page_alloc.c 166.2 KB
Newer Older
L
Linus Torvalds 已提交
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>
L
Linus Torvalds 已提交
23
#include <linux/bootmem.h>
24
#include <linux/memblock.h>
L
Linus Torvalds 已提交
25
#include <linux/compiler.h>
26
#include <linux/kernel.h>
27
#include <linux/kmemcheck.h>
L
Linus Torvalds 已提交
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/ratelimit.h>
34
#include <linux/oom.h>
L
Linus Torvalds 已提交
35 36 37 38 39
#include <linux/notifier.h>
#include <linux/topology.h>
#include <linux/sysctl.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
40
#include <linux/memory_hotplug.h>
L
Linus Torvalds 已提交
41 42
#include <linux/nodemask.h>
#include <linux/vmalloc.h>
43
#include <linux/vmstat.h>
44
#include <linux/mempolicy.h>
45
#include <linux/stop_machine.h>
46 47
#include <linux/sort.h>
#include <linux/pfn.h>
48
#include <linux/backing-dev.h>
49
#include <linux/fault-inject.h>
K
KAMEZAWA Hiroyuki 已提交
50
#include <linux/page-isolation.h>
51
#include <linux/page_cgroup.h>
52
#include <linux/debugobjects.h>
53
#include <linux/kmemleak.h>
54
#include <linux/compaction.h>
55
#include <trace/events/kmem.h>
56
#include <linux/ftrace_event.h>
57
#include <linux/memcontrol.h>
58
#include <linux/prefetch.h>
59
#include <linux/migrate.h>
60
#include <linux/page-debug-flags.h>
L
Linus Torvalds 已提交
61 62

#include <asm/tlbflush.h>
63
#include <asm/div64.h>
L
Linus Torvalds 已提交
64 65
#include "internal.h"

66 67 68 69 70
#ifdef CONFIG_USE_PERCPU_NUMA_NODE_ID
DEFINE_PER_CPU(int, numa_node);
EXPORT_PER_CPU_SYMBOL(numa_node);
#endif

71 72 73 74 75 76 77 78 79 80 81
#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

L
Linus Torvalds 已提交
82
/*
83
 * Array of node states.
L
Linus Torvalds 已提交
84
 */
85 86 87 88 89 90 91 92 93 94 95 96 97
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);

98
unsigned long totalram_pages __read_mostly;
99
unsigned long totalreserve_pages __read_mostly;
100 101 102 103 104 105 106 107
/*
 * When calculating the number of globally allowed dirty pages, there
 * is a certain number of per-zone reserves that should not be
 * considered dirtyable memory.  This is the sum of those reserves
 * over all existing zones that contribute dirtyable memory.
 */
unsigned long dirty_balance_reserve __read_mostly;

108
int percpu_pagelist_fraction;
109
gfp_t gfp_allowed_mask __read_mostly = GFP_BOOT_MASK;
L
Linus Torvalds 已提交
110

111 112 113 114 115 116 117 118 119
#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).
 */
120 121 122 123

static gfp_t saved_gfp_mask;

void pm_restore_gfp_mask(void)
124 125
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
126 127 128 129
	if (saved_gfp_mask) {
		gfp_allowed_mask = saved_gfp_mask;
		saved_gfp_mask = 0;
	}
130 131
}

132
void pm_restrict_gfp_mask(void)
133 134
{
	WARN_ON(!mutex_is_locked(&pm_mutex));
135 136 137
	WARN_ON(saved_gfp_mask);
	saved_gfp_mask = gfp_allowed_mask;
	gfp_allowed_mask &= ~GFP_IOFS;
138
}
139 140 141 142 143 144 145

bool pm_suspended_storage(void)
{
	if ((gfp_allowed_mask & GFP_IOFS) == GFP_IOFS)
		return false;
	return true;
}
146 147
#endif /* CONFIG_PM_SLEEP */

148 149 150 151
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
int pageblock_order __read_mostly;
#endif

152
static void __free_pages_ok(struct page *page, unsigned int order);
153

L
Linus Torvalds 已提交
154 155 156 157 158 159 160
/*
 * 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
A
Andi Kleen 已提交
161 162 163
 *
 * TBD: should special case ZONE_DMA32 machines here - in those we normally
 * don't need any ZONE_NORMAL reservation
L
Linus Torvalds 已提交
164
 */
165
int sysctl_lowmem_reserve_ratio[MAX_NR_ZONES-1] = {
166
#ifdef CONFIG_ZONE_DMA
167
	 256,
168
#endif
169
#ifdef CONFIG_ZONE_DMA32
170
	 256,
171
#endif
172
#ifdef CONFIG_HIGHMEM
M
Mel Gorman 已提交
173
	 32,
174
#endif
M
Mel Gorman 已提交
175
	 32,
176
};
L
Linus Torvalds 已提交
177 178 179

EXPORT_SYMBOL(totalram_pages);

180
static char * const zone_names[MAX_NR_ZONES] = {
181
#ifdef CONFIG_ZONE_DMA
182
	 "DMA",
183
#endif
184
#ifdef CONFIG_ZONE_DMA32
185
	 "DMA32",
186
#endif
187
	 "Normal",
188
#ifdef CONFIG_HIGHMEM
M
Mel Gorman 已提交
189
	 "HighMem",
190
#endif
M
Mel Gorman 已提交
191
	 "Movable",
192 193
};

L
Linus Torvalds 已提交
194 195
int min_free_kbytes = 1024;

196 197
static unsigned long __meminitdata nr_kernel_pages;
static unsigned long __meminitdata nr_all_pages;
198
static unsigned long __meminitdata dma_reserve;
L
Linus Torvalds 已提交
199

T
Tejun Heo 已提交
200 201 202 203 204 205 206 207 208 209 210
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
static unsigned long __meminitdata arch_zone_lowest_possible_pfn[MAX_NR_ZONES];
static unsigned long __meminitdata arch_zone_highest_possible_pfn[MAX_NR_ZONES];
static unsigned long __initdata required_kernelcore;
static unsigned long __initdata required_movablecore;
static unsigned long __meminitdata zone_movable_pfn[MAX_NUMNODES];

/* movable_zone is the "real" zone pages in ZONE_MOVABLE are taken from */
int movable_zone;
EXPORT_SYMBOL(movable_zone);
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
211

M
Miklos Szeredi 已提交
212 213
#if MAX_NUMNODES > 1
int nr_node_ids __read_mostly = MAX_NUMNODES;
214
int nr_online_nodes __read_mostly = 1;
M
Miklos Szeredi 已提交
215
EXPORT_SYMBOL(nr_node_ids);
216
EXPORT_SYMBOL(nr_online_nodes);
M
Miklos Szeredi 已提交
217 218
#endif

219 220
int page_group_by_mobility_disabled __read_mostly;

221 222 223 224 225
/*
 * NOTE:
 * Don't use set_pageblock_migratetype(page, MIGRATE_ISOLATE) directly.
 * Instead, use {un}set_pageblock_isolate.
 */
226
void set_pageblock_migratetype(struct page *page, int migratetype)
227
{
228 229 230 231

	if (unlikely(page_group_by_mobility_disabled))
		migratetype = MIGRATE_UNMOVABLE;

232 233 234 235
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

236 237
bool oom_killer_disabled __read_mostly;

N
Nick Piggin 已提交
238
#ifdef CONFIG_DEBUG_VM
239
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
L
Linus Torvalds 已提交
240
{
241 242 243
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
244

245 246 247 248 249 250 251 252 253
	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;
254 255 256 257
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
258
	if (!pfn_valid_within(page_to_pfn(page)))
259
		return 0;
L
Linus Torvalds 已提交
260
	if (zone != page_zone(page))
261 262 263 264 265 266 267 268 269 270
		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))
L
Linus Torvalds 已提交
271
		return 1;
272 273 274
	if (!page_is_consistent(zone, page))
		return 1;

L
Linus Torvalds 已提交
275 276
	return 0;
}
N
Nick Piggin 已提交
277 278 279 280 281 282 283
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

N
Nick Piggin 已提交
284
static void bad_page(struct page *page)
L
Linus Torvalds 已提交
285
{
286 287 288 289
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

290 291
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
292
		reset_page_mapcount(page); /* remove PageBuddy */
293 294 295
		return;
	}

296 297 298 299 300 301 302 303 304 305
	/*
	 * 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) {
306 307
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
308 309 310 311 312 313 314 315
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

316
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
317
		current->comm, page_to_pfn(page));
318
	dump_page(page);
319

320
	print_modules();
L
Linus Torvalds 已提交
321
	dump_stack();
322
out:
323
	/* Leave bad fields for debug, except PageBuddy could make trouble */
324
	reset_page_mapcount(page); /* remove PageBuddy */
325
	add_taint(TAINT_BAD_PAGE);
L
Linus Torvalds 已提交
326 327 328 329 330 331 332 333 334
}

/*
 * 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".
 *
335 336
 * All pages have PG_compound set.  All tail pages have their ->first_page
 * pointing at the head page.
L
Linus Torvalds 已提交
337
 *
338 339 340
 * 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.
L
Linus Torvalds 已提交
341
 */
342 343 344

static void free_compound_page(struct page *page)
{
345
	__free_pages_ok(page, compound_order(page));
346 347
}

A
Andi Kleen 已提交
348
void prep_compound_page(struct page *page, unsigned long order)
349 350 351 352 353 354 355 356 357 358
{
	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);
359
		set_page_count(p, 0);
360 361 362 363
		p->first_page = page;
	}
}

364
/* update __split_huge_page_refcount if you change this function */
365
static int destroy_compound_page(struct page *page, unsigned long order)
L
Linus Torvalds 已提交
366 367 368
{
	int i;
	int nr_pages = 1 << order;
369
	int bad = 0;
L
Linus Torvalds 已提交
370

371 372
	if (unlikely(compound_order(page) != order) ||
	    unlikely(!PageHead(page))) {
N
Nick Piggin 已提交
373
		bad_page(page);
374 375
		bad++;
	}
L
Linus Torvalds 已提交
376

377
	__ClearPageHead(page);
378

379 380
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
L
Linus Torvalds 已提交
381

382
		if (unlikely(!PageTail(p) || (p->first_page != page))) {
N
Nick Piggin 已提交
383
			bad_page(page);
384 385
			bad++;
		}
386
		__ClearPageTail(p);
L
Linus Torvalds 已提交
387
	}
388 389

	return bad;
L
Linus Torvalds 已提交
390 391
}

N
Nick Piggin 已提交
392 393 394 395
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

396 397 398 399
	/*
	 * clear_highpage() will use KM_USER0, so it's a bug to use __GFP_ZERO
	 * and __GFP_HIGHMEM from hard or soft interrupt context.
	 */
N
Nick Piggin 已提交
400
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
N
Nick Piggin 已提交
401 402 403 404
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435
#ifdef CONFIG_DEBUG_PAGEALLOC
unsigned int _debug_guardpage_minorder;

static int __init debug_guardpage_minorder_setup(char *buf)
{
	unsigned long res;

	if (kstrtoul(buf, 10, &res) < 0 ||  res > MAX_ORDER / 2) {
		printk(KERN_ERR "Bad debug_guardpage_minorder value\n");
		return 0;
	}
	_debug_guardpage_minorder = res;
	printk(KERN_INFO "Setting debug_guardpage_minorder to %lu\n", res);
	return 0;
}
__setup("debug_guardpage_minorder=", debug_guardpage_minorder_setup);

static inline void set_page_guard_flag(struct page *page)
{
	__set_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags);
}

static inline void clear_page_guard_flag(struct page *page)
{
	__clear_bit(PAGE_DEBUG_FLAG_GUARD, &page->debug_flags);
}
#else
static inline void set_page_guard_flag(struct page *page) { }
static inline void clear_page_guard_flag(struct page *page) { }
#endif

436 437
static inline void set_page_order(struct page *page, int order)
{
H
Hugh Dickins 已提交
438
	set_page_private(page, order);
439
	__SetPageBuddy(page);
L
Linus Torvalds 已提交
440 441 442 443
}

static inline void rmv_page_order(struct page *page)
{
444
	__ClearPageBuddy(page);
H
Hugh Dickins 已提交
445
	set_page_private(page, 0);
L
Linus Torvalds 已提交
446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462
}

/*
 * 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)
 *
A
Andreas Mohr 已提交
463
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
L
Linus Torvalds 已提交
464 465
 */
static inline unsigned long
466
__find_buddy_index(unsigned long page_idx, unsigned int order)
L
Linus Torvalds 已提交
467
{
468
	return page_idx ^ (1 << order);
L
Linus Torvalds 已提交
469 470 471 472 473
}

/*
 * This function checks whether a page is free && is the buddy
 * we can do coalesce a page and its buddy if
N
Nick Piggin 已提交
474
 * (a) the buddy is not in a hole &&
475
 * (b) the buddy is in the buddy system &&
476 477
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
478
 *
A
Andrea Arcangeli 已提交
479 480
 * For recording whether a page is in the buddy system, we set ->_mapcount -2.
 * Setting, clearing, and testing _mapcount -2 is serialized by zone->lock.
L
Linus Torvalds 已提交
481
 *
482
 * For recording page's order, we use page_private(page).
L
Linus Torvalds 已提交
483
 */
484 485
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
L
Linus Torvalds 已提交
486
{
487
	if (!pfn_valid_within(page_to_pfn(buddy)))
N
Nick Piggin 已提交
488 489
		return 0;

490 491 492
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

493 494 495 496 497
	if (page_is_guard(buddy) && page_order(buddy) == order) {
		VM_BUG_ON(page_count(buddy) != 0);
		return 1;
	}

498
	if (PageBuddy(buddy) && page_order(buddy) == order) {
499
		VM_BUG_ON(page_count(buddy) != 0);
500
		return 1;
501
	}
502
	return 0;
L
Linus Torvalds 已提交
503 504 505 506 507 508 509 510 511 512 513 514 515 516 517
}

/*
 * 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
A
Andrea Arcangeli 已提交
518
 * free pages of length of (1 << order) and marked with _mapcount -2. Page's
H
Hugh Dickins 已提交
519
 * order is recorded in page_private(page) field.
L
Linus Torvalds 已提交
520
 * So when we are allocating or freeing one, we can derive the state of the
521 522
 * other.  That is, if we allocate a small block, and both were
 * free, the remainder of the region must be split into blocks.
L
Linus Torvalds 已提交
523
 * If a block is freed, and its buddy is also free, then this
524
 * triggers coalescing into a block of larger size.
L
Linus Torvalds 已提交
525 526 527 528
 *
 * -- wli
 */

N
Nick Piggin 已提交
529
static inline void __free_one_page(struct page *page,
530 531
		struct zone *zone, unsigned int order,
		int migratetype)
L
Linus Torvalds 已提交
532 533
{
	unsigned long page_idx;
534
	unsigned long combined_idx;
535
	unsigned long uninitialized_var(buddy_idx);
536
	struct page *buddy;
L
Linus Torvalds 已提交
537

N
Nick Piggin 已提交
538
	if (unlikely(PageCompound(page)))
539 540
		if (unlikely(destroy_compound_page(page, order)))
			return;
L
Linus Torvalds 已提交
541

542 543
	VM_BUG_ON(migratetype == -1);

L
Linus Torvalds 已提交
544 545
	page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);

546
	VM_BUG_ON(page_idx & ((1 << order) - 1));
N
Nick Piggin 已提交
547
	VM_BUG_ON(bad_range(zone, page));
L
Linus Torvalds 已提交
548 549

	while (order < MAX_ORDER-1) {
550 551
		buddy_idx = __find_buddy_index(page_idx, order);
		buddy = page + (buddy_idx - page_idx);
552
		if (!page_is_buddy(page, buddy, order))
553
			break;
554 555 556 557 558 559 560 561 562 563 564 565 566
		/*
		 * Our buddy is free or it is CONFIG_DEBUG_PAGEALLOC guard page,
		 * merge with it and move up one order.
		 */
		if (page_is_guard(buddy)) {
			clear_page_guard_flag(buddy);
			set_page_private(page, 0);
			__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
		} else {
			list_del(&buddy->lru);
			zone->free_area[order].nr_free--;
			rmv_page_order(buddy);
		}
567
		combined_idx = buddy_idx & page_idx;
L
Linus Torvalds 已提交
568 569 570 571 572
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
573 574 575 576 577 578 579 580 581

	/*
	 * 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
	 */
582
	if ((order < MAX_ORDER-2) && pfn_valid_within(page_to_pfn(buddy))) {
583
		struct page *higher_page, *higher_buddy;
584 585 586 587
		combined_idx = buddy_idx & page_idx;
		higher_page = page + (combined_idx - page_idx);
		buddy_idx = __find_buddy_index(combined_idx, order + 1);
		higher_buddy = page + (buddy_idx - combined_idx);
588 589 590 591 592 593 594 595 596
		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:
L
Linus Torvalds 已提交
597 598 599
	zone->free_area[order].nr_free++;
}

600 601 602 603 604 605 606 607 608 609 610
/*
 * 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);
}

N
Nick Piggin 已提交
611
static inline int free_pages_check(struct page *page)
L
Linus Torvalds 已提交
612
{
N
Nick Piggin 已提交
613 614
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
615
		(atomic_read(&page->_count) != 0) |
616 617
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE) |
		(mem_cgroup_bad_page_check(page)))) {
N
Nick Piggin 已提交
618
		bad_page(page);
619
		return 1;
620
	}
621 622 623
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
L
Linus Torvalds 已提交
624 625 626
}

/*
627
 * Frees a number of pages from the PCP lists
L
Linus Torvalds 已提交
628
 * Assumes all pages on list are in same zone, and of same order.
629
 * count is the number of pages to free.
L
Linus Torvalds 已提交
630 631 632 633 634 635 636
 *
 * 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.
 */
637 638
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
L
Linus Torvalds 已提交
639
{
640
	int migratetype = 0;
641
	int batch_free = 0;
642
	int to_free = count;
643

N
Nick Piggin 已提交
644
	spin_lock(&zone->lock);
645
	zone->all_unreclaimable = 0;
L
Linus Torvalds 已提交
646
	zone->pages_scanned = 0;
647

648
	while (to_free) {
N
Nick Piggin 已提交
649
		struct page *page;
650 651 652
		struct list_head *list;

		/*
653 654 655 656 657
		 * 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
658 659
		 */
		do {
660
			batch_free++;
661 662 663 664
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
N
Nick Piggin 已提交
665

666 667 668 669
		/* This is the only non-empty list. Free them all. */
		if (batch_free == MIGRATE_PCPTYPES)
			batch_free = to_free;

670 671 672 673
		do {
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
674 675 676
			/* 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));
677
		} while (--to_free && --batch_free && !list_empty(list));
L
Linus Torvalds 已提交
678
	}
679
	__mod_zone_page_state(zone, NR_FREE_PAGES, count);
N
Nick Piggin 已提交
680
	spin_unlock(&zone->lock);
L
Linus Torvalds 已提交
681 682
}

683 684
static void free_one_page(struct zone *zone, struct page *page, int order,
				int migratetype)
L
Linus Torvalds 已提交
685
{
686
	spin_lock(&zone->lock);
687
	zone->all_unreclaimable = 0;
688
	zone->pages_scanned = 0;
689

690
	__free_one_page(page, zone, order, migratetype);
691
	__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
692
	spin_unlock(&zone->lock);
N
Nick Piggin 已提交
693 694
}

695
static bool free_pages_prepare(struct page *page, unsigned int order)
N
Nick Piggin 已提交
696
{
L
Linus Torvalds 已提交
697
	int i;
698
	int bad = 0;
L
Linus Torvalds 已提交
699

700
	trace_mm_page_free(page, order);
701 702
	kmemcheck_free_shadow(page, order);

A
Andrea Arcangeli 已提交
703 704 705 706
	if (PageAnon(page))
		page->mapping = NULL;
	for (i = 0; i < (1 << order); i++)
		bad += free_pages_check(page + i);
707
	if (bad)
708
		return false;
709

710
	if (!PageHighMem(page)) {
N
Nick Piggin 已提交
711
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
712 713 714
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
N
Nick Piggin 已提交
715
	arch_free_page(page, order);
N
Nick Piggin 已提交
716
	kernel_map_pages(page, 1 << order, 0);
N
Nick Piggin 已提交
717

718 719 720 721 722 723 724 725 726 727 728
	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;

N
Nick Piggin 已提交
729
	local_irq_save(flags);
730
	if (unlikely(wasMlocked))
731
		free_page_mlock(page);
732
	__count_vm_events(PGFREE, 1 << order);
733 734
	free_one_page(page_zone(page), page, order,
					get_pageblock_migratetype(page));
N
Nick Piggin 已提交
735
	local_irq_restore(flags);
L
Linus Torvalds 已提交
736 737
}

738
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
739
{
740 741
	unsigned int nr_pages = 1 << order;
	unsigned int loop;
742

743 744 745 746 747 748 749 750
	prefetchw(page);
	for (loop = 0; loop < nr_pages; loop++) {
		struct page *p = &page[loop];

		if (loop + 1 < nr_pages)
			prefetchw(p + 1);
		__ClearPageReserved(p);
		set_page_count(p, 0);
751
	}
752 753 754

	set_page_refcounted(page);
	__free_pages(page, order);
755 756
}

757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774
#ifdef CONFIG_CMA
/* Free whole pageblock and set it's migration type to MIGRATE_CMA. */
void __init init_cma_reserved_pageblock(struct page *page)
{
	unsigned i = pageblock_nr_pages;
	struct page *p = page;

	do {
		__ClearPageReserved(p);
		set_page_count(p, 0);
	} while (++p, --i);

	set_page_refcounted(page);
	set_pageblock_migratetype(page, MIGRATE_CMA);
	__free_pages(page, pageblock_order);
	totalram_pages += pageblock_nr_pages;
}
#endif
L
Linus Torvalds 已提交
775 776 777 778 779 780 781 782 783 784 785 786 787 788 789

/*
 * 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
 */
N
Nick Piggin 已提交
790
static inline void expand(struct zone *zone, struct page *page,
791 792
	int low, int high, struct free_area *area,
	int migratetype)
L
Linus Torvalds 已提交
793 794 795 796 797 798 799
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
N
Nick Piggin 已提交
800
		VM_BUG_ON(bad_range(zone, &page[size]));
801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817

#ifdef CONFIG_DEBUG_PAGEALLOC
		if (high < debug_guardpage_minorder()) {
			/*
			 * Mark as guard pages (or page), that will allow to
			 * merge back to allocator when buddy will be freed.
			 * Corresponding page table entries will not be touched,
			 * pages will stay not present in virtual address space
			 */
			INIT_LIST_HEAD(&page[size].lru);
			set_page_guard_flag(&page[size]);
			set_page_private(&page[size], high);
			/* Guard pages are not available for any usage */
			__mod_zone_page_state(zone, NR_FREE_PAGES, -(1 << high));
			continue;
		}
#endif
818
		list_add(&page[size].lru, &area->free_list[migratetype]);
L
Linus Torvalds 已提交
819 820 821 822 823 824 825 826
		area->nr_free++;
		set_page_order(&page[size], high);
	}
}

/*
 * This page is about to be returned from the page allocator
 */
827
static inline int check_new_page(struct page *page)
L
Linus Torvalds 已提交
828
{
N
Nick Piggin 已提交
829 830
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
831
		(atomic_read(&page->_count) != 0)  |
832 833
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP) |
		(mem_cgroup_bad_page_check(page)))) {
N
Nick Piggin 已提交
834
		bad_page(page);
835
		return 1;
836
	}
837 838 839 840 841 842 843 844 845 846 847 848
	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;
	}
849

H
Hugh Dickins 已提交
850
	set_page_private(page, 0);
851
	set_page_refcounted(page);
N
Nick Piggin 已提交
852 853

	arch_alloc_page(page, order);
L
Linus Torvalds 已提交
854
	kernel_map_pages(page, 1 << order, 1);
N
Nick Piggin 已提交
855 856 857 858 859 860 861

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

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

862
	return 0;
L
Linus Torvalds 已提交
863 864
}

865 866 867 868
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
869 870
static inline
struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895
						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;
}


896 897 898 899
/*
 * This array describes the order lists are fallen back to when
 * the free lists for the desirable migrate type are depleted
 */
900 901 902 903 904 905 906 907 908
static int fallbacks[MIGRATE_TYPES][4] = {
	[MIGRATE_UNMOVABLE]   = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE,     MIGRATE_RESERVE },
	[MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE,   MIGRATE_MOVABLE,     MIGRATE_RESERVE },
#ifdef CONFIG_CMA
	[MIGRATE_MOVABLE]     = { MIGRATE_CMA,         MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
	[MIGRATE_CMA]         = { MIGRATE_RESERVE }, /* Never used */
#else
	[MIGRATE_MOVABLE]     = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE,   MIGRATE_RESERVE },
#endif
909 910
	[MIGRATE_RESERVE]     = { MIGRATE_RESERVE }, /* Never used */
	[MIGRATE_ISOLATE]     = { MIGRATE_RESERVE }, /* Never used */
911 912
};

913 914
/*
 * Move the free pages in a range to the free lists of the requested type.
915
 * Note that start_page and end_pages are not aligned on a pageblock
916 917
 * boundary. If alignment is required, use move_freepages_block()
 */
A
Adrian Bunk 已提交
918 919 920
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
921 922 923
{
	struct page *page;
	unsigned long order;
924
	int pages_moved = 0;
925 926 927 928 929 930 931

#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
M
Mel Gorman 已提交
932
	 * grouping pages by mobility
933 934 935 936 937
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
938 939 940
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

941 942 943 944 945 946 947 948 949 950 951
		if (!pfn_valid_within(page_to_pfn(page))) {
			page++;
			continue;
		}

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

		order = page_order(page);
952 953
		list_move(&page->lru,
			  &zone->free_area[order].free_list[migratetype]);
954
		page += 1 << order;
955
		pages_moved += 1 << order;
956 957
	}

958
	return pages_moved;
959 960
}

961
int move_freepages_block(struct zone *zone, struct page *page,
962
				int migratetype)
963 964 965 966 967
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
968
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
969
	start_page = pfn_to_page(start_pfn);
970 971
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
972 973 974 975 976 977 978 979 980 981

	/* 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);
}

982 983 984 985 986 987 988 989 990 991 992
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;
	}
}

993
/* Remove an element from the buddy allocator from the fallback list */
994 995
static inline struct page *
__rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
996 997 998 999 1000 1001 1002 1003 1004
{
	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) {
1005
		for (i = 0;; i++) {
1006 1007
			migratetype = fallbacks[start_migratetype][i];

1008 1009
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
1010
				break;
M
Mel Gorman 已提交
1011

1012 1013 1014 1015 1016 1017 1018 1019 1020
			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--;

			/*
1021
			 * If breaking a large block of pages, move all free
1022 1023
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
L
Lucas De Marchi 已提交
1024
			 * aggressive about taking ownership of free pages
1025 1026 1027 1028 1029 1030
			 *
			 * On the other hand, never change migration
			 * type of MIGRATE_CMA pageblocks nor move CMA
			 * pages on different free lists. We don't
			 * want unmovable pages to be allocated from
			 * MIGRATE_CMA areas.
1031
			 */
1032 1033 1034 1035 1036
			if (!is_migrate_cma(migratetype) &&
			    (unlikely(current_order >= pageblock_order / 2) ||
			     start_migratetype == MIGRATE_RECLAIMABLE ||
			     page_group_by_mobility_disabled)) {
				int pages;
1037 1038 1039 1040
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
1041 1042
				if (pages >= (1 << (pageblock_order-1)) ||
						page_group_by_mobility_disabled)
1043 1044 1045
					set_pageblock_migratetype(page,
								start_migratetype);

1046
				migratetype = start_migratetype;
1047
			}
1048 1049 1050 1051 1052

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

1053
			/* Take ownership for orders >= pageblock_order */
1054 1055
			if (current_order >= pageblock_order &&
			    !is_migrate_cma(migratetype))
1056
				change_pageblock_range(page, current_order,
1057 1058
							start_migratetype);

1059 1060 1061
			expand(zone, page, order, current_order, area,
			       is_migrate_cma(migratetype)
			     ? migratetype : start_migratetype);
1062 1063 1064 1065

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

1066 1067 1068 1069
			return page;
		}
	}

1070
	return NULL;
1071 1072
}

1073
/*
L
Linus Torvalds 已提交
1074 1075 1076
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
1077 1078
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
L
Linus Torvalds 已提交
1079 1080 1081
{
	struct page *page;

1082
retry_reserve:
1083
	page = __rmqueue_smallest(zone, order, migratetype);
1084

1085
	if (unlikely(!page) && migratetype != MIGRATE_RESERVE) {
1086
		page = __rmqueue_fallback(zone, order, migratetype);
1087

1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098
		/*
		 * Use MIGRATE_RESERVE rather than fail an allocation. goto
		 * is used because __rmqueue_smallest is an inline function
		 * and we want just one call site
		 */
		if (!page) {
			migratetype = MIGRATE_RESERVE;
			goto retry_reserve;
		}
	}

1099
	trace_mm_page_alloc_zone_locked(page, order, migratetype);
1100
	return page;
L
Linus Torvalds 已提交
1101 1102
}

1103
/*
L
Linus Torvalds 已提交
1104 1105 1106 1107
 * 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.
 */
1108
static int rmqueue_bulk(struct zone *zone, unsigned int order,
1109
			unsigned long count, struct list_head *list,
1110
			int migratetype, int cold)
L
Linus Torvalds 已提交
1111
{
1112
	int mt = migratetype, i;
1113

N
Nick Piggin 已提交
1114
	spin_lock(&zone->lock);
L
Linus Torvalds 已提交
1115
	for (i = 0; i < count; ++i) {
1116
		struct page *page = __rmqueue(zone, order, migratetype);
N
Nick Piggin 已提交
1117
		if (unlikely(page == NULL))
L
Linus Torvalds 已提交
1118
			break;
1119 1120 1121 1122 1123 1124 1125 1126 1127 1128

		/*
		 * Split buddy pages returned by expand() are received here
		 * in physical page order. The page is added to the callers and
		 * list and the list head then moves forward. From the callers
		 * perspective, the linked list is ordered by page number in
		 * some conditions. This is useful for IO devices that can
		 * merge IO requests if the physical pages are ordered
		 * properly.
		 */
1129 1130 1131 1132
		if (likely(cold == 0))
			list_add(&page->lru, list);
		else
			list_add_tail(&page->lru, list);
1133 1134 1135 1136 1137 1138
		if (IS_ENABLED(CONFIG_CMA)) {
			mt = get_pageblock_migratetype(page);
			if (!is_migrate_cma(mt) && mt != MIGRATE_ISOLATE)
				mt = migratetype;
		}
		set_page_private(page, mt);
1139
		list = &page->lru;
L
Linus Torvalds 已提交
1140
	}
1141
	__mod_zone_page_state(zone, NR_FREE_PAGES, -(i << order));
N
Nick Piggin 已提交
1142
	spin_unlock(&zone->lock);
N
Nick Piggin 已提交
1143
	return i;
L
Linus Torvalds 已提交
1144 1145
}

1146
#ifdef CONFIG_NUMA
1147
/*
1148 1149 1150 1151
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
1152 1153
 * Note that this function must be called with the thread pinned to
 * a single processor.
1154
 */
1155
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
1156 1157
{
	unsigned long flags;
1158
	int to_drain;
1159

1160 1161 1162 1163 1164
	local_irq_save(flags);
	if (pcp->count >= pcp->batch)
		to_drain = pcp->batch;
	else
		to_drain = pcp->count;
1165 1166 1167 1168
	if (to_drain > 0) {
		free_pcppages_bulk(zone, to_drain, pcp);
		pcp->count -= to_drain;
	}
1169
	local_irq_restore(flags);
1170 1171 1172
}
#endif

1173 1174 1175 1176 1177 1178 1179 1180
/*
 * Drain pages of the indicated processor.
 *
 * The processor must either be the current processor and the
 * thread pinned to the current processor or a processor that
 * is not online.
 */
static void drain_pages(unsigned int cpu)
L
Linus Torvalds 已提交
1181
{
N
Nick Piggin 已提交
1182
	unsigned long flags;
L
Linus Torvalds 已提交
1183 1184
	struct zone *zone;

1185
	for_each_populated_zone(zone) {
L
Linus Torvalds 已提交
1186
		struct per_cpu_pageset *pset;
1187
		struct per_cpu_pages *pcp;
L
Linus Torvalds 已提交
1188

1189 1190
		local_irq_save(flags);
		pset = per_cpu_ptr(zone->pageset, cpu);
1191 1192

		pcp = &pset->pcp;
1193 1194 1195 1196
		if (pcp->count) {
			free_pcppages_bulk(zone, pcp->count, pcp);
			pcp->count = 0;
		}
1197
		local_irq_restore(flags);
L
Linus Torvalds 已提交
1198 1199 1200
	}
}

1201 1202 1203 1204 1205 1206 1207 1208 1209
/*
 * Spill all of this CPU's per-cpu pages back into the buddy allocator.
 */
void drain_local_pages(void *arg)
{
	drain_pages(smp_processor_id());
}

/*
1210 1211 1212 1213 1214 1215 1216
 * Spill all the per-cpu pages from all CPUs back into the buddy allocator.
 *
 * Note that this code is protected against sending an IPI to an offline
 * CPU but does not guarantee sending an IPI to newly hotplugged CPUs:
 * on_each_cpu_mask() blocks hotplug and won't talk to offlined CPUs but
 * nothing keeps CPUs from showing up after we populated the cpumask and
 * before the call to on_each_cpu_mask().
1217 1218 1219
 */
void drain_all_pages(void)
{
1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250
	int cpu;
	struct per_cpu_pageset *pcp;
	struct zone *zone;

	/*
	 * Allocate in the BSS so we wont require allocation in
	 * direct reclaim path for CONFIG_CPUMASK_OFFSTACK=y
	 */
	static cpumask_t cpus_with_pcps;

	/*
	 * We don't care about racing with CPU hotplug event
	 * as offline notification will cause the notified
	 * cpu to drain that CPU pcps and on_each_cpu_mask
	 * disables preemption as part of its processing
	 */
	for_each_online_cpu(cpu) {
		bool has_pcps = false;
		for_each_populated_zone(zone) {
			pcp = per_cpu_ptr(zone->pageset, cpu);
			if (pcp->pcp.count) {
				has_pcps = true;
				break;
			}
		}
		if (has_pcps)
			cpumask_set_cpu(cpu, &cpus_with_pcps);
		else
			cpumask_clear_cpu(cpu, &cpus_with_pcps);
	}
	on_each_cpu_mask(&cpus_with_pcps, drain_local_pages, NULL, 1);
1251 1252
}

1253
#ifdef CONFIG_HIBERNATION
L
Linus Torvalds 已提交
1254 1255 1256

void mark_free_pages(struct zone *zone)
{
1257 1258
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
1259
	int order, t;
L
Linus Torvalds 已提交
1260 1261 1262 1263 1264 1265
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
1266 1267 1268 1269 1270 1271

	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);

1272 1273
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
1274
		}
L
Linus Torvalds 已提交
1275

1276 1277
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
1278
			unsigned long i;
L
Linus Torvalds 已提交
1279

1280 1281
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
1282
				swsusp_set_page_free(pfn_to_page(pfn + i));
1283
		}
1284
	}
L
Linus Torvalds 已提交
1285 1286
	spin_unlock_irqrestore(&zone->lock, flags);
}
1287
#endif /* CONFIG_PM */
L
Linus Torvalds 已提交
1288 1289 1290

/*
 * Free a 0-order page
L
Li Hong 已提交
1291
 * cold == 1 ? free a cold page : free a hot page
L
Linus Torvalds 已提交
1292
 */
L
Li Hong 已提交
1293
void free_hot_cold_page(struct page *page, int cold)
L
Linus Torvalds 已提交
1294 1295 1296 1297
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;
1298
	int migratetype;
1299
	int wasMlocked = __TestClearPageMlocked(page);
L
Linus Torvalds 已提交
1300

1301
	if (!free_pages_prepare(page, 0))
1302 1303
		return;

1304 1305
	migratetype = get_pageblock_migratetype(page);
	set_page_private(page, migratetype);
L
Linus Torvalds 已提交
1306
	local_irq_save(flags);
1307
	if (unlikely(wasMlocked))
1308
		free_page_mlock(page);
1309
	__count_vm_event(PGFREE);
1310

1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325
	/*
	 * We only track unmovable, reclaimable and movable on pcp lists.
	 * Free ISOLATE pages back to the allocator because they are being
	 * offlined but treat RESERVE as movable pages so we can get those
	 * areas back if necessary. Otherwise, we may have to free
	 * excessively into the page allocator
	 */
	if (migratetype >= MIGRATE_PCPTYPES) {
		if (unlikely(migratetype == MIGRATE_ISOLATE)) {
			free_one_page(zone, page, 0, migratetype);
			goto out;
		}
		migratetype = MIGRATE_MOVABLE;
	}

1326
	pcp = &this_cpu_ptr(zone->pageset)->pcp;
1327
	if (cold)
1328
		list_add_tail(&page->lru, &pcp->lists[migratetype]);
1329
	else
1330
		list_add(&page->lru, &pcp->lists[migratetype]);
L
Linus Torvalds 已提交
1331
	pcp->count++;
N
Nick Piggin 已提交
1332
	if (pcp->count >= pcp->high) {
1333
		free_pcppages_bulk(zone, pcp->batch, pcp);
N
Nick Piggin 已提交
1334 1335
		pcp->count -= pcp->batch;
	}
1336 1337

out:
L
Linus Torvalds 已提交
1338 1339 1340
	local_irq_restore(flags);
}

1341 1342 1343 1344 1345 1346 1347 1348
/*
 * Free a list of 0-order pages
 */
void free_hot_cold_page_list(struct list_head *list, int cold)
{
	struct page *page, *next;

	list_for_each_entry_safe(page, next, list, lru) {
1349
		trace_mm_page_free_batched(page, cold);
1350 1351 1352 1353
		free_hot_cold_page(page, cold);
	}
}

N
Nick Piggin 已提交
1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365
/*
 * 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;

N
Nick Piggin 已提交
1366 1367
	VM_BUG_ON(PageCompound(page));
	VM_BUG_ON(!page_count(page));
1368 1369 1370 1371 1372 1373 1374 1375 1376 1377

#ifdef CONFIG_KMEMCHECK
	/*
	 * Split shadow pages too, because free(page[0]) would
	 * otherwise free the whole shadow.
	 */
	if (kmemcheck_page_is_tracked(page))
		split_page(virt_to_page(page[0].shadow), order);
#endif

1378 1379
	for (i = 1; i < (1 << order); i++)
		set_page_refcounted(page + i);
N
Nick Piggin 已提交
1380 1381
}

1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419
/*
 * Similar to split_page except the page is already free. As this is only
 * being used for migration, the migratetype of the block also changes.
 * As this is called with interrupts disabled, the caller is responsible
 * for calling arch_alloc_page() and kernel_map_page() after interrupts
 * are enabled.
 *
 * Note: this is probably too low level an operation for use in drivers.
 * Please consult with lkml before using this in your driver.
 */
int split_free_page(struct page *page)
{
	unsigned int order;
	unsigned long watermark;
	struct zone *zone;

	BUG_ON(!PageBuddy(page));

	zone = page_zone(page);
	order = page_order(page);

	/* Obey watermarks as if the page was being allocated */
	watermark = low_wmark_pages(zone) + (1 << order);
	if (!zone_watermark_ok(zone, 0, watermark, 0, 0))
		return 0;

	/* Remove page from free list */
	list_del(&page->lru);
	zone->free_area[order].nr_free--;
	rmv_page_order(page);
	__mod_zone_page_state(zone, NR_FREE_PAGES, -(1UL << order));

	/* Split into individual pages */
	set_page_refcounted(page);
	split_page(page, order);

	if (order >= pageblock_order - 1) {
		struct page *endpage = page + (1 << order) - 1;
1420 1421 1422 1423 1424 1425
		for (; page < endpage; page += pageblock_nr_pages) {
			int mt = get_pageblock_migratetype(page);
			if (mt != MIGRATE_ISOLATE && !is_migrate_cma(mt))
				set_pageblock_migratetype(page,
							  MIGRATE_MOVABLE);
		}
1426 1427 1428 1429 1430
	}

	return 1 << order;
}

L
Linus Torvalds 已提交
1431 1432 1433 1434 1435
/*
 * 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.
 */
1436 1437
static inline
struct page *buffered_rmqueue(struct zone *preferred_zone,
1438 1439
			struct zone *zone, int order, gfp_t gfp_flags,
			int migratetype)
L
Linus Torvalds 已提交
1440 1441
{
	unsigned long flags;
1442
	struct page *page;
L
Linus Torvalds 已提交
1443 1444
	int cold = !!(gfp_flags & __GFP_COLD);

1445
again:
N
Nick Piggin 已提交
1446
	if (likely(order == 0)) {
L
Linus Torvalds 已提交
1447
		struct per_cpu_pages *pcp;
1448
		struct list_head *list;
L
Linus Torvalds 已提交
1449 1450

		local_irq_save(flags);
1451 1452
		pcp = &this_cpu_ptr(zone->pageset)->pcp;
		list = &pcp->lists[migratetype];
1453
		if (list_empty(list)) {
1454
			pcp->count += rmqueue_bulk(zone, 0,
1455
					pcp->batch, list,
1456
					migratetype, cold);
1457
			if (unlikely(list_empty(list)))
1458
				goto failed;
1459
		}
1460

1461 1462 1463 1464 1465
		if (cold)
			page = list_entry(list->prev, struct page, lru);
		else
			page = list_entry(list->next, struct page, lru);

1466 1467
		list_del(&page->lru);
		pcp->count--;
R
Rohit Seth 已提交
1468
	} else {
1469 1470 1471 1472 1473 1474 1475 1476
		if (unlikely(gfp_flags & __GFP_NOFAIL)) {
			/*
			 * __GFP_NOFAIL is not to be used in new code.
			 *
			 * All __GFP_NOFAIL callers should be fixed so that they
			 * properly detect and handle allocation failures.
			 *
			 * We most definitely don't want callers attempting to
1477
			 * allocate greater than order-1 page units with
1478 1479
			 * __GFP_NOFAIL.
			 */
1480
			WARN_ON_ONCE(order > 1);
1481
		}
L
Linus Torvalds 已提交
1482
		spin_lock_irqsave(&zone->lock, flags);
1483
		page = __rmqueue(zone, order, migratetype);
N
Nick Piggin 已提交
1484 1485 1486
		spin_unlock(&zone->lock);
		if (!page)
			goto failed;
1487
		__mod_zone_page_state(zone, NR_FREE_PAGES, -(1 << order));
L
Linus Torvalds 已提交
1488 1489
	}

1490
	__count_zone_vm_events(PGALLOC, zone, 1 << order);
A
Andi Kleen 已提交
1491
	zone_statistics(preferred_zone, zone, gfp_flags);
N
Nick Piggin 已提交
1492
	local_irq_restore(flags);
L
Linus Torvalds 已提交
1493

N
Nick Piggin 已提交
1494
	VM_BUG_ON(bad_range(zone, page));
N
Nick Piggin 已提交
1495
	if (prep_new_page(page, order, gfp_flags))
N
Nick Piggin 已提交
1496
		goto again;
L
Linus Torvalds 已提交
1497
	return page;
N
Nick Piggin 已提交
1498 1499 1500 1501

failed:
	local_irq_restore(flags);
	return NULL;
L
Linus Torvalds 已提交
1502 1503
}

1504 1505 1506 1507 1508 1509 1510 1511 1512
/* The ALLOC_WMARK bits are used as an index to zone->watermark */
#define ALLOC_WMARK_MIN		WMARK_MIN
#define ALLOC_WMARK_LOW		WMARK_LOW
#define ALLOC_WMARK_HIGH	WMARK_HIGH
#define ALLOC_NO_WATERMARKS	0x04 /* don't check watermarks at all */

/* Mask to get the watermark bits */
#define ALLOC_WMARK_MASK	(ALLOC_NO_WATERMARKS-1)

1513 1514 1515
#define ALLOC_HARDER		0x10 /* try to alloc harder */
#define ALLOC_HIGH		0x20 /* __GFP_HIGH set */
#define ALLOC_CPUSET		0x40 /* check for correct cpuset */
R
Rohit Seth 已提交
1516

1517 1518
#ifdef CONFIG_FAIL_PAGE_ALLOC

1519
static struct {
1520 1521 1522 1523
	struct fault_attr attr;

	u32 ignore_gfp_highmem;
	u32 ignore_gfp_wait;
1524
	u32 min_order;
1525 1526
} fail_page_alloc = {
	.attr = FAULT_ATTR_INITIALIZER,
1527 1528
	.ignore_gfp_wait = 1,
	.ignore_gfp_highmem = 1,
1529
	.min_order = 1,
1530 1531 1532 1533 1534 1535 1536 1537
};

static int __init setup_fail_page_alloc(char *str)
{
	return setup_fault_attr(&fail_page_alloc.attr, str);
}
__setup("fail_page_alloc=", setup_fail_page_alloc);

1538
static bool should_fail_alloc_page(gfp_t gfp_mask, unsigned int order)
1539
{
1540
	if (order < fail_page_alloc.min_order)
1541
		return false;
1542
	if (gfp_mask & __GFP_NOFAIL)
1543
		return false;
1544
	if (fail_page_alloc.ignore_gfp_highmem && (gfp_mask & __GFP_HIGHMEM))
1545
		return false;
1546
	if (fail_page_alloc.ignore_gfp_wait && (gfp_mask & __GFP_WAIT))
1547
		return false;
1548 1549 1550 1551 1552 1553 1554 1555

	return should_fail(&fail_page_alloc.attr, 1 << order);
}

#ifdef CONFIG_FAULT_INJECTION_DEBUG_FS

static int __init fail_page_alloc_debugfs(void)
{
A
Al Viro 已提交
1556
	umode_t mode = S_IFREG | S_IRUSR | S_IWUSR;
1557 1558
	struct dentry *dir;

1559 1560 1561 1562
	dir = fault_create_debugfs_attr("fail_page_alloc", NULL,
					&fail_page_alloc.attr);
	if (IS_ERR(dir))
		return PTR_ERR(dir);
1563

1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575
	if (!debugfs_create_bool("ignore-gfp-wait", mode, dir,
				&fail_page_alloc.ignore_gfp_wait))
		goto fail;
	if (!debugfs_create_bool("ignore-gfp-highmem", mode, dir,
				&fail_page_alloc.ignore_gfp_highmem))
		goto fail;
	if (!debugfs_create_u32("min-order", mode, dir,
				&fail_page_alloc.min_order))
		goto fail;

	return 0;
fail:
1576
	debugfs_remove_recursive(dir);
1577

1578
	return -ENOMEM;
1579 1580 1581 1582 1583 1584 1585 1586
}

late_initcall(fail_page_alloc_debugfs);

#endif /* CONFIG_FAULT_INJECTION_DEBUG_FS */

#else /* CONFIG_FAIL_PAGE_ALLOC */

1587
static inline bool should_fail_alloc_page(gfp_t gfp_mask, unsigned int order)
1588
{
1589
	return false;
1590 1591 1592 1593
}

#endif /* CONFIG_FAIL_PAGE_ALLOC */

L
Linus Torvalds 已提交
1594
/*
1595
 * Return true if free pages are above 'mark'. This takes into account the order
L
Linus Torvalds 已提交
1596 1597
 * of the allocation.
 */
1598 1599
static bool __zone_watermark_ok(struct zone *z, int order, unsigned long mark,
		      int classzone_idx, int alloc_flags, long free_pages)
L
Linus Torvalds 已提交
1600 1601
{
	/* free_pages my go negative - that's OK */
1602
	long min = mark;
1603
	long lowmem_reserve = z->lowmem_reserve[classzone_idx];
L
Linus Torvalds 已提交
1604 1605
	int o;

1606
	free_pages -= (1 << order) - 1;
R
Rohit Seth 已提交
1607
	if (alloc_flags & ALLOC_HIGH)
L
Linus Torvalds 已提交
1608
		min -= min / 2;
R
Rohit Seth 已提交
1609
	if (alloc_flags & ALLOC_HARDER)
L
Linus Torvalds 已提交
1610 1611
		min -= min / 4;

1612
	if (free_pages <= min + lowmem_reserve)
1613
		return false;
L
Linus Torvalds 已提交
1614 1615 1616 1617 1618 1619 1620 1621
	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)
1622
			return false;
L
Linus Torvalds 已提交
1623
	}
1624 1625 1626
	return true;
}

1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640
#ifdef CONFIG_MEMORY_ISOLATION
static inline unsigned long nr_zone_isolate_freepages(struct zone *zone)
{
	if (unlikely(zone->nr_pageblock_isolate))
		return zone->nr_pageblock_isolate * pageblock_nr_pages;
	return 0;
}
#else
static inline unsigned long nr_zone_isolate_freepages(struct zone *zone)
{
	return 0;
}
#endif

1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655
bool zone_watermark_ok(struct zone *z, int order, unsigned long mark,
		      int classzone_idx, int alloc_flags)
{
	return __zone_watermark_ok(z, order, mark, classzone_idx, alloc_flags,
					zone_page_state(z, NR_FREE_PAGES));
}

bool zone_watermark_ok_safe(struct zone *z, int order, unsigned long mark,
		      int classzone_idx, int alloc_flags)
{
	long free_pages = zone_page_state(z, NR_FREE_PAGES);

	if (z->percpu_drift_mark && free_pages < z->percpu_drift_mark)
		free_pages = zone_page_state_snapshot(z, NR_FREE_PAGES);

1656 1657 1658 1659 1660 1661 1662 1663
	/*
	 * If the zone has MIGRATE_ISOLATE type free pages, we should consider
	 * it.  nr_zone_isolate_freepages is never accurate so kswapd might not
	 * sleep although it could do so.  But this is more desirable for memory
	 * hotplug than sleeping which can cause a livelock in the direct
	 * reclaim path.
	 */
	free_pages -= nr_zone_isolate_freepages(z);
1664 1665
	return __zone_watermark_ok(z, order, mark, classzone_idx, alloc_flags,
								free_pages);
L
Linus Torvalds 已提交
1666 1667
}

1668 1669 1670 1671 1672 1673
#ifdef CONFIG_NUMA
/*
 * zlc_setup - Setup for "zonelist cache".  Uses cached zone data to
 * skip over zones that are not allowed by the cpuset, or that have
 * been recently (in last second) found to be nearly full.  See further
 * comments in mmzone.h.  Reduces cache footprint of zonelist scans
S
Simon Arlott 已提交
1674
 * that have to skip over a lot of full or unallowed zones.
1675 1676 1677
 *
 * If the zonelist cache is present in the passed in zonelist, then
 * returns a pointer to the allowed node mask (either the current
1678
 * tasks mems_allowed, or node_states[N_HIGH_MEMORY].)
1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699
 *
 * If the zonelist cache is not available for this zonelist, does
 * nothing and returns NULL.
 *
 * If the fullzones BITMAP in the zonelist cache is stale (more than
 * a second since last zap'd) then we zap it out (clear its bits.)
 *
 * We hold off even calling zlc_setup, until after we've checked the
 * first zone in the zonelist, on the theory that most allocations will
 * be satisfied from that first zone, so best to examine that zone as
 * quickly as we can.
 */
static nodemask_t *zlc_setup(struct zonelist *zonelist, int alloc_flags)
{
	struct zonelist_cache *zlc;	/* cached zonelist speedup info */
	nodemask_t *allowednodes;	/* zonelist_cache approximation */

	zlc = zonelist->zlcache_ptr;
	if (!zlc)
		return NULL;

S
S.Caglar Onur 已提交
1700
	if (time_after(jiffies, zlc->last_full_zap + HZ)) {
1701 1702 1703 1704 1705 1706
		bitmap_zero(zlc->fullzones, MAX_ZONES_PER_ZONELIST);
		zlc->last_full_zap = jiffies;
	}

	allowednodes = !in_interrupt() && (alloc_flags & ALLOC_CPUSET) ?
					&cpuset_current_mems_allowed :
1707
					&node_states[N_HIGH_MEMORY];
1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732
	return allowednodes;
}

/*
 * Given 'z' scanning a zonelist, run a couple of quick checks to see
 * if it is worth looking at further for free memory:
 *  1) Check that the zone isn't thought to be full (doesn't have its
 *     bit set in the zonelist_cache fullzones BITMAP).
 *  2) Check that the zones node (obtained from the zonelist_cache
 *     z_to_n[] mapping) is allowed in the passed in allowednodes mask.
 * Return true (non-zero) if zone is worth looking at further, or
 * else return false (zero) if it is not.
 *
 * This check -ignores- the distinction between various watermarks,
 * such as GFP_HIGH, GFP_ATOMIC, PF_MEMALLOC, ...  If a zone is
 * found to be full for any variation of these watermarks, it will
 * be considered full for up to one second by all requests, unless
 * we are so low on memory on all allowed nodes that we are forced
 * into the second scan of the zonelist.
 *
 * In the second scan we ignore this zonelist cache and exactly
 * apply the watermarks to all zones, even it is slower to do so.
 * We are low on memory in the second scan, and should leave no stone
 * unturned looking for a free page.
 */
1733
static int zlc_zone_worth_trying(struct zonelist *zonelist, struct zoneref *z,
1734 1735 1736 1737 1738 1739 1740 1741 1742 1743
						nodemask_t *allowednodes)
{
	struct zonelist_cache *zlc;	/* cached zonelist speedup info */
	int i;				/* index of *z in zonelist zones */
	int n;				/* node that zone *z is on */

	zlc = zonelist->zlcache_ptr;
	if (!zlc)
		return 1;

1744
	i = z - zonelist->_zonerefs;
1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755
	n = zlc->z_to_n[i];

	/* This zone is worth trying if it is allowed but not full */
	return node_isset(n, *allowednodes) && !test_bit(i, zlc->fullzones);
}

/*
 * Given 'z' scanning a zonelist, set the corresponding bit in
 * zlc->fullzones, so that subsequent attempts to allocate a page
 * from that zone don't waste time re-examining it.
 */
1756
static void zlc_mark_zone_full(struct zonelist *zonelist, struct zoneref *z)
1757 1758 1759 1760 1761 1762 1763 1764
{
	struct zonelist_cache *zlc;	/* cached zonelist speedup info */
	int i;				/* index of *z in zonelist zones */

	zlc = zonelist->zlcache_ptr;
	if (!zlc)
		return;

1765
	i = z - zonelist->_zonerefs;
1766 1767 1768 1769

	set_bit(i, zlc->fullzones);
}

1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784
/*
 * clear all zones full, called after direct reclaim makes progress so that
 * a zone that was recently full is not skipped over for up to a second
 */
static void zlc_clear_zones_full(struct zonelist *zonelist)
{
	struct zonelist_cache *zlc;	/* cached zonelist speedup info */

	zlc = zonelist->zlcache_ptr;
	if (!zlc)
		return;

	bitmap_zero(zlc->fullzones, MAX_ZONES_PER_ZONELIST);
}

1785 1786 1787 1788 1789 1790 1791
#else	/* CONFIG_NUMA */

static nodemask_t *zlc_setup(struct zonelist *zonelist, int alloc_flags)
{
	return NULL;
}

1792
static int zlc_zone_worth_trying(struct zonelist *zonelist, struct zoneref *z,
1793 1794 1795 1796 1797
				nodemask_t *allowednodes)
{
	return 1;
}

1798
static void zlc_mark_zone_full(struct zonelist *zonelist, struct zoneref *z)
1799 1800
{
}
1801 1802 1803 1804

static void zlc_clear_zones_full(struct zonelist *zonelist)
{
}
1805 1806
#endif	/* CONFIG_NUMA */

R
Rohit Seth 已提交
1807
/*
1808
 * get_page_from_freelist goes through the zonelist trying to allocate
R
Rohit Seth 已提交
1809 1810 1811
 * a page.
 */
static struct page *
1812
get_page_from_freelist(gfp_t gfp_mask, nodemask_t *nodemask, unsigned int order,
1813
		struct zonelist *zonelist, int high_zoneidx, int alloc_flags,
1814
		struct zone *preferred_zone, int migratetype)
M
Martin Hicks 已提交
1815
{
1816
	struct zoneref *z;
R
Rohit Seth 已提交
1817
	struct page *page = NULL;
1818
	int classzone_idx;
1819
	struct zone *zone;
1820 1821 1822
	nodemask_t *allowednodes = NULL;/* zonelist_cache approximation */
	int zlc_active = 0;		/* set if using zonelist_cache */
	int did_zlc_setup = 0;		/* just call zlc_setup() one time */
1823

1824
	classzone_idx = zone_idx(preferred_zone);
1825
zonelist_scan:
R
Rohit Seth 已提交
1826
	/*
1827
	 * Scan zonelist, looking for a zone with enough free.
R
Rohit Seth 已提交
1828 1829
	 * See also cpuset_zone_allowed() comment in kernel/cpuset.c.
	 */
1830 1831
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						high_zoneidx, nodemask) {
1832 1833 1834
		if (NUMA_BUILD && zlc_active &&
			!zlc_zone_worth_trying(zonelist, z, allowednodes))
				continue;
R
Rohit Seth 已提交
1835
		if ((alloc_flags & ALLOC_CPUSET) &&
1836
			!cpuset_zone_allowed_softwall(zone, gfp_mask))
1837
				continue;
1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866
		/*
		 * When allocating a page cache page for writing, we
		 * want to get it from a zone that is within its dirty
		 * limit, such that no single zone holds more than its
		 * proportional share of globally allowed dirty pages.
		 * The dirty limits take into account the zone's
		 * lowmem reserves and high watermark so that kswapd
		 * should be able to balance it without having to
		 * write pages from its LRU list.
		 *
		 * This may look like it could increase pressure on
		 * lower zones by failing allocations in higher zones
		 * before they are full.  But the pages that do spill
		 * over are limited as the lower zones are protected
		 * by this very same mechanism.  It should not become
		 * a practical burden to them.
		 *
		 * XXX: For now, allow allocations to potentially
		 * exceed the per-zone dirty limit in the slowpath
		 * (ALLOC_WMARK_LOW unset) before going into reclaim,
		 * which is important when on a NUMA setup the allowed
		 * zones are together not big enough to reach the
		 * global limit.  The proper fix for these situations
		 * will require awareness of zones in the
		 * dirty-throttling and the flusher threads.
		 */
		if ((alloc_flags & ALLOC_WMARK_LOW) &&
		    (gfp_mask & __GFP_WRITE) && !zone_dirty_ok(zone))
			goto this_zone_full;
R
Rohit Seth 已提交
1867

1868
		BUILD_BUG_ON(ALLOC_NO_WATERMARKS < NR_WMARK);
R
Rohit Seth 已提交
1869
		if (!(alloc_flags & ALLOC_NO_WATERMARKS)) {
1870
			unsigned long mark;
1871 1872
			int ret;

1873
			mark = zone->watermark[alloc_flags & ALLOC_WMARK_MASK];
1874 1875 1876 1877
			if (zone_watermark_ok(zone, order, mark,
				    classzone_idx, alloc_flags))
				goto try_this_zone;

1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888
			if (NUMA_BUILD && !did_zlc_setup && nr_online_nodes > 1) {
				/*
				 * we do zlc_setup if there are multiple nodes
				 * and before considering the first zone allowed
				 * by the cpuset.
				 */
				allowednodes = zlc_setup(zonelist, alloc_flags);
				zlc_active = 1;
				did_zlc_setup = 1;
			}

1889 1890 1891
			if (zone_reclaim_mode == 0)
				goto this_zone_full;

1892 1893 1894 1895 1896 1897 1898 1899
			/*
			 * As we may have just activated ZLC, check if the first
			 * eligible zone has failed zone_reclaim recently.
			 */
			if (NUMA_BUILD && zlc_active &&
				!zlc_zone_worth_trying(zonelist, z, allowednodes))
				continue;

1900 1901 1902 1903
			ret = zone_reclaim(zone, gfp_mask, order);
			switch (ret) {
			case ZONE_RECLAIM_NOSCAN:
				/* did not scan */
1904
				continue;
1905 1906
			case ZONE_RECLAIM_FULL:
				/* scanned but unreclaimable */
1907
				continue;
1908 1909 1910 1911
			default:
				/* did we reclaim enough */
				if (!zone_watermark_ok(zone, order, mark,
						classzone_idx, alloc_flags))
1912
					goto this_zone_full;
1913
			}
R
Rohit Seth 已提交
1914 1915
		}

1916
try_this_zone:
1917 1918
		page = buffered_rmqueue(preferred_zone, zone, order,
						gfp_mask, migratetype);
1919
		if (page)
R
Rohit Seth 已提交
1920
			break;
1921 1922 1923
this_zone_full:
		if (NUMA_BUILD)
			zlc_mark_zone_full(zonelist, z);
1924
	}
1925 1926 1927 1928 1929 1930

	if (unlikely(NUMA_BUILD && page == NULL && zlc_active)) {
		/* Disable zlc cache for second zonelist scan */
		zlc_active = 0;
		goto zonelist_scan;
	}
R
Rohit Seth 已提交
1931
	return page;
M
Martin Hicks 已提交
1932 1933
}

1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947
/*
 * Large machines with many possible nodes should not always dump per-node
 * meminfo in irq context.
 */
static inline bool should_suppress_show_mem(void)
{
	bool ret = false;

#if NODES_SHIFT > 8
	ret = in_interrupt();
#endif
	return ret;
}

1948 1949 1950 1951 1952 1953 1954 1955
static DEFINE_RATELIMIT_STATE(nopage_rs,
		DEFAULT_RATELIMIT_INTERVAL,
		DEFAULT_RATELIMIT_BURST);

void warn_alloc_failed(gfp_t gfp_mask, int order, const char *fmt, ...)
{
	unsigned int filter = SHOW_MEM_FILTER_NODES;

1956 1957
	if ((gfp_mask & __GFP_NOWARN) || !__ratelimit(&nopage_rs) ||
	    debug_guardpage_minorder() > 0)
1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972
		return;

	/*
	 * This documents exceptions given to allocations in certain
	 * contexts that are allowed to allocate outside current's set
	 * of allowed nodes.
	 */
	if (!(gfp_mask & __GFP_NOMEMALLOC))
		if (test_thread_flag(TIF_MEMDIE) ||
		    (current->flags & (PF_MEMALLOC | PF_EXITING)))
			filter &= ~SHOW_MEM_FILTER_NODES;
	if (in_interrupt() || !(gfp_mask & __GFP_WAIT))
		filter &= ~SHOW_MEM_FILTER_NODES;

	if (fmt) {
J
Joe Perches 已提交
1973 1974 1975
		struct va_format vaf;
		va_list args;

1976
		va_start(args, fmt);
J
Joe Perches 已提交
1977 1978 1979 1980 1981 1982

		vaf.fmt = fmt;
		vaf.va = &args;

		pr_warn("%pV", &vaf);

1983 1984 1985
		va_end(args);
	}

J
Joe Perches 已提交
1986 1987
	pr_warn("%s: page allocation failure: order:%d, mode:0x%x\n",
		current->comm, order, gfp_mask);
1988 1989 1990 1991 1992 1993

	dump_stack();
	if (!should_suppress_show_mem())
		show_mem(filter);
}

1994 1995
static inline int
should_alloc_retry(gfp_t gfp_mask, unsigned int order,
1996
				unsigned long did_some_progress,
1997
				unsigned long pages_reclaimed)
L
Linus Torvalds 已提交
1998
{
1999 2000 2001
	/* Do not loop if specifically requested */
	if (gfp_mask & __GFP_NORETRY)
		return 0;
L
Linus Torvalds 已提交
2002

2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
	/* Always retry if specifically requested */
	if (gfp_mask & __GFP_NOFAIL)
		return 1;

	/*
	 * Suspend converts GFP_KERNEL to __GFP_WAIT which can prevent reclaim
	 * making forward progress without invoking OOM. Suspend also disables
	 * storage devices so kswapd will not help. Bail if we are suspending.
	 */
	if (!did_some_progress && pm_suspended_storage())
		return 0;

2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031
	/*
	 * In this implementation, order <= PAGE_ALLOC_COSTLY_ORDER
	 * means __GFP_NOFAIL, but that may not be true in other
	 * implementations.
	 */
	if (order <= PAGE_ALLOC_COSTLY_ORDER)
		return 1;

	/*
	 * For order > PAGE_ALLOC_COSTLY_ORDER, if __GFP_REPEAT is
	 * specified, then we retry until we no longer reclaim any pages
	 * (above), or we've reclaimed an order of pages at least as
	 * large as the allocation's order. In both cases, if the
	 * allocation still fails, we stop retrying.
	 */
	if (gfp_mask & __GFP_REPEAT && pages_reclaimed < (1 << order))
		return 1;
2032

2033 2034
	return 0;
}
2035

2036 2037 2038
static inline struct page *
__alloc_pages_may_oom(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
2039 2040
	nodemask_t *nodemask, struct zone *preferred_zone,
	int migratetype)
2041 2042 2043 2044
{
	struct page *page;

	/* Acquire the OOM killer lock for the zones in zonelist */
2045
	if (!try_set_zonelist_oom(zonelist, gfp_mask)) {
2046
		schedule_timeout_uninterruptible(1);
L
Linus Torvalds 已提交
2047 2048
		return NULL;
	}
2049

2050 2051 2052 2053 2054 2055 2056
	/*
	 * Go through the zonelist yet one more time, keep very high watermark
	 * here, this is only to catch a parallel oom killing, we must fail if
	 * we're still under heavy pressure.
	 */
	page = get_page_from_freelist(gfp_mask|__GFP_HARDWALL, nodemask,
		order, zonelist, high_zoneidx,
2057
		ALLOC_WMARK_HIGH|ALLOC_CPUSET,
2058
		preferred_zone, migratetype);
R
Rohit Seth 已提交
2059
	if (page)
2060 2061
		goto out;

2062 2063 2064 2065
	if (!(gfp_mask & __GFP_NOFAIL)) {
		/* The OOM killer will not help higher order allocs */
		if (order > PAGE_ALLOC_COSTLY_ORDER)
			goto out;
2066 2067 2068
		/* The OOM killer does not needlessly kill tasks for lowmem */
		if (high_zoneidx < ZONE_NORMAL)
			goto out;
2069 2070 2071 2072 2073 2074 2075 2076 2077 2078
		/*
		 * GFP_THISNODE contains __GFP_NORETRY and we never hit this.
		 * Sanity check for bare calls of __GFP_THISNODE, not real OOM.
		 * The caller should handle page allocation failure by itself if
		 * it specifies __GFP_THISNODE.
		 * Note: Hugepage uses it but will hit PAGE_ALLOC_COSTLY_ORDER.
		 */
		if (gfp_mask & __GFP_THISNODE)
			goto out;
	}
2079
	/* Exhausted what can be done so it's blamo time */
2080
	out_of_memory(zonelist, gfp_mask, order, nodemask, false);
2081 2082 2083 2084 2085 2086

out:
	clear_zonelist_oom(zonelist, gfp_mask);
	return page;
}

2087 2088 2089 2090 2091 2092
#ifdef CONFIG_COMPACTION
/* Try memory compaction for high-order allocations before reclaim */
static struct page *
__alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
	nodemask_t *nodemask, int alloc_flags, struct zone *preferred_zone,
2093 2094 2095
	int migratetype, bool sync_migration,
	bool *deferred_compaction,
	unsigned long *did_some_progress)
2096 2097 2098
{
	struct page *page;

2099
	if (!order)
2100 2101
		return NULL;

2102
	if (compaction_deferred(preferred_zone, order)) {
2103 2104 2105 2106
		*deferred_compaction = true;
		return NULL;
	}

2107
	current->flags |= PF_MEMALLOC;
2108
	*did_some_progress = try_to_compact_pages(zonelist, order, gfp_mask,
2109
						nodemask, sync_migration);
2110
	current->flags &= ~PF_MEMALLOC;
2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121
	if (*did_some_progress != COMPACT_SKIPPED) {

		/* Page migration frees to the PCP lists but we want merging */
		drain_pages(get_cpu());
		put_cpu();

		page = get_page_from_freelist(gfp_mask, nodemask,
				order, zonelist, high_zoneidx,
				alloc_flags, preferred_zone,
				migratetype);
		if (page) {
2122 2123
			preferred_zone->compact_considered = 0;
			preferred_zone->compact_defer_shift = 0;
2124 2125
			if (order >= preferred_zone->compact_order_failed)
				preferred_zone->compact_order_failed = order + 1;
2126 2127 2128 2129 2130 2131 2132 2133 2134 2135
			count_vm_event(COMPACTSUCCESS);
			return page;
		}

		/*
		 * It's bad if compaction run occurs and fails.
		 * The most likely reason is that pages exist,
		 * but not enough to satisfy watermarks.
		 */
		count_vm_event(COMPACTFAIL);
2136 2137 2138 2139 2140 2141

		/*
		 * As async compaction considers a subset of pageblocks, only
		 * defer if the failure was a sync compaction failure.
		 */
		if (sync_migration)
2142
			defer_compaction(preferred_zone, order);
2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153

		cond_resched();
	}

	return NULL;
}
#else
static inline struct page *
__alloc_pages_direct_compact(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
	nodemask_t *nodemask, int alloc_flags, struct zone *preferred_zone,
2154 2155 2156
	int migratetype, bool sync_migration,
	bool *deferred_compaction,
	unsigned long *did_some_progress)
2157 2158 2159 2160 2161
{
	return NULL;
}
#endif /* CONFIG_COMPACTION */

2162 2163 2164 2165
/* Perform direct synchronous page reclaim */
static int
__perform_reclaim(gfp_t gfp_mask, unsigned int order, struct zonelist *zonelist,
		  nodemask_t *nodemask)
2166 2167
{
	struct reclaim_state reclaim_state;
2168
	int progress;
2169 2170 2171 2172 2173

	cond_resched();

	/* We now go into synchronous reclaim */
	cpuset_memory_pressure_bump();
2174
	current->flags |= PF_MEMALLOC;
2175 2176
	lockdep_set_current_reclaim_state(gfp_mask);
	reclaim_state.reclaimed_slab = 0;
2177
	current->reclaim_state = &reclaim_state;
2178

2179
	progress = try_to_free_pages(zonelist, order, gfp_mask, nodemask);
2180

2181
	current->reclaim_state = NULL;
2182
	lockdep_clear_current_reclaim_state();
2183
	current->flags &= ~PF_MEMALLOC;
2184 2185 2186

	cond_resched();

2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201
	return progress;
}

/* The really slow allocator path where we enter direct reclaim */
static inline struct page *
__alloc_pages_direct_reclaim(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
	nodemask_t *nodemask, int alloc_flags, struct zone *preferred_zone,
	int migratetype, unsigned long *did_some_progress)
{
	struct page *page = NULL;
	bool drained = false;

	*did_some_progress = __perform_reclaim(gfp_mask, order, zonelist,
					       nodemask);
2202 2203
	if (unlikely(!(*did_some_progress)))
		return NULL;
2204

2205 2206 2207 2208
	/* After successful reclaim, reconsider all zones for allocation */
	if (NUMA_BUILD)
		zlc_clear_zones_full(zonelist);

2209 2210
retry:
	page = get_page_from_freelist(gfp_mask, nodemask, order,
2211
					zonelist, high_zoneidx,
2212 2213
					alloc_flags, preferred_zone,
					migratetype);
2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224

	/*
	 * If an allocation failed after direct reclaim, it could be because
	 * pages are pinned on the per-cpu lists. Drain them and try again
	 */
	if (!page && !drained) {
		drain_all_pages();
		drained = true;
		goto retry;
	}

2225 2226 2227
	return page;
}

L
Linus Torvalds 已提交
2228
/*
2229 2230
 * This is called in the allocator slow-path if the allocation request is of
 * sufficient urgency to ignore watermarks and take other desperate measures
L
Linus Torvalds 已提交
2231
 */
2232 2233 2234
static inline struct page *
__alloc_pages_high_priority(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
2235 2236
	nodemask_t *nodemask, struct zone *preferred_zone,
	int migratetype)
2237 2238 2239 2240 2241
{
	struct page *page;

	do {
		page = get_page_from_freelist(gfp_mask, nodemask, order,
2242
			zonelist, high_zoneidx, ALLOC_NO_WATERMARKS,
2243
			preferred_zone, migratetype);
2244 2245

		if (!page && gfp_mask & __GFP_NOFAIL)
2246
			wait_iff_congested(preferred_zone, BLK_RW_ASYNC, HZ/50);
2247 2248 2249 2250 2251 2252 2253
	} while (!page && (gfp_mask & __GFP_NOFAIL));

	return page;
}

static inline
void wake_all_kswapd(unsigned int order, struct zonelist *zonelist,
2254 2255
						enum zone_type high_zoneidx,
						enum zone_type classzone_idx)
L
Linus Torvalds 已提交
2256
{
2257 2258
	struct zoneref *z;
	struct zone *zone;
L
Linus Torvalds 已提交
2259

2260
	for_each_zone_zonelist(zone, z, zonelist, high_zoneidx)
2261
		wakeup_kswapd(zone, order, classzone_idx);
2262
}
2263

2264 2265 2266 2267 2268
static inline int
gfp_to_alloc_flags(gfp_t gfp_mask)
{
	int alloc_flags = ALLOC_WMARK_MIN | ALLOC_CPUSET;
	const gfp_t wait = gfp_mask & __GFP_WAIT;
L
Linus Torvalds 已提交
2269

2270
	/* __GFP_HIGH is assumed to be the same as ALLOC_HIGH to save a branch. */
2271
	BUILD_BUG_ON(__GFP_HIGH != (__force gfp_t) ALLOC_HIGH);
2272

2273 2274 2275 2276 2277 2278
	/*
	 * The caller may dip into page reserves a bit more if the caller
	 * cannot run direct reclaim, or if the caller has realtime scheduling
	 * policy or is asking for __GFP_HIGH memory.  GFP_ATOMIC requests will
	 * set both ALLOC_HARDER (!wait) and ALLOC_HIGH (__GFP_HIGH).
	 */
2279
	alloc_flags |= (__force int) (gfp_mask & __GFP_HIGH);
L
Linus Torvalds 已提交
2280

2281
	if (!wait) {
2282 2283 2284 2285 2286 2287
		/*
		 * Not worth trying to allocate harder for
		 * __GFP_NOMEMALLOC even if it can't schedule.
		 */
		if  (!(gfp_mask & __GFP_NOMEMALLOC))
			alloc_flags |= ALLOC_HARDER;
2288
		/*
2289 2290
		 * Ignore cpuset if GFP_ATOMIC (!wait) rather than fail alloc.
		 * See also cpuset_zone_allowed() comment in kernel/cpuset.c.
2291
		 */
2292
		alloc_flags &= ~ALLOC_CPUSET;
2293
	} else if (unlikely(rt_task(current)) && !in_interrupt())
2294 2295
		alloc_flags |= ALLOC_HARDER;

2296 2297 2298 2299
	if (likely(!(gfp_mask & __GFP_NOMEMALLOC))) {
		if (gfp_mask & __GFP_MEMALLOC)
			alloc_flags |= ALLOC_NO_WATERMARKS;
		else if (likely(!(gfp_mask & __GFP_NOMEMALLOC)) && !in_interrupt())
2300
			alloc_flags |= ALLOC_NO_WATERMARKS;
L
Linus Torvalds 已提交
2301
	}
2302

2303 2304 2305
	return alloc_flags;
}

2306 2307
bool gfp_pfmemalloc_allowed(gfp_t gfp_mask)
{
2308
	return !!(gfp_to_alloc_flags(gfp_mask) & ALLOC_NO_WATERMARKS);
2309 2310
}

2311 2312 2313
static inline struct page *
__alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
2314 2315
	nodemask_t *nodemask, struct zone *preferred_zone,
	int migratetype)
2316 2317 2318 2319 2320 2321
{
	const gfp_t wait = gfp_mask & __GFP_WAIT;
	struct page *page = NULL;
	int alloc_flags;
	unsigned long pages_reclaimed = 0;
	unsigned long did_some_progress;
2322
	bool sync_migration = false;
2323
	bool deferred_compaction = false;
L
Linus Torvalds 已提交
2324

2325 2326 2327 2328 2329 2330
	/*
	 * In the slowpath, we sanity check order to avoid ever trying to
	 * reclaim >= MAX_ORDER areas which will never succeed. Callers may
	 * be using allocators in order of preference for an area that is
	 * too large.
	 */
2331 2332
	if (order >= MAX_ORDER) {
		WARN_ON_ONCE(!(gfp_mask & __GFP_NOWARN));
2333
		return NULL;
2334
	}
L
Linus Torvalds 已提交
2335

2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346
	/*
	 * GFP_THISNODE (meaning __GFP_THISNODE, __GFP_NORETRY and
	 * __GFP_NOWARN set) should not cause reclaim since the subsystem
	 * (f.e. slab) using GFP_THISNODE may choose to trigger reclaim
	 * using a larger set of nodes after it has established that the
	 * allowed per node queues are empty and that nodes are
	 * over allocated.
	 */
	if (NUMA_BUILD && (gfp_mask & GFP_THISNODE) == GFP_THISNODE)
		goto nopage;

2347
restart:
A
Andrea Arcangeli 已提交
2348 2349
	if (!(gfp_mask & __GFP_NO_KSWAPD))
		wake_all_kswapd(order, zonelist, high_zoneidx,
2350
						zone_idx(preferred_zone));
L
Linus Torvalds 已提交
2351

2352
	/*
R
Rohit Seth 已提交
2353 2354 2355
	 * 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.
2356
	 */
2357
	alloc_flags = gfp_to_alloc_flags(gfp_mask);
L
Linus Torvalds 已提交
2358

2359 2360 2361 2362 2363 2364 2365 2366
	/*
	 * Find the true preferred zone if the allocation is unconstrained by
	 * cpusets.
	 */
	if (!(alloc_flags & ALLOC_CPUSET) && !nodemask)
		first_zones_zonelist(zonelist, high_zoneidx, NULL,
					&preferred_zone);

2367
rebalance:
2368
	/* This is the last chance, in general, before the goto nopage. */
2369
	page = get_page_from_freelist(gfp_mask, nodemask, order, zonelist,
2370 2371
			high_zoneidx, alloc_flags & ~ALLOC_NO_WATERMARKS,
			preferred_zone, migratetype);
R
Rohit Seth 已提交
2372 2373
	if (page)
		goto got_pg;
L
Linus Torvalds 已提交
2374

2375
	/* Allocate without watermarks if the context allows */
2376 2377 2378 2379 2380 2381
	if (alloc_flags & ALLOC_NO_WATERMARKS) {
		page = __alloc_pages_high_priority(gfp_mask, order,
				zonelist, high_zoneidx, nodemask,
				preferred_zone, migratetype);
		if (page)
			goto got_pg;
L
Linus Torvalds 已提交
2382 2383 2384 2385 2386 2387
	}

	/* Atomic allocations - we can't balance anything */
	if (!wait)
		goto nopage;

2388
	/* Avoid recursion of direct reclaim */
2389
	if (current->flags & PF_MEMALLOC)
2390 2391
		goto nopage;

2392 2393 2394 2395
	/* Avoid allocations with no watermarks from looping endlessly */
	if (test_thread_flag(TIF_MEMDIE) && !(gfp_mask & __GFP_NOFAIL))
		goto nopage;

2396 2397 2398 2399
	/*
	 * Try direct compaction. The first pass is asynchronous. Subsequent
	 * attempts after direct reclaim are synchronous
	 */
2400 2401 2402 2403
	page = __alloc_pages_direct_compact(gfp_mask, order,
					zonelist, high_zoneidx,
					nodemask,
					alloc_flags, preferred_zone,
2404 2405 2406
					migratetype, sync_migration,
					&deferred_compaction,
					&did_some_progress);
2407 2408
	if (page)
		goto got_pg;
2409
	sync_migration = true;
2410

2411 2412 2413 2414 2415 2416 2417 2418 2419
	/*
	 * If compaction is deferred for high-order allocations, it is because
	 * sync compaction recently failed. In this is the case and the caller
	 * has requested the system not be heavily disrupted, fail the
	 * allocation now instead of entering direct reclaim
	 */
	if (deferred_compaction && (gfp_mask & __GFP_NO_KSWAPD))
		goto nopage;

2420 2421 2422 2423
	/* Try direct reclaim and then allocating */
	page = __alloc_pages_direct_reclaim(gfp_mask, order,
					zonelist, high_zoneidx,
					nodemask,
2424
					alloc_flags, preferred_zone,
2425
					migratetype, &did_some_progress);
2426 2427
	if (page)
		goto got_pg;
L
Linus Torvalds 已提交
2428

2429
	/*
2430 2431
	 * If we failed to make any progress reclaiming, then we are
	 * running out of options and have to consider going OOM
2432
	 */
2433 2434
	if (!did_some_progress) {
		if ((gfp_mask & __GFP_FS) && !(gfp_mask & __GFP_NORETRY)) {
2435 2436
			if (oom_killer_disabled)
				goto nopage;
2437 2438 2439 2440
			/* Coredumps can quickly deplete all memory reserves */
			if ((current->flags & PF_DUMPCORE) &&
			    !(gfp_mask & __GFP_NOFAIL))
				goto nopage;
2441 2442
			page = __alloc_pages_may_oom(gfp_mask, order,
					zonelist, high_zoneidx,
2443 2444
					nodemask, preferred_zone,
					migratetype);
2445 2446
			if (page)
				goto got_pg;
L
Linus Torvalds 已提交
2447

2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464
			if (!(gfp_mask & __GFP_NOFAIL)) {
				/*
				 * The oom killer is not called for high-order
				 * allocations that may fail, so if no progress
				 * is being made, there are no other options and
				 * retrying is unlikely to help.
				 */
				if (order > PAGE_ALLOC_COSTLY_ORDER)
					goto nopage;
				/*
				 * The oom killer is not called for lowmem
				 * allocations to prevent needlessly killing
				 * innocent tasks.
				 */
				if (high_zoneidx < ZONE_NORMAL)
					goto nopage;
			}
2465

2466 2467
			goto restart;
		}
L
Linus Torvalds 已提交
2468 2469
	}

2470
	/* Check if we should retry the allocation */
2471
	pages_reclaimed += did_some_progress;
2472 2473
	if (should_alloc_retry(gfp_mask, order, did_some_progress,
						pages_reclaimed)) {
2474
		/* Wait for some write requests to complete then retry */
2475
		wait_iff_congested(preferred_zone, BLK_RW_ASYNC, HZ/50);
L
Linus Torvalds 已提交
2476
		goto rebalance;
2477 2478 2479 2480 2481 2482 2483 2484 2485 2486
	} else {
		/*
		 * High-order allocations do not necessarily loop after
		 * direct reclaim and reclaim/compaction depends on compaction
		 * being called after reclaim so call directly if necessary
		 */
		page = __alloc_pages_direct_compact(gfp_mask, order,
					zonelist, high_zoneidx,
					nodemask,
					alloc_flags, preferred_zone,
2487 2488 2489
					migratetype, sync_migration,
					&deferred_compaction,
					&did_some_progress);
2490 2491
		if (page)
			goto got_pg;
L
Linus Torvalds 已提交
2492 2493 2494
	}

nopage:
2495
	warn_alloc_failed(gfp_mask, order, NULL);
2496
	return page;
L
Linus Torvalds 已提交
2497
got_pg:
2498
	/*
2499 2500 2501 2502
	 * page->pfmemalloc is set when the caller had PFMEMALLOC set, is
	 * been OOM killed or specified __GFP_MEMALLOC. The expectation is
	 * that the caller is taking steps that will free more memory. The
	 * caller should avoid the page being used for !PFMEMALLOC purposes.
2503
	 */
2504
	page->pfmemalloc = !!(alloc_flags & ALLOC_NO_WATERMARKS);
2505

2506 2507
	if (kmemcheck_enabled)
		kmemcheck_pagealloc_alloc(page, order, gfp_mask);
2508

2509
	return page;
L
Linus Torvalds 已提交
2510
}
2511 2512 2513 2514 2515 2516 2517 2518 2519

/*
 * This is the 'heart' of the zoned buddy allocator.
 */
struct page *
__alloc_pages_nodemask(gfp_t gfp_mask, unsigned int order,
			struct zonelist *zonelist, nodemask_t *nodemask)
{
	enum zone_type high_zoneidx = gfp_zone(gfp_mask);
2520
	struct zone *preferred_zone;
2521
	struct page *page = NULL;
2522
	int migratetype = allocflags_to_migratetype(gfp_mask);
2523
	unsigned int cpuset_mems_cookie;
2524

2525 2526
	gfp_mask &= gfp_allowed_mask;

2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541
	lockdep_trace_alloc(gfp_mask);

	might_sleep_if(gfp_mask & __GFP_WAIT);

	if (should_fail_alloc_page(gfp_mask, order))
		return NULL;

	/*
	 * Check the zones suitable for the gfp_mask contain at least one
	 * valid zone. It's possible to have an empty zonelist as a result
	 * of GFP_THISNODE and a memoryless node
	 */
	if (unlikely(!zonelist->_zonerefs->zone))
		return NULL;

2542 2543 2544
retry_cpuset:
	cpuset_mems_cookie = get_mems_allowed();

2545
	/* The preferred zone is used for statistics later */
2546 2547 2548
	first_zones_zonelist(zonelist, high_zoneidx,
				nodemask ? : &cpuset_current_mems_allowed,
				&preferred_zone);
2549 2550
	if (!preferred_zone)
		goto out;
2551 2552

	/* First allocation attempt */
2553
	page = get_page_from_freelist(gfp_mask|__GFP_HARDWALL, nodemask, order,
2554
			zonelist, high_zoneidx, ALLOC_WMARK_LOW|ALLOC_CPUSET,
2555
			preferred_zone, migratetype);
2556 2557
	if (unlikely(!page))
		page = __alloc_pages_slowpath(gfp_mask, order,
2558
				zonelist, high_zoneidx, nodemask,
2559
				preferred_zone, migratetype);
2560 2561
	else
		page->pfmemalloc = false;
2562

2563
	trace_mm_page_alloc(page, order, gfp_mask, migratetype);
2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574

out:
	/*
	 * When updating a task's mems_allowed, it is possible to race with
	 * parallel threads in such a way that an allocation can fail while
	 * the mask is being updated. If a page allocation is about to fail,
	 * check if the cpuset changed during allocation and if so, retry.
	 */
	if (unlikely(!put_mems_allowed(cpuset_mems_cookie) && !page))
		goto retry_cpuset;

2575
	return page;
L
Linus Torvalds 已提交
2576
}
2577
EXPORT_SYMBOL(__alloc_pages_nodemask);
L
Linus Torvalds 已提交
2578 2579 2580 2581

/*
 * Common helper functions.
 */
H
Harvey Harrison 已提交
2582
unsigned long __get_free_pages(gfp_t gfp_mask, unsigned int order)
L
Linus Torvalds 已提交
2583
{
2584 2585 2586 2587 2588 2589 2590 2591
	struct page *page;

	/*
	 * __get_free_pages() returns a 32-bit address, which cannot represent
	 * a highmem page
	 */
	VM_BUG_ON((gfp_mask & __GFP_HIGHMEM) != 0);

L
Linus Torvalds 已提交
2592 2593 2594 2595 2596 2597 2598
	page = alloc_pages(gfp_mask, order);
	if (!page)
		return 0;
	return (unsigned long) page_address(page);
}
EXPORT_SYMBOL(__get_free_pages);

H
Harvey Harrison 已提交
2599
unsigned long get_zeroed_page(gfp_t gfp_mask)
L
Linus Torvalds 已提交
2600
{
2601
	return __get_free_pages(gfp_mask | __GFP_ZERO, 0);
L
Linus Torvalds 已提交
2602 2603 2604
}
EXPORT_SYMBOL(get_zeroed_page);

H
Harvey Harrison 已提交
2605
void __free_pages(struct page *page, unsigned int order)
L
Linus Torvalds 已提交
2606
{
N
Nick Piggin 已提交
2607
	if (put_page_testzero(page)) {
L
Linus Torvalds 已提交
2608
		if (order == 0)
L
Li Hong 已提交
2609
			free_hot_cold_page(page, 0);
L
Linus Torvalds 已提交
2610 2611 2612 2613 2614 2615 2616
		else
			__free_pages_ok(page, order);
	}
}

EXPORT_SYMBOL(__free_pages);

H
Harvey Harrison 已提交
2617
void free_pages(unsigned long addr, unsigned int order)
L
Linus Torvalds 已提交
2618 2619
{
	if (addr != 0) {
N
Nick Piggin 已提交
2620
		VM_BUG_ON(!virt_addr_valid((void *)addr));
L
Linus Torvalds 已提交
2621 2622 2623 2624 2625 2626
		__free_pages(virt_to_page((void *)addr), order);
	}
}

EXPORT_SYMBOL(free_pages);

A
Andi Kleen 已提交
2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641
static void *make_alloc_exact(unsigned long addr, unsigned order, size_t size)
{
	if (addr) {
		unsigned long alloc_end = addr + (PAGE_SIZE << order);
		unsigned long used = addr + PAGE_ALIGN(size);

		split_page(virt_to_page((void *)addr), order);
		while (used < alloc_end) {
			free_page(used);
			used += PAGE_SIZE;
		}
	}
	return (void *)addr;
}

2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660
/**
 * alloc_pages_exact - allocate an exact number physically-contiguous pages.
 * @size: the number of bytes to allocate
 * @gfp_mask: GFP flags for the allocation
 *
 * This function is similar to alloc_pages(), except that it allocates the
 * minimum number of pages to satisfy the request.  alloc_pages() can only
 * allocate memory in power-of-two pages.
 *
 * This function is also limited by MAX_ORDER.
 *
 * Memory allocated by this function must be released by free_pages_exact().
 */
void *alloc_pages_exact(size_t size, gfp_t gfp_mask)
{
	unsigned int order = get_order(size);
	unsigned long addr;

	addr = __get_free_pages(gfp_mask, order);
A
Andi Kleen 已提交
2661
	return make_alloc_exact(addr, order, size);
2662 2663 2664
}
EXPORT_SYMBOL(alloc_pages_exact);

A
Andi Kleen 已提交
2665 2666 2667
/**
 * alloc_pages_exact_nid - allocate an exact number of physically-contiguous
 *			   pages on a node.
2668
 * @nid: the preferred node ID where memory should be allocated
A
Andi Kleen 已提交
2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686
 * @size: the number of bytes to allocate
 * @gfp_mask: GFP flags for the allocation
 *
 * Like alloc_pages_exact(), but try to allocate on node nid first before falling
 * back.
 * Note this is not alloc_pages_exact_node() which allocates on a specific node,
 * but is not exact.
 */
void *alloc_pages_exact_nid(int nid, size_t size, gfp_t gfp_mask)
{
	unsigned order = get_order(size);
	struct page *p = alloc_pages_node(nid, gfp_mask, order);
	if (!p)
		return NULL;
	return make_alloc_exact((unsigned long)page_address(p), order, size);
}
EXPORT_SYMBOL(alloc_pages_exact_nid);

2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705
/**
 * free_pages_exact - release memory allocated via alloc_pages_exact()
 * @virt: the value returned by alloc_pages_exact.
 * @size: size of allocation, same value as passed to alloc_pages_exact().
 *
 * Release the memory allocated by a previous call to alloc_pages_exact.
 */
void free_pages_exact(void *virt, size_t size)
{
	unsigned long addr = (unsigned long)virt;
	unsigned long end = addr + PAGE_ALIGN(size);

	while (addr < end) {
		free_page(addr);
		addr += PAGE_SIZE;
	}
}
EXPORT_SYMBOL(free_pages_exact);

L
Linus Torvalds 已提交
2706 2707
static unsigned int nr_free_zone_pages(int offset)
{
2708
	struct zoneref *z;
2709 2710
	struct zone *zone;

2711
	/* Just pick one node, since fallback list is circular */
L
Linus Torvalds 已提交
2712 2713
	unsigned int sum = 0;

2714
	struct zonelist *zonelist = node_zonelist(numa_node_id(), GFP_KERNEL);
L
Linus Torvalds 已提交
2715

2716
	for_each_zone_zonelist(zone, z, zonelist, offset) {
2717
		unsigned long size = zone->present_pages;
2718
		unsigned long high = high_wmark_pages(zone);
2719 2720
		if (size > high)
			sum += size - high;
L
Linus Torvalds 已提交
2721 2722 2723 2724 2725 2726 2727 2728 2729 2730
	}

	return sum;
}

/*
 * Amount of free RAM allocatable within ZONE_DMA and ZONE_NORMAL
 */
unsigned int nr_free_buffer_pages(void)
{
A
Al Viro 已提交
2731
	return nr_free_zone_pages(gfp_zone(GFP_USER));
L
Linus Torvalds 已提交
2732
}
2733
EXPORT_SYMBOL_GPL(nr_free_buffer_pages);
L
Linus Torvalds 已提交
2734 2735 2736 2737 2738 2739

/*
 * Amount of free RAM allocatable within all zones
 */
unsigned int nr_free_pagecache_pages(void)
{
M
Mel Gorman 已提交
2740
	return nr_free_zone_pages(gfp_zone(GFP_HIGHUSER_MOVABLE));
L
Linus Torvalds 已提交
2741
}
2742 2743

static inline void show_node(struct zone *zone)
L
Linus Torvalds 已提交
2744
{
2745
	if (NUMA_BUILD)
2746
		printk("Node %d ", zone_to_nid(zone));
L
Linus Torvalds 已提交
2747 2748 2749 2750 2751 2752
}

void si_meminfo(struct sysinfo *val)
{
	val->totalram = totalram_pages;
	val->sharedram = 0;
2753
	val->freeram = global_page_state(NR_FREE_PAGES);
L
Linus Torvalds 已提交
2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767
	val->bufferram = nr_blockdev_pages();
	val->totalhigh = totalhigh_pages;
	val->freehigh = nr_free_highpages();
	val->mem_unit = PAGE_SIZE;
}

EXPORT_SYMBOL(si_meminfo);

#ifdef CONFIG_NUMA
void si_meminfo_node(struct sysinfo *val, int nid)
{
	pg_data_t *pgdat = NODE_DATA(nid);

	val->totalram = pgdat->node_present_pages;
2768
	val->freeram = node_page_state(nid, NR_FREE_PAGES);
2769
#ifdef CONFIG_HIGHMEM
L
Linus Torvalds 已提交
2770
	val->totalhigh = pgdat->node_zones[ZONE_HIGHMEM].present_pages;
2771 2772
	val->freehigh = zone_page_state(&pgdat->node_zones[ZONE_HIGHMEM],
			NR_FREE_PAGES);
2773 2774 2775 2776
#else
	val->totalhigh = 0;
	val->freehigh = 0;
#endif
L
Linus Torvalds 已提交
2777 2778 2779 2780
	val->mem_unit = PAGE_SIZE;
}
#endif

2781
/*
2782 2783
 * Determine whether the node should be displayed or not, depending on whether
 * SHOW_MEM_FILTER_NODES was passed to show_free_areas().
2784
 */
2785
bool skip_free_areas_node(unsigned int flags, int nid)
2786 2787
{
	bool ret = false;
2788
	unsigned int cpuset_mems_cookie;
2789 2790 2791 2792

	if (!(flags & SHOW_MEM_FILTER_NODES))
		goto out;

2793 2794 2795 2796
	do {
		cpuset_mems_cookie = get_mems_allowed();
		ret = !node_isset(nid, cpuset_current_mems_allowed);
	} while (!put_mems_allowed(cpuset_mems_cookie));
2797 2798 2799 2800
out:
	return ret;
}

L
Linus Torvalds 已提交
2801 2802 2803 2804 2805 2806
#define K(x) ((x) << (PAGE_SHIFT-10))

/*
 * Show free area list (used inside shift_scroll-lock stuff)
 * We also calculate the percentage fragmentation. We do this by counting the
 * memory on each free list with the exception of the first item on the list.
2807 2808
 * Suppresses nodes that are not allowed by current's cpuset if
 * SHOW_MEM_FILTER_NODES is passed.
L
Linus Torvalds 已提交
2809
 */
2810
void show_free_areas(unsigned int filter)
L
Linus Torvalds 已提交
2811
{
2812
	int cpu;
L
Linus Torvalds 已提交
2813 2814
	struct zone *zone;

2815
	for_each_populated_zone(zone) {
2816
		if (skip_free_areas_node(filter, zone_to_nid(zone)))
2817
			continue;
2818 2819
		show_node(zone);
		printk("%s per-cpu:\n", zone->name);
L
Linus Torvalds 已提交
2820

2821
		for_each_online_cpu(cpu) {
L
Linus Torvalds 已提交
2822 2823
			struct per_cpu_pageset *pageset;

2824
			pageset = per_cpu_ptr(zone->pageset, cpu);
L
Linus Torvalds 已提交
2825

2826 2827 2828
			printk("CPU %4d: hi:%5d, btch:%4d usd:%4d\n",
			       cpu, pageset->pcp.high,
			       pageset->pcp.batch, pageset->pcp.count);
L
Linus Torvalds 已提交
2829 2830 2831
		}
	}

K
KOSAKI Motohiro 已提交
2832 2833
	printk("active_anon:%lu inactive_anon:%lu isolated_anon:%lu\n"
		" active_file:%lu inactive_file:%lu isolated_file:%lu\n"
L
Lee Schermerhorn 已提交
2834
		" unevictable:%lu"
2835
		" dirty:%lu writeback:%lu unstable:%lu\n"
2836
		" free:%lu slab_reclaimable:%lu slab_unreclaimable:%lu\n"
2837
		" mapped:%lu shmem:%lu pagetables:%lu bounce:%lu\n",
2838 2839
		global_page_state(NR_ACTIVE_ANON),
		global_page_state(NR_INACTIVE_ANON),
K
KOSAKI Motohiro 已提交
2840 2841
		global_page_state(NR_ISOLATED_ANON),
		global_page_state(NR_ACTIVE_FILE),
2842
		global_page_state(NR_INACTIVE_FILE),
K
KOSAKI Motohiro 已提交
2843
		global_page_state(NR_ISOLATED_FILE),
L
Lee Schermerhorn 已提交
2844
		global_page_state(NR_UNEVICTABLE),
2845
		global_page_state(NR_FILE_DIRTY),
2846
		global_page_state(NR_WRITEBACK),
2847
		global_page_state(NR_UNSTABLE_NFS),
2848
		global_page_state(NR_FREE_PAGES),
2849 2850
		global_page_state(NR_SLAB_RECLAIMABLE),
		global_page_state(NR_SLAB_UNRECLAIMABLE),
2851
		global_page_state(NR_FILE_MAPPED),
2852
		global_page_state(NR_SHMEM),
2853 2854
		global_page_state(NR_PAGETABLE),
		global_page_state(NR_BOUNCE));
L
Linus Torvalds 已提交
2855

2856
	for_each_populated_zone(zone) {
L
Linus Torvalds 已提交
2857 2858
		int i;

2859
		if (skip_free_areas_node(filter, zone_to_nid(zone)))
2860
			continue;
L
Linus Torvalds 已提交
2861 2862 2863 2864 2865 2866
		show_node(zone);
		printk("%s"
			" free:%lukB"
			" min:%lukB"
			" low:%lukB"
			" high:%lukB"
2867 2868 2869 2870
			" active_anon:%lukB"
			" inactive_anon:%lukB"
			" active_file:%lukB"
			" inactive_file:%lukB"
L
Lee Schermerhorn 已提交
2871
			" unevictable:%lukB"
K
KOSAKI Motohiro 已提交
2872 2873
			" isolated(anon):%lukB"
			" isolated(file):%lukB"
L
Linus Torvalds 已提交
2874
			" present:%lukB"
2875 2876 2877 2878
			" mlocked:%lukB"
			" dirty:%lukB"
			" writeback:%lukB"
			" mapped:%lukB"
2879
			" shmem:%lukB"
2880 2881
			" slab_reclaimable:%lukB"
			" slab_unreclaimable:%lukB"
2882
			" kernel_stack:%lukB"
2883 2884 2885 2886
			" pagetables:%lukB"
			" unstable:%lukB"
			" bounce:%lukB"
			" writeback_tmp:%lukB"
L
Linus Torvalds 已提交
2887 2888 2889 2890
			" pages_scanned:%lu"
			" all_unreclaimable? %s"
			"\n",
			zone->name,
2891
			K(zone_page_state(zone, NR_FREE_PAGES)),
2892 2893 2894
			K(min_wmark_pages(zone)),
			K(low_wmark_pages(zone)),
			K(high_wmark_pages(zone)),
2895 2896 2897 2898
			K(zone_page_state(zone, NR_ACTIVE_ANON)),
			K(zone_page_state(zone, NR_INACTIVE_ANON)),
			K(zone_page_state(zone, NR_ACTIVE_FILE)),
			K(zone_page_state(zone, NR_INACTIVE_FILE)),
L
Lee Schermerhorn 已提交
2899
			K(zone_page_state(zone, NR_UNEVICTABLE)),
K
KOSAKI Motohiro 已提交
2900 2901
			K(zone_page_state(zone, NR_ISOLATED_ANON)),
			K(zone_page_state(zone, NR_ISOLATED_FILE)),
L
Linus Torvalds 已提交
2902
			K(zone->present_pages),
2903 2904 2905 2906
			K(zone_page_state(zone, NR_MLOCK)),
			K(zone_page_state(zone, NR_FILE_DIRTY)),
			K(zone_page_state(zone, NR_WRITEBACK)),
			K(zone_page_state(zone, NR_FILE_MAPPED)),
2907
			K(zone_page_state(zone, NR_SHMEM)),
2908 2909
			K(zone_page_state(zone, NR_SLAB_RECLAIMABLE)),
			K(zone_page_state(zone, NR_SLAB_UNRECLAIMABLE)),
2910 2911
			zone_page_state(zone, NR_KERNEL_STACK) *
				THREAD_SIZE / 1024,
2912 2913 2914 2915
			K(zone_page_state(zone, NR_PAGETABLE)),
			K(zone_page_state(zone, NR_UNSTABLE_NFS)),
			K(zone_page_state(zone, NR_BOUNCE)),
			K(zone_page_state(zone, NR_WRITEBACK_TEMP)),
L
Linus Torvalds 已提交
2916
			zone->pages_scanned,
2917
			(zone->all_unreclaimable ? "yes" : "no")
L
Linus Torvalds 已提交
2918 2919 2920 2921 2922 2923 2924
			);
		printk("lowmem_reserve[]:");
		for (i = 0; i < MAX_NR_ZONES; i++)
			printk(" %lu", zone->lowmem_reserve[i]);
		printk("\n");
	}

2925
	for_each_populated_zone(zone) {
2926
 		unsigned long nr[MAX_ORDER], flags, order, total = 0;
L
Linus Torvalds 已提交
2927

2928
		if (skip_free_areas_node(filter, zone_to_nid(zone)))
2929
			continue;
L
Linus Torvalds 已提交
2930 2931 2932 2933 2934
		show_node(zone);
		printk("%s: ", zone->name);

		spin_lock_irqsave(&zone->lock, flags);
		for (order = 0; order < MAX_ORDER; order++) {
2935 2936
			nr[order] = zone->free_area[order].nr_free;
			total += nr[order] << order;
L
Linus Torvalds 已提交
2937 2938
		}
		spin_unlock_irqrestore(&zone->lock, flags);
2939 2940
		for (order = 0; order < MAX_ORDER; order++)
			printk("%lu*%lukB ", nr[order], K(1UL) << order);
L
Linus Torvalds 已提交
2941 2942 2943
		printk("= %lukB\n", K(total));
	}

2944 2945
	printk("%ld total pagecache pages\n", global_page_state(NR_FILE_PAGES));

L
Linus Torvalds 已提交
2946 2947 2948
	show_swap_cache_info();
}

2949 2950 2951 2952 2953 2954
static void zoneref_set_zone(struct zone *zone, struct zoneref *zoneref)
{
	zoneref->zone = zone;
	zoneref->zone_idx = zone_idx(zone);
}

L
Linus Torvalds 已提交
2955 2956
/*
 * Builds allocation fallback zone lists.
2957 2958
 *
 * Add all populated zones of a node to the zonelist.
L
Linus Torvalds 已提交
2959
 */
2960 2961
static int build_zonelists_node(pg_data_t *pgdat, struct zonelist *zonelist,
				int nr_zones, enum zone_type zone_type)
L
Linus Torvalds 已提交
2962
{
2963 2964
	struct zone *zone;

2965
	BUG_ON(zone_type >= MAX_NR_ZONES);
2966
	zone_type++;
2967 2968

	do {
2969
		zone_type--;
2970
		zone = pgdat->node_zones + zone_type;
2971
		if (populated_zone(zone)) {
2972 2973
			zoneref_set_zone(zone,
				&zonelist->_zonerefs[nr_zones++]);
2974
			check_highest_zone(zone_type);
L
Linus Torvalds 已提交
2975
		}
2976

2977
	} while (zone_type);
2978
	return nr_zones;
L
Linus Torvalds 已提交
2979 2980
}

2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001

/*
 *  zonelist_order:
 *  0 = automatic detection of better ordering.
 *  1 = order by ([node] distance, -zonetype)
 *  2 = order by (-zonetype, [node] distance)
 *
 *  If not NUMA, ZONELIST_ORDER_ZONE and ZONELIST_ORDER_NODE will create
 *  the same zonelist. So only NUMA can configure this param.
 */
#define ZONELIST_ORDER_DEFAULT  0
#define ZONELIST_ORDER_NODE     1
#define ZONELIST_ORDER_ZONE     2

/* zonelist order in the kernel.
 * set_zonelist_order() will set this to NODE or ZONE.
 */
static int current_zonelist_order = ZONELIST_ORDER_DEFAULT;
static char zonelist_order_name[3][8] = {"Default", "Node", "Zone"};


L
Linus Torvalds 已提交
3002
#ifdef CONFIG_NUMA
3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032 3033 3034 3035
/* The value user specified ....changed by config */
static int user_zonelist_order = ZONELIST_ORDER_DEFAULT;
/* string for sysctl */
#define NUMA_ZONELIST_ORDER_LEN	16
char numa_zonelist_order[16] = "default";

/*
 * interface for configure zonelist ordering.
 * command line option "numa_zonelist_order"
 *	= "[dD]efault	- default, automatic configuration.
 *	= "[nN]ode 	- order by node locality, then by zone within node
 *	= "[zZ]one      - order by zone, then by locality within zone
 */

static int __parse_numa_zonelist_order(char *s)
{
	if (*s == 'd' || *s == 'D') {
		user_zonelist_order = ZONELIST_ORDER_DEFAULT;
	} else if (*s == 'n' || *s == 'N') {
		user_zonelist_order = ZONELIST_ORDER_NODE;
	} else if (*s == 'z' || *s == 'Z') {
		user_zonelist_order = ZONELIST_ORDER_ZONE;
	} else {
		printk(KERN_WARNING
			"Ignoring invalid numa_zonelist_order value:  "
			"%s\n", s);
		return -EINVAL;
	}
	return 0;
}

static __init int setup_numa_zonelist_order(char *s)
{
3036 3037 3038 3039 3040 3041 3042 3043 3044 3045
	int ret;

	if (!s)
		return 0;

	ret = __parse_numa_zonelist_order(s);
	if (ret == 0)
		strlcpy(numa_zonelist_order, s, NUMA_ZONELIST_ORDER_LEN);

	return ret;
3046 3047 3048 3049 3050 3051 3052
}
early_param("numa_zonelist_order", setup_numa_zonelist_order);

/*
 * sysctl handler for numa_zonelist_order
 */
int numa_zonelist_order_handler(ctl_table *table, int write,
3053
		void __user *buffer, size_t *length,
3054 3055 3056 3057
		loff_t *ppos)
{
	char saved_string[NUMA_ZONELIST_ORDER_LEN];
	int ret;
3058
	static DEFINE_MUTEX(zl_order_mutex);
3059

3060
	mutex_lock(&zl_order_mutex);
3061
	if (write)
3062
		strcpy(saved_string, (char*)table->data);
3063
	ret = proc_dostring(table, write, buffer, length, ppos);
3064
	if (ret)
3065
		goto out;
3066 3067 3068 3069 3070 3071 3072 3073 3074
	if (write) {
		int oldval = user_zonelist_order;
		if (__parse_numa_zonelist_order((char*)table->data)) {
			/*
			 * bogus value.  restore saved string
			 */
			strncpy((char*)table->data, saved_string,
				NUMA_ZONELIST_ORDER_LEN);
			user_zonelist_order = oldval;
3075 3076
		} else if (oldval != user_zonelist_order) {
			mutex_lock(&zonelists_mutex);
3077
			build_all_zonelists(NULL, NULL);
3078 3079
			mutex_unlock(&zonelists_mutex);
		}
3080
	}
3081 3082 3083
out:
	mutex_unlock(&zl_order_mutex);
	return ret;
3084 3085 3086
}


3087
#define MAX_NODE_LOAD (nr_online_nodes)
3088 3089
static int node_load[MAX_NUMNODES];

L
Linus Torvalds 已提交
3090
/**
3091
 * find_next_best_node - find the next node that should appear in a given node's fallback list
L
Linus Torvalds 已提交
3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103
 * @node: node whose fallback list we're appending
 * @used_node_mask: nodemask_t of already used nodes
 *
 * We use a number of factors to determine which is the next node that should
 * appear on a given node's fallback list.  The node should not have appeared
 * already in @node's fallback list, and it should be the next closest node
 * according to the distance array (which contains arbitrary distance values
 * from each node to each node in the system), and should also prefer nodes
 * with no CPUs, since presumably they'll have very little allocation pressure
 * on them otherwise.
 * It returns -1 if no node is found.
 */
3104
static int find_next_best_node(int node, nodemask_t *used_node_mask)
L
Linus Torvalds 已提交
3105
{
3106
	int n, val;
L
Linus Torvalds 已提交
3107 3108
	int min_val = INT_MAX;
	int best_node = -1;
3109
	const struct cpumask *tmp = cpumask_of_node(0);
L
Linus Torvalds 已提交
3110

3111 3112 3113 3114 3115
	/* Use the local node if we haven't already */
	if (!node_isset(node, *used_node_mask)) {
		node_set(node, *used_node_mask);
		return node;
	}
L
Linus Torvalds 已提交
3116

3117
	for_each_node_state(n, N_HIGH_MEMORY) {
L
Linus Torvalds 已提交
3118 3119 3120 3121 3122 3123 3124 3125

		/* Don't want a node to appear more than once */
		if (node_isset(n, *used_node_mask))
			continue;

		/* Use the distance array to find the distance */
		val = node_distance(node, n);

3126 3127 3128
		/* Penalize nodes under us ("prefer the next node") */
		val += (n < node);

L
Linus Torvalds 已提交
3129
		/* Give preference to headless and unused nodes */
3130 3131
		tmp = cpumask_of_node(n);
		if (!cpumask_empty(tmp))
L
Linus Torvalds 已提交
3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149
			val += PENALTY_FOR_NODE_WITH_CPUS;

		/* Slight preference for less loaded node */
		val *= (MAX_NODE_LOAD*MAX_NUMNODES);
		val += node_load[n];

		if (val < min_val) {
			min_val = val;
			best_node = n;
		}
	}

	if (best_node >= 0)
		node_set(best_node, *used_node_mask);

	return best_node;
}

3150 3151 3152 3153 3154 3155 3156

/*
 * Build zonelists ordered by node and zones within node.
 * This results in maximum locality--normal zone overflows into local
 * DMA zone, if any--but risks exhausting DMA zone.
 */
static void build_zonelists_in_node_order(pg_data_t *pgdat, int node)
L
Linus Torvalds 已提交
3157
{
3158
	int j;
L
Linus Torvalds 已提交
3159
	struct zonelist *zonelist;
3160

3161
	zonelist = &pgdat->node_zonelists[0];
3162
	for (j = 0; zonelist->_zonerefs[j].zone != NULL; j++)
3163 3164 3165
		;
	j = build_zonelists_node(NODE_DATA(node), zonelist, j,
							MAX_NR_ZONES - 1);
3166 3167
	zonelist->_zonerefs[j].zone = NULL;
	zonelist->_zonerefs[j].zone_idx = 0;
3168 3169
}

3170 3171 3172 3173 3174 3175 3176 3177
/*
 * Build gfp_thisnode zonelists
 */
static void build_thisnode_zonelists(pg_data_t *pgdat)
{
	int j;
	struct zonelist *zonelist;

3178 3179
	zonelist = &pgdat->node_zonelists[1];
	j = build_zonelists_node(pgdat, zonelist, 0, MAX_NR_ZONES - 1);
3180 3181
	zonelist->_zonerefs[j].zone = NULL;
	zonelist->_zonerefs[j].zone_idx = 0;
3182 3183
}

3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198
/*
 * Build zonelists ordered by zone and nodes within zones.
 * This results in conserving DMA zone[s] until all Normal memory is
 * exhausted, but results in overflowing to remote node while memory
 * may still exist in local DMA zone.
 */
static int node_order[MAX_NUMNODES];

static void build_zonelists_in_zone_order(pg_data_t *pgdat, int nr_nodes)
{
	int pos, j, node;
	int zone_type;		/* needs to be signed */
	struct zone *z;
	struct zonelist *zonelist;

3199 3200 3201 3202 3203 3204 3205
	zonelist = &pgdat->node_zonelists[0];
	pos = 0;
	for (zone_type = MAX_NR_ZONES - 1; zone_type >= 0; zone_type--) {
		for (j = 0; j < nr_nodes; j++) {
			node = node_order[j];
			z = &NODE_DATA(node)->node_zones[zone_type];
			if (populated_zone(z)) {
3206 3207
				zoneref_set_zone(z,
					&zonelist->_zonerefs[pos++]);
3208
				check_highest_zone(zone_type);
3209 3210 3211
			}
		}
	}
3212 3213
	zonelist->_zonerefs[pos].zone = NULL;
	zonelist->_zonerefs[pos].zone_idx = 0;
3214 3215 3216 3217 3218 3219 3220 3221 3222
}

static int default_zonelist_order(void)
{
	int nid, zone_type;
	unsigned long low_kmem_size,total_size;
	struct zone *z;
	int average_size;
	/*
T
Thomas Weber 已提交
3223
         * ZONE_DMA and ZONE_DMA32 can be very small area in the system.
3224 3225
	 * If they are really small and used heavily, the system can fall
	 * into OOM very easily.
3226
	 * This function detect ZONE_DMA/DMA32 size and configures zone order.
3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237
	 */
	/* Is there ZONE_NORMAL ? (ex. ppc has only DMA zone..) */
	low_kmem_size = 0;
	total_size = 0;
	for_each_online_node(nid) {
		for (zone_type = 0; zone_type < MAX_NR_ZONES; zone_type++) {
			z = &NODE_DATA(nid)->node_zones[zone_type];
			if (populated_zone(z)) {
				if (zone_type < ZONE_NORMAL)
					low_kmem_size += z->present_pages;
				total_size += z->present_pages;
3238 3239 3240 3241 3242 3243 3244 3245 3246
			} else if (zone_type == ZONE_NORMAL) {
				/*
				 * If any node has only lowmem, then node order
				 * is preferred to allow kernel allocations
				 * locally; otherwise, they can easily infringe
				 * on other nodes when there is an abundance of
				 * lowmem available to allocate from.
				 */
				return ZONELIST_ORDER_NODE;
3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257
			}
		}
	}
	if (!low_kmem_size ||  /* there are no DMA area. */
	    low_kmem_size > total_size/2) /* DMA/DMA32 is big. */
		return ZONELIST_ORDER_NODE;
	/*
	 * look into each node's config.
  	 * If there is a node whose DMA/DMA32 memory is very big area on
 	 * local memory, NODE_ORDER may be suitable.
         */
3258 3259
	average_size = total_size /
				(nodes_weight(node_states[N_HIGH_MEMORY]) + 1);
3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290
	for_each_online_node(nid) {
		low_kmem_size = 0;
		total_size = 0;
		for (zone_type = 0; zone_type < MAX_NR_ZONES; zone_type++) {
			z = &NODE_DATA(nid)->node_zones[zone_type];
			if (populated_zone(z)) {
				if (zone_type < ZONE_NORMAL)
					low_kmem_size += z->present_pages;
				total_size += z->present_pages;
			}
		}
		if (low_kmem_size &&
		    total_size > average_size && /* ignore small node */
		    low_kmem_size > total_size * 70/100)
			return ZONELIST_ORDER_NODE;
	}
	return ZONELIST_ORDER_ZONE;
}

static void set_zonelist_order(void)
{
	if (user_zonelist_order == ZONELIST_ORDER_DEFAULT)
		current_zonelist_order = default_zonelist_order();
	else
		current_zonelist_order = user_zonelist_order;
}

static void build_zonelists(pg_data_t *pgdat)
{
	int j, node, load;
	enum zone_type i;
L
Linus Torvalds 已提交
3291
	nodemask_t used_mask;
3292 3293 3294
	int local_node, prev_node;
	struct zonelist *zonelist;
	int order = current_zonelist_order;
L
Linus Torvalds 已提交
3295 3296

	/* initialize zonelists */
3297
	for (i = 0; i < MAX_ZONELISTS; i++) {
L
Linus Torvalds 已提交
3298
		zonelist = pgdat->node_zonelists + i;
3299 3300
		zonelist->_zonerefs[0].zone = NULL;
		zonelist->_zonerefs[0].zone_idx = 0;
L
Linus Torvalds 已提交
3301 3302 3303 3304
	}

	/* NUMA-aware ordering of nodes */
	local_node = pgdat->node_id;
3305
	load = nr_online_nodes;
L
Linus Torvalds 已提交
3306 3307
	prev_node = local_node;
	nodes_clear(used_mask);
3308 3309 3310 3311

	memset(node_order, 0, sizeof(node_order));
	j = 0;

L
Linus Torvalds 已提交
3312
	while ((node = find_next_best_node(local_node, &used_mask)) >= 0) {
3313 3314 3315 3316 3317 3318 3319 3320 3321
		int distance = node_distance(local_node, node);

		/*
		 * If another node is sufficiently far away then it is better
		 * to reclaim pages in a zone before going off node.
		 */
		if (distance > RECLAIM_DISTANCE)
			zone_reclaim_mode = 1;

L
Linus Torvalds 已提交
3322 3323 3324 3325 3326
		/*
		 * We don't want to pressure a particular node.
		 * So adding penalty to the first node in same
		 * distance group to make it round-robin.
		 */
3327
		if (distance != node_distance(local_node, prev_node))
3328 3329
			node_load[node] = load;

L
Linus Torvalds 已提交
3330 3331
		prev_node = node;
		load--;
3332 3333 3334 3335 3336
		if (order == ZONELIST_ORDER_NODE)
			build_zonelists_in_node_order(pgdat, node);
		else
			node_order[j++] = node;	/* remember order */
	}
L
Linus Torvalds 已提交
3337

3338 3339 3340
	if (order == ZONELIST_ORDER_ZONE) {
		/* calculate node order -- i.e., DMA last! */
		build_zonelists_in_zone_order(pgdat, j);
L
Linus Torvalds 已提交
3341
	}
3342 3343

	build_thisnode_zonelists(pgdat);
L
Linus Torvalds 已提交
3344 3345
}

3346
/* Construct the zonelist performance cache - see further mmzone.h */
3347
static void build_zonelist_cache(pg_data_t *pgdat)
3348
{
3349 3350
	struct zonelist *zonelist;
	struct zonelist_cache *zlc;
3351
	struct zoneref *z;
3352

3353 3354 3355
	zonelist = &pgdat->node_zonelists[0];
	zonelist->zlcache_ptr = zlc = &zonelist->zlcache;
	bitmap_zero(zlc->fullzones, MAX_ZONES_PER_ZONELIST);
3356 3357
	for (z = zonelist->_zonerefs; z->zone; z++)
		zlc->z_to_n[z - zonelist->_zonerefs] = zonelist_node_idx(z);
3358 3359
}

3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377
#ifdef CONFIG_HAVE_MEMORYLESS_NODES
/*
 * Return node id of node used for "local" allocations.
 * I.e., first node id of first zone in arg node's generic zonelist.
 * Used for initializing percpu 'numa_mem', which is used primarily
 * for kernel allocations, so use GFP_KERNEL flags to locate zonelist.
 */
int local_memory_node(int node)
{
	struct zone *zone;

	(void)first_zones_zonelist(node_zonelist(node, GFP_KERNEL),
				   gfp_zone(GFP_KERNEL),
				   NULL,
				   &zone);
	return zone->node;
}
#endif
3378

L
Linus Torvalds 已提交
3379 3380
#else	/* CONFIG_NUMA */

3381 3382 3383 3384 3385 3386
static void set_zonelist_order(void)
{
	current_zonelist_order = ZONELIST_ORDER_ZONE;
}

static void build_zonelists(pg_data_t *pgdat)
L
Linus Torvalds 已提交
3387
{
3388
	int node, local_node;
3389 3390
	enum zone_type j;
	struct zonelist *zonelist;
L
Linus Torvalds 已提交
3391 3392 3393

	local_node = pgdat->node_id;

3394 3395
	zonelist = &pgdat->node_zonelists[0];
	j = build_zonelists_node(pgdat, zonelist, 0, MAX_NR_ZONES - 1);
L
Linus Torvalds 已提交
3396

3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409
	/*
	 * Now we build the zonelist so that it contains the zones
	 * of all the other nodes.
	 * We don't want to pressure a particular node, so when
	 * building the zones for node N, we make sure that the
	 * zones coming right after the local ones are those from
	 * node N+1 (modulo N)
	 */
	for (node = local_node + 1; node < MAX_NUMNODES; node++) {
		if (!node_online(node))
			continue;
		j = build_zonelists_node(NODE_DATA(node), zonelist, j,
							MAX_NR_ZONES - 1);
L
Linus Torvalds 已提交
3410
	}
3411 3412 3413 3414 3415 3416 3417
	for (node = 0; node < local_node; node++) {
		if (!node_online(node))
			continue;
		j = build_zonelists_node(NODE_DATA(node), zonelist, j,
							MAX_NR_ZONES - 1);
	}

3418 3419
	zonelist->_zonerefs[j].zone = NULL;
	zonelist->_zonerefs[j].zone_idx = 0;
L
Linus Torvalds 已提交
3420 3421
}

3422
/* non-NUMA variant of zonelist performance cache - just NULL zlcache_ptr */
3423
static void build_zonelist_cache(pg_data_t *pgdat)
3424
{
3425
	pgdat->node_zonelists[0].zlcache_ptr = NULL;
3426 3427
}

L
Linus Torvalds 已提交
3428 3429
#endif	/* CONFIG_NUMA */

3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446
/*
 * Boot pageset table. One per cpu which is going to be used for all
 * zones and all nodes. The parameters will be set in such a way
 * that an item put on a list will immediately be handed over to
 * the buddy list. This is safe since pageset manipulation is done
 * with interrupts disabled.
 *
 * The boot_pagesets must be kept even after bootup is complete for
 * unused processors and/or zones. They do play a role for bootstrapping
 * hotplugged processors.
 *
 * zoneinfo_show() and maybe other functions do
 * not check if the processor is online before following the pageset pointer.
 * Other parts of the kernel may not check if the zone is available.
 */
static void setup_pageset(struct per_cpu_pageset *p, unsigned long batch);
static DEFINE_PER_CPU(struct per_cpu_pageset, boot_pageset);
3447
static void setup_zone_pageset(struct zone *zone);
3448

3449 3450 3451 3452 3453 3454
/*
 * Global mutex to protect against size modification of zonelists
 * as well as to serialize pageset setup for the new populated zone.
 */
DEFINE_MUTEX(zonelists_mutex);

3455
/* return values int ....just for stop_machine() */
3456
static int __build_all_zonelists(void *data)
L
Linus Torvalds 已提交
3457
{
3458
	int nid;
3459
	int cpu;
3460
	pg_data_t *self = data;
3461

3462 3463 3464
#ifdef CONFIG_NUMA
	memset(node_load, 0, sizeof(node_load));
#endif
3465 3466 3467 3468 3469 3470

	if (self && !node_online(self->node_id)) {
		build_zonelists(self);
		build_zonelist_cache(self);
	}

3471
	for_each_online_node(nid) {
3472 3473 3474 3475
		pg_data_t *pgdat = NODE_DATA(nid);

		build_zonelists(pgdat);
		build_zonelist_cache(pgdat);
3476
	}
3477 3478 3479 3480 3481 3482 3483 3484 3485 3486 3487 3488 3489 3490

	/*
	 * Initialize the boot_pagesets that are going to be used
	 * for bootstrapping processors. The real pagesets for
	 * each zone will be allocated later when the per cpu
	 * allocator is available.
	 *
	 * boot_pagesets are used also for bootstrapping offline
	 * cpus if the system is already booted because the pagesets
	 * are needed to initialize allocators on a specific cpu too.
	 * F.e. the percpu allocator needs the page allocator which
	 * needs the percpu allocator in order to allocate its pagesets
	 * (a chicken-egg dilemma).
	 */
3491
	for_each_possible_cpu(cpu) {
3492 3493
		setup_pageset(&per_cpu(boot_pageset, cpu), 0);

3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507
#ifdef CONFIG_HAVE_MEMORYLESS_NODES
		/*
		 * We now know the "local memory node" for each node--
		 * i.e., the node of the first zone in the generic zonelist.
		 * Set up numa_mem percpu variable for on-line cpus.  During
		 * boot, only the boot cpu should be on-line;  we'll init the
		 * secondary cpus' numa_mem as they come on-line.  During
		 * node/memory hotplug, we'll fixup all on-line cpus.
		 */
		if (cpu_online(cpu))
			set_cpu_numa_mem(cpu, local_memory_node(cpu_to_node(cpu)));
#endif
	}

3508 3509 3510
	return 0;
}

3511 3512 3513 3514
/*
 * Called with zonelists_mutex held always
 * unless system_state == SYSTEM_BOOTING.
 */
3515
void __ref build_all_zonelists(pg_data_t *pgdat, struct zone *zone)
3516
{
3517 3518
	set_zonelist_order();

3519
	if (system_state == SYSTEM_BOOTING) {
3520
		__build_all_zonelists(NULL);
3521
		mminit_verify_zonelist();
3522 3523
		cpuset_init_current_mems_allowed();
	} else {
S
Simon Arlott 已提交
3524
		/* we have to stop all cpus to guarantee there is no user
3525
		   of zonelist */
3526
#ifdef CONFIG_MEMORY_HOTPLUG
3527 3528
		if (zone)
			setup_zone_pageset(zone);
3529
#endif
3530
		stop_machine(__build_all_zonelists, pgdat, NULL);
3531 3532
		/* cpuset refresh routine should be here */
	}
3533
	vm_total_pages = nr_free_pagecache_pages();
3534 3535 3536 3537 3538 3539 3540
	/*
	 * Disable grouping by mobility if the number of pages in the
	 * system is too low to allow the mechanism to work. It would be
	 * more accurate, but expensive to check per-zone. This check is
	 * made on memory-hotadd so a system can start with mobility
	 * disabled and enable it later
	 */
3541
	if (vm_total_pages < (pageblock_nr_pages * MIGRATE_TYPES))
3542 3543 3544 3545 3546 3547
		page_group_by_mobility_disabled = 1;
	else
		page_group_by_mobility_disabled = 0;

	printk("Built %i zonelists in %s order, mobility grouping %s.  "
		"Total pages: %ld\n",
3548
			nr_online_nodes,
3549
			zonelist_order_name[current_zonelist_order],
3550
			page_group_by_mobility_disabled ? "off" : "on",
3551 3552 3553 3554
			vm_total_pages);
#ifdef CONFIG_NUMA
	printk("Policy zone: %s\n", zone_names[policy_zone]);
#endif
L
Linus Torvalds 已提交
3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569
}

/*
 * Helper functions to size the waitqueue hash table.
 * Essentially these want to choose hash table sizes sufficiently
 * large so that collisions trying to wait on pages are rare.
 * But in fact, the number of active page waitqueues on typical
 * systems is ridiculously low, less than 200. So this is even
 * conservative, even though it seems large.
 *
 * The constant PAGES_PER_WAITQUEUE specifies the ratio of pages to
 * waitqueues, i.e. the size of the waitq table given the number of pages.
 */
#define PAGES_PER_WAITQUEUE	256

3570
#ifndef CONFIG_MEMORY_HOTPLUG
3571
static inline unsigned long wait_table_hash_nr_entries(unsigned long pages)
L
Linus Torvalds 已提交
3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588
{
	unsigned long size = 1;

	pages /= PAGES_PER_WAITQUEUE;

	while (size < pages)
		size <<= 1;

	/*
	 * Once we have dozens or even hundreds of threads sleeping
	 * on IO we've got bigger problems than wait queue collision.
	 * Limit the size of the wait table to a reasonable size.
	 */
	size = min(size, 4096UL);

	return max(size, 4UL);
}
3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611
#else
/*
 * A zone's size might be changed by hot-add, so it is not possible to determine
 * a suitable size for its wait_table.  So we use the maximum size now.
 *
 * The max wait table size = 4096 x sizeof(wait_queue_head_t).   ie:
 *
 *    i386 (preemption config)    : 4096 x 16 = 64Kbyte.
 *    ia64, x86-64 (no preemption): 4096 x 20 = 80Kbyte.
 *    ia64, x86-64 (preemption)   : 4096 x 24 = 96Kbyte.
 *
 * The maximum entries are prepared when a zone's memory is (512K + 256) pages
 * or more by the traditional way. (See above).  It equals:
 *
 *    i386, x86-64, powerpc(4K page size) : =  ( 2G + 1M)byte.
 *    ia64(16K page size)                 : =  ( 8G + 4M)byte.
 *    powerpc (64K page size)             : =  (32G +16M)byte.
 */
static inline unsigned long wait_table_hash_nr_entries(unsigned long pages)
{
	return 4096UL;
}
#endif
L
Linus Torvalds 已提交
3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624

/*
 * This is an integer logarithm so that shifts can be used later
 * to extract the more random high bits from the multiplicative
 * hash function before the remainder is taken.
 */
static inline unsigned long wait_table_bits(unsigned long size)
{
	return ffz(~size);
}

#define LONG_ALIGN(x) (((x)+(sizeof(long))-1)&~((sizeof(long))-1))

3625 3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638
/*
 * Check if a pageblock contains reserved pages
 */
static int pageblock_is_reserved(unsigned long start_pfn, unsigned long end_pfn)
{
	unsigned long pfn;

	for (pfn = start_pfn; pfn < end_pfn; pfn++) {
		if (!pfn_valid_within(pfn) || PageReserved(pfn_to_page(pfn)))
			return 1;
	}
	return 0;
}

3639
/*
3640
 * Mark a number of pageblocks as MIGRATE_RESERVE. The number
3641 3642
 * of blocks reserved is based on min_wmark_pages(zone). The memory within
 * the reserve will tend to store contiguous free pages. Setting min_free_kbytes
3643 3644 3645 3646 3647
 * higher will lead to a bigger reserve which will get freed as contiguous
 * blocks as reclaim kicks in
 */
static void setup_zone_migrate_reserve(struct zone *zone)
{
3648
	unsigned long start_pfn, pfn, end_pfn, block_end_pfn;
3649
	struct page *page;
3650 3651
	unsigned long block_migratetype;
	int reserve;
3652

3653 3654 3655 3656 3657 3658
	/*
	 * Get the start pfn, end pfn and the number of blocks to reserve
	 * We have to be careful to be aligned to pageblock_nr_pages to
	 * make sure that we always check pfn_valid for the first page in
	 * the block.
	 */
3659 3660
	start_pfn = zone->zone_start_pfn;
	end_pfn = start_pfn + zone->spanned_pages;
3661
	start_pfn = roundup(start_pfn, pageblock_nr_pages);
3662
	reserve = roundup(min_wmark_pages(zone), pageblock_nr_pages) >>
3663
							pageblock_order;
3664

3665 3666 3667 3668 3669 3670 3671 3672 3673
	/*
	 * Reserve blocks are generally in place to help high-order atomic
	 * allocations that are short-lived. A min_free_kbytes value that
	 * would result in more than 2 reserve blocks for atomic allocations
	 * is assumed to be in place to help anti-fragmentation for the
	 * future allocation of hugepages at runtime.
	 */
	reserve = min(2, reserve);

3674
	for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) {
3675 3676 3677 3678
		if (!pfn_valid(pfn))
			continue;
		page = pfn_to_page(pfn);

3679 3680 3681 3682
		/* Watch out for overlapping nodes */
		if (page_to_nid(page) != zone_to_nid(zone))
			continue;

3683 3684
		block_migratetype = get_pageblock_migratetype(page);

3685 3686 3687 3688 3689 3690 3691 3692 3693
		/* Only test what is necessary when the reserves are not met */
		if (reserve > 0) {
			/*
			 * Blocks with reserved pages will never free, skip
			 * them.
			 */
			block_end_pfn = min(pfn + pageblock_nr_pages, end_pfn);
			if (pageblock_is_reserved(pfn, block_end_pfn))
				continue;
3694

3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709
			/* If this block is reserved, account for it */
			if (block_migratetype == MIGRATE_RESERVE) {
				reserve--;
				continue;
			}

			/* Suitable for reserving if this block is movable */
			if (block_migratetype == MIGRATE_MOVABLE) {
				set_pageblock_migratetype(page,
							MIGRATE_RESERVE);
				move_freepages_block(zone, page,
							MIGRATE_RESERVE);
				reserve--;
				continue;
			}
3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721
		}

		/*
		 * If the reserve is met and this is a previous reserved block,
		 * take it back
		 */
		if (block_migratetype == MIGRATE_RESERVE) {
			set_pageblock_migratetype(page, MIGRATE_MOVABLE);
			move_freepages_block(zone, page, MIGRATE_MOVABLE);
		}
	}
}
M
Mel Gorman 已提交
3722

L
Linus Torvalds 已提交
3723 3724 3725 3726 3727
/*
 * Initially all pages are reserved - free ones are freed
 * up by free_all_bootmem() once the early boot process is
 * done. Non-atomic initialization, single-pass.
 */
3728
void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
D
Dave Hansen 已提交
3729
		unsigned long start_pfn, enum memmap_context context)
L
Linus Torvalds 已提交
3730 3731
{
	struct page *page;
A
Andy Whitcroft 已提交
3732 3733
	unsigned long end_pfn = start_pfn + size;
	unsigned long pfn;
3734
	struct zone *z;
L
Linus Torvalds 已提交
3735

3736 3737 3738
	if (highest_memmap_pfn < end_pfn - 1)
		highest_memmap_pfn = end_pfn - 1;

3739
	z = &NODE_DATA(nid)->node_zones[zone];
3740
	for (pfn = start_pfn; pfn < end_pfn; pfn++) {
D
Dave Hansen 已提交
3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751
		/*
		 * There can be holes in boot-time mem_map[]s
		 * handed to this function.  They do not
		 * exist on hotplugged memory.
		 */
		if (context == MEMMAP_EARLY) {
			if (!early_pfn_valid(pfn))
				continue;
			if (!early_pfn_in_nid(pfn, nid))
				continue;
		}
A
Andy Whitcroft 已提交
3752 3753
		page = pfn_to_page(pfn);
		set_page_links(page, zone, nid, pfn);
3754
		mminit_verify_page_links(page, zone, nid, pfn);
3755
		init_page_count(page);
L
Linus Torvalds 已提交
3756 3757
		reset_page_mapcount(page);
		SetPageReserved(page);
3758 3759 3760 3761 3762
		/*
		 * Mark the block movable so that blocks are reserved for
		 * movable at startup. This will force kernel allocations
		 * to reserve their blocks rather than leaking throughout
		 * the address space during boot when many long-lived
3763 3764 3765
		 * kernel allocations are made. Later some blocks near
		 * the start are marked MIGRATE_RESERVE by
		 * setup_zone_migrate_reserve()
3766 3767 3768 3769 3770
		 *
		 * bitmap is created for zone's valid pfn range. but memmap
		 * can be created for invalid pages (for alignment)
		 * check here not to call set_pageblock_migratetype() against
		 * pfn out of zone.
3771
		 */
3772 3773 3774
		if ((z->zone_start_pfn <= pfn)
		    && (pfn < z->zone_start_pfn + z->spanned_pages)
		    && !(pfn & (pageblock_nr_pages - 1)))
3775
			set_pageblock_migratetype(page, MIGRATE_MOVABLE);
3776

L
Linus Torvalds 已提交
3777 3778 3779 3780
		INIT_LIST_HEAD(&page->lru);
#ifdef WANT_PAGE_VIRTUAL
		/* The shift won't overflow because ZONE_NORMAL is below 4G. */
		if (!is_highmem_idx(zone))
3781
			set_page_address(page, __va(pfn << PAGE_SHIFT));
L
Linus Torvalds 已提交
3782 3783 3784 3785
#endif
	}
}

3786
static void __meminit zone_init_free_lists(struct zone *zone)
L
Linus Torvalds 已提交
3787
{
3788 3789 3790
	int order, t;
	for_each_migratetype_order(order, t) {
		INIT_LIST_HEAD(&zone->free_area[order].free_list[t]);
L
Linus Torvalds 已提交
3791 3792 3793 3794 3795 3796
		zone->free_area[order].nr_free = 0;
	}
}

#ifndef __HAVE_ARCH_MEMMAP_INIT
#define memmap_init(size, nid, zone, start_pfn) \
D
Dave Hansen 已提交
3797
	memmap_init_zone((size), (nid), (zone), (start_pfn), MEMMAP_EARLY)
L
Linus Torvalds 已提交
3798 3799
#endif

3800
static int __meminit zone_batchsize(struct zone *zone)
3801
{
3802
#ifdef CONFIG_MMU
3803 3804 3805 3806
	int batch;

	/*
	 * The per-cpu-pages pools are set to around 1000th of the
3807
	 * size of the zone.  But no more than 1/2 of a meg.
3808 3809 3810 3811
	 *
	 * OK, so we don't know how big the cache is.  So guess.
	 */
	batch = zone->present_pages / 1024;
3812 3813
	if (batch * PAGE_SIZE > 512 * 1024)
		batch = (512 * 1024) / PAGE_SIZE;
3814 3815 3816 3817 3818
	batch /= 4;		/* We effectively *= 4 below */
	if (batch < 1)
		batch = 1;

	/*
3819 3820 3821
	 * Clamp the batch to a 2^n - 1 value. Having a power
	 * of 2 value was found to be more likely to have
	 * suboptimal cache aliasing properties in some cases.
3822
	 *
3823 3824 3825 3826
	 * For example if 2 tasks are alternately allocating
	 * batches of pages, one task can end up with a lot
	 * of pages of one half of the possible page colors
	 * and the other with pages of the other colors.
3827
	 */
3828
	batch = rounddown_pow_of_two(batch + batch/2) - 1;
3829

3830
	return batch;
3831 3832 3833 3834 3835 3836 3837 3838 3839 3840 3841 3842 3843 3844 3845 3846 3847

#else
	/* The deferral and batching of frees should be suppressed under NOMMU
	 * conditions.
	 *
	 * The problem is that NOMMU needs to be able to allocate large chunks
	 * of contiguous memory as there's no hardware page translation to
	 * assemble apparent contiguous memory from discontiguous pages.
	 *
	 * Queueing large contiguous runs of pages for batching, however,
	 * causes the pages to actually be freed in smaller chunks.  As there
	 * can be a significant delay between the individual batches being
	 * recycled, this leads to the once large chunks of space being
	 * fragmented and becoming unavailable for high-order allocations.
	 */
	return 0;
#endif
3848 3849
}

A
Adrian Bunk 已提交
3850
static void setup_pageset(struct per_cpu_pageset *p, unsigned long batch)
3851 3852
{
	struct per_cpu_pages *pcp;
3853
	int migratetype;
3854

3855 3856
	memset(p, 0, sizeof(*p));

3857
	pcp = &p->pcp;
3858 3859 3860
	pcp->count = 0;
	pcp->high = 6 * batch;
	pcp->batch = max(1UL, 1 * batch);
3861 3862
	for (migratetype = 0; migratetype < MIGRATE_PCPTYPES; migratetype++)
		INIT_LIST_HEAD(&pcp->lists[migratetype]);
3863 3864
}

3865 3866 3867 3868 3869 3870 3871 3872 3873 3874
/*
 * setup_pagelist_highmark() sets the high water mark for hot per_cpu_pagelist
 * to the value high for the pageset p.
 */

static void setup_pagelist_highmark(struct per_cpu_pageset *p,
				unsigned long high)
{
	struct per_cpu_pages *pcp;

3875
	pcp = &p->pcp;
3876 3877 3878 3879 3880 3881
	pcp->high = high;
	pcp->batch = max(1UL, high/4);
	if ((high/4) > (PAGE_SHIFT * 8))
		pcp->batch = PAGE_SHIFT * 8;
}

3882
static void __meminit setup_zone_pageset(struct zone *zone)
3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899
{
	int cpu;

	zone->pageset = alloc_percpu(struct per_cpu_pageset);

	for_each_possible_cpu(cpu) {
		struct per_cpu_pageset *pcp = per_cpu_ptr(zone->pageset, cpu);

		setup_pageset(pcp, zone_batchsize(zone));

		if (percpu_pagelist_fraction)
			setup_pagelist_highmark(pcp,
				(zone->present_pages /
					percpu_pagelist_fraction));
	}
}

3900
/*
3901 3902
 * Allocate per cpu pagesets and initialize them.
 * Before this call only boot pagesets were available.
3903
 */
3904
void __init setup_per_cpu_pageset(void)
3905
{
3906
	struct zone *zone;
3907

3908 3909
	for_each_populated_zone(zone)
		setup_zone_pageset(zone);
3910 3911
}

S
Sam Ravnborg 已提交
3912
static noinline __init_refok
3913
int zone_wait_table_init(struct zone *zone, unsigned long zone_size_pages)
3914 3915 3916
{
	int i;
	struct pglist_data *pgdat = zone->zone_pgdat;
3917
	size_t alloc_size;
3918 3919 3920 3921 3922

	/*
	 * The per-page waitqueue mechanism uses hashed waitqueues
	 * per zone.
	 */
3923 3924 3925 3926
	zone->wait_table_hash_nr_entries =
		 wait_table_hash_nr_entries(zone_size_pages);
	zone->wait_table_bits =
		wait_table_bits(zone->wait_table_hash_nr_entries);
3927 3928 3929
	alloc_size = zone->wait_table_hash_nr_entries
					* sizeof(wait_queue_head_t);

3930
	if (!slab_is_available()) {
3931
		zone->wait_table = (wait_queue_head_t *)
3932
			alloc_bootmem_node_nopanic(pgdat, alloc_size);
3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943
	} else {
		/*
		 * This case means that a zone whose size was 0 gets new memory
		 * via memory hot-add.
		 * But it may be the case that a new node was hot-added.  In
		 * this case vmalloc() will not be able to use this new node's
		 * memory - this wait_table must be initialized to use this new
		 * node itself as well.
		 * To use this new node's memory, further consideration will be
		 * necessary.
		 */
3944
		zone->wait_table = vmalloc(alloc_size);
3945 3946 3947
	}
	if (!zone->wait_table)
		return -ENOMEM;
3948

3949
	for(i = 0; i < zone->wait_table_hash_nr_entries; ++i)
3950
		init_waitqueue_head(zone->wait_table + i);
3951 3952

	return 0;
3953 3954
}

3955
static __meminit void zone_pcp_init(struct zone *zone)
3956
{
3957 3958 3959 3960 3961 3962
	/*
	 * per cpu subsystem is not up at this point. The following code
	 * relies on the ability of the linker to provide the
	 * offset of a (static) per cpu variable into the per cpu area.
	 */
	zone->pageset = &boot_pageset;
3963

A
Anton Blanchard 已提交
3964
	if (zone->present_pages)
3965 3966 3967
		printk(KERN_DEBUG "  %s zone: %lu pages, LIFO batch:%u\n",
			zone->name, zone->present_pages,
					 zone_batchsize(zone));
3968 3969
}

3970
int __meminit init_currently_empty_zone(struct zone *zone,
3971
					unsigned long zone_start_pfn,
D
Dave Hansen 已提交
3972 3973
					unsigned long size,
					enum memmap_context context)
3974 3975
{
	struct pglist_data *pgdat = zone->zone_pgdat;
3976 3977 3978 3979
	int ret;
	ret = zone_wait_table_init(zone, size);
	if (ret)
		return ret;
3980 3981 3982 3983
	pgdat->nr_zones = zone_idx(zone) + 1;

	zone->zone_start_pfn = zone_start_pfn;

3984 3985 3986 3987 3988 3989
	mminit_dprintk(MMINIT_TRACE, "memmap_init",
			"Initialising map node %d zone %lu pfns %lu -> %lu\n",
			pgdat->node_id,
			(unsigned long)zone_idx(zone),
			zone_start_pfn, (zone_start_pfn + size));

3990
	zone_init_free_lists(zone);
3991 3992

	return 0;
3993 3994
}

T
Tejun Heo 已提交
3995
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
3996 3997 3998 3999 4000 4001 4002
#ifndef CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID
/*
 * Required by SPARSEMEM. Given a PFN, return what node the PFN is on.
 * Architectures may implement their own version but if add_active_range()
 * was used and there are no special requirements, this is a convenient
 * alternative
 */
4003
int __meminit __early_pfn_to_nid(unsigned long pfn)
4004
{
4005 4006
	unsigned long start_pfn, end_pfn;
	int i, nid;
4007

4008
	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid)
4009
		if (start_pfn <= pfn && pfn < end_pfn)
4010
			return nid;
4011 4012
	/* This is a memory hole */
	return -1;
4013 4014 4015
}
#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */

4016 4017
int __meminit early_pfn_to_nid(unsigned long pfn)
{
4018 4019 4020 4021 4022 4023 4024
	int nid;

	nid = __early_pfn_to_nid(pfn);
	if (nid >= 0)
		return nid;
	/* just returns 0 */
	return 0;
4025 4026
}

4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037
#ifdef CONFIG_NODES_SPAN_OTHER_NODES
bool __meminit early_pfn_in_nid(unsigned long pfn, int node)
{
	int nid;

	nid = __early_pfn_to_nid(pfn);
	if (nid >= 0 && nid != node)
		return false;
	return true;
}
#endif
4038

4039 4040
/**
 * free_bootmem_with_active_regions - Call free_bootmem_node for each active range
4041 4042
 * @nid: The node to free memory on. If MAX_NUMNODES, all nodes are freed.
 * @max_low_pfn: The highest PFN that will be passed to free_bootmem_node
4043 4044 4045 4046 4047
 *
 * If an architecture guarantees that all ranges registered with
 * add_active_ranges() contain no holes and may be freed, this
 * this function may be used instead of calling free_bootmem() manually.
 */
4048
void __init free_bootmem_with_active_regions(int nid, unsigned long max_low_pfn)
4049
{
4050 4051
	unsigned long start_pfn, end_pfn;
	int i, this_nid;
4052

4053 4054 4055
	for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, &this_nid) {
		start_pfn = min(start_pfn, max_low_pfn);
		end_pfn = min(end_pfn, max_low_pfn);
4056

4057 4058 4059 4060
		if (start_pfn < end_pfn)
			free_bootmem_node(NODE_DATA(this_nid),
					  PFN_PHYS(start_pfn),
					  (end_pfn - start_pfn) << PAGE_SHIFT);
4061 4062 4063
	}
}

4064 4065
/**
 * sparse_memory_present_with_active_regions - Call memory_present for each active range
4066
 * @nid: The node to call memory_present for. If MAX_NUMNODES, all nodes will be used.
4067 4068 4069
 *
 * If an architecture guarantees that all ranges registered with
 * add_active_ranges() contain no holes and may be freed, this
4070
 * function may be used instead of calling memory_present() manually.
4071 4072 4073
 */
void __init sparse_memory_present_with_active_regions(int nid)
{
4074 4075
	unsigned long start_pfn, end_pfn;
	int i, this_nid;
4076

4077 4078
	for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, &this_nid)
		memory_present(this_nid, start_pfn, end_pfn);
4079 4080 4081 4082
}

/**
 * get_pfn_range_for_nid - Return the start and end page frames for a node
4083 4084 4085
 * @nid: The nid to return the range for. If MAX_NUMNODES, the min and max PFN are returned.
 * @start_pfn: Passed by reference. On return, it will have the node start_pfn.
 * @end_pfn: Passed by reference. On return, it will have the node end_pfn.
4086 4087 4088 4089
 *
 * It returns the start and end page frame of a node based on information
 * provided by an arch calling add_active_range(). If called for a node
 * with no available memory, a warning is printed and the start and end
4090
 * PFNs will be 0.
4091
 */
4092
void __meminit get_pfn_range_for_nid(unsigned int nid,
4093 4094
			unsigned long *start_pfn, unsigned long *end_pfn)
{
4095
	unsigned long this_start_pfn, this_end_pfn;
4096
	int i;
4097

4098 4099 4100
	*start_pfn = -1UL;
	*end_pfn = 0;

4101 4102 4103
	for_each_mem_pfn_range(i, nid, &this_start_pfn, &this_end_pfn, NULL) {
		*start_pfn = min(*start_pfn, this_start_pfn);
		*end_pfn = max(*end_pfn, this_end_pfn);
4104 4105
	}

4106
	if (*start_pfn == -1UL)
4107 4108 4109
		*start_pfn = 0;
}

M
Mel Gorman 已提交
4110 4111 4112 4113 4114
/*
 * This finds a zone that can be used for ZONE_MOVABLE pages. The
 * assumption is made that zones within a node are ordered in monotonic
 * increasing memory addresses so that the "highest" populated zone is used
 */
A
Adrian Bunk 已提交
4115
static void __init find_usable_zone_for_movable(void)
M
Mel Gorman 已提交
4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132
{
	int zone_index;
	for (zone_index = MAX_NR_ZONES - 1; zone_index >= 0; zone_index--) {
		if (zone_index == ZONE_MOVABLE)
			continue;

		if (arch_zone_highest_possible_pfn[zone_index] >
				arch_zone_lowest_possible_pfn[zone_index])
			break;
	}

	VM_BUG_ON(zone_index == -1);
	movable_zone = zone_index;
}

/*
 * The zone ranges provided by the architecture do not include ZONE_MOVABLE
L
Lucas De Marchi 已提交
4133
 * because it is sized independent of architecture. Unlike the other zones,
M
Mel Gorman 已提交
4134 4135 4136 4137 4138 4139 4140
 * the starting point for ZONE_MOVABLE is not fixed. It may be different
 * in each node depending on the size of each node and how evenly kernelcore
 * is distributed. This helper function adjusts the zone ranges
 * provided by the architecture for a given node by using the end of the
 * highest usable zone for ZONE_MOVABLE. This preserves the assumption that
 * zones within a node are in order of monotonic increases memory addresses
 */
A
Adrian Bunk 已提交
4141
static void __meminit adjust_zone_range_for_zone_movable(int nid,
M
Mel Gorman 已提交
4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166
					unsigned long zone_type,
					unsigned long node_start_pfn,
					unsigned long node_end_pfn,
					unsigned long *zone_start_pfn,
					unsigned long *zone_end_pfn)
{
	/* Only adjust if ZONE_MOVABLE is on this node */
	if (zone_movable_pfn[nid]) {
		/* Size ZONE_MOVABLE */
		if (zone_type == ZONE_MOVABLE) {
			*zone_start_pfn = zone_movable_pfn[nid];
			*zone_end_pfn = min(node_end_pfn,
				arch_zone_highest_possible_pfn[movable_zone]);

		/* Adjust for ZONE_MOVABLE starting within this range */
		} else if (*zone_start_pfn < zone_movable_pfn[nid] &&
				*zone_end_pfn > zone_movable_pfn[nid]) {
			*zone_end_pfn = zone_movable_pfn[nid];

		/* Check if this whole range is within ZONE_MOVABLE */
		} else if (*zone_start_pfn >= zone_movable_pfn[nid])
			*zone_start_pfn = *zone_end_pfn;
	}
}

4167 4168 4169 4170
/*
 * Return the number of pages a zone spans in a node, including holes
 * present_pages = zone_spanned_pages_in_node() - zone_absent_pages_in_node()
 */
P
Paul Mundt 已提交
4171
static unsigned long __meminit zone_spanned_pages_in_node(int nid,
4172 4173 4174 4175 4176 4177 4178 4179 4180 4181
					unsigned long zone_type,
					unsigned long *ignored)
{
	unsigned long node_start_pfn, node_end_pfn;
	unsigned long zone_start_pfn, zone_end_pfn;

	/* Get the start and end of the node and zone */
	get_pfn_range_for_nid(nid, &node_start_pfn, &node_end_pfn);
	zone_start_pfn = arch_zone_lowest_possible_pfn[zone_type];
	zone_end_pfn = arch_zone_highest_possible_pfn[zone_type];
M
Mel Gorman 已提交
4182 4183 4184
	adjust_zone_range_for_zone_movable(nid, zone_type,
				node_start_pfn, node_end_pfn,
				&zone_start_pfn, &zone_end_pfn);
4185 4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199

	/* Check that this node has pages within the zone's required range */
	if (zone_end_pfn < node_start_pfn || zone_start_pfn > node_end_pfn)
		return 0;

	/* Move the zone boundaries inside the node if necessary */
	zone_end_pfn = min(zone_end_pfn, node_end_pfn);
	zone_start_pfn = max(zone_start_pfn, node_start_pfn);

	/* Return the spanned pages */
	return zone_end_pfn - zone_start_pfn;
}

/*
 * Return the number of holes in a range on a node. If nid is MAX_NUMNODES,
4200
 * then all holes in the requested range will be accounted for.
4201
 */
4202
unsigned long __meminit __absent_pages_in_range(int nid,
4203 4204 4205
				unsigned long range_start_pfn,
				unsigned long range_end_pfn)
{
4206 4207 4208
	unsigned long nr_absent = range_end_pfn - range_start_pfn;
	unsigned long start_pfn, end_pfn;
	int i;
4209

4210 4211 4212 4213
	for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, NULL) {
		start_pfn = clamp(start_pfn, range_start_pfn, range_end_pfn);
		end_pfn = clamp(end_pfn, range_start_pfn, range_end_pfn);
		nr_absent -= end_pfn - start_pfn;
4214
	}
4215
	return nr_absent;
4216 4217 4218 4219 4220 4221 4222
}

/**
 * absent_pages_in_range - Return number of page frames in holes within a range
 * @start_pfn: The start PFN to start searching for holes
 * @end_pfn: The end PFN to stop searching for holes
 *
4223
 * It returns the number of pages frames in memory holes within a range.
4224 4225 4226 4227 4228 4229 4230 4231
 */
unsigned long __init absent_pages_in_range(unsigned long start_pfn,
							unsigned long end_pfn)
{
	return __absent_pages_in_range(MAX_NUMNODES, start_pfn, end_pfn);
}

/* Return the number of page frames in holes in a zone on a node */
P
Paul Mundt 已提交
4232
static unsigned long __meminit zone_absent_pages_in_node(int nid,
4233 4234 4235
					unsigned long zone_type,
					unsigned long *ignored)
{
4236 4237
	unsigned long zone_low = arch_zone_lowest_possible_pfn[zone_type];
	unsigned long zone_high = arch_zone_highest_possible_pfn[zone_type];
4238 4239 4240 4241
	unsigned long node_start_pfn, node_end_pfn;
	unsigned long zone_start_pfn, zone_end_pfn;

	get_pfn_range_for_nid(nid, &node_start_pfn, &node_end_pfn);
4242 4243
	zone_start_pfn = clamp(node_start_pfn, zone_low, zone_high);
	zone_end_pfn = clamp(node_end_pfn, zone_low, zone_high);
4244

M
Mel Gorman 已提交
4245 4246 4247
	adjust_zone_range_for_zone_movable(nid, zone_type,
			node_start_pfn, node_end_pfn,
			&zone_start_pfn, &zone_end_pfn);
4248
	return __absent_pages_in_range(nid, zone_start_pfn, zone_end_pfn);
4249
}
4250

T
Tejun Heo 已提交
4251
#else /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
P
Paul Mundt 已提交
4252
static inline unsigned long __meminit zone_spanned_pages_in_node(int nid,
4253 4254 4255 4256 4257 4258
					unsigned long zone_type,
					unsigned long *zones_size)
{
	return zones_size[zone_type];
}

P
Paul Mundt 已提交
4259
static inline unsigned long __meminit zone_absent_pages_in_node(int nid,
4260 4261 4262 4263 4264 4265 4266 4267
						unsigned long zone_type,
						unsigned long *zholes_size)
{
	if (!zholes_size)
		return 0;

	return zholes_size[zone_type];
}
4268

T
Tejun Heo 已提交
4269
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
4270

4271
static void __meminit calculate_node_totalpages(struct pglist_data *pgdat,
4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291
		unsigned long *zones_size, unsigned long *zholes_size)
{
	unsigned long realtotalpages, totalpages = 0;
	enum zone_type i;

	for (i = 0; i < MAX_NR_ZONES; i++)
		totalpages += zone_spanned_pages_in_node(pgdat->node_id, i,
								zones_size);
	pgdat->node_spanned_pages = totalpages;

	realtotalpages = totalpages;
	for (i = 0; i < MAX_NR_ZONES; i++)
		realtotalpages -=
			zone_absent_pages_in_node(pgdat->node_id, i,
								zholes_size);
	pgdat->node_present_pages = realtotalpages;
	printk(KERN_DEBUG "On node %d totalpages: %lu\n", pgdat->node_id,
							realtotalpages);
}

4292 4293 4294
#ifndef CONFIG_SPARSEMEM
/*
 * Calculate the size of the zone->blockflags rounded to an unsigned long
4295 4296
 * Start by making sure zonesize is a multiple of pageblock_order by rounding
 * up. Then use 1 NR_PAGEBLOCK_BITS worth of bits per pageblock, finally
4297 4298 4299 4300 4301 4302 4303
 * round what is now in bits to nearest long in bits, then return it in
 * bytes.
 */
static unsigned long __init usemap_size(unsigned long zonesize)
{
	unsigned long usemapsize;

4304 4305
	usemapsize = roundup(zonesize, pageblock_nr_pages);
	usemapsize = usemapsize >> pageblock_order;
4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316
	usemapsize *= NR_PAGEBLOCK_BITS;
	usemapsize = roundup(usemapsize, 8 * sizeof(unsigned long));

	return usemapsize / 8;
}

static void __init setup_usemap(struct pglist_data *pgdat,
				struct zone *zone, unsigned long zonesize)
{
	unsigned long usemapsize = usemap_size(zonesize);
	zone->pageblock_flags = NULL;
4317
	if (usemapsize)
4318 4319
		zone->pageblock_flags = alloc_bootmem_node_nopanic(pgdat,
								   usemapsize);
4320 4321
}
#else
4322
static inline void setup_usemap(struct pglist_data *pgdat,
4323 4324 4325
				struct zone *zone, unsigned long zonesize) {}
#endif /* CONFIG_SPARSEMEM */

4326
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
4327

4328
/* Initialise the number of pages represented by NR_PAGEBLOCK_BITS */
4329
void __init set_pageblock_order(void)
4330
{
4331 4332
	unsigned int order;

4333 4334 4335 4336
	/* Check that pageblock_nr_pages has not already been setup */
	if (pageblock_order)
		return;

4337 4338 4339 4340 4341
	if (HPAGE_SHIFT > PAGE_SHIFT)
		order = HUGETLB_PAGE_ORDER;
	else
		order = MAX_ORDER - 1;

4342 4343
	/*
	 * Assume the largest contiguous order of interest is a huge page.
4344 4345
	 * This value may be variable depending on boot parameters on IA64 and
	 * powerpc.
4346 4347 4348 4349 4350
	 */
	pageblock_order = order;
}
#else /* CONFIG_HUGETLB_PAGE_SIZE_VARIABLE */

4351 4352
/*
 * When CONFIG_HUGETLB_PAGE_SIZE_VARIABLE is not set, set_pageblock_order()
4353 4354 4355
 * is unused as pageblock_order is set at compile-time. See
 * include/linux/pageblock-flags.h for the values of pageblock_order based on
 * the kernel config
4356
 */
4357
void __init set_pageblock_order(void)
4358 4359
{
}
4360 4361 4362

#endif /* CONFIG_HUGETLB_PAGE_SIZE_VARIABLE */

L
Linus Torvalds 已提交
4363 4364 4365 4366 4367 4368
/*
 * Set up the zone data structures:
 *   - mark all pages reserved
 *   - mark all memory queues empty
 *   - clear the memory bitmaps
 */
4369
static void __paginginit free_area_init_core(struct pglist_data *pgdat,
L
Linus Torvalds 已提交
4370 4371
		unsigned long *zones_size, unsigned long *zholes_size)
{
4372
	enum zone_type j;
4373
	int nid = pgdat->node_id;
L
Linus Torvalds 已提交
4374
	unsigned long zone_start_pfn = pgdat->node_start_pfn;
4375
	int ret;
L
Linus Torvalds 已提交
4376

4377
	pgdat_resize_init(pgdat);
L
Linus Torvalds 已提交
4378 4379 4380
	pgdat->nr_zones = 0;
	init_waitqueue_head(&pgdat->kswapd_wait);
	pgdat->kswapd_max_order = 0;
4381
	pgdat_page_cgroup_init(pgdat);
4382

L
Linus Torvalds 已提交
4383 4384
	for (j = 0; j < MAX_NR_ZONES; j++) {
		struct zone *zone = pgdat->node_zones + j;
4385
		unsigned long size, realsize, memmap_pages;
L
Linus Torvalds 已提交
4386

4387 4388 4389
		size = zone_spanned_pages_in_node(nid, j, zones_size);
		realsize = size - zone_absent_pages_in_node(nid, j,
								zholes_size);
L
Linus Torvalds 已提交
4390

4391 4392 4393 4394 4395
		/*
		 * Adjust realsize so that it accounts for how much memory
		 * is used by this zone for memmap. This affects the watermark
		 * and per-cpu initialisations
		 */
4396 4397
		memmap_pages =
			PAGE_ALIGN(size * sizeof(struct page)) >> PAGE_SHIFT;
4398 4399
		if (realsize >= memmap_pages) {
			realsize -= memmap_pages;
4400 4401 4402 4403
			if (memmap_pages)
				printk(KERN_DEBUG
				       "  %s zone: %lu pages used for memmap\n",
				       zone_names[j], memmap_pages);
4404 4405 4406 4407 4408
		} else
			printk(KERN_WARNING
				"  %s zone: %lu pages exceeds realsize %lu\n",
				zone_names[j], memmap_pages, realsize);

4409 4410
		/* Account for reserved pages */
		if (j == 0 && realsize > dma_reserve) {
4411
			realsize -= dma_reserve;
Y
Yinghai Lu 已提交
4412
			printk(KERN_DEBUG "  %s zone: %lu pages reserved\n",
4413
					zone_names[0], dma_reserve);
4414 4415
		}

4416
		if (!is_highmem_idx(j))
L
Linus Torvalds 已提交
4417 4418 4419 4420 4421
			nr_kernel_pages += realsize;
		nr_all_pages += realsize;

		zone->spanned_pages = size;
		zone->present_pages = realsize;
4422 4423 4424 4425 4426
#if defined CONFIG_COMPACTION || defined CONFIG_CMA
		zone->compact_cached_free_pfn = zone->zone_start_pfn +
						zone->spanned_pages;
		zone->compact_cached_free_pfn &= ~(pageblock_nr_pages-1);
#endif
4427
#ifdef CONFIG_NUMA
4428
		zone->node = nid;
4429
		zone->min_unmapped_pages = (realsize*sysctl_min_unmapped_ratio)
4430
						/ 100;
4431
		zone->min_slab_pages = (realsize * sysctl_min_slab_ratio) / 100;
4432
#endif
L
Linus Torvalds 已提交
4433 4434 4435
		zone->name = zone_names[j];
		spin_lock_init(&zone->lock);
		spin_lock_init(&zone->lru_lock);
4436
		zone_seqlock_init(zone);
L
Linus Torvalds 已提交
4437 4438
		zone->zone_pgdat = pgdat;

4439
		zone_pcp_init(zone);
4440
		lruvec_init(&zone->lruvec, zone);
4441
		zap_zone_vm_stats(zone);
4442
		zone->flags = 0;
4443 4444 4445
#ifdef CONFIG_MEMORY_ISOLATION
		zone->nr_pageblock_isolate = 0;
#endif
L
Linus Torvalds 已提交
4446 4447 4448
		if (!size)
			continue;

4449
		set_pageblock_order();
4450
		setup_usemap(pgdat, zone, size);
D
Dave Hansen 已提交
4451 4452
		ret = init_currently_empty_zone(zone, zone_start_pfn,
						size, MEMMAP_EARLY);
4453
		BUG_ON(ret);
4454
		memmap_init(size, nid, j, zone_start_pfn);
L
Linus Torvalds 已提交
4455 4456 4457 4458
		zone_start_pfn += size;
	}
}

S
Sam Ravnborg 已提交
4459
static void __init_refok alloc_node_mem_map(struct pglist_data *pgdat)
L
Linus Torvalds 已提交
4460 4461 4462 4463 4464
{
	/* Skip empty nodes */
	if (!pgdat->node_spanned_pages)
		return;

A
Andy Whitcroft 已提交
4465
#ifdef CONFIG_FLAT_NODE_MEM_MAP
L
Linus Torvalds 已提交
4466 4467
	/* ia64 gets its own node_mem_map, before this, without bootmem */
	if (!pgdat->node_mem_map) {
4468
		unsigned long size, start, end;
A
Andy Whitcroft 已提交
4469 4470
		struct page *map;

4471 4472 4473 4474 4475 4476 4477 4478 4479
		/*
		 * The zone's endpoints aren't required to be MAX_ORDER
		 * aligned but the node_mem_map endpoints must be in order
		 * for the buddy allocator to function correctly.
		 */
		start = pgdat->node_start_pfn & ~(MAX_ORDER_NR_PAGES - 1);
		end = pgdat->node_start_pfn + pgdat->node_spanned_pages;
		end = ALIGN(end, MAX_ORDER_NR_PAGES);
		size =  (end - start) * sizeof(struct page);
4480 4481
		map = alloc_remap(pgdat->node_id, size);
		if (!map)
4482
			map = alloc_bootmem_node_nopanic(pgdat, size);
4483
		pgdat->node_mem_map = map + (pgdat->node_start_pfn - start);
L
Linus Torvalds 已提交
4484
	}
4485
#ifndef CONFIG_NEED_MULTIPLE_NODES
L
Linus Torvalds 已提交
4486 4487 4488
	/*
	 * With no DISCONTIG, the global mem_map is just set as node 0's
	 */
4489
	if (pgdat == NODE_DATA(0)) {
L
Linus Torvalds 已提交
4490
		mem_map = NODE_DATA(0)->node_mem_map;
T
Tejun Heo 已提交
4491
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
4492
		if (page_to_pfn(mem_map) != pgdat->node_start_pfn)
4493
			mem_map -= (pgdat->node_start_pfn - ARCH_PFN_OFFSET);
T
Tejun Heo 已提交
4494
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
4495
	}
L
Linus Torvalds 已提交
4496
#endif
A
Andy Whitcroft 已提交
4497
#endif /* CONFIG_FLAT_NODE_MEM_MAP */
L
Linus Torvalds 已提交
4498 4499
}

4500 4501
void __paginginit free_area_init_node(int nid, unsigned long *zones_size,
		unsigned long node_start_pfn, unsigned long *zholes_size)
L
Linus Torvalds 已提交
4502
{
4503 4504
	pg_data_t *pgdat = NODE_DATA(nid);

L
Linus Torvalds 已提交
4505 4506
	pgdat->node_id = nid;
	pgdat->node_start_pfn = node_start_pfn;
4507
	calculate_node_totalpages(pgdat, zones_size, zholes_size);
L
Linus Torvalds 已提交
4508 4509

	alloc_node_mem_map(pgdat);
4510 4511 4512 4513 4514
#ifdef CONFIG_FLAT_NODE_MEM_MAP
	printk(KERN_DEBUG "free_area_init_node: node %d, pgdat %08lx, node_mem_map %08lx\n",
		nid, (unsigned long)pgdat,
		(unsigned long)pgdat->node_mem_map);
#endif
L
Linus Torvalds 已提交
4515 4516 4517 4518

	free_area_init_core(pgdat, zones_size, zholes_size);
}

T
Tejun Heo 已提交
4519
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
M
Miklos Szeredi 已提交
4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538 4539

#if MAX_NUMNODES > 1
/*
 * Figure out the number of possible node ids.
 */
static void __init setup_nr_node_ids(void)
{
	unsigned int node;
	unsigned int highest = 0;

	for_each_node_mask(node, node_possible_map)
		highest = node;
	nr_node_ids = highest + 1;
}
#else
static inline void setup_nr_node_ids(void)
{
}
#endif

4540 4541 4542 4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559 4560 4561
/**
 * node_map_pfn_alignment - determine the maximum internode alignment
 *
 * This function should be called after node map is populated and sorted.
 * It calculates the maximum power of two alignment which can distinguish
 * all the nodes.
 *
 * For example, if all nodes are 1GiB and aligned to 1GiB, the return value
 * would indicate 1GiB alignment with (1 << (30 - PAGE_SHIFT)).  If the
 * nodes are shifted by 256MiB, 256MiB.  Note that if only the last node is
 * shifted, 1GiB is enough and this function will indicate so.
 *
 * This is used to test whether pfn -> nid mapping of the chosen memory
 * model has fine enough granularity to avoid incorrect mapping for the
 * populated node map.
 *
 * Returns the determined alignment in pfn's.  0 if there is no alignment
 * requirement (single node).
 */
unsigned long __init node_map_pfn_alignment(void)
{
	unsigned long accl_mask = 0, last_end = 0;
4562
	unsigned long start, end, mask;
4563
	int last_nid = -1;
4564
	int i, nid;
4565

4566
	for_each_mem_pfn_range(i, MAX_NUMNODES, &start, &end, &nid) {
4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589
		if (!start || last_nid < 0 || last_nid == nid) {
			last_nid = nid;
			last_end = end;
			continue;
		}

		/*
		 * Start with a mask granular enough to pin-point to the
		 * start pfn and tick off bits one-by-one until it becomes
		 * too coarse to separate the current node from the last.
		 */
		mask = ~((1 << __ffs(start)) - 1);
		while (mask && last_end <= (start & (mask << 1)))
			mask <<= 1;

		/* accumulate all internode masks */
		accl_mask |= mask;
	}

	/* convert mask to number of pages */
	return ~accl_mask + 1;
}

4590
/* Find the lowest pfn for a node */
A
Adrian Bunk 已提交
4591
static unsigned long __init find_min_pfn_for_node(int nid)
4592
{
4593
	unsigned long min_pfn = ULONG_MAX;
4594 4595
	unsigned long start_pfn;
	int i;
4596

4597 4598
	for_each_mem_pfn_range(i, nid, &start_pfn, NULL, NULL)
		min_pfn = min(min_pfn, start_pfn);
4599

4600 4601
	if (min_pfn == ULONG_MAX) {
		printk(KERN_WARNING
4602
			"Could not find start_pfn for node %d\n", nid);
4603 4604 4605 4606
		return 0;
	}

	return min_pfn;
4607 4608 4609 4610 4611 4612
}

/**
 * find_min_pfn_with_active_regions - Find the minimum PFN registered
 *
 * It returns the minimum PFN based on information provided via
4613
 * add_active_range().
4614 4615 4616 4617 4618 4619
 */
unsigned long __init find_min_pfn_with_active_regions(void)
{
	return find_min_pfn_for_node(MAX_NUMNODES);
}

4620 4621 4622 4623 4624
/*
 * early_calculate_totalpages()
 * Sum pages in active regions for movable zone.
 * Populate N_HIGH_MEMORY for calculating usable_nodes.
 */
A
Adrian Bunk 已提交
4625
static unsigned long __init early_calculate_totalpages(void)
4626 4627
{
	unsigned long totalpages = 0;
4628 4629 4630 4631 4632
	unsigned long start_pfn, end_pfn;
	int i, nid;

	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid) {
		unsigned long pages = end_pfn - start_pfn;
4633

4634 4635
		totalpages += pages;
		if (pages)
4636
			node_set_state(nid, N_HIGH_MEMORY);
4637 4638
	}
  	return totalpages;
4639 4640
}

M
Mel Gorman 已提交
4641 4642 4643 4644 4645 4646
/*
 * Find the PFN the Movable zone begins in each node. Kernel memory
 * is spread evenly between nodes as long as the nodes have enough
 * memory. When they don't, some nodes will have more kernelcore than
 * others
 */
4647
static void __init find_zone_movable_pfns_for_nodes(void)
M
Mel Gorman 已提交
4648 4649 4650 4651
{
	int i, nid;
	unsigned long usable_startpfn;
	unsigned long kernelcore_node, kernelcore_remaining;
4652 4653
	/* save the state before borrow the nodemask */
	nodemask_t saved_node_state = node_states[N_HIGH_MEMORY];
4654 4655
	unsigned long totalpages = early_calculate_totalpages();
	int usable_nodes = nodes_weight(node_states[N_HIGH_MEMORY]);
M
Mel Gorman 已提交
4656

4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678
	/*
	 * If movablecore was specified, calculate what size of
	 * kernelcore that corresponds so that memory usable for
	 * any allocation type is evenly spread. If both kernelcore
	 * and movablecore are specified, then the value of kernelcore
	 * will be used for required_kernelcore if it's greater than
	 * what movablecore would have allowed.
	 */
	if (required_movablecore) {
		unsigned long corepages;

		/*
		 * Round-up so that ZONE_MOVABLE is at least as large as what
		 * was requested by the user
		 */
		required_movablecore =
			roundup(required_movablecore, MAX_ORDER_NR_PAGES);
		corepages = totalpages - required_movablecore;

		required_kernelcore = max(required_kernelcore, corepages);
	}

M
Mel Gorman 已提交
4679 4680
	/* If kernelcore was not specified, there is no ZONE_MOVABLE */
	if (!required_kernelcore)
4681
		goto out;
M
Mel Gorman 已提交
4682 4683 4684 4685 4686 4687 4688 4689

	/* usable_startpfn is the lowest possible pfn ZONE_MOVABLE can be at */
	find_usable_zone_for_movable();
	usable_startpfn = arch_zone_lowest_possible_pfn[movable_zone];

restart:
	/* Spread kernelcore memory as evenly as possible throughout nodes */
	kernelcore_node = required_kernelcore / usable_nodes;
4690
	for_each_node_state(nid, N_HIGH_MEMORY) {
4691 4692
		unsigned long start_pfn, end_pfn;

M
Mel Gorman 已提交
4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708
		/*
		 * Recalculate kernelcore_node if the division per node
		 * now exceeds what is necessary to satisfy the requested
		 * amount of memory for the kernel
		 */
		if (required_kernelcore < kernelcore_node)
			kernelcore_node = required_kernelcore / usable_nodes;

		/*
		 * As the map is walked, we track how much memory is usable
		 * by the kernel using kernelcore_remaining. When it is
		 * 0, the rest of the node is usable by ZONE_MOVABLE
		 */
		kernelcore_remaining = kernelcore_node;

		/* Go through each range of PFNs within this node */
4709
		for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, NULL) {
M
Mel Gorman 已提交
4710 4711
			unsigned long size_pages;

4712
			start_pfn = max(start_pfn, zone_movable_pfn[nid]);
M
Mel Gorman 已提交
4713 4714 4715 4716 4717 4718 4719 4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737 4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778
			if (start_pfn >= end_pfn)
				continue;

			/* Account for what is only usable for kernelcore */
			if (start_pfn < usable_startpfn) {
				unsigned long kernel_pages;
				kernel_pages = min(end_pfn, usable_startpfn)
								- start_pfn;

				kernelcore_remaining -= min(kernel_pages,
							kernelcore_remaining);
				required_kernelcore -= min(kernel_pages,
							required_kernelcore);

				/* Continue if range is now fully accounted */
				if (end_pfn <= usable_startpfn) {

					/*
					 * Push zone_movable_pfn to the end so
					 * that if we have to rebalance
					 * kernelcore across nodes, we will
					 * not double account here
					 */
					zone_movable_pfn[nid] = end_pfn;
					continue;
				}
				start_pfn = usable_startpfn;
			}

			/*
			 * The usable PFN range for ZONE_MOVABLE is from
			 * start_pfn->end_pfn. Calculate size_pages as the
			 * number of pages used as kernelcore
			 */
			size_pages = end_pfn - start_pfn;
			if (size_pages > kernelcore_remaining)
				size_pages = kernelcore_remaining;
			zone_movable_pfn[nid] = start_pfn + size_pages;

			/*
			 * Some kernelcore has been met, update counts and
			 * break if the kernelcore for this node has been
			 * satisified
			 */
			required_kernelcore -= min(required_kernelcore,
								size_pages);
			kernelcore_remaining -= size_pages;
			if (!kernelcore_remaining)
				break;
		}
	}

	/*
	 * If there is still required_kernelcore, we do another pass with one
	 * less node in the count. This will push zone_movable_pfn[nid] further
	 * along on the nodes that still have memory until kernelcore is
	 * satisified
	 */
	usable_nodes--;
	if (usable_nodes && required_kernelcore > usable_nodes)
		goto restart;

	/* Align start of ZONE_MOVABLE on all nids to MAX_ORDER_NR_PAGES */
	for (nid = 0; nid < MAX_NUMNODES; nid++)
		zone_movable_pfn[nid] =
			roundup(zone_movable_pfn[nid], MAX_ORDER_NR_PAGES);
4779 4780 4781 4782

out:
	/* restore the node_state */
	node_states[N_HIGH_MEMORY] = saved_node_state;
M
Mel Gorman 已提交
4783 4784
}

4785
/* Any regular memory on that node ? */
4786
static void __init check_for_regular_memory(pg_data_t *pgdat)
4787 4788 4789 4790 4791 4792
{
#ifdef CONFIG_HIGHMEM
	enum zone_type zone_type;

	for (zone_type = 0; zone_type <= ZONE_NORMAL; zone_type++) {
		struct zone *zone = &pgdat->node_zones[zone_type];
4793
		if (zone->present_pages) {
4794
			node_set_state(zone_to_nid(zone), N_NORMAL_MEMORY);
4795 4796
			break;
		}
4797 4798 4799 4800
	}
#endif
}

4801 4802
/**
 * free_area_init_nodes - Initialise all pg_data_t and zone data
4803
 * @max_zone_pfn: an array of max PFNs for each zone
4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815
 *
 * This will call free_area_init_node() for each active node in the system.
 * Using the page ranges provided by add_active_range(), the size of each
 * zone in each node and their holes is calculated. If the maximum PFN
 * between two adjacent zones match, it is assumed that the zone is empty.
 * For example, if arch_max_dma_pfn == arch_max_dma32_pfn, it is assumed
 * that arch_max_dma32_pfn has no pages. It is also assumed that a zone
 * starts where the previous one ended. For example, ZONE_DMA32 starts
 * at arch_max_dma_pfn.
 */
void __init free_area_init_nodes(unsigned long *max_zone_pfn)
{
4816 4817
	unsigned long start_pfn, end_pfn;
	int i, nid;
4818

4819 4820 4821 4822 4823 4824 4825 4826
	/* Record where the zone boundaries are */
	memset(arch_zone_lowest_possible_pfn, 0,
				sizeof(arch_zone_lowest_possible_pfn));
	memset(arch_zone_highest_possible_pfn, 0,
				sizeof(arch_zone_highest_possible_pfn));
	arch_zone_lowest_possible_pfn[0] = find_min_pfn_with_active_regions();
	arch_zone_highest_possible_pfn[0] = max_zone_pfn[0];
	for (i = 1; i < MAX_NR_ZONES; i++) {
M
Mel Gorman 已提交
4827 4828
		if (i == ZONE_MOVABLE)
			continue;
4829 4830 4831 4832 4833
		arch_zone_lowest_possible_pfn[i] =
			arch_zone_highest_possible_pfn[i-1];
		arch_zone_highest_possible_pfn[i] =
			max(max_zone_pfn[i], arch_zone_lowest_possible_pfn[i]);
	}
M
Mel Gorman 已提交
4834 4835 4836 4837 4838
	arch_zone_lowest_possible_pfn[ZONE_MOVABLE] = 0;
	arch_zone_highest_possible_pfn[ZONE_MOVABLE] = 0;

	/* Find the PFNs that ZONE_MOVABLE begins at in each node */
	memset(zone_movable_pfn, 0, sizeof(zone_movable_pfn));
4839
	find_zone_movable_pfns_for_nodes();
4840 4841

	/* Print out the zone ranges */
4842
	printk("Zone ranges:\n");
M
Mel Gorman 已提交
4843 4844 4845
	for (i = 0; i < MAX_NR_ZONES; i++) {
		if (i == ZONE_MOVABLE)
			continue;
4846
		printk(KERN_CONT "  %-8s ", zone_names[i]);
4847 4848
		if (arch_zone_lowest_possible_pfn[i] ==
				arch_zone_highest_possible_pfn[i])
4849
			printk(KERN_CONT "empty\n");
4850
		else
4851 4852 4853 4854
			printk(KERN_CONT "[mem %0#10lx-%0#10lx]\n",
				arch_zone_lowest_possible_pfn[i] << PAGE_SHIFT,
				(arch_zone_highest_possible_pfn[i]
					<< PAGE_SHIFT) - 1);
M
Mel Gorman 已提交
4855 4856 4857
	}

	/* Print out the PFNs ZONE_MOVABLE begins at in each node */
4858
	printk("Movable zone start for each node\n");
M
Mel Gorman 已提交
4859 4860
	for (i = 0; i < MAX_NUMNODES; i++) {
		if (zone_movable_pfn[i])
4861 4862
			printk("  Node %d: %#010lx\n", i,
			       zone_movable_pfn[i] << PAGE_SHIFT);
M
Mel Gorman 已提交
4863
	}
4864 4865

	/* Print out the early_node_map[] */
4866
	printk("Early memory node ranges\n");
4867
	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid)
4868 4869
		printk("  node %3d: [mem %#010lx-%#010lx]\n", nid,
		       start_pfn << PAGE_SHIFT, (end_pfn << PAGE_SHIFT) - 1);
4870 4871

	/* Initialise every node */
4872
	mminit_verify_pageflags_layout();
4873
	setup_nr_node_ids();
4874 4875
	for_each_online_node(nid) {
		pg_data_t *pgdat = NODE_DATA(nid);
4876
		free_area_init_node(nid, NULL,
4877
				find_min_pfn_for_node(nid), NULL);
4878 4879 4880 4881 4882

		/* Any memory on that node */
		if (pgdat->node_present_pages)
			node_set_state(nid, N_HIGH_MEMORY);
		check_for_regular_memory(pgdat);
4883 4884
	}
}
M
Mel Gorman 已提交
4885

4886
static int __init cmdline_parse_core(char *p, unsigned long *core)
M
Mel Gorman 已提交
4887 4888 4889 4890 4891 4892
{
	unsigned long long coremem;
	if (!p)
		return -EINVAL;

	coremem = memparse(p, &p);
4893
	*core = coremem >> PAGE_SHIFT;
M
Mel Gorman 已提交
4894

4895
	/* Paranoid check that UL is enough for the coremem value */
M
Mel Gorman 已提交
4896 4897 4898 4899
	WARN_ON((coremem >> PAGE_SHIFT) > ULONG_MAX);

	return 0;
}
M
Mel Gorman 已提交
4900

4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918
/*
 * kernelcore=size sets the amount of memory for use for allocations that
 * cannot be reclaimed or migrated.
 */
static int __init cmdline_parse_kernelcore(char *p)
{
	return cmdline_parse_core(p, &required_kernelcore);
}

/*
 * movablecore=size sets the amount of memory for use for allocations that
 * can be reclaimed or migrated.
 */
static int __init cmdline_parse_movablecore(char *p)
{
	return cmdline_parse_core(p, &required_movablecore);
}

M
Mel Gorman 已提交
4919
early_param("kernelcore", cmdline_parse_kernelcore);
4920
early_param("movablecore", cmdline_parse_movablecore);
M
Mel Gorman 已提交
4921

T
Tejun Heo 已提交
4922
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
4923

4924
/**
4925 4926
 * set_dma_reserve - set the specified number of pages reserved in the first zone
 * @new_dma_reserve: The number of pages to mark reserved
4927 4928 4929 4930
 *
 * The per-cpu batchsize and zone watermarks are determined by present_pages.
 * In the DMA zone, a significant percentage may be consumed by kernel image
 * and other unfreeable allocations which can skew the watermarks badly. This
4931 4932 4933
 * function may optionally be used to account for unfreeable pages in the
 * first zone (e.g., ZONE_DMA). The effect will be lower watermarks and
 * smaller per-cpu batchsize.
4934 4935 4936 4937 4938 4939
 */
void __init set_dma_reserve(unsigned long new_dma_reserve)
{
	dma_reserve = new_dma_reserve;
}

L
Linus Torvalds 已提交
4940 4941
void __init free_area_init(unsigned long *zones_size)
{
4942
	free_area_init_node(0, zones_size,
L
Linus Torvalds 已提交
4943 4944 4945 4946 4947 4948 4949 4950
			__pa(PAGE_OFFSET) >> PAGE_SHIFT, NULL);
}

static int page_alloc_cpu_notify(struct notifier_block *self,
				 unsigned long action, void *hcpu)
{
	int cpu = (unsigned long)hcpu;

4951
	if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) {
4952
		lru_add_drain_cpu(cpu);
4953 4954 4955 4956 4957 4958 4959 4960
		drain_pages(cpu);

		/*
		 * Spill the event counters of the dead processor
		 * into the current processors event counters.
		 * This artificially elevates the count of the current
		 * processor.
		 */
4961
		vm_events_fold_cpu(cpu);
4962 4963 4964 4965 4966 4967 4968 4969

		/*
		 * Zero the differential counters of the dead processor
		 * so that the vm statistics are consistent.
		 *
		 * This is only okay since the processor is dead and cannot
		 * race with what we are doing.
		 */
4970
		refresh_cpu_vm_stats(cpu);
L
Linus Torvalds 已提交
4971 4972 4973 4974 4975 4976 4977 4978 4979
	}
	return NOTIFY_OK;
}

void __init page_alloc_init(void)
{
	hotcpu_notifier(page_alloc_cpu_notify, 0);
}

4980 4981 4982 4983 4984 4985 4986 4987
/*
 * calculate_totalreserve_pages - called when sysctl_lower_zone_reserve_ratio
 *	or min_free_kbytes changes.
 */
static void calculate_totalreserve_pages(void)
{
	struct pglist_data *pgdat;
	unsigned long reserve_pages = 0;
4988
	enum zone_type i, j;
4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000

	for_each_online_pgdat(pgdat) {
		for (i = 0; i < MAX_NR_ZONES; i++) {
			struct zone *zone = pgdat->node_zones + i;
			unsigned long max = 0;

			/* Find valid and maximum lowmem_reserve in the zone */
			for (j = i; j < MAX_NR_ZONES; j++) {
				if (zone->lowmem_reserve[j] > max)
					max = zone->lowmem_reserve[j];
			}

5001 5002
			/* we treat the high watermark as reserved pages. */
			max += high_wmark_pages(zone);
5003 5004 5005 5006

			if (max > zone->present_pages)
				max = zone->present_pages;
			reserve_pages += max;
5007 5008 5009 5010 5011 5012 5013 5014 5015 5016
			/*
			 * Lowmem reserves are not available to
			 * GFP_HIGHUSER page cache allocations and
			 * kswapd tries to balance zones to their high
			 * watermark.  As a result, neither should be
			 * regarded as dirtyable memory, to prevent a
			 * situation where reclaim has to clean pages
			 * in order to balance the zones.
			 */
			zone->dirty_balance_reserve = max;
5017 5018
		}
	}
5019
	dirty_balance_reserve = reserve_pages;
5020 5021 5022
	totalreserve_pages = reserve_pages;
}

L
Linus Torvalds 已提交
5023 5024 5025 5026 5027 5028 5029 5030 5031
/*
 * setup_per_zone_lowmem_reserve - called whenever
 *	sysctl_lower_zone_reserve_ratio changes.  Ensures that each zone
 *	has a correct pages reserved value, so an adequate number of
 *	pages are left in the zone after a successful __alloc_pages().
 */
static void setup_per_zone_lowmem_reserve(void)
{
	struct pglist_data *pgdat;
5032
	enum zone_type j, idx;
L
Linus Torvalds 已提交
5033

5034
	for_each_online_pgdat(pgdat) {
L
Linus Torvalds 已提交
5035 5036 5037 5038 5039 5040
		for (j = 0; j < MAX_NR_ZONES; j++) {
			struct zone *zone = pgdat->node_zones + j;
			unsigned long present_pages = zone->present_pages;

			zone->lowmem_reserve[j] = 0;

5041 5042
			idx = j;
			while (idx) {
L
Linus Torvalds 已提交
5043 5044
				struct zone *lower_zone;

5045 5046
				idx--;

L
Linus Torvalds 已提交
5047 5048 5049 5050 5051 5052 5053 5054 5055 5056
				if (sysctl_lowmem_reserve_ratio[idx] < 1)
					sysctl_lowmem_reserve_ratio[idx] = 1;

				lower_zone = pgdat->node_zones + idx;
				lower_zone->lowmem_reserve[j] = present_pages /
					sysctl_lowmem_reserve_ratio[idx];
				present_pages += lower_zone->present_pages;
			}
		}
	}
5057 5058 5059

	/* update totalreserve_pages */
	calculate_totalreserve_pages();
L
Linus Torvalds 已提交
5060 5061
}

5062
static void __setup_per_zone_wmarks(void)
L
Linus Torvalds 已提交
5063 5064 5065 5066 5067 5068 5069 5070 5071 5072 5073 5074 5075
{
	unsigned long pages_min = min_free_kbytes >> (PAGE_SHIFT - 10);
	unsigned long lowmem_pages = 0;
	struct zone *zone;
	unsigned long flags;

	/* Calculate total number of !ZONE_HIGHMEM pages */
	for_each_zone(zone) {
		if (!is_highmem(zone))
			lowmem_pages += zone->present_pages;
	}

	for_each_zone(zone) {
5076 5077
		u64 tmp;

5078
		spin_lock_irqsave(&zone->lock, flags);
5079 5080
		tmp = (u64)pages_min * zone->present_pages;
		do_div(tmp, lowmem_pages);
L
Linus Torvalds 已提交
5081 5082
		if (is_highmem(zone)) {
			/*
N
Nick Piggin 已提交
5083 5084 5085 5086
			 * __GFP_HIGH and PF_MEMALLOC allocations usually don't
			 * need highmem pages, so cap pages_min to a small
			 * value here.
			 *
5087
			 * The WMARK_HIGH-WMARK_LOW and (WMARK_LOW-WMARK_MIN)
N
Nick Piggin 已提交
5088 5089
			 * deltas controls asynch page reclaim, and so should
			 * not be capped for highmem.
L
Linus Torvalds 已提交
5090 5091 5092 5093 5094 5095 5096 5097
			 */
			int min_pages;

			min_pages = zone->present_pages / 1024;
			if (min_pages < SWAP_CLUSTER_MAX)
				min_pages = SWAP_CLUSTER_MAX;
			if (min_pages > 128)
				min_pages = 128;
5098
			zone->watermark[WMARK_MIN] = min_pages;
L
Linus Torvalds 已提交
5099
		} else {
N
Nick Piggin 已提交
5100 5101
			/*
			 * If it's a lowmem zone, reserve a number of pages
L
Linus Torvalds 已提交
5102 5103
			 * proportionate to the zone's size.
			 */
5104
			zone->watermark[WMARK_MIN] = tmp;
L
Linus Torvalds 已提交
5105 5106
		}

5107 5108
		zone->watermark[WMARK_LOW]  = min_wmark_pages(zone) + (tmp >> 2);
		zone->watermark[WMARK_HIGH] = min_wmark_pages(zone) + (tmp >> 1);
5109 5110 5111 5112 5113

		zone->watermark[WMARK_MIN] += cma_wmark_pages(zone);
		zone->watermark[WMARK_LOW] += cma_wmark_pages(zone);
		zone->watermark[WMARK_HIGH] += cma_wmark_pages(zone);

5114
		setup_zone_migrate_reserve(zone);
5115
		spin_unlock_irqrestore(&zone->lock, flags);
L
Linus Torvalds 已提交
5116
	}
5117 5118 5119

	/* update totalreserve_pages */
	calculate_totalreserve_pages();
L
Linus Torvalds 已提交
5120 5121
}

5122 5123 5124 5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135
/**
 * setup_per_zone_wmarks - called when min_free_kbytes changes
 * or when memory is hot-{added|removed}
 *
 * Ensures that the watermark[min,low,high] values for each zone are set
 * correctly with respect to min_free_kbytes.
 */
void setup_per_zone_wmarks(void)
{
	mutex_lock(&zonelists_mutex);
	__setup_per_zone_wmarks();
	mutex_unlock(&zonelists_mutex);
}

5136
/*
5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156
 * The inactive anon list should be small enough that the VM never has to
 * do too much work, but large enough that each inactive page has a chance
 * to be referenced again before it is swapped out.
 *
 * The inactive_anon ratio is the target ratio of ACTIVE_ANON to
 * INACTIVE_ANON pages on this zone's LRU, maintained by the
 * pageout code. A zone->inactive_ratio of 3 means 3:1 or 25% of
 * the anonymous pages are kept on the inactive list.
 *
 * total     target    max
 * memory    ratio     inactive anon
 * -------------------------------------
 *   10MB       1         5MB
 *  100MB       1        50MB
 *    1GB       3       250MB
 *   10GB      10       0.9GB
 *  100GB      31         3GB
 *    1TB     101        10GB
 *   10TB     320        32GB
 */
5157
static void __meminit calculate_zone_inactive_ratio(struct zone *zone)
5158
{
5159
	unsigned int gb, ratio;
5160

5161 5162 5163
	/* Zone size in gigabytes */
	gb = zone->present_pages >> (30 - PAGE_SHIFT);
	if (gb)
5164
		ratio = int_sqrt(10 * gb);
5165 5166
	else
		ratio = 1;
5167

5168 5169
	zone->inactive_ratio = ratio;
}
5170

5171
static void __meminit setup_per_zone_inactive_ratio(void)
5172 5173 5174 5175 5176
{
	struct zone *zone;

	for_each_zone(zone)
		calculate_zone_inactive_ratio(zone);
5177 5178
}

L
Linus Torvalds 已提交
5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202
/*
 * Initialise min_free_kbytes.
 *
 * For small machines we want it small (128k min).  For large machines
 * we want it large (64MB max).  But it is not linear, because network
 * bandwidth does not increase linearly with machine size.  We use
 *
 * 	min_free_kbytes = 4 * sqrt(lowmem_kbytes), for better accuracy:
 *	min_free_kbytes = sqrt(lowmem_kbytes * 16)
 *
 * which yields
 *
 * 16MB:	512k
 * 32MB:	724k
 * 64MB:	1024k
 * 128MB:	1448k
 * 256MB:	2048k
 * 512MB:	2896k
 * 1024MB:	4096k
 * 2048MB:	5792k
 * 4096MB:	8192k
 * 8192MB:	11584k
 * 16384MB:	16384k
 */
5203
int __meminit init_per_zone_wmark_min(void)
L
Linus Torvalds 已提交
5204 5205 5206 5207 5208 5209 5210 5211 5212 5213
{
	unsigned long lowmem_kbytes;

	lowmem_kbytes = nr_free_buffer_pages() * (PAGE_SIZE >> 10);

	min_free_kbytes = int_sqrt(lowmem_kbytes * 16);
	if (min_free_kbytes < 128)
		min_free_kbytes = 128;
	if (min_free_kbytes > 65536)
		min_free_kbytes = 65536;
5214
	setup_per_zone_wmarks();
5215
	refresh_zone_stat_thresholds();
L
Linus Torvalds 已提交
5216
	setup_per_zone_lowmem_reserve();
5217
	setup_per_zone_inactive_ratio();
L
Linus Torvalds 已提交
5218 5219
	return 0;
}
5220
module_init(init_per_zone_wmark_min)
L
Linus Torvalds 已提交
5221 5222 5223 5224 5225 5226 5227

/*
 * min_free_kbytes_sysctl_handler - just a wrapper around proc_dointvec() so 
 *	that we can call two helper functions whenever min_free_kbytes
 *	changes.
 */
int min_free_kbytes_sysctl_handler(ctl_table *table, int write, 
5228
	void __user *buffer, size_t *length, loff_t *ppos)
L
Linus Torvalds 已提交
5229
{
5230
	proc_dointvec(table, write, buffer, length, ppos);
5231
	if (write)
5232
		setup_per_zone_wmarks();
L
Linus Torvalds 已提交
5233 5234 5235
	return 0;
}

5236 5237
#ifdef CONFIG_NUMA
int sysctl_min_unmapped_ratio_sysctl_handler(ctl_table *table, int write,
5238
	void __user *buffer, size_t *length, loff_t *ppos)
5239 5240 5241 5242
{
	struct zone *zone;
	int rc;

5243
	rc = proc_dointvec_minmax(table, write, buffer, length, ppos);
5244 5245 5246 5247
	if (rc)
		return rc;

	for_each_zone(zone)
5248
		zone->min_unmapped_pages = (zone->present_pages *
5249 5250 5251
				sysctl_min_unmapped_ratio) / 100;
	return 0;
}
5252 5253

int sysctl_min_slab_ratio_sysctl_handler(ctl_table *table, int write,
5254
	void __user *buffer, size_t *length, loff_t *ppos)
5255 5256 5257 5258
{
	struct zone *zone;
	int rc;

5259
	rc = proc_dointvec_minmax(table, write, buffer, length, ppos);
5260 5261 5262 5263 5264 5265 5266 5267
	if (rc)
		return rc;

	for_each_zone(zone)
		zone->min_slab_pages = (zone->present_pages *
				sysctl_min_slab_ratio) / 100;
	return 0;
}
5268 5269
#endif

L
Linus Torvalds 已提交
5270 5271 5272 5273 5274 5275
/*
 * lowmem_reserve_ratio_sysctl_handler - just a wrapper around
 *	proc_dointvec() so that we can call setup_per_zone_lowmem_reserve()
 *	whenever sysctl_lowmem_reserve_ratio changes.
 *
 * The reserve ratio obviously has absolutely no relation with the
5276
 * minimum watermarks. The lowmem reserve ratio can only make sense
L
Linus Torvalds 已提交
5277 5278 5279
 * if in function of the boot time zone sizes.
 */
int lowmem_reserve_ratio_sysctl_handler(ctl_table *table, int write,
5280
	void __user *buffer, size_t *length, loff_t *ppos)
L
Linus Torvalds 已提交
5281
{
5282
	proc_dointvec_minmax(table, write, buffer, length, ppos);
L
Linus Torvalds 已提交
5283 5284 5285 5286
	setup_per_zone_lowmem_reserve();
	return 0;
}

5287 5288 5289 5290 5291 5292 5293
/*
 * percpu_pagelist_fraction - changes the pcp->high for each zone on each
 * cpu.  It is the fraction of total pages in each zone that a hot per cpu pagelist
 * can have before it gets flushed back to buddy allocator.
 */

int percpu_pagelist_fraction_sysctl_handler(ctl_table *table, int write,
5294
	void __user *buffer, size_t *length, loff_t *ppos)
5295 5296 5297 5298 5299
{
	struct zone *zone;
	unsigned int cpu;
	int ret;

5300
	ret = proc_dointvec_minmax(table, write, buffer, length, ppos);
5301
	if (!write || (ret < 0))
5302
		return ret;
5303
	for_each_populated_zone(zone) {
5304
		for_each_possible_cpu(cpu) {
5305 5306
			unsigned long  high;
			high = zone->present_pages / percpu_pagelist_fraction;
5307 5308
			setup_pagelist_highmark(
				per_cpu_ptr(zone->pageset, cpu), high);
5309 5310 5311 5312 5313
		}
	}
	return 0;
}

5314
int hashdist = HASHDIST_DEFAULT;
L
Linus Torvalds 已提交
5315 5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339

#ifdef CONFIG_NUMA
static int __init set_hashdist(char *str)
{
	if (!str)
		return 0;
	hashdist = simple_strtoul(str, &str, 0);
	return 1;
}
__setup("hashdist=", set_hashdist);
#endif

/*
 * allocate a large system hash table from bootmem
 * - it is assumed that the hash table must contain an exact power-of-2
 *   quantity of entries
 * - limit is the number of hash buckets, not the total allocation size
 */
void *__init alloc_large_system_hash(const char *tablename,
				     unsigned long bucketsize,
				     unsigned long numentries,
				     int scale,
				     int flags,
				     unsigned int *_hash_shift,
				     unsigned int *_hash_mask,
5340 5341
				     unsigned long low_limit,
				     unsigned long high_limit)
L
Linus Torvalds 已提交
5342
{
5343
	unsigned long long max = high_limit;
L
Linus Torvalds 已提交
5344 5345 5346 5347 5348 5349
	unsigned long log2qty, size;
	void *table = NULL;

	/* allow the kernel cmdline to have a say */
	if (!numentries) {
		/* round applicable memory size up to nearest megabyte */
A
Andrew Morton 已提交
5350
		numentries = nr_kernel_pages;
L
Linus Torvalds 已提交
5351 5352 5353 5354 5355 5356 5357 5358 5359
		numentries += (1UL << (20 - PAGE_SHIFT)) - 1;
		numentries >>= 20 - PAGE_SHIFT;
		numentries <<= 20 - PAGE_SHIFT;

		/* limit to 1 bucket per 2^scale bytes of low memory */
		if (scale > PAGE_SHIFT)
			numentries >>= (scale - PAGE_SHIFT);
		else
			numentries <<= (PAGE_SHIFT - scale);
5360 5361

		/* Make sure we've got at least a 0-order allocation.. */
5362 5363 5364 5365 5366 5367 5368 5369
		if (unlikely(flags & HASH_SMALL)) {
			/* Makes no sense without HASH_EARLY */
			WARN_ON(!(flags & HASH_EARLY));
			if (!(numentries >> *_hash_shift)) {
				numentries = 1UL << *_hash_shift;
				BUG_ON(!numentries);
			}
		} else if (unlikely((numentries * bucketsize) < PAGE_SIZE))
5370
			numentries = PAGE_SIZE / bucketsize;
L
Linus Torvalds 已提交
5371
	}
5372
	numentries = roundup_pow_of_two(numentries);
L
Linus Torvalds 已提交
5373 5374 5375 5376 5377 5378

	/* limit allocation size to 1/16 total memory by default */
	if (max == 0) {
		max = ((unsigned long long)nr_all_pages << PAGE_SHIFT) >> 4;
		do_div(max, bucketsize);
	}
5379
	max = min(max, 0x80000000ULL);
L
Linus Torvalds 已提交
5380

5381 5382
	if (numentries < low_limit)
		numentries = low_limit;
L
Linus Torvalds 已提交
5383 5384 5385
	if (numentries > max)
		numentries = max;

5386
	log2qty = ilog2(numentries);
L
Linus Torvalds 已提交
5387 5388 5389 5390

	do {
		size = bucketsize << log2qty;
		if (flags & HASH_EARLY)
5391
			table = alloc_bootmem_nopanic(size);
L
Linus Torvalds 已提交
5392 5393 5394
		else if (hashdist)
			table = __vmalloc(size, GFP_ATOMIC, PAGE_KERNEL);
		else {
5395 5396
			/*
			 * If bucketsize is not a power-of-two, we may free
5397 5398
			 * some pages at the end of hash table which
			 * alloc_pages_exact() automatically does
5399
			 */
5400
			if (get_order(size) < MAX_ORDER) {
5401
				table = alloc_pages_exact(size, GFP_ATOMIC);
5402 5403
				kmemleak_alloc(table, size, 1, GFP_ATOMIC);
			}
L
Linus Torvalds 已提交
5404 5405 5406 5407 5408 5409
		}
	} while (!table && size > PAGE_SIZE && --log2qty);

	if (!table)
		panic("Failed to allocate %s hash table\n", tablename);

5410
	printk(KERN_INFO "%s hash table entries: %ld (order: %d, %lu bytes)\n",
L
Linus Torvalds 已提交
5411
	       tablename,
5412
	       (1UL << log2qty),
5413
	       ilog2(size) - PAGE_SHIFT,
L
Linus Torvalds 已提交
5414 5415 5416 5417 5418 5419 5420 5421 5422
	       size);

	if (_hash_shift)
		*_hash_shift = log2qty;
	if (_hash_mask)
		*_hash_mask = (1 << log2qty) - 1;

	return table;
}
5423

5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438
/* Return a pointer to the bitmap storing bits affecting a block of pages */
static inline unsigned long *get_pageblock_bitmap(struct zone *zone,
							unsigned long pfn)
{
#ifdef CONFIG_SPARSEMEM
	return __pfn_to_section(pfn)->pageblock_flags;
#else
	return zone->pageblock_flags;
#endif /* CONFIG_SPARSEMEM */
}

static inline int pfn_to_bitidx(struct zone *zone, unsigned long pfn)
{
#ifdef CONFIG_SPARSEMEM
	pfn &= (PAGES_PER_SECTION-1);
5439
	return (pfn >> pageblock_order) * NR_PAGEBLOCK_BITS;
5440 5441
#else
	pfn = pfn - zone->zone_start_pfn;
5442
	return (pfn >> pageblock_order) * NR_PAGEBLOCK_BITS;
5443 5444 5445 5446
#endif /* CONFIG_SPARSEMEM */
}

/**
5447
 * get_pageblock_flags_group - Return the requested group of flags for the pageblock_nr_pages block of pages
5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469
 * @page: The page within the block of interest
 * @start_bitidx: The first bit of interest to retrieve
 * @end_bitidx: The last bit of interest
 * returns pageblock_bits flags
 */
unsigned long get_pageblock_flags_group(struct page *page,
					int start_bitidx, int end_bitidx)
{
	struct zone *zone;
	unsigned long *bitmap;
	unsigned long pfn, bitidx;
	unsigned long flags = 0;
	unsigned long value = 1;

	zone = page_zone(page);
	pfn = page_to_pfn(page);
	bitmap = get_pageblock_bitmap(zone, pfn);
	bitidx = pfn_to_bitidx(zone, pfn);

	for (; start_bitidx <= end_bitidx; start_bitidx++, value <<= 1)
		if (test_bit(bitidx + start_bitidx, bitmap))
			flags |= value;
5470

5471 5472 5473 5474
	return flags;
}

/**
5475
 * set_pageblock_flags_group - Set the requested group of flags for a pageblock_nr_pages block of pages
5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492
 * @page: The page within the block of interest
 * @start_bitidx: The first bit of interest
 * @end_bitidx: The last bit of interest
 * @flags: The flags to set
 */
void set_pageblock_flags_group(struct page *page, unsigned long flags,
					int start_bitidx, int end_bitidx)
{
	struct zone *zone;
	unsigned long *bitmap;
	unsigned long pfn, bitidx;
	unsigned long value = 1;

	zone = page_zone(page);
	pfn = page_to_pfn(page);
	bitmap = get_pageblock_bitmap(zone, pfn);
	bitidx = pfn_to_bitidx(zone, pfn);
5493 5494
	VM_BUG_ON(pfn < zone->zone_start_pfn);
	VM_BUG_ON(pfn >= zone->zone_start_pfn + zone->spanned_pages);
5495 5496 5497 5498 5499 5500 5501

	for (; start_bitidx <= end_bitidx; start_bitidx++, value <<= 1)
		if (flags & value)
			__set_bit(bitidx + start_bitidx, bitmap);
		else
			__clear_bit(bitidx + start_bitidx, bitmap);
}
K
KAMEZAWA Hiroyuki 已提交
5502 5503

/*
5504 5505 5506 5507 5508 5509
 * This function checks whether pageblock includes unmovable pages or not.
 * If @count is not zero, it is okay to include less @count unmovable pages
 *
 * PageLRU check wihtout isolation or lru_lock could race so that
 * MIGRATE_MOVABLE block might include unmovable pages. It means you can't
 * expect this function should be exact.
K
KAMEZAWA Hiroyuki 已提交
5510
 */
5511
bool has_unmovable_pages(struct zone *zone, struct page *page, int count)
5512 5513
{
	unsigned long pfn, iter, found;
5514 5515
	int mt;

5516 5517
	/*
	 * For avoiding noise data, lru_add_drain_all() should be called
5518
	 * If ZONE_MOVABLE, the zone never contains unmovable pages
5519 5520
	 */
	if (zone_idx(zone) == ZONE_MOVABLE)
5521
		return false;
5522 5523
	mt = get_pageblock_migratetype(page);
	if (mt == MIGRATE_MOVABLE || is_migrate_cma(mt))
5524
		return false;
5525 5526 5527 5528 5529

	pfn = page_to_pfn(page);
	for (found = 0, iter = 0; iter < pageblock_nr_pages; iter++) {
		unsigned long check = pfn + iter;

5530
		if (!pfn_valid_within(check))
5531
			continue;
5532

5533
		page = pfn_to_page(check);
5534 5535 5536 5537 5538 5539 5540
		/*
		 * We can't use page_count without pin a page
		 * because another CPU can free compound page.
		 * This check already skips compound tails of THP
		 * because their page->_count is zero at all time.
		 */
		if (!atomic_read(&page->_count)) {
5541 5542 5543 5544
			if (PageBuddy(page))
				iter += (1 << page_order(page)) - 1;
			continue;
		}
5545

5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560 5561
		if (!PageLRU(page))
			found++;
		/*
		 * If there are RECLAIMABLE pages, we need to check it.
		 * But now, memory offline itself doesn't call shrink_slab()
		 * and it still to be fixed.
		 */
		/*
		 * If the page is not RAM, page_count()should be 0.
		 * we don't need more check. This is an _used_ not-movable page.
		 *
		 * The problematic thing here is PG_reserved pages. PG_reserved
		 * is set to both of a memory hole page and a _used_ kernel
		 * page at boot.
		 */
		if (found > count)
5562
			return true;
5563
	}
5564
	return false;
5565 5566 5567 5568
}

bool is_pageblock_removable_nolock(struct page *page)
{
5569 5570
	struct zone *zone;
	unsigned long pfn;
5571 5572 5573 5574 5575

	/*
	 * We have to be careful here because we are iterating over memory
	 * sections which are not zone aware so we might end up outside of
	 * the zone but still within the section.
5576 5577
	 * We have to take care about the node as well. If the node is offline
	 * its NODE_DATA will be NULL - see page_zone.
5578
	 */
5579 5580 5581 5582 5583 5584
	if (!node_online(page_to_nid(page)))
		return false;

	zone = page_zone(page);
	pfn = page_to_pfn(page);
	if (zone->zone_start_pfn > pfn ||
5585 5586 5587
			zone->zone_start_pfn + zone->spanned_pages <= pfn)
		return false;

5588
	return !has_unmovable_pages(zone, page, 0);
K
KAMEZAWA Hiroyuki 已提交
5589
}
K
KAMEZAWA Hiroyuki 已提交
5590

5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608
#ifdef CONFIG_CMA

static unsigned long pfn_max_align_down(unsigned long pfn)
{
	return pfn & ~(max_t(unsigned long, MAX_ORDER_NR_PAGES,
			     pageblock_nr_pages) - 1);
}

static unsigned long pfn_max_align_up(unsigned long pfn)
{
	return ALIGN(pfn, max_t(unsigned long, MAX_ORDER_NR_PAGES,
				pageblock_nr_pages));
}

static struct page *
__alloc_contig_migrate_alloc(struct page *page, unsigned long private,
			     int **resultp)
{
5609 5610 5611 5612 5613 5614
	gfp_t gfp_mask = GFP_USER | __GFP_MOVABLE;

	if (PageHighMem(page))
		gfp_mask |= __GFP_HIGHMEM;

	return alloc_page(gfp_mask);
5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629
}

/* [start, end) must belong to a single zone. */
static int __alloc_contig_migrate_range(unsigned long start, unsigned long end)
{
	/* This function is based on compact_zone() from compaction.c. */

	unsigned long pfn = start;
	unsigned int tries = 0;
	int ret = 0;

	struct compact_control cc = {
		.nr_migratepages = 0,
		.order = -1,
		.zone = page_zone(pfn_to_page(start)),
5630
		.sync = true,
5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648 5649 5650 5651 5652 5653 5654 5655 5656 5657
	};
	INIT_LIST_HEAD(&cc.migratepages);

	migrate_prep_local();

	while (pfn < end || !list_empty(&cc.migratepages)) {
		if (fatal_signal_pending(current)) {
			ret = -EINTR;
			break;
		}

		if (list_empty(&cc.migratepages)) {
			cc.nr_migratepages = 0;
			pfn = isolate_migratepages_range(cc.zone, &cc,
							 pfn, end);
			if (!pfn) {
				ret = -EINTR;
				break;
			}
			tries = 0;
		} else if (++tries == 5) {
			ret = ret < 0 ? ret : -EBUSY;
			break;
		}

		ret = migrate_pages(&cc.migratepages,
				    __alloc_contig_migrate_alloc,
M
Minchan Kim 已提交
5658
				    0, false, MIGRATE_SYNC);
5659 5660 5661 5662 5663 5664
	}

	putback_lru_pages(&cc.migratepages);
	return ret > 0 ? 0 : ret;
}

5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712
/*
 * Update zone's cma pages counter used for watermark level calculation.
 */
static inline void __update_cma_watermarks(struct zone *zone, int count)
{
	unsigned long flags;
	spin_lock_irqsave(&zone->lock, flags);
	zone->min_cma_pages += count;
	spin_unlock_irqrestore(&zone->lock, flags);
	setup_per_zone_wmarks();
}

/*
 * Trigger memory pressure bump to reclaim some pages in order to be able to
 * allocate 'count' pages in single page units. Does similar work as
 *__alloc_pages_slowpath() function.
 */
static int __reclaim_pages(struct zone *zone, gfp_t gfp_mask, int count)
{
	enum zone_type high_zoneidx = gfp_zone(gfp_mask);
	struct zonelist *zonelist = node_zonelist(0, gfp_mask);
	int did_some_progress = 0;
	int order = 1;

	/*
	 * Increase level of watermarks to force kswapd do his job
	 * to stabilise at new watermark level.
	 */
	__update_cma_watermarks(zone, count);

	/* Obey watermarks as if the page was being allocated */
	while (!zone_watermark_ok(zone, 0, low_wmark_pages(zone), 0, 0)) {
		wake_all_kswapd(order, zonelist, high_zoneidx, zone_idx(zone));

		did_some_progress = __perform_reclaim(gfp_mask, order, zonelist,
						      NULL);
		if (!did_some_progress) {
			/* Exhausted what can be done so it's blamo time */
			out_of_memory(zonelist, gfp_mask, order, NULL, false);
		}
	}

	/* Restore original watermark levels. */
	__update_cma_watermarks(zone, -count);

	return count;
}

5713 5714 5715 5716
/**
 * alloc_contig_range() -- tries to allocate given range of pages
 * @start:	start PFN to allocate
 * @end:	one-past-the-last PFN to allocate
5717 5718 5719 5720
 * @migratetype:	migratetype of the underlaying pageblocks (either
 *			#MIGRATE_MOVABLE or #MIGRATE_CMA).  All pageblocks
 *			in range must have the same migratetype and it must
 *			be either of the two.
5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732
 *
 * The PFN range does not have to be pageblock or MAX_ORDER_NR_PAGES
 * aligned, however it's the caller's responsibility to guarantee that
 * we are the only thread that changes migrate type of pageblocks the
 * pages fall in.
 *
 * The PFN range must belong to a single zone.
 *
 * Returns zero on success or negative error code.  On success all
 * pages which PFN is in [start, end) are allocated for the caller and
 * need to be freed with free_contig_range().
 */
5733 5734
int alloc_contig_range(unsigned long start, unsigned long end,
		       unsigned migratetype)
5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764
{
	struct zone *zone = page_zone(pfn_to_page(start));
	unsigned long outer_start, outer_end;
	int ret = 0, order;

	/*
	 * What we do here is we mark all pageblocks in range as
	 * MIGRATE_ISOLATE.  Because pageblock and max order pages may
	 * have different sizes, and due to the way page allocator
	 * work, we align the range to biggest of the two pages so
	 * that page allocator won't try to merge buddies from
	 * different pageblocks and change MIGRATE_ISOLATE to some
	 * other migration type.
	 *
	 * Once the pageblocks are marked as MIGRATE_ISOLATE, we
	 * migrate the pages from an unaligned range (ie. pages that
	 * we are interested in).  This will put all the pages in
	 * range back to page allocator as MIGRATE_ISOLATE.
	 *
	 * When this is done, we take the pages in range from page
	 * allocator removing them from the buddy system.  This way
	 * page allocator will never consider using them.
	 *
	 * This lets us mark the pageblocks back as
	 * MIGRATE_CMA/MIGRATE_MOVABLE so that free pages in the
	 * aligned range but not in the unaligned, original range are
	 * put back to page allocator so that buddy can use them.
	 */

	ret = start_isolate_page_range(pfn_max_align_down(start),
5765
				       pfn_max_align_up(end), migratetype);
5766 5767 5768 5769 5770 5771 5772 5773 5774 5775 5776 5777 5778 5779 5780 5781 5782 5783 5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799 5800 5801 5802 5803 5804 5805 5806 5807 5808 5809 5810
	if (ret)
		goto done;

	ret = __alloc_contig_migrate_range(start, end);
	if (ret)
		goto done;

	/*
	 * Pages from [start, end) are within a MAX_ORDER_NR_PAGES
	 * aligned blocks that are marked as MIGRATE_ISOLATE.  What's
	 * more, all pages in [start, end) are free in page allocator.
	 * What we are going to do is to allocate all pages from
	 * [start, end) (that is remove them from page allocator).
	 *
	 * The only problem is that pages at the beginning and at the
	 * end of interesting range may be not aligned with pages that
	 * page allocator holds, ie. they can be part of higher order
	 * pages.  Because of this, we reserve the bigger range and
	 * once this is done free the pages we are not interested in.
	 *
	 * We don't have to hold zone->lock here because the pages are
	 * isolated thus they won't get removed from buddy.
	 */

	lru_add_drain_all();
	drain_all_pages();

	order = 0;
	outer_start = start;
	while (!PageBuddy(pfn_to_page(outer_start))) {
		if (++order >= MAX_ORDER) {
			ret = -EBUSY;
			goto done;
		}
		outer_start &= ~0UL << order;
	}

	/* Make sure the range is really isolated. */
	if (test_pages_isolated(outer_start, end)) {
		pr_warn("alloc_contig_range test_pages_isolated(%lx, %lx) failed\n",
		       outer_start, end);
		ret = -EBUSY;
		goto done;
	}

5811 5812 5813 5814 5815 5816 5817
	/*
	 * Reclaim enough pages to make sure that contiguous allocation
	 * will not starve the system.
	 */
	__reclaim_pages(zone, GFP_HIGHUSER_MOVABLE, end-start);

	/* Grab isolated pages from freelists. */
5818 5819 5820 5821 5822 5823 5824 5825 5826 5827 5828 5829 5830 5831
	outer_end = isolate_freepages_range(outer_start, end);
	if (!outer_end) {
		ret = -EBUSY;
		goto done;
	}

	/* Free head and tail (if any) */
	if (start != outer_start)
		free_contig_range(outer_start, start - outer_start);
	if (end != outer_end)
		free_contig_range(end, outer_end - end);

done:
	undo_isolate_page_range(pfn_max_align_down(start),
5832
				pfn_max_align_up(end), migratetype);
5833 5834 5835 5836 5837 5838 5839 5840 5841 5842
	return ret;
}

void free_contig_range(unsigned long pfn, unsigned nr_pages)
{
	for (; nr_pages--; ++pfn)
		__free_page(pfn_to_page(pfn));
}
#endif

5843 5844 5845 5846 5847 5848 5849 5850 5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871
#ifdef CONFIG_MEMORY_HOTPLUG
static int __meminit __zone_pcp_update(void *data)
{
	struct zone *zone = data;
	int cpu;
	unsigned long batch = zone_batchsize(zone), flags;

	for_each_possible_cpu(cpu) {
		struct per_cpu_pageset *pset;
		struct per_cpu_pages *pcp;

		pset = per_cpu_ptr(zone->pageset, cpu);
		pcp = &pset->pcp;

		local_irq_save(flags);
		if (pcp->count > 0)
			free_pcppages_bulk(zone, pcp->count, pcp);
		setup_pageset(pset, batch);
		local_irq_restore(flags);
	}
	return 0;
}

void __meminit zone_pcp_update(struct zone *zone)
{
	stop_machine(__zone_pcp_update, zone, NULL);
}
#endif

K
KAMEZAWA Hiroyuki 已提交
5872
#ifdef CONFIG_MEMORY_HOTREMOVE
5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885
void zone_pcp_reset(struct zone *zone)
{
	unsigned long flags;

	/* avoid races with drain_pages()  */
	local_irq_save(flags);
	if (zone->pageset != &boot_pageset) {
		free_percpu(zone->pageset);
		zone->pageset = &boot_pageset;
	}
	local_irq_restore(flags);
}

K
KAMEZAWA Hiroyuki 已提交
5886 5887 5888 5889 5890 5891 5892 5893 5894 5895 5896 5897 5898 5899 5900 5901 5902 5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916 5917 5918 5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929 5930
/*
 * All pages in the range must be isolated before calling this.
 */
void
__offline_isolated_pages(unsigned long start_pfn, unsigned long end_pfn)
{
	struct page *page;
	struct zone *zone;
	int order, i;
	unsigned long pfn;
	unsigned long flags;
	/* find the first valid pfn */
	for (pfn = start_pfn; pfn < end_pfn; pfn++)
		if (pfn_valid(pfn))
			break;
	if (pfn == end_pfn)
		return;
	zone = page_zone(pfn_to_page(pfn));
	spin_lock_irqsave(&zone->lock, flags);
	pfn = start_pfn;
	while (pfn < end_pfn) {
		if (!pfn_valid(pfn)) {
			pfn++;
			continue;
		}
		page = pfn_to_page(pfn);
		BUG_ON(page_count(page));
		BUG_ON(!PageBuddy(page));
		order = page_order(page);
#ifdef CONFIG_DEBUG_VM
		printk(KERN_INFO "remove from free list %lx %d %lx\n",
		       pfn, 1 << order, end_pfn);
#endif
		list_del(&page->lru);
		rmv_page_order(page);
		zone->free_area[order].nr_free--;
		__mod_zone_page_state(zone, NR_FREE_PAGES,
				      - (1UL << order));
		for (i = 0; i < (1 << order); i++)
			SetPageReserved((page+i));
		pfn += (1 << order);
	}
	spin_unlock_irqrestore(&zone->lock, flags);
}
#endif
5931 5932 5933 5934 5935 5936 5937 5938 5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950 5951

#ifdef CONFIG_MEMORY_FAILURE
bool is_free_buddy_page(struct page *page)
{
	struct zone *zone = page_zone(page);
	unsigned long pfn = page_to_pfn(page);
	unsigned long flags;
	int order;

	spin_lock_irqsave(&zone->lock, flags);
	for (order = 0; order < MAX_ORDER; order++) {
		struct page *page_head = page - (pfn & ((1 << order) - 1));

		if (PageBuddy(page_head) && page_order(page_head) >= order)
			break;
	}
	spin_unlock_irqrestore(&zone->lock, flags);

	return order < MAX_ORDER;
}
#endif
5952

A
Andrew Morton 已提交
5953
static const struct trace_print_flags pageflag_names[] = {
5954 5955 5956 5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967 5968 5969 5970 5971 5972 5973 5974 5975 5976 5977 5978 5979 5980 5981 5982 5983 5984 5985 5986
	{1UL << PG_locked,		"locked"	},
	{1UL << PG_error,		"error"		},
	{1UL << PG_referenced,		"referenced"	},
	{1UL << PG_uptodate,		"uptodate"	},
	{1UL << PG_dirty,		"dirty"		},
	{1UL << PG_lru,			"lru"		},
	{1UL << PG_active,		"active"	},
	{1UL << PG_slab,		"slab"		},
	{1UL << PG_owner_priv_1,	"owner_priv_1"	},
	{1UL << PG_arch_1,		"arch_1"	},
	{1UL << PG_reserved,		"reserved"	},
	{1UL << PG_private,		"private"	},
	{1UL << PG_private_2,		"private_2"	},
	{1UL << PG_writeback,		"writeback"	},
#ifdef CONFIG_PAGEFLAGS_EXTENDED
	{1UL << PG_head,		"head"		},
	{1UL << PG_tail,		"tail"		},
#else
	{1UL << PG_compound,		"compound"	},
#endif
	{1UL << PG_swapcache,		"swapcache"	},
	{1UL << PG_mappedtodisk,	"mappedtodisk"	},
	{1UL << PG_reclaim,		"reclaim"	},
	{1UL << PG_swapbacked,		"swapbacked"	},
	{1UL << PG_unevictable,		"unevictable"	},
#ifdef CONFIG_MMU
	{1UL << PG_mlocked,		"mlocked"	},
#endif
#ifdef CONFIG_ARCH_USES_PG_UNCACHED
	{1UL << PG_uncached,		"uncached"	},
#endif
#ifdef CONFIG_MEMORY_FAILURE
	{1UL << PG_hwpoison,		"hwpoison"	},
5987 5988 5989
#endif
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	{1UL << PG_compound_lock,	"compound_lock"	},
5990 5991 5992 5993 5994 5995 5996 5997 5998
#endif
};

static void dump_page_flags(unsigned long flags)
{
	const char *delim = "";
	unsigned long mask;
	int i;

A
Andrew Morton 已提交
5999
	BUILD_BUG_ON(ARRAY_SIZE(pageflag_names) != __NR_PAGEFLAGS);
6000

6001 6002 6003 6004 6005
	printk(KERN_ALERT "page flags: %#lx(", flags);

	/* remove zone id */
	flags &= (1UL << NR_PAGEFLAGS) - 1;

A
Andrew Morton 已提交
6006
	for (i = 0; i < ARRAY_SIZE(pageflag_names) && flags; i++) {
6007 6008 6009 6010 6011 6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027

		mask = pageflag_names[i].mask;
		if ((flags & mask) != mask)
			continue;

		flags &= ~mask;
		printk("%s%s", delim, pageflag_names[i].name);
		delim = "|";
	}

	/* check for left over flags */
	if (flags)
		printk("%s%#lx", delim, flags);

	printk(")\n");
}

void dump_page(struct page *page)
{
	printk(KERN_ALERT
	       "page:%p count:%d mapcount:%d mapping:%p index:%#lx\n",
6028
		page, atomic_read(&page->_count), page_mapcount(page),
6029 6030
		page->mapping, page->index);
	dump_page_flags(page->flags);
6031
	mem_cgroup_print_bad_page(page);
6032
}