddy(page, buddy, order))
Linus Torvalds's avatar
Linus Torvalds committed
434
			break;		/* Move the buddy up one level. */
Nick Piggin's avatar
Nick Piggin committed
435

Linus Torvalds's avatar
Linus Torvalds committed
436
		list_del(&buddy->lru);
437
		zone->free_area[order].nr_free--;
Linus Torvalds's avatar
Linus Torvalds committed
438
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
439
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
440 441 442 443 444
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
445 446
	list_add(&page->lru,
		&zone->free_area[order].free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
447 448 449
	zone->free_area[order].nr_free++;
}

450
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
451
{
Nick Piggin's avatar
Nick Piggin committed
452 453 454
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
Linus Torvalds's avatar
Linus Torvalds committed
455 456 457 458 459 460 461
		(page->flags & (
			1 << PG_lru	|
			1 << PG_private |
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_slab	|
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
462
			1 << PG_writeback |
463 464
			1 << PG_reserved |
			1 << PG_buddy ))))
465
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
466
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
467
		__ClearPageDirty(page);
468 469 470 471 472 473
	/*
	 * For now, we report if PG_reserved was found set, but do not
	 * clear it, and do not free the page.  But we shall soon need
	 * to do more, for when the ZERO_PAGE count wraps negative.
	 */
	return PageReserved(page);
Linus Torvalds's avatar
Linus Torvalds committed
474 475 476 477 478
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
479
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
480 481 482 483 484 485 486
 *
 * 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
487 488
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
489
{
Nick Piggin's avatar
Nick Piggin committed
490
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
491 492
	zone->all_unreclaimable = 0;
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
493 494 495
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
496
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
497
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
498
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
499
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
500
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
501
	}
Nick Piggin's avatar
Nick Piggin committed
502
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
503 504
}

Nick Piggin's avatar
Nick Piggin committed
505
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
506
{
507 508 509
	spin_lock(&zone->lock);
	zone->all_unreclaimable = 0;
	zone->pages_scanned = 0;
510
	__free_one_page(page, zone, order);
511
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
512 513 514 515 516
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
517
	int i;
518
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
519 520

	for (i = 0 ; i < (1 << order) ; ++i)
521
		reserved += free_pages_check(page + i);
522 523 524
	if (reserved)
		return;

Nick Piggin's avatar
Nick Piggin committed
525 526
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
Nick Piggin's avatar
Nick Piggin committed
527
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
528
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
529

Nick Piggin's avatar
Nick Piggin committed
530
	local_irq_save(flags);
531
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
532
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
533
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
534 535
}

536 537 538 539 540 541 542 543
/*
 * permit the bootmem allocator to evade page validation on high-order frees
 */
void fastcall __init __free_pages_bootmem(struct page *page, unsigned int order)
{
	if (order == 0) {
		__ClearPageReserved(page);
		set_page_count(page, 0);
544
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
545
		__free_page(page);
546 547 548
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
549
		prefetchw(page);
550 551 552
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
553 554
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
555 556 557 558
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

559
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
560
		__free_pages(page, order);
561 562 563
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
564 565 566 567 568 569 570 571 572 573 574 575 576 577 578

/*
 * 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
579
static inline void expand(struct zone *zone, struct page *page,
580 581
	int low, int high, struct free_area *area,
	int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
582 583 584 585 586 587 588
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
589
		VM_BUG_ON(bad_range(zone, &page[size]));
590
		list_add(&page[size].lru, &area->free_list[migratetype]);
Linus Torvalds's avatar
Linus Torvalds committed
591 592 593 594 595 596 597 598
		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
599
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
600
{
Nick Piggin's avatar
Nick Piggin committed
601 602 603
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
604 605
		(page->flags & (
			1 << PG_lru	|
Linus Torvalds's avatar
Linus Torvalds committed
606 607 608 609
			1 << PG_private	|
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
610
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
611
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
612
			1 << PG_writeback |
613 614
			1 << PG_reserved |
			1 << PG_buddy ))))
615
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
616

617 618 619 620 621 622 623
	/*
	 * For now, we report if PG_reserved was found set, but do not
	 * clear it, and do not allocate the page: as a safety net.
	 */
	if (PageReserved(page))
		return 1;

624
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error | 1 << PG_readahead |
Linus Torvalds's avatar
Linus Torvalds committed
625
			1 << PG_referenced | 1 << PG_arch_1 |
626
			1 << PG_owner_priv_1 | 1 << PG_mappedtodisk);
627
	set_page_private(page, 0);
628
	set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
629 630

	arch_alloc_page(page, order);
Linus Torvalds's avatar
Linus Torvalds committed
631
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
632 633 634 635 636 637 638

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

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

639
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
640 641
}

642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672
/*
 * 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;
}


673 674 675 676 677
/*
 * 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] = {
678 679 680 681
	[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 */
682 683
};

684 685
/*
 * Move the free pages in a range to the free lists of the requested type.
686
 * Note that start_page and end_pages are not aligned on a pageblock
687 688 689 690 691 692 693 694
 * boundary. If alignment is required, use move_freepages_block()
 */
int move_freepages(struct zone *zone,
			struct page *start_page, struct page *end_page,
			int migratetype)
{
	struct page *page;
	unsigned long order;
695
	int pages_moved = 0;
696 697 698 699 700 701 702

#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
703
	 * grouping pages by mobility
704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
		if (!pfn_valid_within(page_to_pfn(page))) {
			page++;
			continue;
		}

		if (!PageBuddy(page)) {
			page++;
			continue;
		}

		order = page_order(page);
		list_del(&page->lru);
		list_add(&page->lru,
			&zone->free_area[order].free_list[migratetype]);
		page += 1 << order;
724
		pages_moved += 1 << order;
725 726
	}

727
	return pages_moved;
728 729 730 731 732 733 734 735
}

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

	start_pfn = page_to_pfn(page);
736
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
737
	start_page = pfn_to_page(start_pfn);
738 739
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
740 741 742 743 744 745 746 747 748 749

	/* 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);
}

750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766
/* Return the page with the lowest PFN in the list */
static struct page *min_page(struct list_head *list)
{
	unsigned long min_pfn = -1UL;
	struct page *min_page = NULL, *page;;

	list_for_each_entry(page, list, lru) {
		unsigned long pfn = page_to_pfn(page);
		if (pfn < min_pfn) {
			min_pfn = pfn;
			min_page = page;
		}
	}

	return min_page;
}

767 768 769 770 771 772 773 774 775 776 777 778 779 780 781
/* 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];

782 783 784
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
785

786 787 788 789
			area = &(zone->free_area[current_order]);
			if (list_empty(&area->free_list[migratetype]))
				continue;

790
			/* Bias kernel allocations towards low pfns */
791 792
			page = list_entry(area->free_list[migratetype].next,
					struct page, lru);
793 794
			if (unlikely(start_migratetype != MIGRATE_MOVABLE))
				page = min_page(&area->free_list[migratetype]);
795 796 797
			area->nr_free--;

			/*
798
			 * If breaking a large block of pages, move all free
799 800 801
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
			 * agressive about taking ownership of free pages
802
			 */
803
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
804 805 806 807 808 809
					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 */
810
				if (pages >= (1 << (pageblock_order-1)))
811 812 813
					set_pageblock_migratetype(page,
								start_migratetype);

814
				migratetype = start_migratetype;
815
			}
816 817 818 819 820 821 822

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

823
			if (current_order == pageblock_order)
824 825 826 827 828 829 830 831
				set_pageblock_migratetype(page,
							start_migratetype);

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

832 833
	/* Use MIGRATE_RESERVE rather than fail an allocation */
	return __rmqueue_smallest(zone, order, MIGRATE_RESERVE);
834 835
}

836
/*
Linus Torvalds's avatar
Linus Torvalds committed
837 838 839
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
840 841
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
Linus Torvalds's avatar
Linus Torvalds committed
842 843 844
{
	struct page *page;

845
	page = __rmqueue_smallest(zone, order, migratetype);
846

847 848
	if (unlikely(!page))
		page = __rmqueue_fallback(zone, order, migratetype);
849 850

	return page;
Linus Torvalds's avatar
Linus Torvalds committed
851 852 853 854 855 856 857 858
}

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

876
#ifdef CONFIG_NUMA
877
/*
878 879 880 881
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
882 883
 * Note that this function must be called with the thread pinned to
 * a single processor.
884
 */
885
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
886 887
{
	unsigned long flags;
888
	int to_drain;
889

890 891 892 893 894 895 896 897
	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);
898 899 900
}
#endif

Linus Torvalds's avatar
Linus Torvalds committed
901 902
static void __drain_pages(unsigned int cpu)
{
Nick Piggin's avatar
Nick Piggin committed
903
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
904 905 906 907 908 909
	struct zone *zone;
	int i;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;

910 911 912
		if (!populated_zone(zone))
			continue;

913
		pset = zone_pcp(zone, cpu);
Linus Torvalds's avatar
Linus Torvalds committed
914 915 916 917
		for (i = 0; i < ARRAY_SIZE(pset->pcp); i++) {
			struct per_cpu_pages *pcp;

			pcp = &pset->pcp[i];
Nick Piggin's avatar
Nick Piggin committed
918
			local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
919 920
			free_pages_bulk(zone, pcp->count, &pcp->list, 0);
			pcp->count = 0;
Nick Piggin's avatar
Nick Piggin committed
921
			local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
922 923 924 925
		}
	}
}

926
#ifdef CONFIG_HIBERNATION
Linus Torvalds's avatar
Linus Torvalds committed
927 928 929

void mark_free_pages(struct zone *zone)
{
930 931
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
932
	int order, t;
Linus Torvalds's avatar
Linus Torvalds committed
933 934 935 936 937 938
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
939 940 941 942 943 944

	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);

945 946
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
947
		}
Linus Torvalds's avatar
Linus Torvalds committed
948

949 950
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
951
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
952

953 954
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
955
				swsusp_set_page_free(pfn_to_page(pfn + i));
956
		}
957
	}
Linus Torvalds's avatar
Linus Torvalds committed
958 959
	spin_unlock_irqrestore(&zone->lock, flags);
}
960
#endif /* CONFIG_PM */
Linus Torvalds's avatar
Linus Torvalds committed
961 962 963 964 965 966 967 968 969 970 971 972

/*
 * Spill all of this CPU's per-cpu pages back into the buddy allocator.
 */
void drain_local_pages(void)
{
	unsigned long flags;

	local_irq_save(flags);	
	__drain_pages(smp_processor_id());
	local_irq_restore(flags);	
}
973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991

void smp_drain_local_pages(void *arg)
{
	drain_local_pages();
}

/*
 * Spill all the per-cpu pages from all CPUs back into the buddy allocator
 */
void drain_all_local_pages(void)
{
	unsigned long flags;

	local_irq_save(flags);
	__drain_pages(smp_processor_id());
	local_irq_restore(flags);

	smp_call_function(smp_drain_local_pages, NULL, 0, 1);
}
Linus Torvalds's avatar
Linus Torvalds committed
992 993 994 995 996 997 998 999 1000 1001 1002 1003

/*
 * Free a 0-order page
 */
static void fastcall free_hot_cold_page(struct page *page, int cold)
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;

	if (PageAnon(page))
		page->mapping = NULL;
1004
	if (free_pages_check(page))
1005 1006
		return;

Nick Piggin's avatar
Nick Piggin committed
1007 1008
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
Nick Piggin's avatar
Nick Piggin committed
1009
	arch_free_page(page, 0);
1010 1011
	kernel_map_pages(page, 1, 0);

1012
	pcp = &zone_pcp(zone, get_cpu())->pcp[cold];
Linus Torvalds's avatar
Linus Torvalds committed
1013
	local_irq_save(flags);
1014
	__count_vm_event(PGFREE);
Linus Torvalds's avatar
Linus Torvalds committed
1015
	list_add(&page->lru, &pcp->list);
1016
	set_page_private(page, get_pageblock_migratetype(page));
Linus Torvalds's avatar
Linus Torvalds committed
1017
	pcp->count++;
Nick Piggin's avatar
Nick Piggin committed