page_alloc.c 83.2 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>
Linus Torvalds's avatar
Linus Torvalds committed
42 43

#include <asm/tlbflush.h>
44
#include <asm/div64.h>
Linus Torvalds's avatar
Linus Torvalds committed
45 46 47 48 49 50
#include "internal.h"

/*
 * MCD - HACK: Find somewhere to initialize this EARLY, or make this
 * initializer cleaner
 */
51
nodemask_t node_online_map __read_mostly = { { [0] = 1UL } };
52
EXPORT_SYMBOL(node_online_map);
53
nodemask_t node_possible_map __read_mostly = NODE_MASK_ALL;
54
EXPORT_SYMBOL(node_possible_map);
55
unsigned long totalram_pages __read_mostly;
56
unsigned long totalreserve_pages __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
57
long nr_swap_pages;
58
int percpu_pagelist_fraction;
Linus Torvalds's avatar
Linus Torvalds committed
59

60
static void __free_pages_ok(struct page *page, unsigned int order);
61

Linus Torvalds's avatar
Linus Torvalds committed
62 63 64 65 66 67 68
/*
 * 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
69 70 71
 *
 * 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
72
 */
73 74
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
	 256,
75
#ifdef CONFIG_ZONE_DMA32
76
	 256,
77
#endif
78
#ifdef CONFIG_HIGHMEM
79
	 32
80
#endif
81
};
Linus Torvalds's avatar
Linus Torvalds committed
82 83 84 85 86 87 88

EXPORT_SYMBOL(totalram_pages);

/*
 * Used by page_zone() to look up the address of the struct zone whose
 * id is encoded in the upper bits of page->flags
 */
89
struct zone *zone_table[1 << ZONETABLE_SHIFT] __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
90 91
EXPORT_SYMBOL(zone_table);

92 93
static char *zone_names[MAX_NR_ZONES] = {
	 "DMA",
94
#ifdef CONFIG_ZONE_DMA32
95
	 "DMA32",
96
#endif
97
	 "Normal",
98
#ifdef CONFIG_HIGHMEM
99
	 "HighMem"
100
#endif
101 102
};

Linus Torvalds's avatar
Linus Torvalds committed
103 104
int min_free_kbytes = 1024;

105 106
unsigned long __meminitdata nr_kernel_pages;
unsigned long __meminitdata nr_all_pages;
107
static unsigned long __initdata dma_reserve;
Linus Torvalds's avatar
Linus Torvalds committed
108

109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133
#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

  struct node_active_region __initdata early_node_map[MAX_ACTIVE_REGIONS];
  int __initdata nr_nodemap_entries;
  unsigned long __initdata arch_zone_lowest_possible_pfn[MAX_NR_ZONES];
  unsigned long __initdata arch_zone_highest_possible_pfn[MAX_NR_ZONES];
134 135 136 137
#ifdef CONFIG_MEMORY_HOTPLUG_RESERVE
  unsigned long __initdata node_boundary_start_pfn[MAX_NUMNODES];
  unsigned long __initdata node_boundary_end_pfn[MAX_NUMNODES];
#endif /* CONFIG_MEMORY_HOTPLUG_RESERVE */
138 139
#endif /* CONFIG_ARCH_POPULATES_NODE_MAP */

Nick Piggin's avatar
Nick Piggin committed
140
#ifdef CONFIG_DEBUG_VM
141
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
142
{
143 144 145
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
146

147 148 149 150 151 152 153 154 155
	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;
156 157 158 159
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
Linus Torvalds's avatar
Linus Torvalds committed
160 161
#ifdef CONFIG_HOLES_IN_ZONE
	if (!pfn_valid(page_to_pfn(page)))
162
		return 0;
Linus Torvalds's avatar
Linus Torvalds committed
163 164
#endif
	if (zone != page_zone(page))
165 166 167 168 169 170 171 172 173 174
		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
175
		return 1;
176 177 178
	if (!page_is_consistent(zone, page))
		return 1;

Linus Torvalds's avatar
Linus Torvalds committed
179 180
	return 0;
}
Nick Piggin's avatar
Nick Piggin committed
181 182 183 184 185 186 187
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

188
static void bad_page(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
189
{
190
	printk(KERN_EMERG "Bad page state in process '%s'\n"
191 192 193
		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",
194 195 196
		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
197
	dump_stack();
198 199
	page->flags &= ~(1 << PG_lru	|
			1 << PG_private |
Linus Torvalds's avatar
Linus Torvalds committed
200 201 202
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
203 204
			1 << PG_reclaim |
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
205
			1 << PG_swapcache |
206 207
			1 << PG_writeback |
			1 << PG_buddy );
Linus Torvalds's avatar
Linus Torvalds committed
208 209 210
	set_page_count(page, 0);
	reset_page_mapcount(page);
	page->mapping = NULL;
211
	add_taint(TAINT_BAD_PAGE);
Linus Torvalds's avatar
Linus Torvalds committed
212 213 214 215 216 217 218 219 220 221 222 223
}

/*
 * 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).
 *
224 225 226
 * 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
227
 */
228 229 230 231 232 233

static void free_compound_page(struct page *page)
{
	__free_pages_ok(page, (unsigned long)page[1].lru.prev);
}

Linus Torvalds's avatar
Linus Torvalds committed
234 235 236 237 238
static void prep_compound_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;

239
	page[1].lru.next = (void *)free_compound_page;	/* set dtor */
240
	page[1].lru.prev = (void *)order;
Linus Torvalds's avatar
Linus Torvalds committed
241 242 243
	for (i = 0; i < nr_pages; i++) {
		struct page *p = page + i;

244
		__SetPageCompound(p);
245
		set_page_private(p, (unsigned long)page);
Linus Torvalds's avatar
Linus Torvalds committed
246 247 248 249 250 251 252 253
	}
}

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

254
	if (unlikely((unsigned long)page[1].lru.prev != order))
255
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
256 257 258 259

	for (i = 0; i < nr_pages; i++) {
		struct page *p = page + i;

260 261 262
		if (unlikely(!PageCompound(p) |
				(page_private(p) != (unsigned long)page)))
			bad_page(page);
263
		__ClearPageCompound(p);
Linus Torvalds's avatar
Linus Torvalds committed
264 265 266
	}
}

Nick Piggin's avatar
Nick Piggin committed
267 268 269 270
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

Nick Piggin's avatar
Nick Piggin committed
271
	VM_BUG_ON((gfp_flags & (__GFP_WAIT | __GFP_HIGHMEM)) == __GFP_HIGHMEM);
272 273 274 275
	/*
	 * 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
276
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
Nick Piggin's avatar
Nick Piggin committed
277 278 279 280
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

Linus Torvalds's avatar
Linus Torvalds committed
281 282 283 284 285
/*
 * function for dealing with page's order in buddy system.
 * zone->lock is already acquired when we use these.
 * So, we don't need atomic page->flags operations here.
 */
286 287
static inline unsigned long page_order(struct page *page)
{
288
	return page_private(page);
Linus Torvalds's avatar
Linus Torvalds committed
289 290
}

291 292
static inline void set_page_order(struct page *page, int order)
{
293
	set_page_private(page, order);
294
	__SetPageBuddy(page);
Linus Torvalds's avatar
Linus Torvalds committed
295 296 297 298
}

static inline void rmv_page_order(struct page *page)
{
299
	__ClearPageBuddy(page);
300
	set_page_private(page, 0);
Linus Torvalds's avatar
Linus Torvalds committed
301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317
}

/*
 * 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
318
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
Linus Torvalds's avatar
Linus Torvalds committed
319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
 */
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
337
 * (a) the buddy is not in a hole &&
338
 * (b) the buddy is in the buddy system &&
339 340
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
341 342 343
 *
 * 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
344
 *
345
 * For recording page's order, we use page_private(page).
Linus Torvalds's avatar
Linus Torvalds committed
346
 */
347 348
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
Linus Torvalds's avatar
Linus Torvalds committed
349
{
Nick Piggin's avatar
Nick Piggin committed
350
#ifdef CONFIG_HOLES_IN_ZONE
351
	if (!pfn_valid(page_to_pfn(buddy)))
Nick Piggin's avatar
Nick Piggin committed
352 353 354
		return 0;
#endif

355 356 357 358 359
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

	if (PageBuddy(buddy) && page_order(buddy) == order) {
		BUG_ON(page_count(buddy) != 0);
360
		return 1;
361
	}
362
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
363 364 365 366 367 368 369 370 371 372 373 374 375 376 377
}

/*
 * 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
378
 * free pages of length of (1 << order) and marked with PG_buddy. Page's
379
 * order is recorded in page_private(page) field.
Linus Torvalds's avatar
Linus Torvalds committed
380 381 382 383 384 385 386 387 388
 * 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
389
static inline void __free_one_page(struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
390 391 392 393 394
		struct zone *zone, unsigned int order)
{
	unsigned long page_idx;
	int order_size = 1 << order;

395
	if (unlikely(PageCompound(page)))
Linus Torvalds's avatar
Linus Torvalds committed
396 397 398 399
		destroy_compound_page(page, order);

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

Nick Piggin's avatar
Nick Piggin committed
400 401
	VM_BUG_ON(page_idx & (order_size - 1));
	VM_BUG_ON(bad_range(zone, page));
Linus Torvalds's avatar
Linus Torvalds committed
402 403 404 405 406 407 408 409

	zone->free_pages += order_size;
	while (order < MAX_ORDER-1) {
		unsigned long combined_idx;
		struct free_area *area;
		struct page *buddy;

		buddy = __page_find_buddy(page, page_idx, order);
410
		if (!page_is_buddy(page, buddy, order))
Linus Torvalds's avatar
Linus Torvalds committed
411
			break;		/* Move the buddy up one level. */
Nick Piggin's avatar
Nick Piggin committed
412

Linus Torvalds's avatar
Linus Torvalds committed
413 414 415 416
		list_del(&buddy->lru);
		area = zone->free_area + order;
		area->nr_free--;
		rmv_page_order(buddy);
Nick Piggin's avatar
Nick Piggin committed
417
		combined_idx = __find_combined_index(page_idx, order);
Linus Torvalds's avatar
Linus Torvalds committed
418 419 420 421 422 423 424 425 426
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
	list_add(&page->lru, &zone->free_area[order].free_list);
	zone->free_area[order].nr_free++;
}

427
static inline int free_pages_check(struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
428
{
Nick Piggin's avatar
Nick Piggin committed
429 430 431
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
Linus Torvalds's avatar
Linus Torvalds committed
432 433 434 435 436 437 438 439
		(page->flags & (
			1 << PG_lru	|
			1 << PG_private |
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_reclaim	|
			1 << PG_slab	|
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
440
			1 << PG_writeback |
441 442
			1 << PG_reserved |
			1 << PG_buddy ))))
443
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
444
	if (PageDirty(page))
Nick Piggin's avatar
Nick Piggin committed
445
		__ClearPageDirty(page);
446 447 448 449 450 451
	/*
	 * 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
452 453 454 455 456
}

/*
 * Frees a list of pages. 
 * Assumes all pages on list are in same zone, and of same order.
457
 * count is the number of pages to free.
Linus Torvalds's avatar
Linus Torvalds committed
458 459 460 461 462 463 464
 *
 * 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
465 466
static void free_pages_bulk(struct zone *zone, int count,
					struct list_head *list, int order)
Linus Torvalds's avatar
Linus Torvalds committed
467
{
Nick Piggin's avatar
Nick Piggin committed
468
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
469 470
	zone->all_unreclaimable = 0;
	zone->pages_scanned = 0;
Nick Piggin's avatar
Nick Piggin committed
471 472 473
	while (count--) {
		struct page *page;

Nick Piggin's avatar
Nick Piggin committed
474
		VM_BUG_ON(list_empty(list));
Linus Torvalds's avatar
Linus Torvalds committed
475
		page = list_entry(list->prev, struct page, lru);
Nick Piggin's avatar
Nick Piggin committed
476
		/* have to delete it as __free_one_page list manipulates */
Linus Torvalds's avatar
Linus Torvalds committed
477
		list_del(&page->lru);
Nick Piggin's avatar
Nick Piggin committed
478
		__free_one_page(page, zone, order);
Linus Torvalds's avatar
Linus Torvalds committed
479
	}
Nick Piggin's avatar
Nick Piggin committed
480
	spin_unlock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
481 482
}

Nick Piggin's avatar
Nick Piggin committed
483
static void free_one_page(struct zone *zone, struct page *page, int order)
Linus Torvalds's avatar
Linus Torvalds committed
484
{
485 486 487 488 489
	spin_lock(&zone->lock);
	zone->all_unreclaimable = 0;
	zone->pages_scanned = 0;
	__free_one_page(page, zone ,order);
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
490 491 492 493 494
}

static void __free_pages_ok(struct page *page, unsigned int order)
{
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
495
	int i;
496
	int reserved = 0;
Linus Torvalds's avatar
Linus Torvalds committed
497 498

	for (i = 0 ; i < (1 << order) ; ++i)
499
		reserved += free_pages_check(page + i);
500 501 502
	if (reserved)
		return;

Nick Piggin's avatar
Nick Piggin committed
503 504
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
Nick Piggin's avatar
Nick Piggin committed
505
	arch_free_page(page, order);
Nick Piggin's avatar
Nick Piggin committed
506
	kernel_map_pages(page, 1 << order, 0);
Nick Piggin's avatar
Nick Piggin committed
507

Nick Piggin's avatar
Nick Piggin committed
508
	local_irq_save(flags);
509
	__count_vm_events(PGFREE, 1 << order);
Nick Piggin's avatar
Nick Piggin committed
510
	free_one_page(page_zone(page), page, order);
Nick Piggin's avatar
Nick Piggin committed
511
	local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
512 513
}

514 515 516 517 518 519 520 521
/*
 * 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);
522
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
523
		__free_page(page);
524 525 526
	} else {
		int loop;

Nick Piggin's avatar
Nick Piggin committed
527
		prefetchw(page);
528 529 530
		for (loop = 0; loop < BITS_PER_LONG; loop++) {
			struct page *p = &page[loop];

Nick Piggin's avatar
Nick Piggin committed
531 532
			if (loop + 1 < BITS_PER_LONG)
				prefetchw(p + 1);
533 534 535 536
			__ClearPageReserved(p);
			set_page_count(p, 0);
		}

537
		set_page_refcounted(page);
Nick Piggin's avatar
Nick Piggin committed
538
		__free_pages(page, order);
539 540 541
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
542 543 544 545 546 547 548 549 550 551 552 553 554 555 556

/*
 * 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
557
static inline void expand(struct zone *zone, struct page *page,
Linus Torvalds's avatar
Linus Torvalds committed
558 559 560 561 562 563 564 565
 	int low, int high, struct free_area *area)
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
Nick Piggin's avatar
Nick Piggin committed
566
		VM_BUG_ON(bad_range(zone, &page[size]));
Linus Torvalds's avatar
Linus Torvalds committed
567 568 569 570 571 572 573 574 575
		list_add(&page[size].lru, &area->free_list);
		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
576
static int prep_new_page(struct page *page, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
577
{
Nick Piggin's avatar
Nick Piggin committed
578 579 580
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
		(page_count(page) != 0)  |
581 582
		(page->flags & (
			1 << PG_lru	|
Linus Torvalds's avatar
Linus Torvalds committed
583 584 585 586 587
			1 << PG_private	|
			1 << PG_locked	|
			1 << PG_active	|
			1 << PG_dirty	|
			1 << PG_reclaim	|
588
			1 << PG_slab    |
Linus Torvalds's avatar
Linus Torvalds committed
589
			1 << PG_swapcache |
Nick Piggin's avatar
Nick Piggin committed
590
			1 << PG_writeback |
591 592
			1 << PG_reserved |
			1 << PG_buddy ))))
593
		bad_page(page);
Linus Torvalds's avatar
Linus Torvalds committed
594

595 596 597 598 599 600 601
	/*
	 * 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;

Linus Torvalds's avatar
Linus Torvalds committed
602 603 604
	page->flags &= ~(1 << PG_uptodate | 1 << PG_error |
			1 << PG_referenced | 1 << PG_arch_1 |
			1 << PG_checked | 1 << PG_mappedtodisk);
605
	set_page_private(page, 0);
606
	set_page_refcounted(page);
Linus Torvalds's avatar
Linus Torvalds committed
607
	kernel_map_pages(page, 1 << order, 1);
Nick Piggin's avatar
Nick Piggin committed
608 609 610 611 612 613 614

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

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

615
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637
}

/* 
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
static struct page *__rmqueue(struct zone *zone, unsigned int order)
{
	struct free_area * area;
	unsigned int current_order;
	struct page *page;

	for (current_order = order; current_order < MAX_ORDER; ++current_order) {
		area = zone->free_area + current_order;
		if (list_empty(&area->free_list))
			continue;

		page = list_entry(area->free_list.next, struct page, lru);
		list_del(&page->lru);
		rmv_page_order(page);
		area->nr_free--;
		zone->free_pages -= 1UL << order;
Nick Piggin's avatar
Nick Piggin committed
638 639
		expand(zone, page, order, current_order, area);
		return page;
Linus Torvalds's avatar
Linus Torvalds committed
640 641 642 643 644 645 646 647 648 649 650 651 652 653 654
	}

	return NULL;
}

/* 
 * 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, 
			unsigned long count, struct list_head *list)
{
	int i;
	
Nick Piggin's avatar
Nick Piggin committed
655
	spin_lock(&zone->lock);
Linus Torvalds's avatar
Linus Torvalds committed
656
	for (i = 0; i < count; ++i) {
Nick Piggin's avatar
Nick Piggin committed
657 658
		struct page *page = __rmqueue(zone, order);
		if (unlikely(page == NULL))
Linus Torvalds's avatar
Linus Torvalds committed
659 660 661
			break;
		list_add_tail(&page->lru, list);
	}
Nick Piggin's avatar
Nick Piggin committed
662
	spin_unlock(&zone->lock);
Nick Piggin's avatar
Nick Piggin committed
663
	return i;
Linus Torvalds's avatar
Linus Torvalds committed
664 665
}

666
#ifdef CONFIG_NUMA
667 668
/*
 * Called from the slab reaper to drain pagesets on a particular node that
669
 * belongs to the currently executing processor.
670 671
 * Note that this function must be called with the thread pinned to
 * a single processor.
672 673
 */
void drain_node_pages(int nodeid)
674
{
675 676
	int i;
	enum zone_type z;
677 678
	unsigned long flags;

679 680
	for (z = 0; z < MAX_NR_ZONES; z++) {
		struct zone *zone = NODE_DATA(nodeid)->node_zones + z;
681 682
		struct per_cpu_pageset *pset;

683 684 685
		if (!populated_zone(zone))
			continue;

Nick Piggin's avatar
Nick Piggin committed
686
		pset = zone_pcp(zone, smp_processor_id());
687 688 689 690
		for (i = 0; i < ARRAY_SIZE(pset->pcp); i++) {
			struct per_cpu_pages *pcp;

			pcp = &pset->pcp[i];
691 692 693 694 695 696
			if (pcp->count) {
				local_irq_save(flags);
				free_pages_bulk(zone, pcp->count, &pcp->list, 0);
				pcp->count = 0;
				local_irq_restore(flags);
			}
697 698 699 700 701
		}
	}
}
#endif

Linus Torvalds's avatar
Linus Torvalds committed
702 703 704
#if defined(CONFIG_PM) || defined(CONFIG_HOTPLUG_CPU)
static void __drain_pages(unsigned int cpu)
{
Nick Piggin's avatar
Nick Piggin committed
705
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
706 707 708 709 710 711
	struct zone *zone;
	int i;

	for_each_zone(zone) {
		struct per_cpu_pageset *pset;

712
		pset = zone_pcp(zone, cpu);
Linus Torvalds's avatar
Linus Torvalds committed
713 714 715 716
		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
717
			local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
718 719
			free_pages_bulk(zone, pcp->count, &pcp->list, 0);
			pcp->count = 0;
Nick Piggin's avatar
Nick Piggin committed
720
			local_irq_restore(flags);
Linus Torvalds's avatar
Linus Torvalds committed
721 722 723 724 725 726 727 728 729
		}
	}
}
#endif /* CONFIG_PM || CONFIG_HOTPLUG_CPU */

#ifdef CONFIG_PM

void mark_free_pages(struct zone *zone)
{
730 731
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
Linus Torvalds's avatar
Linus Torvalds committed
732 733 734 735 736 737 738
	int order;
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
739 740 741 742 743 744 745 746 747

	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);

			if (!PageNosave(page))
				ClearPageNosaveFree(page);
		}
Linus Torvalds's avatar
Linus Torvalds committed
748 749 750

	for (order = MAX_ORDER - 1; order >= 0; --order)
		list_for_each(curr, &zone->free_area[order].free_list) {
751
			unsigned long i;
Linus Torvalds's avatar
Linus Torvalds committed
752

753 754 755 756
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
				SetPageNosaveFree(pfn_to_page(pfn + i));
		}
Linus Torvalds's avatar
Linus Torvalds committed
757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784

	spin_unlock_irqrestore(&zone->lock, flags);
}

/*
 * 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);	
}
#endif /* CONFIG_PM */

/*
 * 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;
785
	if (free_pages_check(page))
786 787
		return;

Nick Piggin's avatar
Nick Piggin committed
788 789
	if (!PageHighMem(page))
		debug_check_no_locks_freed(page_address(page), PAGE_SIZE);
Nick Piggin's avatar
Nick Piggin committed
790
	arch_free_page(page, 0);
791 792
	kernel_map_pages(page, 1, 0);

793
	pcp = &zone_pcp(zone, get_cpu())->pcp[cold];
Linus Torvalds's avatar
Linus Torvalds committed
794
	local_irq_save(flags);
795
	__count_vm_event(PGFREE);
Linus Torvalds's avatar
Linus Torvalds committed
796 797
	list_add(&page->lru, &pcp->list);
	pcp->count++;
Nick Piggin's avatar
Nick Piggin committed
798 799 800 801
	if (pcp->count >= pcp->high) {
		free_pages_bulk(zone, pcp->batch, &pcp->list, 0);
		pcp->count -= pcp->batch;
	}
Linus Torvalds's avatar
Linus Torvalds committed
802 803 804 805 806 807 808 809 810 811 812 813 814 815
	local_irq_restore(flags);
	put_cpu();
}

void fastcall free_hot_page(struct page *page)
{
	free_hot_cold_page(page, 0);
}
	
void fastcall free_cold_page(struct page *page)
{
	free_hot_cold_page(page, 1);
}

816 817 818 819 820 821 822 823 824 825 826 827
/*
 * split_page takes a non-compound higher-order page, and splits it into
 * n (1<<order) sub-pages: page[0..n]
 * Each sub-page must be freed individually.
 *
 * Note: this is probably too low level an operation for use in drivers.
 * Please consult with lkml before using this in your driver.
 */
void split_page(struct page *page, unsigned int order)
{
	int i;

Nick Piggin's avatar
Nick Piggin committed
828 829
	VM_BUG_ON(PageCompound(page));
	VM_BUG_ON(!page_count(page));
830 831
	for (i = 1; i < (1 << order); i++)
		set_page_refcounted(page + i);
832 833
}

Linus Torvalds's avatar
Linus Torvalds committed
834 835 836 837 838
/*
 * Really, prep_compound_page() should be called from __rmqueue_bulk().  But
 * we cheat by calling it from here, in the order > 0 path.  Saves a branch
 * or two.
 */
Nick Piggin's avatar
Nick Piggin committed
839 840
static struct page *buffered_rmqueue(struct zonelist *zonelist,
			struct zone *zone, int order, gfp_t gfp_flags)
Linus Torvalds's avatar
Linus Torvalds committed
841 842
{
	unsigned long flags;
843
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
844
	int cold = !!(gfp_flags & __GFP_COLD);
Nick Piggin's avatar
Nick Piggin committed
845
	int cpu;
Linus Torvalds's avatar
Linus Torvalds committed
846

847
again:
Nick Piggin's avatar
Nick Piggin committed
848
	cpu  = get_cpu();
Nick Piggin's avatar
Nick Piggin committed
849
	if (likely(order == 0)) {
Linus Torvalds's avatar
Linus Torvalds committed
850 851
		struct per_cpu_pages *pcp;

Nick Piggin's avatar
Nick Piggin committed
852
		pcp = &zone_pcp(zone, cpu)->pcp[cold];
Linus Torvalds's avatar
Linus Torvalds committed
853
		local_irq_save(flags);
Nick Piggin's avatar
Nick Piggin committed
854
		if (!pcp->count) {
Linus Torvalds's avatar
Linus Torvalds committed
855 856
			pcp->count += rmqueue_bulk(zone, 0,
						pcp->batch, &pcp->list);
Nick Piggin's avatar
Nick Piggin committed
857 858
			if (unlikely(!pcp->count))
				goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
859
		}
Nick Piggin's avatar
Nick Piggin committed
860 861 862
		page = list_entry(pcp->list.next, struct page, lru);
		list_del(&page->lru);
		pcp->count--;
Rohit Seth's avatar
Rohit Seth committed
863
	} else {
Linus Torvalds's avatar
Linus Torvalds committed
864 865
		spin_lock_irqsave(&zone->lock, flags);
		page = __rmqueue(zone, order);
Nick Piggin's avatar
Nick Piggin committed
866 867 868
		spin_unlock(&zone->lock);
		if (!page)
			goto failed;
Linus Torvalds's avatar
Linus Torvalds committed
869 870
	}

871
	__count_zone_vm_events(PGALLOC, zone, 1 << order);
872
	zone_statistics(zonelist, zone);
Nick Piggin's avatar
Nick Piggin committed
873 874
	local_irq_restore(flags);
	put_cpu();
Linus Torvalds's avatar
Linus Torvalds committed
875

Nick Piggin's avatar
Nick Piggin committed
876
	VM_BUG_ON(bad_range(zone, page));
Nick Piggin's avatar
Nick Piggin committed
877
	if (prep_new_page(page, order, gfp_flags))
Nick Piggin's avatar
Nick Piggin committed
878
		goto again;
Linus Torvalds's avatar
Linus Torvalds committed
879
	return page;
Nick Piggin's avatar
Nick Piggin committed
880 881 882 883 884

failed:
	local_irq_restore(flags);
	put_cpu();
	return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
885 886
}

Rohit Seth's avatar
Rohit Seth committed
887
#define ALLOC_NO_WATERMARKS	0x01 /* don't check watermarks at all */
888 889 890 891 892 893
#define ALLOC_WMARK_MIN		0x02 /* use pages_min watermark */
#define ALLOC_WMARK_LOW		0x04 /* use pages_low watermark */
#define ALLOC_WMARK_HIGH	0x08 /* use pages_high watermark */
#define ALLOC_HARDER		0x10 /* try to alloc harder */
#define ALLOC_HIGH		0x20 /* __GFP_HIGH set */
#define ALLOC_CPUSET		0x40 /* check for correct cpuset */
Rohit Seth's avatar
Rohit Seth committed
894

Linus Torvalds's avatar
Linus Torvalds committed
895 896 897 898 899
/*
 * Return 1 if free pages are above 'mark'. This takes into account the order
 * of the allocation.
 */
int zone_watermark_ok(struct zone *z, int order, unsigned long mark,
Rohit Seth's avatar
Rohit Seth committed
900
		      int classzone_idx, int alloc_flags)
Linus Torvalds's avatar
Linus Torvalds committed
901 902
{
	/* free_pages my go negative - that's OK */
903 904
	unsigned long min = mark;
	long free_pages = z->free_pages - (1 << order) + 1;
Linus Torvalds's avatar
Linus Torvalds committed
905 906
	int o;

Rohit Seth's avatar
Rohit Seth committed
907
	if (alloc_flags & ALLOC_HIGH)
Linus Torvalds's avatar
Linus Torvalds committed
908
		min -= min / 2;
Rohit Seth's avatar
Rohit Seth committed
909
	if (alloc_flags & ALLOC_HARDER)
Linus Torvalds's avatar
Linus Torvalds committed
910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926
		min -= min / 4;

	if (free_pages <= min + z->lowmem_reserve[classzone_idx])
		return 0;
	for (o = 0; o < order; o++) {
		/* At the next order, this order's pages become unavailable */
		free_pages -= z->free_area[o].nr_free << o;

		/* Require fewer higher order pages to be free */
		min >>= 1;

		if (free_pages <= min)
			return 0;
	}
	return 1;
}

Rohit Seth's avatar
Rohit Seth committed
927 928 929 930 931 932 933
/*
 * get_page_from_freeliest goes through the zonelist trying to allocate
 * a page.
 */
static struct page *
get_page_from_freelist(gfp_t gfp_mask, unsigned int order,
		struct zonelist *zonelist, int alloc_flags)
Martin Hicks's avatar
Martin Hicks committed
934
{
Rohit Seth's avatar
Rohit Seth committed
935 936 937
	struct zone **z = zonelist->zones;
	struct page *page = NULL;
	int classzone_idx = zone_idx(*z);
938
	struct zone *zone;
Rohit Seth's avatar
Rohit Seth committed
939 940 941 942 943 944

	/*
	 * Go through the zonelist once, looking for a zone with enough free.
	 * See also cpuset_zone_allowed() comment in kernel/cpuset.c.
	 */
	do {
945
		zone = *z;
946
		if (unlikely(NUMA_BUILD && (gfp_mask & __GFP_THISNODE) &&
947
			zone->zone_pgdat != zonelist->zones[0]->zone_pgdat))
948
				break;
Rohit Seth's avatar
Rohit Seth committed
949
		if ((alloc_flags & ALLOC_CPUSET) &&
950
				!cpuset_zone_allowed(zone, gfp_mask))
Rohit Seth's avatar
Rohit Seth committed
951 952 953
			continue;

		if (!(alloc_flags & ALLOC_NO_WATERMARKS)) {
954 955
			unsigned long mark;
			if (alloc_flags & ALLOC_WMARK_MIN)
956
				mark = zone->pages_min;
957
			else if (alloc_flags & ALLOC_WMARK_LOW)
958
				mark = zone->pages_low;
959
			else
960 961
				mark = zone->pages_high;
			if (!zone_watermark_ok(zone , order, mark,
Rohit Seth's avatar
Rohit Seth committed
962
				    classzone_idx, alloc_flags))
963
				if (!zone_reclaim_mode ||
964
				    !zone_reclaim(zone, gfp_mask, order))
965
					continue;
Rohit Seth's avatar
Rohit Seth committed
966 967
		}

968
		page = buffered_rmqueue(zonelist, zone, order, gfp_mask);
Rohit Seth's avatar
Rohit Seth committed
969 970 971 972 973
		if (page) {
			break;
		}
	} while (*(++z) != NULL);
	return page;
Martin Hicks's avatar
Martin Hicks committed
974 975
}

Linus Torvalds's avatar
Linus Torvalds committed
976 977 978 979
/*
 * This is the 'heart' of the zoned buddy allocator.
 */
struct page * fastcall
980
__alloc_pages(gfp_t gfp_mask, unsigned int order,
Linus Torvalds's avatar
Linus Torvalds committed
981 982
		struct zonelist *zonelist)
{
Al Viro's avatar
Al Viro committed
983
	const gfp_t wait = gfp_mask & __GFP_WAIT;
Rohit Seth's avatar
Rohit Seth committed
984
	struct zone **z;
Linus Torvalds's avatar
Linus Torvalds committed
985 986 987 988
	struct page *page;
	struct reclaim_state reclaim_state;
	struct task_struct *p = current;
	int do_retry;
Rohit Seth's avatar
Rohit Seth committed
989
	int alloc_flags;
Linus Torvalds's avatar
Linus Torvalds committed
990 991 992 993
	int did_some_progress;

	might_sleep_if(wait);

994
restart:
Rohit Seth's avatar
Rohit Seth committed
995
	z = zonelist->zones;  /* the list of zones suitable for gfp_mask */
Linus Torvalds's avatar
Linus Torvalds committed
996

Rohit Seth's avatar
Rohit Seth committed
997
	if (unlikely(*z == NULL)) {
Linus Torvalds's avatar
Linus Torvalds committed
998 999 1000
		/* Should this ever happen?? */
		return NULL;
	}
1001

Rohit Seth's avatar
Rohit Seth committed
1002
	page = get_page_from_freelist(gfp_mask|__GFP_HARDWALL, order,
1003
				zonelist, ALLOC_WMARK_LOW|ALLOC_CPUSET);
Rohit Seth's avatar
Rohit Seth committed
1004 1005
	if (page)
		goto got_pg;
Linus Torvalds's avatar
Linus Torvalds committed
1006

1007
	do {
1008
		wakeup_kswapd(*z, order);
1009
	} while (*(++z));
Linus Torvalds's avatar
Linus Torvalds committed
1010

1011
	/*
Rohit Seth's avatar
Rohit Seth committed
1012 1013 1014 1015 1016 1017
	 * OK, we're below the kswapd watermark and have kicked background
	 * reclaim. Now things get more complex, so set up alloc_flags according
	 * to how we want to proceed.
	 *
	 * The caller may dip into page reserves a bit more if the caller
	 * cannot run direct reclaim, or if the caller has realtime scheduling
Paul Jackson's avatar
Paul Jackson committed
1018 1019
	 * policy or is asking for __GFP_HIGH memory.  GFP_ATOMIC requests will
	 * set both ALLOC_HARDER (!wait) and ALLOC_HIGH (__GFP_HIGH).
1020
	 */
1021
	alloc_flags = ALLOC_WMARK_MIN;
Rohit Seth's avatar
Rohit Seth committed
1022 1023 1024 1025
	if ((unlikely(rt_task(p)) && !in_interrupt()) || !wait)
		alloc_flags |= ALLOC_HARDER;
	if (gfp_mask & __GFP_HIGH)
		alloc_flags |= ALLOC_HIGH;
1026 1027
	if (wait)
		alloc_flags |= ALLOC_CPUSET;
Linus Torvalds's avatar
Linus Torvalds committed
1028 1029 1030

	/*
	 * Go through the zonelist again. Let __GFP_HIGH and allocations
Rohit Seth's avatar
Rohit Seth committed
1031
	 * coming from realtime tasks go deeper into reserves.