page_alloc.c 156.8 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/memory.h>
55
#include <linux/compaction.h>
56
#include <trace/events/kmem.h>
57
#include <linux/ftrace_event.h>
58
#include <linux/memcontrol.h>
59
#include <linux/prefetch.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
static void set_pageblock_migratetype(struct page *page, int migratetype)
{
223 224 225 226

	if (unlikely(page_group_by_mobility_disabled))
		migratetype = MIGRATE_UNMOVABLE;

227 228 229 230
	set_pageblock_flags_group(page, (unsigned long)migratetype,
					PB_migrate, PB_migrate_end);
}

231 232
bool oom_killer_disabled __read_mostly;

N
Nick Piggin 已提交
233
#ifdef CONFIG_DEBUG_VM
234
static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
L
Linus Torvalds 已提交
235
{
236 237 238
	int ret = 0;
	unsigned seq;
	unsigned long pfn = page_to_pfn(page);
239

240 241 242 243 244 245 246 247 248
	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;
249 250 251 252
}

static int page_is_consistent(struct zone *zone, struct page *page)
{
253
	if (!pfn_valid_within(page_to_pfn(page)))
254
		return 0;
L
Linus Torvalds 已提交
255
	if (zone != page_zone(page))
256 257 258 259 260 261 262 263 264 265
		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
266
		return 1;
267 268 269
	if (!page_is_consistent(zone, page))
		return 1;

L
Linus Torvalds 已提交
270 271
	return 0;
}
N
Nick Piggin 已提交
272 273 274 275 276 277 278
#else
static inline int bad_range(struct zone *zone, struct page *page)
{
	return 0;
}
#endif

N
Nick Piggin 已提交
279
static void bad_page(struct page *page)
L
Linus Torvalds 已提交
280
{
281 282 283 284
	static unsigned long resume;
	static unsigned long nr_shown;
	static unsigned long nr_unshown;

285 286
	/* Don't complain about poisoned pages */
	if (PageHWPoison(page)) {
287
		reset_page_mapcount(page); /* remove PageBuddy */
288 289 290
		return;
	}

291 292 293 294 295 296 297 298 299 300
	/*
	 * 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) {
301 302
			printk(KERN_ALERT
			      "BUG: Bad page state: %lu messages suppressed\n",
303 304 305 306 307 308 309 310
				nr_unshown);
			nr_unshown = 0;
		}
		nr_shown = 0;
	}
	if (nr_shown++ == 0)
		resume = jiffies + 60 * HZ;

311
	printk(KERN_ALERT "BUG: Bad page state in process %s  pfn:%05lx\n",
312
		current->comm, page_to_pfn(page));
313
	dump_page(page);
314

315
	print_modules();
L
Linus Torvalds 已提交
316
	dump_stack();
317
out:
318
	/* Leave bad fields for debug, except PageBuddy could make trouble */
319
	reset_page_mapcount(page); /* remove PageBuddy */
320
	add_taint(TAINT_BAD_PAGE);
L
Linus Torvalds 已提交
321 322 323 324 325 326 327 328 329
}

/*
 * 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".
 *
330 331
 * All pages have PG_compound set.  All tail pages have their ->first_page
 * pointing at the head page.
L
Linus Torvalds 已提交
332
 *
333 334 335
 * 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
336
 */
337 338 339

static void free_compound_page(struct page *page)
{
340
	__free_pages_ok(page, compound_order(page));
341 342
}

A
Andi Kleen 已提交
343
void prep_compound_page(struct page *page, unsigned long order)
344 345 346 347 348 349 350 351 352 353
{
	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);
354
		set_page_count(p, 0);
355 356 357 358
		p->first_page = page;
	}
}

359
/* update __split_huge_page_refcount if you change this function */
360
static int destroy_compound_page(struct page *page, unsigned long order)
L
Linus Torvalds 已提交
361 362 363
{
	int i;
	int nr_pages = 1 << order;
364
	int bad = 0;
L
Linus Torvalds 已提交
365

366 367
	if (unlikely(compound_order(page) != order) ||
	    unlikely(!PageHead(page))) {
N
Nick Piggin 已提交
368
		bad_page(page);
369 370
		bad++;
	}
L
Linus Torvalds 已提交
371

372
	__ClearPageHead(page);
373

374 375
	for (i = 1; i < nr_pages; i++) {
		struct page *p = page + i;
L
Linus Torvalds 已提交
376

377
		if (unlikely(!PageTail(p) || (p->first_page != page))) {
N
Nick Piggin 已提交
378
			bad_page(page);
379 380
			bad++;
		}
381
		__ClearPageTail(p);
L
Linus Torvalds 已提交
382
	}
383 384

	return bad;
L
Linus Torvalds 已提交
385 386
}

N
Nick Piggin 已提交
387 388 389 390
static inline void prep_zero_page(struct page *page, int order, gfp_t gfp_flags)
{
	int i;

391 392 393 394
	/*
	 * 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
395
	VM_BUG_ON((gfp_flags & __GFP_HIGHMEM) && in_interrupt());
N
Nick Piggin 已提交
396 397 398 399
	for (i = 0; i < (1 << order); i++)
		clear_highpage(page + i);
}

400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430
#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

431 432
static inline void set_page_order(struct page *page, int order)
{
H
Hugh Dickins 已提交
433
	set_page_private(page, order);
434
	__SetPageBuddy(page);
L
Linus Torvalds 已提交
435 436 437 438
}

static inline void rmv_page_order(struct page *page)
{
439
	__ClearPageBuddy(page);
H
Hugh Dickins 已提交
440
	set_page_private(page, 0);
L
Linus Torvalds 已提交
441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
}

/*
 * 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
458
 * Assumption: *_mem_map is contiguous at least up to MAX_ORDER
L
Linus Torvalds 已提交
459 460
 */
static inline unsigned long
461
__find_buddy_index(unsigned long page_idx, unsigned int order)
L
Linus Torvalds 已提交
462
{
463
	return page_idx ^ (1 << order);
L
Linus Torvalds 已提交
464 465 466 467 468
}

/*
 * 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
469
 * (a) the buddy is not in a hole &&
470
 * (b) the buddy is in the buddy system &&
471 472
 * (c) a page and its buddy have the same order &&
 * (d) a page and its buddy are in the same zone.
473
 *
A
Andrea Arcangeli 已提交
474 475
 * 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
476
 *
477
 * For recording page's order, we use page_private(page).
L
Linus Torvalds 已提交
478
 */
479 480
static inline int page_is_buddy(struct page *page, struct page *buddy,
								int order)
L
Linus Torvalds 已提交
481
{
482
	if (!pfn_valid_within(page_to_pfn(buddy)))
N
Nick Piggin 已提交
483 484
		return 0;

485 486 487
	if (page_zone_id(page) != page_zone_id(buddy))
		return 0;

488 489 490 491 492
	if (page_is_guard(buddy) && page_order(buddy) == order) {
		VM_BUG_ON(page_count(buddy) != 0);
		return 1;
	}

493
	if (PageBuddy(buddy) && page_order(buddy) == order) {
494
		VM_BUG_ON(page_count(buddy) != 0);
495
		return 1;
496
	}
497
	return 0;
L
Linus Torvalds 已提交
498 499 500 501 502 503 504 505 506 507 508 509 510 511 512
}

/*
 * 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
513
 * free pages of length of (1 << order) and marked with _mapcount -2. Page's
H
Hugh Dickins 已提交
514
 * order is recorded in page_private(page) field.
L
Linus Torvalds 已提交
515 516 517 518 519 520 521 522 523
 * So when we are allocating or freeing one, we can derive the state of the
 * other.  That is, if we allocate a small block, and both were   
 * free, the remainder of the region must be split into blocks.   
 * If a block is freed, and its buddy is also free, then this
 * triggers coalescing into a block of larger size.            
 *
 * -- wli
 */

N
Nick Piggin 已提交
524
static inline void __free_one_page(struct page *page,
525 526
		struct zone *zone, unsigned int order,
		int migratetype)
L
Linus Torvalds 已提交
527 528
{
	unsigned long page_idx;
529
	unsigned long combined_idx;
530
	unsigned long uninitialized_var(buddy_idx);
531
	struct page *buddy;
L
Linus Torvalds 已提交
532

N
Nick Piggin 已提交
533
	if (unlikely(PageCompound(page)))
534 535
		if (unlikely(destroy_compound_page(page, order)))
			return;
L
Linus Torvalds 已提交
536

537 538
	VM_BUG_ON(migratetype == -1);

L
Linus Torvalds 已提交
539 540
	page_idx = page_to_pfn(page) & ((1 << MAX_ORDER) - 1);

541
	VM_BUG_ON(page_idx & ((1 << order) - 1));
N
Nick Piggin 已提交
542
	VM_BUG_ON(bad_range(zone, page));
L
Linus Torvalds 已提交
543 544

	while (order < MAX_ORDER-1) {
545 546
		buddy_idx = __find_buddy_index(page_idx, order);
		buddy = page + (buddy_idx - page_idx);
547
		if (!page_is_buddy(page, buddy, order))
548
			break;
549 550 551 552 553 554 555 556 557 558 559 560 561
		/*
		 * 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);
		}
562
		combined_idx = buddy_idx & page_idx;
L
Linus Torvalds 已提交
563 564 565 566 567
		page = page + (combined_idx - page_idx);
		page_idx = combined_idx;
		order++;
	}
	set_page_order(page, order);
568 569 570 571 572 573 574 575 576

	/*
	 * 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
	 */
577
	if ((order < MAX_ORDER-2) && pfn_valid_within(page_to_pfn(buddy))) {
578
		struct page *higher_page, *higher_buddy;
579 580 581 582
		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);
583 584 585 586 587 588 589 590 591
		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
592 593 594
	zone->free_area[order].nr_free++;
}

595 596 597 598 599 600 601 602 603 604 605
/*
 * 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
606
static inline int free_pages_check(struct page *page)
L
Linus Torvalds 已提交
607
{
N
Nick Piggin 已提交
608 609
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
610
		(atomic_read(&page->_count) != 0) |
611 612
		(page->flags & PAGE_FLAGS_CHECK_AT_FREE) |
		(mem_cgroup_bad_page_check(page)))) {
N
Nick Piggin 已提交
613
		bad_page(page);
614
		return 1;
615
	}
616 617 618
	if (page->flags & PAGE_FLAGS_CHECK_AT_PREP)
		page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;
	return 0;
L
Linus Torvalds 已提交
619 620 621
}

/*
622
 * Frees a number of pages from the PCP lists
L
Linus Torvalds 已提交
623
 * Assumes all pages on list are in same zone, and of same order.
624
 * count is the number of pages to free.
L
Linus Torvalds 已提交
625 626 627 628 629 630 631
 *
 * 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.
 */
632 633
static void free_pcppages_bulk(struct zone *zone, int count,
					struct per_cpu_pages *pcp)
L
Linus Torvalds 已提交
634
{
635
	int migratetype = 0;
636
	int batch_free = 0;
637
	int to_free = count;
638

N
Nick Piggin 已提交
639
	spin_lock(&zone->lock);
640
	zone->all_unreclaimable = 0;
L
Linus Torvalds 已提交
641
	zone->pages_scanned = 0;
642

643
	while (to_free) {
N
Nick Piggin 已提交
644
		struct page *page;
645 646 647
		struct list_head *list;

		/*
648 649 650 651 652
		 * 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
653 654
		 */
		do {
655
			batch_free++;
656 657 658 659
			if (++migratetype == MIGRATE_PCPTYPES)
				migratetype = 0;
			list = &pcp->lists[migratetype];
		} while (list_empty(list));
N
Nick Piggin 已提交
660

661 662 663 664
		/* This is the only non-empty list. Free them all. */
		if (batch_free == MIGRATE_PCPTYPES)
			batch_free = to_free;

665 666 667 668
		do {
			page = list_entry(list->prev, struct page, lru);
			/* must delete as __free_one_page list manipulates */
			list_del(&page->lru);
669 670 671
			/* 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));
672
		} while (--to_free && --batch_free && !list_empty(list));
L
Linus Torvalds 已提交
673
	}
674
	__mod_zone_page_state(zone, NR_FREE_PAGES, count);
N
Nick Piggin 已提交
675
	spin_unlock(&zone->lock);
L
Linus Torvalds 已提交
676 677
}

678 679
static void free_one_page(struct zone *zone, struct page *page, int order,
				int migratetype)
L
Linus Torvalds 已提交
680
{
681
	spin_lock(&zone->lock);
682
	zone->all_unreclaimable = 0;
683
	zone->pages_scanned = 0;
684

685
	__free_one_page(page, zone, order, migratetype);
686
	__mod_zone_page_state(zone, NR_FREE_PAGES, 1 << order);
687
	spin_unlock(&zone->lock);
N
Nick Piggin 已提交
688 689
}

690
static bool free_pages_prepare(struct page *page, unsigned int order)
N
Nick Piggin 已提交
691
{
L
Linus Torvalds 已提交
692
	int i;
693
	int bad = 0;
L
Linus Torvalds 已提交
694

695
	trace_mm_page_free(page, order);
696 697
	kmemcheck_free_shadow(page, order);

A
Andrea Arcangeli 已提交
698 699 700 701
	if (PageAnon(page))
		page->mapping = NULL;
	for (i = 0; i < (1 << order); i++)
		bad += free_pages_check(page + i);
702
	if (bad)
703
		return false;
704

705
	if (!PageHighMem(page)) {
N
Nick Piggin 已提交
706
		debug_check_no_locks_freed(page_address(page),PAGE_SIZE<<order);
707 708 709
		debug_check_no_obj_freed(page_address(page),
					   PAGE_SIZE << order);
	}
N
Nick Piggin 已提交
710
	arch_free_page(page, order);
N
Nick Piggin 已提交
711
	kernel_map_pages(page, 1 << order, 0);
N
Nick Piggin 已提交
712

713 714 715 716 717 718 719 720 721 722 723
	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
724
	local_irq_save(flags);
725
	if (unlikely(wasMlocked))
726
		free_page_mlock(page);
727
	__count_vm_events(PGFREE, 1 << order);
728 729
	free_one_page(page_zone(page), page, order,
					get_pageblock_migratetype(page));
N
Nick Piggin 已提交
730
	local_irq_restore(flags);
L
Linus Torvalds 已提交
731 732
}

733
void __meminit __free_pages_bootmem(struct page *page, unsigned int order)
734
{
735 736
	unsigned int nr_pages = 1 << order;
	unsigned int loop;
737

738 739 740 741 742 743 744 745
	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);
746
	}
747 748 749

	set_page_refcounted(page);
	__free_pages(page, order);
750 751
}

L
Linus Torvalds 已提交
752 753 754 755 756 757 758 759 760 761 762 763 764 765 766

/*
 * 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
767
static inline void expand(struct zone *zone, struct page *page,
768 769
	int low, int high, struct free_area *area,
	int migratetype)
L
Linus Torvalds 已提交
770 771 772 773 774 775 776
{
	unsigned long size = 1 << high;

	while (high > low) {
		area--;
		high--;
		size >>= 1;
N
Nick Piggin 已提交
777
		VM_BUG_ON(bad_range(zone, &page[size]));
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794

#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
795
		list_add(&page[size].lru, &area->free_list[migratetype]);
L
Linus Torvalds 已提交
796 797 798 799 800 801 802 803
		area->nr_free++;
		set_page_order(&page[size], high);
	}
}

/*
 * This page is about to be returned from the page allocator
 */
804
static inline int check_new_page(struct page *page)
L
Linus Torvalds 已提交
805
{
N
Nick Piggin 已提交
806 807
	if (unlikely(page_mapcount(page) |
		(page->mapping != NULL)  |
808
		(atomic_read(&page->_count) != 0)  |
809 810
		(page->flags & PAGE_FLAGS_CHECK_AT_PREP) |
		(mem_cgroup_bad_page_check(page)))) {
N
Nick Piggin 已提交
811
		bad_page(page);
812
		return 1;
813
	}
814 815 816 817 818 819 820 821 822 823 824 825
	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;
	}
826

H
Hugh Dickins 已提交
827
	set_page_private(page, 0);
828
	set_page_refcounted(page);
N
Nick Piggin 已提交
829 830

	arch_alloc_page(page, order);
L
Linus Torvalds 已提交
831
	kernel_map_pages(page, 1 << order, 1);
N
Nick Piggin 已提交
832 833 834 835 836 837 838

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

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

839
	return 0;
L
Linus Torvalds 已提交
840 841
}

842 843 844 845
/*
 * Go through the free lists for the given migratetype and remove
 * the smallest available page from the freelists
 */
846 847
static inline
struct page *__rmqueue_smallest(struct zone *zone, unsigned int order,
848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872
						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;
}


873 874 875 876 877
/*
 * This array describes the order lists are fallen back to when
 * the free lists for the desirable migrate type are depleted
 */
static int fallbacks[MIGRATE_TYPES][MIGRATE_TYPES-1] = {
878 879 880 881
	[MIGRATE_UNMOVABLE]   = { MIGRATE_RECLAIMABLE, MIGRATE_MOVABLE,   MIGRATE_RESERVE },
	[MIGRATE_RECLAIMABLE] = { MIGRATE_UNMOVABLE,   MIGRATE_MOVABLE,   MIGRATE_RESERVE },
	[MIGRATE_MOVABLE]     = { MIGRATE_RECLAIMABLE, MIGRATE_UNMOVABLE, MIGRATE_RESERVE },
	[MIGRATE_RESERVE]     = { MIGRATE_RESERVE,     MIGRATE_RESERVE,   MIGRATE_RESERVE }, /* Never used */
882 883
};

884 885
/*
 * Move the free pages in a range to the free lists of the requested type.
886
 * Note that start_page and end_pages are not aligned on a pageblock
887 888
 * boundary. If alignment is required, use move_freepages_block()
 */
A
Adrian Bunk 已提交
889 890 891
static int move_freepages(struct zone *zone,
			  struct page *start_page, struct page *end_page,
			  int migratetype)
892 893 894
{
	struct page *page;
	unsigned long order;
895
	int pages_moved = 0;
896 897 898 899 900 901 902

#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
903
	 * grouping pages by mobility
904 905 906 907 908
	 */
	BUG_ON(page_zone(start_page) != page_zone(end_page));
#endif

	for (page = start_page; page <= end_page;) {
909 910 911
		/* Make sure we are not inadvertently changing nodes */
		VM_BUG_ON(page_to_nid(page) != zone_to_nid(zone));

912 913 914 915 916 917 918 919 920 921 922
		if (!pfn_valid_within(page_to_pfn(page))) {
			page++;
			continue;
		}

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

		order = page_order(page);
923 924
		list_move(&page->lru,
			  &zone->free_area[order].free_list[migratetype]);
925
		page += 1 << order;
926
		pages_moved += 1 << order;
927 928
	}

929
	return pages_moved;
930 931
}

A
Adrian Bunk 已提交
932 933
static int move_freepages_block(struct zone *zone, struct page *page,
				int migratetype)
934 935 936 937 938
{
	unsigned long start_pfn, end_pfn;
	struct page *start_page, *end_page;

	start_pfn = page_to_pfn(page);
939
	start_pfn = start_pfn & ~(pageblock_nr_pages-1);
940
	start_page = pfn_to_page(start_pfn);
941 942
	end_page = start_page + pageblock_nr_pages - 1;
	end_pfn = start_pfn + pageblock_nr_pages - 1;
943 944 945 946 947 948 949 950 951 952

	/* 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);
}

953 954 955 956 957 958 959 960 961 962 963
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;
	}
}

964
/* Remove an element from the buddy allocator from the fallback list */
965 966
static inline struct page *
__rmqueue_fallback(struct zone *zone, int order, int start_migratetype)
967 968 969 970 971 972 973 974 975 976 977 978
{
	struct free_area * area;
	int current_order;
	struct page *page;
	int migratetype, i;

	/* Find the largest possible block of pages in the other list */
	for (current_order = MAX_ORDER-1; current_order >= order;
						--current_order) {
		for (i = 0; i < MIGRATE_TYPES - 1; i++) {
			migratetype = fallbacks[start_migratetype][i];

979 980 981
			/* MIGRATE_RESERVE handled later if necessary */
			if (migratetype == MIGRATE_RESERVE)
				continue;
M
Mel Gorman 已提交
982

983 984 985 986 987 988 989 990 991
			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--;

			/*
992
			 * If breaking a large block of pages, move all free
993 994
			 * pages to the preferred allocation list. If falling
			 * back for a reclaimable kernel allocation, be more
L
Lucas De Marchi 已提交
995
			 * aggressive about taking ownership of free pages
996
			 */
997
			if (unlikely(current_order >= (pageblock_order >> 1)) ||
998 999
					start_migratetype == MIGRATE_RECLAIMABLE ||
					page_group_by_mobility_disabled) {
1000 1001 1002 1003 1004
				unsigned long pages;
				pages = move_freepages_block(zone, page,
								start_migratetype);

				/* Claim the whole block if over half of it is free */
1005 1006
				if (pages >= (1 << (pageblock_order-1)) ||
						page_group_by_mobility_disabled)
1007 1008 1009
					set_pageblock_migratetype(page,
								start_migratetype);

1010
				migratetype = start_migratetype;
1011
			}
1012 1013 1014 1015 1016

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

1017 1018 1019
			/* Take ownership for orders >= pageblock_order */
			if (current_order >= pageblock_order)
				change_pageblock_range(page, current_order,
1020 1021 1022
							start_migratetype);

			expand(zone, page, order, current_order, area, migratetype);
1023 1024 1025 1026

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

1027 1028 1029 1030
			return page;
		}
	}

1031
	return NULL;
1032 1033
}

1034
/*
L
Linus Torvalds 已提交
1035 1036 1037
 * Do the hard work of removing an element from the buddy allocator.
 * Call me with the zone->lock already held.
 */
1038 1039
static struct page *__rmqueue(struct zone *zone, unsigned int order,
						int migratetype)
L
Linus Torvalds 已提交
1040 1041 1042
{
	struct page *page;

1043
retry_reserve:
1044
	page = __rmqueue_smallest(zone, order, migratetype);
1045

1046
	if (unlikely(!page) && migratetype != MIGRATE_RESERVE) {
1047
		page = __rmqueue_fallback(zone, order, migratetype);
1048

1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059
		/*
		 * 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;
		}
	}

1060
	trace_mm_page_alloc_zone_locked(page, order, migratetype);
1061
	return page;
L
Linus Torvalds 已提交
1062 1063 1064 1065 1066 1067 1068 1069
}

/* 
 * Obtain a specified number of elements from the buddy allocator, all under
 * a single hold of the lock, for efficiency.  Add them to the supplied list.
 * Returns the number of new pages which were placed at *list.
 */
static int rmqueue_bulk(struct zone *zone, unsigned int order, 
1070
			unsigned long count, struct list_head *list,
1071
			int migratetype, int cold)
L
Linus Torvalds 已提交
1072 1073 1074
{
	int i;
	
N
Nick Piggin 已提交
1075
	spin_lock(&zone->lock);
L
Linus Torvalds 已提交
1076
	for (i = 0; i < count; ++i) {
1077
		struct page *page = __rmqueue(zone, order, migratetype);
N
Nick Piggin 已提交
1078
		if (unlikely(page == NULL))
L
Linus Torvalds 已提交
1079
			break;
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089

		/*
		 * 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.
		 */
1090 1091 1092 1093
		if (likely(cold == 0))
			list_add(&page->lru, list);
		else
			list_add_tail(&page->lru, list);
1094
		set_page_private(page, migratetype);
1095
		list = &page->lru;
L
Linus Torvalds 已提交
1096
	}
1097
	__mod_zone_page_state(zone, NR_FREE_PAGES, -(i << order));
N
Nick Piggin 已提交
1098
	spin_unlock(&zone->lock);
N
Nick Piggin 已提交
1099
	return i;
L
Linus Torvalds 已提交
1100 1101
}

1102
#ifdef CONFIG_NUMA
1103
/*
1104 1105 1106 1107
 * Called from the vmstat counter updater to drain pagesets of this
 * currently executing processor on remote nodes after they have
 * expired.
 *
1108 1109
 * Note that this function must be called with the thread pinned to
 * a single processor.
1110
 */
1111
void drain_zone_pages(struct zone *zone, struct per_cpu_pages *pcp)
1112 1113
{
	unsigned long flags;
1114
	int to_drain;
1115

1116 1117 1118 1119 1120
	local_irq_save(flags);
	if (pcp->count >= pcp->batch)
		to_drain = pcp->batch;
	else
		to_drain = pcp->count;
1121
	free_pcppages_bulk(zone, to_drain, pcp);
1122 1123
	pcp->count -= to_drain;
	local_irq_restore(flags);
1124 1125 1126
}
#endif

1127 1128 1129 1130 1131 1132 1133 1134
/*
 * 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
1135
{
N
Nick Piggin 已提交
1136
	unsigned long flags;
L
Linus Torvalds 已提交
1137 1138
	struct zone *zone;

1139
	for_each_populated_zone(zone) {
L
Linus Torvalds 已提交
1140
		struct per_cpu_pageset *pset;
1141
		struct per_cpu_pages *pcp;
L
Linus Torvalds 已提交
1142

1143 1144
		local_irq_save(flags);
		pset = per_cpu_ptr(zone->pageset, cpu);
1145 1146

		pcp = &pset->pcp;
1147 1148 1149 1150
		if (pcp->count) {
			free_pcppages_bulk(zone, pcp->count, pcp);
			pcp->count = 0;
		}
1151
		local_irq_restore(flags);
L
Linus Torvalds 已提交
1152 1153 1154
	}
}

1155 1156 1157 1158 1159 1160 1161 1162 1163
/*
 * 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());
}

/*
1164 1165 1166 1167 1168 1169 1170
 * 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().
1171 1172 1173
 */
void drain_all_pages(void)
{
1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204
	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);
1205 1206
}

1207
#ifdef CONFIG_HIBERNATION
L
Linus Torvalds 已提交
1208 1209 1210

void mark_free_pages(struct zone *zone)
{
1211 1212
	unsigned long pfn, max_zone_pfn;
	unsigned long flags;
1213
	int order, t;
L
Linus Torvalds 已提交
1214 1215 1216 1217 1218 1219
	struct list_head *curr;

	if (!zone->spanned_pages)
		return;

	spin_lock_irqsave(&zone->lock, flags);
1220 1221 1222 1223 1224 1225

	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);

1226 1227
			if (!swsusp_page_is_forbidden(page))
				swsusp_unset_page_free(page);
1228
		}
L
Linus Torvalds 已提交
1229

1230 1231
	for_each_migratetype_order(order, t) {
		list_for_each(curr, &zone->free_area[order].free_list[t]) {
1232
			unsigned long i;
L
Linus Torvalds 已提交
1233

1234 1235
			pfn = page_to_pfn(list_entry(curr, struct page, lru));
			for (i = 0; i < (1UL << order); i++)
1236
				swsusp_set_page_free(pfn_to_page(pfn + i));
1237
		}
1238
	}
L
Linus Torvalds 已提交
1239 1240
	spin_unlock_irqrestore(&zone->lock, flags);
}
1241
#endif /* CONFIG_PM */
L
Linus Torvalds 已提交
1242 1243 1244

/*
 * Free a 0-order page
L
Li Hong 已提交
1245
 * cold == 1 ? free a cold page : free a hot page
L
Linus Torvalds 已提交
1246
 */
L
Li Hong 已提交
1247
void free_hot_cold_page(struct page *page, int cold)
L
Linus Torvalds 已提交
1248 1249 1250 1251
{
	struct zone *zone = page_zone(page);
	struct per_cpu_pages *pcp;
	unsigned long flags;
1252
	int migratetype;
1253
	int wasMlocked = __TestClearPageMlocked(page);
L
Linus Torvalds 已提交
1254

1255
	if (!free_pages_prepare(page, 0))
1256 1257
		return;

1258 1259
	migratetype = get_pageblock_migratetype(page);
	set_page_private(page, migratetype);
L
Linus Torvalds 已提交
1260
	local_irq_save(flags);
1261
	if (unlikely(wasMlocked))
1262
		free_page_mlock(page);
1263
	__count_vm_event(PGFREE);
1264

1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279
	/*
	 * 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;
	}

1280
	pcp = &this_cpu_ptr(zone->pageset)->pcp;
1281
	if (cold)
1282
		list_add_tail(&page->lru, &pcp->lists[migratetype]);
1283
	else
1284
		list_add(&page->lru, &pcp->lists[migratetype]);
L
Linus Torvalds 已提交
1285
	pcp->count++;
N
Nick Piggin 已提交
1286
	if (pcp->count >= pcp->high) {
1287
		free_pcppages_bulk(zone, pcp->batch, pcp);
N
Nick Piggin 已提交
1288 1289
		pcp->count -= pcp->batch;
	}
1290 1291

out:
L
Linus Torvalds 已提交
1292 1293 1294
	local_irq_restore(flags);
}

1295 1296 1297 1298 1299 1300 1301 1302
/*
 * 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) {
1303
		trace_mm_page_free_batched(page, cold);
1304 1305 1306 1307
		free_hot_cold_page(page, cold);
	}
}

N
Nick Piggin 已提交
1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319
/*
 * 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
1320 1321
	VM_BUG_ON(PageCompound(page));
	VM_BUG_ON(!page_count(page));
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331

#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

1332 1333
	for (i = 1; i < (1 << order); i++)
		set_page_refcounted(page + i);
N
Nick Piggin 已提交
1334 1335
}

1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380
/*
 * 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;
		for (; page < endpage; page += pageblock_nr_pages)
			set_pageblock_migratetype(page, MIGRATE_MOVABLE);
	}

	return 1 << order;
}

L
Linus Torvalds 已提交
1381 1382 1383 1384 1385
/*
 * 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.
 */
1386 1387
static inline
struct page *buffered_rmqueue(struct zone *preferred_zone,
1388 1389
			struct zone *zone, int order, gfp_t gfp_flags,
			int migratetype)
L
Linus Torvalds 已提交
1390 1391
{
	unsigned long flags;
1392
	struct page *page;
L
Linus Torvalds 已提交
1393 1394
	int cold = !!(gfp_flags & __GFP_COLD);

1395
again:
N
Nick Piggin 已提交
1396
	if (likely(order == 0)) {
L
Linus Torvalds 已提交
1397
		struct per_cpu_pages *pcp;
1398
		struct list_head *list;
L
Linus Torvalds 已提交
1399 1400

		local_irq_save(flags);
1401 1402
		pcp = &this_cpu_ptr(zone->pageset)->pcp;
		list = &pcp->lists[migratetype];
1403
		if (list_empty(list)) {
1404
			pcp->count += rmqueue_bulk(zone, 0,
1405
					pcp->batch, list,
1406
					migratetype, cold);
1407
			if (unlikely(list_empty(list)))
1408
				goto failed;
1409
		}
1410

1411 1412 1413 1414 1415
		if (cold)
			page = list_entry(list->prev, struct page, lru);
		else
			page = list_entry(list->next, struct page, lru);

1416 1417
		list_del(&page->lru);
		pcp->count--;
R
Rohit Seth 已提交
1418
	} else {
1419 1420 1421 1422 1423 1424 1425 1426
		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
1427
			 * allocate greater than order-1 page units with
1428 1429
			 * __GFP_NOFAIL.
			 */
1430
			WARN_ON_ONCE(order > 1);
1431
		}
L
Linus Torvalds 已提交
1432
		spin_lock_irqsave(&zone->lock, flags);
1433
		page = __rmqueue(zone, order, migratetype);
N
Nick Piggin 已提交
1434 1435 1436
		spin_unlock(&zone->lock);
		if (!page)
			goto failed;
1437
		__mod_zone_page_state(zone, NR_FREE_PAGES, -(1 << order));
L
Linus Torvalds 已提交
1438 1439
	}

1440
	__count_zone_vm_events(PGALLOC, zone, 1 << order);
A
Andi Kleen 已提交
1441
	zone_statistics(preferred_zone, zone, gfp_flags);
N
Nick Piggin 已提交
1442
	local_irq_restore(flags);
L
Linus Torvalds 已提交
1443

N
Nick Piggin 已提交
1444
	VM_BUG_ON(bad_range(zone, page));
N
Nick Piggin 已提交
1445
	if (prep_new_page(page, order, gfp_flags))
N
Nick Piggin 已提交
1446
		goto again;
L
Linus Torvalds 已提交
1447
	return page;
N
Nick Piggin 已提交
1448 1449 1450 1451

failed:
	local_irq_restore(flags);
	return NULL;
L
Linus Torvalds 已提交
1452 1453
}

1454 1455 1456 1457 1458 1459 1460 1461 1462
/* 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)

1463 1464 1465
#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
1466

1467 1468
#ifdef CONFIG_FAIL_PAGE_ALLOC

1469
static struct {
1470 1471 1472 1473
	struct fault_attr attr;

	u32 ignore_gfp_highmem;
	u32 ignore_gfp_wait;
1474
	u32 min_order;
1475 1476
} fail_page_alloc = {
	.attr = FAULT_ATTR_INITIALIZER,
1477 1478
	.ignore_gfp_wait = 1,
	.ignore_gfp_highmem = 1,
1479
	.min_order = 1,
1480 1481 1482 1483 1484 1485 1486 1487 1488 1489
};

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);

static int should_fail_alloc_page(gfp_t gfp_mask, unsigned int order)
{
1490 1491
	if (order < fail_page_alloc.min_order)
		return 0;
1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505
	if (gfp_mask & __GFP_NOFAIL)
		return 0;
	if (fail_page_alloc.ignore_gfp_highmem && (gfp_mask & __GFP_HIGHMEM))
		return 0;
	if (fail_page_alloc.ignore_gfp_wait && (gfp_mask & __GFP_WAIT))
		return 0;

	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
1506
	umode_t mode = S_IFREG | S_IRUSR | S_IWUSR;
1507 1508
	struct dentry *dir;

1509 1510 1511 1512
	dir = fault_create_debugfs_attr("fail_page_alloc", NULL,
					&fail_page_alloc.attr);
	if (IS_ERR(dir))
		return PTR_ERR(dir);
1513

1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525
	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:
1526
	debugfs_remove_recursive(dir);
1527

1528
	return -ENOMEM;
1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543
}

late_initcall(fail_page_alloc_debugfs);

#endif /* CONFIG_FAULT_INJECTION_DEBUG_FS */

#else /* CONFIG_FAIL_PAGE_ALLOC */

static inline int should_fail_alloc_page(gfp_t gfp_mask, unsigned int order)
{
	return 0;
}

#endif /* CONFIG_FAIL_PAGE_ALLOC */

L
Linus Torvalds 已提交
1544
/*
1545
 * Return true if free pages are above 'mark'. This takes into account the order
L
Linus Torvalds 已提交
1546 1547
 * of the allocation.
 */
1548 1549
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
1550 1551
{
	/* free_pages my go negative - that's OK */
1552
	long min = mark;
L
Linus Torvalds 已提交
1553 1554
	int o;

1555
	free_pages -= (1 << order) - 1;
R
Rohit Seth 已提交
1556
	if (alloc_flags & ALLOC_HIGH)
L
Linus Torvalds 已提交
1557
		min -= min / 2;
R
Rohit Seth 已提交
1558
	if (alloc_flags & ALLOC_HARDER)
L
Linus Torvalds 已提交
1559 1560 1561
		min -= min / 4;

	if (free_pages <= min + z->lowmem_reserve[classzone_idx])
1562
		return false;
L
Linus Torvalds 已提交
1563 1564 1565 1566 1567 1568 1569 1570
	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)
1571
			return false;
L
Linus Torvalds 已提交
1572
	}
1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592
	return true;
}

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);

	return __zone_watermark_ok(z, order, mark, classzone_idx, alloc_flags,
								free_pages);
L
Linus Torvalds 已提交
1593 1594
}

1595 1596 1597 1598 1599 1600
#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
1601
 * that have to skip over a lot of full or unallowed zones.
1602 1603 1604
 *
 * If the zonelist cache is present in the passed in zonelist, then
 * returns a pointer to the allowed node mask (either the current
1605
 * tasks mems_allowed, or node_states[N_HIGH_MEMORY].)
1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626
 *
 * 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
1627
	if (time_after(jiffies, zlc->last_full_zap + HZ)) {
1628 1629 1630 1631 1632 1633
		bitmap_zero(zlc->fullzones, MAX_ZONES_PER_ZONELIST);
		zlc->last_full_zap = jiffies;
	}

	allowednodes = !in_interrupt() && (alloc_flags & ALLOC_CPUSET) ?
					&cpuset_current_mems_allowed :
1634
					&node_states[N_HIGH_MEMORY];
1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659
	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.
 */
1660
static int zlc_zone_worth_trying(struct zonelist *zonelist, struct zoneref *z,
1661 1662 1663 1664 1665 1666 1667 1668 1669 1670
						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;

1671
	i = z - zonelist->_zonerefs;
1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682
	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.
 */
1683
static void zlc_mark_zone_full(struct zonelist *zonelist, struct zoneref *z)
1684 1685 1686 1687 1688 1689 1690 1691
{
	struct zonelist_cache *zlc;	/* cached zonelist speedup info */
	int i;				/* index of *z in zonelist zones */

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

1692
	i = z - zonelist->_zonerefs;
1693 1694 1695 1696

	set_bit(i, zlc->fullzones);
}

1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711
/*
 * 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);
}

1712 1713 1714 1715 1716 1717 1718
#else	/* CONFIG_NUMA */

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

1719
static int zlc_zone_worth_trying(struct zonelist *zonelist, struct zoneref *z,
1720 1721 1722 1723 1724
				nodemask_t *allowednodes)
{
	return 1;
}

1725
static void zlc_mark_zone_full(struct zonelist *zonelist, struct zoneref *z)
1726 1727
{
}
1728 1729 1730 1731

static void zlc_clear_zones_full(struct zonelist *zonelist)
{
}
1732 1733
#endif	/* CONFIG_NUMA */

R
Rohit Seth 已提交
1734
/*
1735
 * get_page_from_freelist goes through the zonelist trying to allocate
R
Rohit Seth 已提交
1736 1737 1738
 * a page.
 */
static struct page *
1739
get_page_from_freelist(gfp_t gfp_mask, nodemask_t *nodemask, unsigned int order,
1740
		struct zonelist *zonelist, int high_zoneidx, int alloc_flags,
1741
		struct zone *preferred_zone, int migratetype)
M
Martin Hicks 已提交
1742
{
1743
	struct zoneref *z;
R
Rohit Seth 已提交
1744
	struct page *page = NULL;
1745
	int classzone_idx;
1746
	struct zone *zone;
1747 1748 1749
	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 */
1750

1751
	classzone_idx = zone_idx(preferred_zone);
1752
zonelist_scan:
R
Rohit Seth 已提交
1753
	/*
1754
	 * Scan zonelist, looking for a zone with enough free.
R
Rohit Seth 已提交
1755 1756
	 * See also cpuset_zone_allowed() comment in kernel/cpuset.c.
	 */
1757 1758
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						high_zoneidx, nodemask) {
1759 1760 1761
		if (NUMA_BUILD && zlc_active &&
			!zlc_zone_worth_trying(zonelist, z, allowednodes))
				continue;
R
Rohit Seth 已提交
1762
		if ((alloc_flags & ALLOC_CPUSET) &&
1763
			!cpuset_zone_allowed_softwall(zone, gfp_mask))
1764
				continue;
1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793
		/*
		 * 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
1794

1795
		BUILD_BUG_ON(ALLOC_NO_WATERMARKS < NR_WMARK);
R
Rohit Seth 已提交
1796
		if (!(alloc_flags & ALLOC_NO_WATERMARKS)) {
1797
			unsigned long mark;
1798 1799
			int ret;

1800
			mark = zone->watermark[alloc_flags & ALLOC_WMARK_MASK];
1801 1802 1803 1804
			if (zone_watermark_ok(zone, order, mark,
				    classzone_idx, alloc_flags))
				goto try_this_zone;

1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815
			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;
			}

1816 1817 1818
			if (zone_reclaim_mode == 0)
				goto this_zone_full;

1819 1820 1821 1822 1823 1824 1825 1826
			/*
			 * 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;

1827 1828 1829 1830
			ret = zone_reclaim(zone, gfp_mask, order);
			switch (ret) {
			case ZONE_RECLAIM_NOSCAN:
				/* did not scan */
1831
				continue;
1832 1833
			case ZONE_RECLAIM_FULL:
				/* scanned but unreclaimable */
1834
				continue;
1835 1836 1837 1838
			default:
				/* did we reclaim enough */
				if (!zone_watermark_ok(zone, order, mark,
						classzone_idx, alloc_flags))
1839
					goto this_zone_full;
1840
			}
R
Rohit Seth 已提交
1841 1842
		}

1843
try_this_zone:
1844 1845
		page = buffered_rmqueue(preferred_zone, zone, order,
						gfp_mask, migratetype);
1846
		if (page)
R
Rohit Seth 已提交
1847
			break;
1848 1849 1850
this_zone_full:
		if (NUMA_BUILD)
			zlc_mark_zone_full(zonelist, z);
1851
	}
1852 1853 1854 1855 1856 1857

	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
1858
	return page;
M
Martin Hicks 已提交
1859 1860
}

1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
/*
 * 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;
}

1875 1876 1877 1878 1879 1880 1881 1882
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;

1883 1884
	if ((gfp_mask & __GFP_NOWARN) || !__ratelimit(&nopage_rs) ||
	    debug_guardpage_minorder() > 0)
1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
		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
1900 1901 1902
		struct va_format vaf;
		va_list args;

1903
		va_start(args, fmt);
J
Joe Perches 已提交
1904 1905 1906 1907 1908 1909

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

		pr_warn("%pV", &vaf);

1910 1911 1912
		va_end(args);
	}

J
Joe Perches 已提交
1913 1914
	pr_warn("%s: page allocation failure: order:%d, mode:0x%x\n",
		current->comm, order, gfp_mask);
1915 1916 1917 1918 1919 1920

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

1921 1922
static inline int
should_alloc_retry(gfp_t gfp_mask, unsigned int order,
1923
				unsigned long did_some_progress,
1924
				unsigned long pages_reclaimed)
L
Linus Torvalds 已提交
1925
{
1926 1927 1928
	/* Do not loop if specifically requested */
	if (gfp_mask & __GFP_NORETRY)
		return 0;
L
Linus Torvalds 已提交
1929

1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941
	/* 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;

1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958
	/*
	 * 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;
1959

1960 1961
	return 0;
}
1962

1963 1964 1965
static inline struct page *
__alloc_pages_may_oom(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
1966 1967
	nodemask_t *nodemask, struct zone *preferred_zone,
	int migratetype)
1968 1969 1970 1971
{
	struct page *page;

	/* Acquire the OOM killer lock for the zones in zonelist */
1972
	if (!try_set_zonelist_oom(zonelist, gfp_mask)) {
1973
		schedule_timeout_uninterruptible(1);
L
Linus Torvalds 已提交
1974 1975
		return NULL;
	}
1976

1977 1978 1979 1980 1981 1982 1983
	/*
	 * 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,
1984
		ALLOC_WMARK_HIGH|ALLOC_CPUSET,
1985
		preferred_zone, migratetype);
R
Rohit Seth 已提交
1986
	if (page)
1987 1988
		goto out;

1989 1990 1991 1992
	if (!(gfp_mask & __GFP_NOFAIL)) {
		/* The OOM killer will not help higher order allocs */
		if (order > PAGE_ALLOC_COSTLY_ORDER)
			goto out;
1993 1994 1995
		/* The OOM killer does not needlessly kill tasks for lowmem */
		if (high_zoneidx < ZONE_NORMAL)
			goto out;
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005
		/*
		 * 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;
	}
2006
	/* Exhausted what can be done so it's blamo time */
2007
	out_of_memory(zonelist, gfp_mask, order, nodemask, false);
2008 2009 2010 2011 2012 2013

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

2014 2015 2016 2017 2018 2019
#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,
2020 2021 2022
	int migratetype, bool sync_migration,
	bool *deferred_compaction,
	unsigned long *did_some_progress)
2023 2024 2025
{
	struct page *page;

2026
	if (!order)
2027 2028
		return NULL;

2029
	if (compaction_deferred(preferred_zone, order)) {
2030 2031 2032 2033
		*deferred_compaction = true;
		return NULL;
	}

2034
	current->flags |= PF_MEMALLOC;
2035
	*did_some_progress = try_to_compact_pages(zonelist, order, gfp_mask,
2036
						nodemask, sync_migration);
2037
	current->flags &= ~PF_MEMALLOC;
2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048
	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) {
2049 2050
			preferred_zone->compact_considered = 0;
			preferred_zone->compact_defer_shift = 0;
2051 2052
			if (order >= preferred_zone->compact_order_failed)
				preferred_zone->compact_order_failed = order + 1;
2053 2054 2055 2056 2057 2058 2059 2060 2061 2062
			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);
2063 2064 2065 2066 2067 2068

		/*
		 * As async compaction considers a subset of pageblocks, only
		 * defer if the failure was a sync compaction failure.
		 */
		if (sync_migration)
2069
			defer_compaction(preferred_zone, order);
2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080

		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,
2081 2082 2083
	int migratetype, bool sync_migration,
	bool *deferred_compaction,
	unsigned long *did_some_progress)
2084 2085 2086 2087 2088
{
	return NULL;
}
#endif /* CONFIG_COMPACTION */

2089 2090 2091 2092
/* 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,
2093
	nodemask_t *nodemask, int alloc_flags, struct zone *preferred_zone,
2094
	int migratetype, unsigned long *did_some_progress)
2095 2096 2097
{
	struct page *page = NULL;
	struct reclaim_state reclaim_state;
2098
	bool drained = false;
2099 2100 2101 2102 2103

	cond_resched();

	/* We now go into synchronous reclaim */
	cpuset_memory_pressure_bump();
2104
	current->flags |= PF_MEMALLOC;
2105 2106
	lockdep_set_current_reclaim_state(gfp_mask);
	reclaim_state.reclaimed_slab = 0;
2107
	current->reclaim_state = &reclaim_state;
2108 2109 2110

	*did_some_progress = try_to_free_pages(zonelist, order, gfp_mask, nodemask);

2111
	current->reclaim_state = NULL;
2112
	lockdep_clear_current_reclaim_state();
2113
	current->flags &= ~PF_MEMALLOC;
2114 2115 2116

	cond_resched();

2117 2118
	if (unlikely(!(*did_some_progress)))
		return NULL;
2119

2120 2121 2122 2123
	/* After successful reclaim, reconsider all zones for allocation */
	if (NUMA_BUILD)
		zlc_clear_zones_full(zonelist);

2124 2125
retry:
	page = get_page_from_freelist(gfp_mask, nodemask, order,
2126
					zonelist, high_zoneidx,
2127 2128
					alloc_flags, preferred_zone,
					migratetype);
2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139

	/*
	 * 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;
	}

2140 2141 2142
	return page;
}

L
Linus Torvalds 已提交
2143
/*
2144 2145
 * 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
2146
 */
2147 2148 2149
static inline struct page *
__alloc_pages_high_priority(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
2150 2151
	nodemask_t *nodemask, struct zone *preferred_zone,
	int migratetype)
2152 2153 2154 2155 2156
{
	struct page *page;

	do {
		page = get_page_from_freelist(gfp_mask, nodemask, order,
2157
			zonelist, high_zoneidx, ALLOC_NO_WATERMARKS,
2158
			preferred_zone, migratetype);
2159 2160

		if (!page && gfp_mask & __GFP_NOFAIL)
2161
			wait_iff_congested(preferred_zone, BLK_RW_ASYNC, HZ/50);
2162 2163 2164 2165 2166 2167 2168
	} while (!page && (gfp_mask & __GFP_NOFAIL));

	return page;
}

static inline
void wake_all_kswapd(unsigned int order, struct zonelist *zonelist,
2169 2170
						enum zone_type high_zoneidx,
						enum zone_type classzone_idx)
L
Linus Torvalds 已提交
2171
{
2172 2173
	struct zoneref *z;
	struct zone *zone;
L
Linus Torvalds 已提交
2174

2175
	for_each_zone_zonelist(zone, z, zonelist, high_zoneidx)
2176
		wakeup_kswapd(zone, order, classzone_idx);
2177
}
2178

2179 2180 2181 2182 2183
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
2184

2185
	/* __GFP_HIGH is assumed to be the same as ALLOC_HIGH to save a branch. */
2186
	BUILD_BUG_ON(__GFP_HIGH != (__force gfp_t) ALLOC_HIGH);
2187

2188 2189 2190 2191 2192 2193
	/*
	 * 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).
	 */
2194
	alloc_flags |= (__force int) (gfp_mask & __GFP_HIGH);
L
Linus Torvalds 已提交
2195

2196
	if (!wait) {
2197 2198 2199 2200 2201 2202
		/*
		 * Not worth trying to allocate harder for
		 * __GFP_NOMEMALLOC even if it can't schedule.
		 */
		if  (!(gfp_mask & __GFP_NOMEMALLOC))
			alloc_flags |= ALLOC_HARDER;
2203
		/*
2204 2205
		 * Ignore cpuset if GFP_ATOMIC (!wait) rather than fail alloc.
		 * See also cpuset_zone_allowed() comment in kernel/cpuset.c.
2206
		 */
2207
		alloc_flags &= ~ALLOC_CPUSET;
2208
	} else if (unlikely(rt_task(current)) && !in_interrupt())
2209 2210 2211 2212
		alloc_flags |= ALLOC_HARDER;

	if (likely(!(gfp_mask & __GFP_NOMEMALLOC))) {
		if (!in_interrupt() &&
2213
		    ((current->flags & PF_MEMALLOC) ||
2214 2215
		     unlikely(test_thread_flag(TIF_MEMDIE))))
			alloc_flags |= ALLOC_NO_WATERMARKS;
L
Linus Torvalds 已提交
2216
	}
2217

2218 2219 2220
	return alloc_flags;
}

2221 2222 2223
static inline struct page *
__alloc_pages_slowpath(gfp_t gfp_mask, unsigned int order,
	struct zonelist *zonelist, enum zone_type high_zoneidx,
2224 2225
	nodemask_t *nodemask, struct zone *preferred_zone,
	int migratetype)
2226 2227 2228 2229 2230 2231
{
	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;
2232
	bool sync_migration = false;
2233
	bool deferred_compaction = false;
L
Linus Torvalds 已提交
2234

2235 2236 2237 2238 2239 2240
	/*
	 * 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.
	 */
2241 2242
	if (order >= MAX_ORDER) {
		WARN_ON_ONCE(!(gfp_mask & __GFP_NOWARN));
2243
		return NULL;
2244
	}
L
Linus Torvalds 已提交
2245

2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
	/*
	 * 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;

2257
restart:
A
Andrea Arcangeli 已提交
2258 2259
	if (!(gfp_mask & __GFP_NO_KSWAPD))
		wake_all_kswapd(order, zonelist, high_zoneidx,
2260
						zone_idx(preferred_zone));
L
Linus Torvalds 已提交
2261

2262
	/*
R
Rohit Seth 已提交
2263 2264 2265
	 * 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.
2266
	 */
2267
	alloc_flags = gfp_to_alloc_flags(gfp_mask);
L
Linus Torvalds 已提交
2268

2269 2270 2271 2272 2273 2274 2275 2276
	/*
	 * 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);

2277
rebalance:
2278
	/* This is the last chance, in general, before the goto nopage. */
2279
	page = get_page_from_freelist(gfp_mask, nodemask, order, zonelist,
2280 2281
			high_zoneidx, alloc_flags & ~ALLOC_NO_WATERMARKS,
			preferred_zone, migratetype);
R
Rohit Seth 已提交
2282 2283
	if (page)
		goto got_pg;
L
Linus Torvalds 已提交
2284

2285
	/* Allocate without watermarks if the context allows */
2286 2287 2288 2289 2290 2291
	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
2292 2293 2294 2295 2296 2297
	}

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

2298
	/* Avoid recursion of direct reclaim */
2299
	if (current->flags & PF_MEMALLOC)
2300 2301
		goto nopage;

2302 2303 2304 2305
	/* Avoid allocations with no watermarks from looping endlessly */
	if (test_thread_flag(TIF_MEMDIE) && !(gfp_mask & __GFP_NOFAIL))
		goto nopage;

2306 2307 2308 2309
	/*
	 * Try direct compaction. The first pass is asynchronous. Subsequent
	 * attempts after direct reclaim are synchronous
	 */
2310 2311 2312 2313
	page = __alloc_pages_direct_compact(gfp_mask, order,
					zonelist, high_zoneidx,
					nodemask,
					alloc_flags, preferred_zone,
2314 2315 2316
					migratetype, sync_migration,
					&deferred_compaction,
					&did_some_progress);
2317 2318
	if (page)
		goto got_pg;
2319
	sync_migration = true;
2320

2321 2322 2323 2324 2325 2326 2327 2328 2329
	/*
	 * 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;

2330 2331 2332 2333
	/* Try direct reclaim and then allocating */
	page = __alloc_pages_direct_reclaim(gfp_mask, order,
					zonelist, high_zoneidx,
					nodemask,
2334
					alloc_flags, preferred_zone,
2335
					migratetype, &did_some_progress);
2336 2337
	if (page)
		goto got_pg;
L
Linus Torvalds 已提交
2338

2339
	/*
2340 2341
	 * If we failed to make any progress reclaiming, then we are
	 * running out of options and have to consider going OOM
2342
	 */
2343 2344
	if (!did_some_progress) {
		if ((gfp_mask & __GFP_FS) && !(gfp_mask & __GFP_NORETRY)) {
2345 2346
			if (oom_killer_disabled)
				goto nopage;
2347 2348 2349 2350
			/* Coredumps can quickly deplete all memory reserves */
			if ((current->flags & PF_DUMPCORE) &&
			    !(gfp_mask & __GFP_NOFAIL))
				goto nopage;
2351 2352
			page = __alloc_pages_may_oom(gfp_mask, order,
					zonelist, high_zoneidx,
2353 2354
					nodemask, preferred_zone,
					migratetype);
2355 2356
			if (page)
				goto got_pg;
L
Linus Torvalds 已提交
2357

2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374
			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;
			}
2375

2376 2377
			goto restart;
		}
L
Linus Torvalds 已提交
2378 2379
	}

2380
	/* Check if we should retry the allocation */
2381
	pages_reclaimed += did_some_progress;
2382 2383
	if (should_alloc_retry(gfp_mask, order, did_some_progress,
						pages_reclaimed)) {
2384
		/* Wait for some write requests to complete then retry */
2385
		wait_iff_congested(preferred_zone, BLK_RW_ASYNC, HZ/50);
L
Linus Torvalds 已提交
2386
		goto rebalance;
2387 2388 2389 2390 2391 2392 2393 2394 2395 2396
	} 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,
2397 2398 2399
					migratetype, sync_migration,
					&deferred_compaction,
					&did_some_progress);
2400 2401
		if (page)
			goto got_pg;
L
Linus Torvalds 已提交
2402 2403 2404
	}

nopage:
2405
	warn_alloc_failed(gfp_mask, order, NULL);
2406
	return page;
L
Linus Torvalds 已提交
2407
got_pg:
2408 2409
	if (kmemcheck_enabled)
		kmemcheck_pagealloc_alloc(page, order, gfp_mask);
L
Linus Torvalds 已提交
2410
	return page;
2411

L
Linus Torvalds 已提交
2412
}
2413 2414 2415 2416 2417 2418 2419 2420 2421

/*
 * 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);
2422
	struct zone *preferred_zone;
2423
	struct page *page = NULL;
2424
	int migratetype = allocflags_to_migratetype(gfp_mask);
2425
	unsigned int cpuset_mems_cookie;
2426

2427 2428
	gfp_mask &= gfp_allowed_mask;

2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443
	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;

2444 2445 2446
retry_cpuset:
	cpuset_mems_cookie = get_mems_allowed();

2447
	/* The preferred zone is used for statistics later */
2448 2449 2450
	first_zones_zonelist(zonelist, high_zoneidx,
				nodemask ? : &cpuset_current_mems_allowed,
				&preferred_zone);
2451 2452
	if (!preferred_zone)
		goto out;
2453 2454

	/* First allocation attempt */
2455
	page = get_page_from_freelist(gfp_mask|__GFP_HARDWALL, nodemask, order,
2456
			zonelist, high_zoneidx, ALLOC_WMARK_LOW|ALLOC_CPUSET,
2457
			preferred_zone, migratetype);
2458 2459
	if (unlikely(!page))
		page = __alloc_pages_slowpath(gfp_mask, order,
2460
				zonelist, high_zoneidx, nodemask,
2461
				preferred_zone, migratetype);
2462

2463
	trace_mm_page_alloc(page, order, gfp_mask, migratetype);
2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474

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;

2475
	return page;
L
Linus Torvalds 已提交
2476
}
2477
EXPORT_SYMBOL(__alloc_pages_nodemask);
L
Linus Torvalds 已提交
2478 2479 2480 2481

/*
 * Common helper functions.
 */
H
Harvey Harrison 已提交
2482
unsigned long __get_free_pages(gfp_t gfp_mask, unsigned int order)
L
Linus Torvalds 已提交
2483
{
2484 2485 2486 2487 2488 2489 2490 2491
	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
2492 2493 2494 2495 2496 2497 2498
	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
2499
unsigned long get_zeroed_page(gfp_t gfp_mask)
L
Linus Torvalds 已提交
2500
{
2501
	return __get_free_pages(gfp_mask | __GFP_ZERO, 0);
L
Linus Torvalds 已提交
2502 2503 2504
}
EXPORT_SYMBOL(get_zeroed_page);

H
Harvey Harrison 已提交
2505
void __free_pages(struct page *page, unsigned int order)
L
Linus Torvalds 已提交
2506
{
N
Nick Piggin 已提交
2507
	if (put_page_testzero(page)) {
L
Linus Torvalds 已提交
2508
		if (order == 0)
L
Li Hong 已提交
2509
			free_hot_cold_page(page, 0);
L
Linus Torvalds 已提交
2510 2511 2512 2513 2514 2515 2516
		else
			__free_pages_ok(page, order);
	}
}

EXPORT_SYMBOL(__free_pages);

H
Harvey Harrison 已提交
2517
void free_pages(unsigned long addr, unsigned int order)
L
Linus Torvalds 已提交
2518 2519
{
	if (addr != 0) {
N
Nick Piggin 已提交
2520
		VM_BUG_ON(!virt_addr_valid((void *)addr));
L
Linus Torvalds 已提交
2521 2522 2523 2524 2525 2526
		__free_pages(virt_to_page((void *)addr), order);
	}
}

EXPORT_SYMBOL(free_pages);

A
Andi Kleen 已提交
2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541
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;
}

2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560
/**
 * 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
2561
	return make_alloc_exact(addr, order, size);
2562 2563 2564
}
EXPORT_SYMBOL(alloc_pages_exact);

A
Andi Kleen 已提交
2565 2566 2567
/**
 * alloc_pages_exact_nid - allocate an exact number of physically-contiguous
 *			   pages on a node.
2568
 * @nid: the preferred node ID where memory should be allocated
A
Andi Kleen 已提交
2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586
 * @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);

2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605
/**
 * 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
2606 2607
static unsigned int nr_free_zone_pages(int offset)
{
2608
	struct zoneref *z;
2609 2610
	struct zone *zone;

2611
	/* Just pick one node, since fallback list is circular */
L
Linus Torvalds 已提交
2612 2613
	unsigned int sum = 0;

2614
	struct zonelist *zonelist = node_zonelist(numa_node_id(), GFP_KERNEL);
L
Linus Torvalds 已提交
2615

2616
	for_each_zone_zonelist(zone, z, zonelist, offset) {
2617
		unsigned long size = zone->present_pages;
2618
		unsigned long high = high_wmark_pages(zone);
2619 2620
		if (size > high)
			sum += size - high;
L
Linus Torvalds 已提交
2621 2622 2623 2624 2625 2626 2627 2628 2629 2630
	}

	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
2631
	return nr_free_zone_pages(gfp_zone(GFP_USER));
L
Linus Torvalds 已提交
2632
}
2633
EXPORT_SYMBOL_GPL(nr_free_buffer_pages);
L
Linus Torvalds 已提交
2634 2635 2636 2637 2638 2639

/*
 * Amount of free RAM allocatable within all zones
 */
unsigned int nr_free_pagecache_pages(void)
{
M
Mel Gorman 已提交
2640
	return nr_free_zone_pages(gfp_zone(GFP_HIGHUSER_MOVABLE));
L
Linus Torvalds 已提交
2641
}
2642 2643

static inline void show_node(struct zone *zone)
L
Linus Torvalds 已提交
2644
{
2645
	if (NUMA_BUILD)
2646
		printk("Node %d ", zone_to_nid(zone));
L
Linus Torvalds 已提交
2647 2648 2649 2650 2651 2652
}

void si_meminfo(struct sysinfo *val)
{
	val->totalram = totalram_pages;
	val->sharedram = 0;
2653
	val->freeram = global_page_state(NR_FREE_PAGES);
L
Linus Torvalds 已提交
2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667
	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;
2668
	val->freeram = node_page_state(nid, NR_FREE_PAGES);
2669
#ifdef CONFIG_HIGHMEM
L
Linus Torvalds 已提交
2670
	val->totalhigh = pgdat->node_zones[ZONE_HIGHMEM].present_pages;
2671 2672
	val->freehigh = zone_page_state(&pgdat->node_zones[ZONE_HIGHMEM],
			NR_FREE_PAGES);
2673 2674 2675 2676
#else
	val->totalhigh = 0;
	val->freehigh = 0;
#endif
L
Linus Torvalds 已提交
2677 2678 2679 2680
	val->mem_unit = PAGE_SIZE;
}
#endif

2681
/*
2682 2683
 * Determine whether the node should be displayed or not, depending on whether
 * SHOW_MEM_FILTER_NODES was passed to show_free_areas().
2684
 */
2685
bool skip_free_areas_node(unsigned int flags, int nid)
2686 2687
{
	bool ret = false;
2688
	unsigned int cpuset_mems_cookie;
2689 2690 2691 2692

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

2693 2694 2695 2696
	do {
		cpuset_mems_cookie = get_mems_allowed();
		ret = !node_isset(nid, cpuset_current_mems_allowed);
	} while (!put_mems_allowed(cpuset_mems_cookie));
2697 2698 2699 2700
out:
	return ret;
}

L
Linus Torvalds 已提交
2701 2702 2703 2704 2705 2706
#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.
2707 2708
 * Suppresses nodes that are not allowed by current's cpuset if
 * SHOW_MEM_FILTER_NODES is passed.
L
Linus Torvalds 已提交
2709
 */
2710
void show_free_areas(unsigned int filter)
L
Linus Torvalds 已提交
2711
{
2712
	int cpu;
L
Linus Torvalds 已提交
2713 2714
	struct zone *zone;

2715
	for_each_populated_zone(zone) {
2716
		if (skip_free_areas_node(filter, zone_to_nid(zone)))
2717
			continue;
2718 2719
		show_node(zone);
		printk("%s per-cpu:\n", zone->name);
L
Linus Torvalds 已提交
2720

2721
		for_each_online_cpu(cpu) {
L
Linus Torvalds 已提交
2722 2723
			struct per_cpu_pageset *pageset;

2724
			pageset = per_cpu_ptr(zone->pageset, cpu);
L
Linus Torvalds 已提交
2725

2726 2727 2728
			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
2729 2730 2731
		}
	}

K
KOSAKI Motohiro 已提交
2732 2733
	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
2734
		" unevictable:%lu"
2735
		" dirty:%lu writeback:%lu unstable:%lu\n"
2736
		" free:%lu slab_reclaimable:%lu slab_unreclaimable:%lu\n"
2737
		" mapped:%lu shmem:%lu pagetables:%lu bounce:%lu\n",
2738 2739
		global_page_state(NR_ACTIVE_ANON),
		global_page_state(NR_INACTIVE_ANON),
K
KOSAKI Motohiro 已提交
2740 2741
		global_page_state(NR_ISOLATED_ANON),
		global_page_state(NR_ACTIVE_FILE),
2742
		global_page_state(NR_INACTIVE_FILE),
K
KOSAKI Motohiro 已提交
2743
		global_page_state(NR_ISOLATED_FILE),
L
Lee Schermerhorn 已提交
2744
		global_page_state(NR_UNEVICTABLE),
2745
		global_page_state(NR_FILE_DIRTY),
2746
		global_page_state(NR_WRITEBACK),
2747
		global_page_state(NR_UNSTABLE_NFS),
2748
		global_page_state(NR_FREE_PAGES),
2749 2750
		global_page_state(NR_SLAB_RECLAIMABLE),
		global_page_state(NR_SLAB_UNRECLAIMABLE),
2751
		global_page_state(NR_FILE_MAPPED),
2752
		global_page_state(NR_SHMEM),
2753 2754
		global_page_state(NR_PAGETABLE),
		global_page_state(NR_BOUNCE));
L
Linus Torvalds 已提交
2755

2756
	for_each_populated_zone(zone) {
L
Linus Torvalds 已提交
2757 2758
		int i;

2759
		if (skip_free_areas_node(filter, zone_to_nid(zone)))
2760
			continue;
L
Linus Torvalds 已提交
2761 2762 2763 2764 2765 2766
		show_node(zone);
		printk("%s"
			" free:%lukB"
			" min:%lukB"
			" low:%lukB"
			" high:%lukB"
2767 2768 2769 2770
			" active_anon:%lukB"
			" inactive_anon:%lukB"
			" active_file:%lukB"
			" inactive_file:%lukB"
L
Lee Schermerhorn 已提交
2771
			" unevictable:%lukB"
K
KOSAKI Motohiro 已提交
2772 2773
			" isolated(anon):%lukB"
			" isolated(file):%lukB"
L
Linus Torvalds 已提交
2774
			" present:%lukB"
2775 2776 2777 2778
			" mlocked:%lukB"
			" dirty:%lukB"
			" writeback:%lukB"
			" mapped:%lukB"
2779
			" shmem:%lukB"
2780 2781
			" slab_reclaimable:%lukB"
			" slab_unreclaimable:%lukB"
2782
			" kernel_stack:%lukB"
2783 2784 2785 2786
			" pagetables:%lukB"
			" unstable:%lukB"
			" bounce:%lukB"
			" writeback_tmp:%lukB"
L
Linus Torvalds 已提交
2787 2788 2789 2790
			" pages_scanned:%lu"
			" all_unreclaimable? %s"
			"\n",
			zone->name,
2791
			K(zone_page_state(zone, NR_FREE_PAGES)),
2792 2793 2794
			K(min_wmark_pages(zone)),
			K(low_wmark_pages(zone)),
			K(high_wmark_pages(zone)),
2795 2796 2797 2798
			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
2799
			K(zone_page_state(zone, NR_UNEVICTABLE)),
K
KOSAKI Motohiro 已提交
2800 2801
			K(zone_page_state(zone, NR_ISOLATED_ANON)),
			K(zone_page_state(zone, NR_ISOLATED_FILE)),
L
Linus Torvalds 已提交
2802
			K(zone->present_pages),
2803 2804 2805 2806
			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)),
2807
			K(zone_page_state(zone, NR_SHMEM)),
2808 2809
			K(zone_page_state(zone, NR_SLAB_RECLAIMABLE)),
			K(zone_page_state(zone, NR_SLAB_UNRECLAIMABLE)),
2810 2811
			zone_page_state(zone, NR_KERNEL_STACK) *
				THREAD_SIZE / 1024,
2812 2813 2814 2815
			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
2816
			zone->pages_scanned,
2817
			(zone->all_unreclaimable ? "yes" : "no")
L
Linus Torvalds 已提交
2818 2819 2820 2821 2822 2823 2824
			);
		printk("lowmem_reserve[]:");
		for (i = 0; i < MAX_NR_ZONES; i++)
			printk(" %lu", zone->lowmem_reserve[i]);
		printk("\n");
	}

2825
	for_each_populated_zone(zone) {
2826
 		unsigned long nr[MAX_ORDER], flags, order, total = 0;
L
Linus Torvalds 已提交
2827

2828
		if (skip_free_areas_node(filter, zone_to_nid(zone)))
2829
			continue;
L
Linus Torvalds 已提交
2830 2831 2832 2833 2834
		show_node(zone);
		printk("%s: ", zone->name);

		spin_lock_irqsave(&zone->lock, flags);
		for (order = 0; order < MAX_ORDER; order++) {
2835 2836
			nr[order] = zone->free_area[order].nr_free;
			total += nr[order] << order;
L
Linus Torvalds 已提交
2837 2838
		}
		spin_unlock_irqrestore(&zone->lock, flags);
2839 2840
		for (order = 0; order < MAX_ORDER; order++)
			printk("%lu*%lukB ", nr[order], K(1UL) << order);
L
Linus Torvalds 已提交
2841 2842 2843
		printk("= %lukB\n", K(total));
	}

2844 2845
	printk("%ld total pagecache pages\n", global_page_state(NR_FILE_PAGES));

L
Linus Torvalds 已提交
2846 2847 2848
	show_swap_cache_info();
}

2849 2850 2851 2852 2853 2854
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
2855 2856
/*
 * Builds allocation fallback zone lists.
2857 2858
 *
 * Add all populated zones of a node to the zonelist.
L
Linus Torvalds 已提交
2859
 */
2860 2861
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
2862
{
2863 2864
	struct zone *zone;

2865
	BUG_ON(zone_type >= MAX_NR_ZONES);
2866
	zone_type++;
2867 2868

	do {
2869
		zone_type--;
2870
		zone = pgdat->node_zones + zone_type;
2871
		if (populated_zone(zone)) {
2872 2873
			zoneref_set_zone(zone,
				&zonelist->_zonerefs[nr_zones++]);
2874
			check_highest_zone(zone_type);
L
Linus Torvalds 已提交
2875
		}
2876

2877
	} while (zone_type);
2878
	return nr_zones;
L
Linus Torvalds 已提交
2879 2880
}

2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901

/*
 *  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
2902
#ifdef CONFIG_NUMA
2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935
/* 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)
{
2936 2937 2938 2939 2940 2941 2942 2943 2944 2945
	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;
2946 2947 2948 2949 2950 2951 2952
}
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,
2953
		void __user *buffer, size_t *length,
2954 2955 2956 2957
		loff_t *ppos)
{
	char saved_string[NUMA_ZONELIST_ORDER_LEN];
	int ret;
2958
	static DEFINE_MUTEX(zl_order_mutex);
2959

2960
	mutex_lock(&zl_order_mutex);
2961
	if (write)
2962
		strcpy(saved_string, (char*)table->data);
2963
	ret = proc_dostring(table, write, buffer, length, ppos);
2964
	if (ret)
2965
		goto out;
2966 2967 2968 2969 2970 2971 2972 2973 2974
	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;
2975 2976
		} else if (oldval != user_zonelist_order) {
			mutex_lock(&zonelists_mutex);
2977
			build_all_zonelists(NULL);
2978 2979
			mutex_unlock(&zonelists_mutex);
		}
2980
	}
2981 2982 2983
out:
	mutex_unlock(&zl_order_mutex);
	return ret;
2984 2985 2986
}


2987
#define MAX_NODE_LOAD (nr_online_nodes)
2988 2989
static int node_load[MAX_NUMNODES];

L
Linus Torvalds 已提交
2990
/**
2991
 * find_next_best_node - find the next node that should appear in a given node's fallback list
L
Linus Torvalds 已提交
2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003
 * @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.
 */
3004
static int find_next_best_node(int node, nodemask_t *used_node_mask)
L
Linus Torvalds 已提交
3005
{
3006
	int n, val;
L
Linus Torvalds 已提交
3007 3008
	int min_val = INT_MAX;
	int best_node = -1;
3009
	const struct cpumask *tmp = cpumask_of_node(0);
L
Linus Torvalds 已提交
3010

3011 3012 3013 3014 3015
	/* 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
3016

3017
	for_each_node_state(n, N_HIGH_MEMORY) {
L
Linus Torvalds 已提交
3018 3019 3020 3021 3022 3023 3024 3025

		/* 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);

3026 3027 3028
		/* Penalize nodes under us ("prefer the next node") */
		val += (n < node);

L
Linus Torvalds 已提交
3029
		/* Give preference to headless and unused nodes */
3030 3031
		tmp = cpumask_of_node(n);
		if (!cpumask_empty(tmp))
L
Linus Torvalds 已提交
3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049
			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;
}

3050 3051 3052 3053 3054 3055 3056

/*
 * 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
3057
{
3058
	int j;
L
Linus Torvalds 已提交
3059
	struct zonelist *zonelist;
3060

3061
	zonelist = &pgdat->node_zonelists[0];
3062
	for (j = 0; zonelist->_zonerefs[j].zone != NULL; j++)
3063 3064 3065
		;
	j = build_zonelists_node(NODE_DATA(node), zonelist, j,
							MAX_NR_ZONES - 1);
3066 3067
	zonelist->_zonerefs[j].zone = NULL;
	zonelist->_zonerefs[j].zone_idx = 0;
3068 3069
}

3070 3071 3072 3073 3074 3075 3076 3077
/*
 * Build gfp_thisnode zonelists
 */
static void build_thisnode_zonelists(pg_data_t *pgdat)
{
	int j;
	struct zonelist *zonelist;

3078 3079
	zonelist = &pgdat->node_zonelists[1];
	j = build_zonelists_node(pgdat, zonelist, 0, MAX_NR_ZONES - 1);
3080 3081
	zonelist->_zonerefs[j].zone = NULL;
	zonelist->_zonerefs[j].zone_idx = 0;
3082 3083
}

3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098
/*
 * 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;

3099 3100 3101 3102 3103 3104 3105
	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)) {
3106 3107
				zoneref_set_zone(z,
					&zonelist->_zonerefs[pos++]);
3108
				check_highest_zone(zone_type);
3109 3110 3111
			}
		}
	}
3112 3113
	zonelist->_zonerefs[pos].zone = NULL;
	zonelist->_zonerefs[pos].zone_idx = 0;
3114 3115 3116 3117 3118 3119 3120 3121 3122
}

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
3123
         * ZONE_DMA and ZONE_DMA32 can be very small area in the system.
3124 3125
	 * If they are really small and used heavily, the system can fall
	 * into OOM very easily.
3126
	 * This function detect ZONE_DMA/DMA32 size and configures zone order.
3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137
	 */
	/* 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;
3138 3139 3140 3141 3142 3143 3144 3145 3146
			} 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;
3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157
			}
		}
	}
	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.
         */
3158 3159
	average_size = total_size /
				(nodes_weight(node_states[N_HIGH_MEMORY]) + 1);
3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190
	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
3191
	nodemask_t used_mask;
3192 3193 3194
	int local_node, prev_node;
	struct zonelist *zonelist;
	int order = current_zonelist_order;
L
Linus Torvalds 已提交
3195 3196

	/* initialize zonelists */
3197
	for (i = 0; i < MAX_ZONELISTS; i++) {
L
Linus Torvalds 已提交
3198
		zonelist = pgdat->node_zonelists + i;
3199 3200
		zonelist->_zonerefs[0].zone = NULL;
		zonelist->_zonerefs[0].zone_idx = 0;
L
Linus Torvalds 已提交
3201 3202 3203 3204
	}

	/* NUMA-aware ordering of nodes */
	local_node = pgdat->node_id;
3205
	load = nr_online_nodes;
L
Linus Torvalds 已提交
3206 3207
	prev_node = local_node;
	nodes_clear(used_mask);
3208 3209 3210 3211

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

L
Linus Torvalds 已提交
3212
	while ((node = find_next_best_node(local_node, &used_mask)) >= 0) {
3213 3214 3215 3216 3217 3218 3219 3220 3221
		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
3222 3223 3224 3225 3226
		/*
		 * 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.
		 */
3227
		if (distance != node_distance(local_node, prev_node))
3228 3229
			node_load[node] = load;

L
Linus Torvalds 已提交
3230 3231
		prev_node = node;
		load--;
3232 3233 3234 3235 3236
		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
3237

3238 3239 3240
	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
3241
	}
3242 3243

	build_thisnode_zonelists(pgdat);
L
Linus Torvalds 已提交
3244 3245
}

3246
/* Construct the zonelist performance cache - see further mmzone.h */
3247
static void build_zonelist_cache(pg_data_t *pgdat)
3248
{
3249 3250
	struct zonelist *zonelist;
	struct zonelist_cache *zlc;
3251
	struct zoneref *z;
3252

3253 3254 3255
	zonelist = &pgdat->node_zonelists[0];
	zonelist->zlcache_ptr = zlc = &zonelist->zlcache;
	bitmap_zero(zlc->fullzones, MAX_ZONES_PER_ZONELIST);
3256 3257
	for (z = zonelist->_zonerefs; z->zone; z++)
		zlc->z_to_n[z - zonelist->_zonerefs] = zonelist_node_idx(z);
3258 3259
}

3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277
#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
3278

L
Linus Torvalds 已提交
3279 3280
#else	/* CONFIG_NUMA */

3281 3282 3283 3284 3285 3286
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
3287
{
3288
	int node, local_node;
3289 3290
	enum zone_type j;
	struct zonelist *zonelist;
L
Linus Torvalds 已提交
3291 3292 3293

	local_node = pgdat->node_id;

3294 3295
	zonelist = &pgdat->node_zonelists[0];
	j = build_zonelists_node(pgdat, zonelist, 0, MAX_NR_ZONES - 1);
L
Linus Torvalds 已提交
3296

3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309
	/*
	 * 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
3310
	}
3311 3312 3313 3314 3315 3316 3317
	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);
	}

3318 3319
	zonelist->_zonerefs[j].zone = NULL;
	zonelist->_zonerefs[j].zone_idx = 0;
L
Linus Torvalds 已提交
3320 3321
}

3322
/* non-NUMA variant of zonelist performance cache - just NULL zlcache_ptr */
3323
static void build_zonelist_cache(pg_data_t *pgdat)
3324
{
3325
	pgdat->node_zonelists[0].zlcache_ptr = NULL;
3326 3327
}

L
Linus Torvalds 已提交
3328 3329
#endif	/* CONFIG_NUMA */

3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346
/*
 * 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);
3347
static void setup_zone_pageset(struct zone *zone);
3348

3349 3350 3351 3352 3353 3354
/*
 * 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);

3355
/* return values int ....just for stop_machine() */
3356
static __init_refok int __build_all_zonelists(void *data)
L
Linus Torvalds 已提交
3357
{
3358
	int nid;
3359
	int cpu;
3360

3361 3362 3363
#ifdef CONFIG_NUMA
	memset(node_load, 0, sizeof(node_load));
#endif
3364
	for_each_online_node(nid) {
3365 3366 3367 3368
		pg_data_t *pgdat = NODE_DATA(nid);

		build_zonelists(pgdat);
		build_zonelist_cache(pgdat);
3369
	}
3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383

	/*
	 * 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).
	 */
3384
	for_each_possible_cpu(cpu) {
3385 3386
		setup_pageset(&per_cpu(boot_pageset, cpu), 0);

3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400
#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
	}

3401 3402 3403
	return 0;
}

3404 3405 3406 3407
/*
 * Called with zonelists_mutex held always
 * unless system_state == SYSTEM_BOOTING.
 */
3408
void __ref build_all_zonelists(void *data)
3409
{
3410 3411
	set_zonelist_order();

3412
	if (system_state == SYSTEM_BOOTING) {
3413
		__build_all_zonelists(NULL);
3414
		mminit_verify_zonelist();
3415 3416
		cpuset_init_current_mems_allowed();
	} else {
S
Simon Arlott 已提交
3417
		/* we have to stop all cpus to guarantee there is no user
3418
		   of zonelist */
3419 3420 3421 3422 3423
#ifdef CONFIG_MEMORY_HOTPLUG
		if (data)
			setup_zone_pageset((struct zone *)data);
#endif
		stop_machine(__build_all_zonelists, NULL, NULL);
3424 3425
		/* cpuset refresh routine should be here */
	}
3426
	vm_total_pages = nr_free_pagecache_pages();
3427 3428 3429 3430 3431 3432 3433
	/*
	 * 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
	 */
3434
	if (vm_total_pages < (pageblock_nr_pages * MIGRATE_TYPES))
3435 3436 3437 3438 3439 3440
		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",
3441
			nr_online_nodes,
3442
			zonelist_order_name[current_zonelist_order],
3443
			page_group_by_mobility_disabled ? "off" : "on",
3444 3445 3446 3447
			vm_total_pages);
#ifdef CONFIG_NUMA
	printk("Policy zone: %s\n", zone_names[policy_zone]);
#endif
L
Linus Torvalds 已提交
3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462
}

/*
 * 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

3463
#ifndef CONFIG_MEMORY_HOTPLUG
3464
static inline unsigned long wait_table_hash_nr_entries(unsigned long pages)
L
Linus Torvalds 已提交
3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481
{
	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);
}
3482 3483 3484 3485 3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504
#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
3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516 3517

/*
 * 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))

3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531
/*
 * 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;
}

3532
/*
3533
 * Mark a number of pageblocks as MIGRATE_RESERVE. The number
3534 3535
 * 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
3536 3537 3538 3539 3540
 * 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)
{
3541
	unsigned long start_pfn, pfn, end_pfn, block_end_pfn;
3542
	struct page *page;
3543 3544
	unsigned long block_migratetype;
	int reserve;
3545

3546 3547 3548 3549 3550 3551
	/*
	 * 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.
	 */
3552 3553
	start_pfn = zone->zone_start_pfn;
	end_pfn = start_pfn + zone->spanned_pages;
3554
	start_pfn = roundup(start_pfn, pageblock_nr_pages);
3555
	reserve = roundup(min_wmark_pages(zone), pageblock_nr_pages) >>
3556
							pageblock_order;
3557

3558 3559 3560 3561 3562 3563 3564 3565 3566
	/*
	 * 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);

3567
	for (pfn = start_pfn; pfn < end_pfn; pfn += pageblock_nr_pages) {
3568 3569 3570 3571
		if (!pfn_valid(pfn))
			continue;
		page = pfn_to_page(pfn);

3572 3573 3574 3575
		/* Watch out for overlapping nodes */
		if (page_to_nid(page) != zone_to_nid(zone))
			continue;

3576 3577
		block_migratetype = get_pageblock_migratetype(page);

3578 3579 3580 3581 3582 3583 3584 3585 3586
		/* 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;
3587

3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601 3602
			/* 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;
			}
3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614
		}

		/*
		 * 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
3615

L
Linus Torvalds 已提交
3616 3617 3618 3619 3620
/*
 * 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.
 */
3621
void __meminit memmap_init_zone(unsigned long size, int nid, unsigned long zone,
D
Dave Hansen 已提交
3622
		unsigned long start_pfn, enum memmap_context context)
L
Linus Torvalds 已提交
3623 3624
{
	struct page *page;
A
Andy Whitcroft 已提交
3625 3626
	unsigned long end_pfn = start_pfn + size;
	unsigned long pfn;
3627
	struct zone *z;
L
Linus Torvalds 已提交
3628

3629 3630 3631
	if (highest_memmap_pfn < end_pfn - 1)
		highest_memmap_pfn = end_pfn - 1;

3632
	z = &NODE_DATA(nid)->node_zones[zone];
3633
	for (pfn = start_pfn; pfn < end_pfn; pfn++) {
D
Dave Hansen 已提交
3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644
		/*
		 * 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
3645 3646
		page = pfn_to_page(pfn);
		set_page_links(page, zone, nid, pfn);
3647
		mminit_verify_page_links(page, zone, nid, pfn);
3648
		init_page_count(page);
L
Linus Torvalds 已提交
3649 3650
		reset_page_mapcount(page);
		SetPageReserved(page);
3651 3652 3653 3654 3655
		/*
		 * 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
3656 3657 3658
		 * kernel allocations are made. Later some blocks near
		 * the start are marked MIGRATE_RESERVE by
		 * setup_zone_migrate_reserve()
3659 3660 3661 3662 3663
		 *
		 * 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.
3664
		 */
3665 3666 3667
		if ((z->zone_start_pfn <= pfn)
		    && (pfn < z->zone_start_pfn + z->spanned_pages)
		    && !(pfn & (pageblock_nr_pages - 1)))
3668
			set_pageblock_migratetype(page, MIGRATE_MOVABLE);
3669

L
Linus Torvalds 已提交
3670 3671 3672 3673
		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))
3674
			set_page_address(page, __va(pfn << PAGE_SHIFT));
L
Linus Torvalds 已提交
3675 3676 3677 3678
#endif
	}
}

3679
static void __meminit zone_init_free_lists(struct zone *zone)
L
Linus Torvalds 已提交
3680
{
3681 3682 3683
	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
3684 3685 3686 3687 3688 3689
		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
3690
	memmap_init_zone((size), (nid), (zone), (start_pfn), MEMMAP_EARLY)
L
Linus Torvalds 已提交
3691 3692
#endif

3693
static int zone_batchsize(struct zone *zone)
3694
{
3695
#ifdef CONFIG_MMU
3696 3697 3698 3699
	int batch;

	/*
	 * The per-cpu-pages pools are set to around 1000th of the
3700
	 * size of the zone.  But no more than 1/2 of a meg.
3701 3702 3703 3704
	 *
	 * OK, so we don't know how big the cache is.  So guess.
	 */
	batch = zone->present_pages / 1024;
3705 3706
	if (batch * PAGE_SIZE > 512 * 1024)
		batch = (512 * 1024) / PAGE_SIZE;
3707 3708 3709 3710 3711
	batch /= 4;		/* We effectively *= 4 below */
	if (batch < 1)
		batch = 1;

	/*
3712 3713 3714
	 * 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.
3715
	 *
3716 3717 3718 3719
	 * 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.
3720
	 */
3721
	batch = rounddown_pow_of_two(batch + batch/2) - 1;
3722

3723
	return batch;
3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740

#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
3741 3742
}

A
Adrian Bunk 已提交
3743
static void setup_pageset(struct per_cpu_pageset *p, unsigned long batch)
3744 3745
{
	struct per_cpu_pages *pcp;
3746
	int migratetype;
3747

3748 3749
	memset(p, 0, sizeof(*p));

3750
	pcp = &p->pcp;
3751 3752 3753
	pcp->count = 0;
	pcp->high = 6 * batch;
	pcp->batch = max(1UL, 1 * batch);
3754 3755
	for (migratetype = 0; migratetype < MIGRATE_PCPTYPES; migratetype++)
		INIT_LIST_HEAD(&pcp->lists[migratetype]);
3756 3757
}

3758 3759 3760 3761 3762 3763 3764 3765 3766 3767
/*
 * 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;

3768
	pcp = &p->pcp;
3769 3770 3771 3772 3773 3774
	pcp->high = high;
	pcp->batch = max(1UL, high/4);
	if ((high/4) > (PAGE_SHIFT * 8))
		pcp->batch = PAGE_SHIFT * 8;
}

3775
static void setup_zone_pageset(struct zone *zone)
3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792
{
	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));
	}
}

3793
/*
3794 3795
 * Allocate per cpu pagesets and initialize them.
 * Before this call only boot pagesets were available.
3796
 */
3797
void __init setup_per_cpu_pageset(void)
3798
{
3799
	struct zone *zone;
3800

3801 3802
	for_each_populated_zone(zone)
		setup_zone_pageset(zone);
3803 3804
}

S
Sam Ravnborg 已提交
3805
static noinline __init_refok
3806
int zone_wait_table_init(struct zone *zone, unsigned long zone_size_pages)
3807 3808 3809
{
	int i;
	struct pglist_data *pgdat = zone->zone_pgdat;
3810
	size_t alloc_size;
3811 3812 3813 3814 3815

	/*
	 * The per-page waitqueue mechanism uses hashed waitqueues
	 * per zone.
	 */
3816 3817 3818 3819
	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);
3820 3821 3822
	alloc_size = zone->wait_table_hash_nr_entries
					* sizeof(wait_queue_head_t);

3823
	if (!slab_is_available()) {
3824
		zone->wait_table = (wait_queue_head_t *)
3825
			alloc_bootmem_node_nopanic(pgdat, alloc_size);
3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836
	} 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.
		 */
3837
		zone->wait_table = vmalloc(alloc_size);
3838 3839 3840
	}
	if (!zone->wait_table)
		return -ENOMEM;
3841

3842
	for(i = 0; i < zone->wait_table_hash_nr_entries; ++i)
3843
		init_waitqueue_head(zone->wait_table + i);
3844 3845

	return 0;
3846 3847
}

3848 3849 3850 3851 3852 3853
static int __zone_pcp_update(void *data)
{
	struct zone *zone = data;
	int cpu;
	unsigned long batch = zone_batchsize(zone), flags;

3854
	for_each_possible_cpu(cpu) {
3855 3856 3857
		struct per_cpu_pageset *pset;
		struct per_cpu_pages *pcp;

3858
		pset = per_cpu_ptr(zone->pageset, cpu);
3859 3860 3861
		pcp = &pset->pcp;

		local_irq_save(flags);
3862
		free_pcppages_bulk(zone, pcp->count, pcp);
3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873
		setup_pageset(pset, batch);
		local_irq_restore(flags);
	}
	return 0;
}

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

3874
static __meminit void zone_pcp_init(struct zone *zone)
3875
{
3876 3877 3878 3879 3880 3881
	/*
	 * 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;
3882

A
Anton Blanchard 已提交
3883
	if (zone->present_pages)
3884 3885 3886
		printk(KERN_DEBUG "  %s zone: %lu pages, LIFO batch:%u\n",
			zone->name, zone->present_pages,
					 zone_batchsize(zone));
3887 3888
}

3889 3890
__meminit int init_currently_empty_zone(struct zone *zone,
					unsigned long zone_start_pfn,
D
Dave Hansen 已提交
3891 3892
					unsigned long size,
					enum memmap_context context)
3893 3894
{
	struct pglist_data *pgdat = zone->zone_pgdat;
3895 3896 3897 3898
	int ret;
	ret = zone_wait_table_init(zone, size);
	if (ret)
		return ret;
3899 3900 3901 3902
	pgdat->nr_zones = zone_idx(zone) + 1;

	zone->zone_start_pfn = zone_start_pfn;

3903 3904 3905 3906 3907 3908
	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));

3909
	zone_init_free_lists(zone);
3910 3911

	return 0;
3912 3913
}

T
Tejun Heo 已提交
3914
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
3915 3916 3917 3918 3919 3920 3921
#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
 */
3922
int __meminit __early_pfn_to_nid(unsigned long pfn)
3923
{
3924 3925
	unsigned long start_pfn, end_pfn;
	int i, nid;
3926

3927
	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid)
3928
		if (start_pfn <= pfn && pfn < end_pfn)
3929
			return nid;
3930 3931
	/* This is a memory hole */
	return -1;
3932 3933 3934
}
#endif /* CONFIG_HAVE_ARCH_EARLY_PFN_TO_NID */

3935 3936
int __meminit early_pfn_to_nid(unsigned long pfn)
{
3937 3938 3939 3940 3941 3942 3943
	int nid;

	nid = __early_pfn_to_nid(pfn);
	if (nid >= 0)
		return nid;
	/* just returns 0 */
	return 0;
3944 3945
}

3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956
#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
3957

3958 3959
/**
 * free_bootmem_with_active_regions - Call free_bootmem_node for each active range
3960 3961
 * @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
3962 3963 3964 3965 3966
 *
 * 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.
 */
3967
void __init free_bootmem_with_active_regions(int nid, unsigned long max_low_pfn)
3968
{
3969 3970
	unsigned long start_pfn, end_pfn;
	int i, this_nid;
3971

3972 3973 3974
	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);
3975

3976 3977 3978 3979
		if (start_pfn < end_pfn)
			free_bootmem_node(NODE_DATA(this_nid),
					  PFN_PHYS(start_pfn),
					  (end_pfn - start_pfn) << PAGE_SHIFT);
3980 3981 3982
	}
}

3983 3984
/**
 * sparse_memory_present_with_active_regions - Call memory_present for each active range
3985
 * @nid: The node to call memory_present for. If MAX_NUMNODES, all nodes will be used.
3986 3987 3988
 *
 * If an architecture guarantees that all ranges registered with
 * add_active_ranges() contain no holes and may be freed, this
3989
 * function may be used instead of calling memory_present() manually.
3990 3991 3992
 */
void __init sparse_memory_present_with_active_regions(int nid)
{
3993 3994
	unsigned long start_pfn, end_pfn;
	int i, this_nid;
3995

3996 3997
	for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, &this_nid)
		memory_present(this_nid, start_pfn, end_pfn);
3998 3999 4000 4001
}

/**
 * get_pfn_range_for_nid - Return the start and end page frames for a node
4002 4003 4004
 * @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.
4005 4006 4007 4008
 *
 * 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
4009
 * PFNs will be 0.
4010
 */
4011
void __meminit get_pfn_range_for_nid(unsigned int nid,
4012 4013
			unsigned long *start_pfn, unsigned long *end_pfn)
{
4014
	unsigned long this_start_pfn, this_end_pfn;
4015
	int i;
4016

4017 4018 4019
	*start_pfn = -1UL;
	*end_pfn = 0;

4020 4021 4022
	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);
4023 4024
	}

4025
	if (*start_pfn == -1UL)
4026 4027 4028
		*start_pfn = 0;
}

M
Mel Gorman 已提交
4029 4030 4031 4032 4033
/*
 * 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
4034
static void __init find_usable_zone_for_movable(void)
M
Mel Gorman 已提交
4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051
{
	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
4052
 * because it is sized independent of architecture. Unlike the other zones,
M
Mel Gorman 已提交
4053 4054 4055 4056 4057 4058 4059
 * 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
4060
static void __meminit adjust_zone_range_for_zone_movable(int nid,
M
Mel Gorman 已提交
4061 4062 4063 4064 4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079 4080 4081 4082 4083 4084 4085
					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;
	}
}

4086 4087 4088 4089
/*
 * 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
4090
static unsigned long __meminit zone_spanned_pages_in_node(int nid,
4091 4092 4093 4094 4095 4096 4097 4098 4099 4100
					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
4101 4102 4103
	adjust_zone_range_for_zone_movable(nid, zone_type,
				node_start_pfn, node_end_pfn,
				&zone_start_pfn, &zone_end_pfn);
4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118

	/* 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,
4119
 * then all holes in the requested range will be accounted for.
4120
 */
4121
unsigned long __meminit __absent_pages_in_range(int nid,
4122 4123 4124
				unsigned long range_start_pfn,
				unsigned long range_end_pfn)
{
4125 4126 4127
	unsigned long nr_absent = range_end_pfn - range_start_pfn;
	unsigned long start_pfn, end_pfn;
	int i;
4128

4129 4130 4131 4132
	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;
4133
	}
4134
	return nr_absent;
4135 4136 4137 4138 4139 4140 4141
}

/**
 * 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
 *
4142
 * It returns the number of pages frames in memory holes within a range.
4143 4144 4145 4146 4147 4148 4149 4150
 */
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
4151
static unsigned long __meminit zone_absent_pages_in_node(int nid,
4152 4153 4154
					unsigned long zone_type,
					unsigned long *ignored)
{
4155 4156
	unsigned long zone_low = arch_zone_lowest_possible_pfn[zone_type];
	unsigned long zone_high = arch_zone_highest_possible_pfn[zone_type];
4157 4158 4159 4160
	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);
4161 4162
	zone_start_pfn = clamp(node_start_pfn, zone_low, zone_high);
	zone_end_pfn = clamp(node_end_pfn, zone_low, zone_high);
4163

M
Mel Gorman 已提交
4164 4165 4166
	adjust_zone_range_for_zone_movable(nid, zone_type,
			node_start_pfn, node_end_pfn,
			&zone_start_pfn, &zone_end_pfn);
4167
	return __absent_pages_in_range(nid, zone_start_pfn, zone_end_pfn);
4168
}
4169

T
Tejun Heo 已提交
4170
#else /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
P
Paul Mundt 已提交
4171
static inline unsigned long __meminit zone_spanned_pages_in_node(int nid,
4172 4173 4174 4175 4176 4177
					unsigned long zone_type,
					unsigned long *zones_size)
{
	return zones_size[zone_type];
}

P
Paul Mundt 已提交
4178
static inline unsigned long __meminit zone_absent_pages_in_node(int nid,
4179 4180 4181 4182 4183 4184 4185 4186
						unsigned long zone_type,
						unsigned long *zholes_size)
{
	if (!zholes_size)
		return 0;

	return zholes_size[zone_type];
}
4187

T
Tejun Heo 已提交
4188
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
4189

4190
static void __meminit calculate_node_totalpages(struct pglist_data *pgdat,
4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210
		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);
}

4211 4212 4213
#ifndef CONFIG_SPARSEMEM
/*
 * Calculate the size of the zone->blockflags rounded to an unsigned long
4214 4215
 * 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
4216 4217 4218 4219 4220 4221 4222
 * 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;

4223 4224
	usemapsize = roundup(zonesize, pageblock_nr_pages);
	usemapsize = usemapsize >> pageblock_order;
4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235
	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;
4236
	if (usemapsize)
4237 4238
		zone->pageblock_flags = alloc_bootmem_node_nopanic(pgdat,
								   usemapsize);
4239 4240
}
#else
4241
static inline void setup_usemap(struct pglist_data *pgdat,
4242 4243 4244
				struct zone *zone, unsigned long zonesize) {}
#endif /* CONFIG_SPARSEMEM */

4245
#ifdef CONFIG_HUGETLB_PAGE_SIZE_VARIABLE
4246 4247 4248 4249 4250 4251 4252 4253 4254 4255

/* Return a sensible default order for the pageblock size. */
static inline int pageblock_default_order(void)
{
	if (HPAGE_SHIFT > PAGE_SHIFT)
		return HUGETLB_PAGE_ORDER;

	return MAX_ORDER-1;
}

4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270
/* Initialise the number of pages represented by NR_PAGEBLOCK_BITS */
static inline void __init set_pageblock_order(unsigned int order)
{
	/* Check that pageblock_nr_pages has not already been setup */
	if (pageblock_order)
		return;

	/*
	 * Assume the largest contiguous order of interest is a huge page.
	 * This value may be variable depending on boot parameters on IA64
	 */
	pageblock_order = order;
}
#else /* CONFIG_HUGETLB_PAGE_SIZE_VARIABLE */

4271 4272 4273 4274 4275 4276 4277 4278 4279 4280
/*
 * When CONFIG_HUGETLB_PAGE_SIZE_VARIABLE is not set, set_pageblock_order()
 * and pageblock_default_order() are 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
 */
static inline int pageblock_default_order(unsigned int order)
{
	return MAX_ORDER-1;
}
4281 4282 4283 4284
#define set_pageblock_order(x)	do {} while (0)

#endif /* CONFIG_HUGETLB_PAGE_SIZE_VARIABLE */

L
Linus Torvalds 已提交
4285 4286 4287 4288 4289 4290
/*
 * Set up the zone data structures:
 *   - mark all pages reserved
 *   - mark all memory queues empty
 *   - clear the memory bitmaps
 */
4291
static void __paginginit free_area_init_core(struct pglist_data *pgdat,
L
Linus Torvalds 已提交
4292 4293
		unsigned long *zones_size, unsigned long *zholes_size)
{
4294
	enum zone_type j;
4295
	int nid = pgdat->node_id;
L
Linus Torvalds 已提交
4296
	unsigned long zone_start_pfn = pgdat->node_start_pfn;
4297
	int ret;
L
Linus Torvalds 已提交
4298

4299
	pgdat_resize_init(pgdat);
L
Linus Torvalds 已提交
4300 4301 4302
	pgdat->nr_zones = 0;
	init_waitqueue_head(&pgdat->kswapd_wait);
	pgdat->kswapd_max_order = 0;
4303
	pgdat_page_cgroup_init(pgdat);
L
Linus Torvalds 已提交
4304 4305 4306
	
	for (j = 0; j < MAX_NR_ZONES; j++) {
		struct zone *zone = pgdat->node_zones + j;
4307
		unsigned long size, realsize, memmap_pages;
H
Hugh Dickins 已提交
4308
		enum lru_list lru;
L
Linus Torvalds 已提交
4309

4310 4311 4312
		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
4313

4314 4315 4316 4317 4318
		/*
		 * 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
		 */
4319 4320
		memmap_pages =
			PAGE_ALIGN(size * sizeof(struct page)) >> PAGE_SHIFT;
4321 4322
		if (realsize >= memmap_pages) {
			realsize -= memmap_pages;
4323 4324 4325 4326
			if (memmap_pages)
				printk(KERN_DEBUG
				       "  %s zone: %lu pages used for memmap\n",
				       zone_names[j], memmap_pages);
4327 4328 4329 4330 4331
		} else
			printk(KERN_WARNING
				"  %s zone: %lu pages exceeds realsize %lu\n",
				zone_names[j], memmap_pages, realsize);

4332 4333
		/* Account for reserved pages */
		if (j == 0 && realsize > dma_reserve) {
4334
			realsize -= dma_reserve;
Y
Yinghai Lu 已提交
4335
			printk(KERN_DEBUG "  %s zone: %lu pages reserved\n",
4336
					zone_names[0], dma_reserve);
4337 4338
		}

4339
		if (!is_highmem_idx(j))
L
Linus Torvalds 已提交
4340 4341 4342 4343 4344
			nr_kernel_pages += realsize;
		nr_all_pages += realsize;

		zone->spanned_pages = size;
		zone->present_pages = realsize;
4345
#ifdef CONFIG_NUMA
4346
		zone->node = nid;
4347
		zone->min_unmapped_pages = (realsize*sysctl_min_unmapped_ratio)
4348
						/ 100;
4349
		zone->min_slab_pages = (realsize * sysctl_min_slab_ratio) / 100;
4350
#endif
L
Linus Torvalds 已提交
4351 4352 4353
		zone->name = zone_names[j];
		spin_lock_init(&zone->lock);
		spin_lock_init(&zone->lru_lock);
4354
		zone_seqlock_init(zone);
L
Linus Torvalds 已提交
4355 4356
		zone->zone_pgdat = pgdat;

4357
		zone_pcp_init(zone);
H
Hugh Dickins 已提交
4358 4359
		for_each_lru(lru)
			INIT_LIST_HEAD(&zone->lruvec.lists[lru]);
4360 4361 4362 4363
		zone->reclaim_stat.recent_rotated[0] = 0;
		zone->reclaim_stat.recent_rotated[1] = 0;
		zone->reclaim_stat.recent_scanned[0] = 0;
		zone->reclaim_stat.recent_scanned[1] = 0;
4364
		zap_zone_vm_stats(zone);
4365
		zone->flags = 0;
L
Linus Torvalds 已提交
4366 4367 4368
		if (!size)
			continue;

4369
		set_pageblock_order(pageblock_default_order());
4370
		setup_usemap(pgdat, zone, size);
D
Dave Hansen 已提交
4371 4372
		ret = init_currently_empty_zone(zone, zone_start_pfn,
						size, MEMMAP_EARLY);
4373
		BUG_ON(ret);
4374
		memmap_init(size, nid, j, zone_start_pfn);
L
Linus Torvalds 已提交
4375 4376 4377 4378
		zone_start_pfn += size;
	}
}

S
Sam Ravnborg 已提交
4379
static void __init_refok alloc_node_mem_map(struct pglist_data *pgdat)
L
Linus Torvalds 已提交
4380 4381 4382 4383 4384
{
	/* Skip empty nodes */
	if (!pgdat->node_spanned_pages)
		return;

A
Andy Whitcroft 已提交
4385
#ifdef CONFIG_FLAT_NODE_MEM_MAP
L
Linus Torvalds 已提交
4386 4387
	/* ia64 gets its own node_mem_map, before this, without bootmem */
	if (!pgdat->node_mem_map) {
4388
		unsigned long size, start, end;
A
Andy Whitcroft 已提交
4389 4390
		struct page *map;

4391 4392 4393 4394 4395 4396 4397 4398 4399
		/*
		 * 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);
4400 4401
		map = alloc_remap(pgdat->node_id, size);
		if (!map)
4402
			map = alloc_bootmem_node_nopanic(pgdat, size);
4403
		pgdat->node_mem_map = map + (pgdat->node_start_pfn - start);
L
Linus Torvalds 已提交
4404
	}
4405
#ifndef CONFIG_NEED_MULTIPLE_NODES
L
Linus Torvalds 已提交
4406 4407 4408
	/*
	 * With no DISCONTIG, the global mem_map is just set as node 0's
	 */
4409
	if (pgdat == NODE_DATA(0)) {
L
Linus Torvalds 已提交
4410
		mem_map = NODE_DATA(0)->node_mem_map;
T
Tejun Heo 已提交
4411
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
4412
		if (page_to_pfn(mem_map) != pgdat->node_start_pfn)
4413
			mem_map -= (pgdat->node_start_pfn - ARCH_PFN_OFFSET);
T
Tejun Heo 已提交
4414
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
4415
	}
L
Linus Torvalds 已提交
4416
#endif
A
Andy Whitcroft 已提交
4417
#endif /* CONFIG_FLAT_NODE_MEM_MAP */
L
Linus Torvalds 已提交
4418 4419
}

4420 4421
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
4422
{
4423 4424
	pg_data_t *pgdat = NODE_DATA(nid);

L
Linus Torvalds 已提交
4425 4426
	pgdat->node_id = nid;
	pgdat->node_start_pfn = node_start_pfn;
4427
	calculate_node_totalpages(pgdat, zones_size, zholes_size);
L
Linus Torvalds 已提交
4428 4429

	alloc_node_mem_map(pgdat);
4430 4431 4432 4433 4434
#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
4435 4436 4437 4438

	free_area_init_core(pgdat, zones_size, zholes_size);
}

T
Tejun Heo 已提交
4439
#ifdef CONFIG_HAVE_MEMBLOCK_NODE_MAP
M
Miklos Szeredi 已提交
4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459

#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

4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481
/**
 * 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;
4482
	unsigned long start, end, mask;
4483
	int last_nid = -1;
4484
	int i, nid;
4485

4486
	for_each_mem_pfn_range(i, MAX_NUMNODES, &start, &end, &nid) {
4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509
		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;
}

4510
/* Find the lowest pfn for a node */
A
Adrian Bunk 已提交
4511
static unsigned long __init find_min_pfn_for_node(int nid)
4512
{
4513
	unsigned long min_pfn = ULONG_MAX;
4514 4515
	unsigned long start_pfn;
	int i;
4516

4517 4518
	for_each_mem_pfn_range(i, nid, &start_pfn, NULL, NULL)
		min_pfn = min(min_pfn, start_pfn);
4519

4520 4521
	if (min_pfn == ULONG_MAX) {
		printk(KERN_WARNING
4522
			"Could not find start_pfn for node %d\n", nid);
4523 4524 4525 4526
		return 0;
	}

	return min_pfn;
4527 4528 4529 4530 4531 4532
}

/**
 * find_min_pfn_with_active_regions - Find the minimum PFN registered
 *
 * It returns the minimum PFN based on information provided via
4533
 * add_active_range().
4534 4535 4536 4537 4538 4539
 */
unsigned long __init find_min_pfn_with_active_regions(void)
{
	return find_min_pfn_for_node(MAX_NUMNODES);
}

4540 4541 4542 4543 4544
/*
 * 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
4545
static unsigned long __init early_calculate_totalpages(void)
4546 4547
{
	unsigned long totalpages = 0;
4548 4549 4550 4551 4552
	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;
4553

4554 4555
		totalpages += pages;
		if (pages)
4556
			node_set_state(nid, N_HIGH_MEMORY);
4557 4558
	}
  	return totalpages;
4559 4560
}

M
Mel Gorman 已提交
4561 4562 4563 4564 4565 4566
/*
 * 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
 */
4567
static void __init find_zone_movable_pfns_for_nodes(void)
M
Mel Gorman 已提交
4568 4569 4570 4571
{
	int i, nid;
	unsigned long usable_startpfn;
	unsigned long kernelcore_node, kernelcore_remaining;
4572 4573
	/* save the state before borrow the nodemask */
	nodemask_t saved_node_state = node_states[N_HIGH_MEMORY];
4574 4575
	unsigned long totalpages = early_calculate_totalpages();
	int usable_nodes = nodes_weight(node_states[N_HIGH_MEMORY]);
M
Mel Gorman 已提交
4576

4577 4578 4579 4580 4581 4582 4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598
	/*
	 * 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
4599 4600
	/* If kernelcore was not specified, there is no ZONE_MOVABLE */
	if (!required_kernelcore)
4601
		goto out;
M
Mel Gorman 已提交
4602 4603 4604 4605 4606 4607 4608 4609

	/* 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;
4610
	for_each_node_state(nid, N_HIGH_MEMORY) {
4611 4612
		unsigned long start_pfn, end_pfn;

M
Mel Gorman 已提交
4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628
		/*
		 * 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 */
4629
		for_each_mem_pfn_range(i, nid, &start_pfn, &end_pfn, NULL) {
M
Mel Gorman 已提交
4630 4631
			unsigned long size_pages;

4632
			start_pfn = max(start_pfn, zone_movable_pfn[nid]);
M
Mel Gorman 已提交
4633 4634 4635 4636 4637 4638 4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698
			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);
4699 4700 4701 4702

out:
	/* restore the node_state */
	node_states[N_HIGH_MEMORY] = saved_node_state;
M
Mel Gorman 已提交
4703 4704
}

4705 4706 4707 4708 4709 4710 4711 4712
/* Any regular memory on that node ? */
static void check_for_regular_memory(pg_data_t *pgdat)
{
#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];
4713
		if (zone->present_pages) {
4714
			node_set_state(zone_to_nid(zone), N_NORMAL_MEMORY);
4715 4716
			break;
		}
4717 4718 4719 4720
	}
#endif
}

4721 4722
/**
 * free_area_init_nodes - Initialise all pg_data_t and zone data
4723
 * @max_zone_pfn: an array of max PFNs for each zone
4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735
 *
 * 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)
{
4736 4737
	unsigned long start_pfn, end_pfn;
	int i, nid;
4738

4739 4740 4741 4742 4743 4744 4745 4746
	/* 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
4747 4748
		if (i == ZONE_MOVABLE)
			continue;
4749 4750 4751 4752 4753
		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
4754 4755 4756 4757 4758
	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));
4759
	find_zone_movable_pfns_for_nodes();
4760 4761 4762

	/* Print out the zone ranges */
	printk("Zone PFN ranges:\n");
M
Mel Gorman 已提交
4763 4764 4765
	for (i = 0; i < MAX_NR_ZONES; i++) {
		if (i == ZONE_MOVABLE)
			continue;
4766 4767 4768 4769 4770 4771
		printk("  %-8s ", zone_names[i]);
		if (arch_zone_lowest_possible_pfn[i] ==
				arch_zone_highest_possible_pfn[i])
			printk("empty\n");
		else
			printk("%0#10lx -> %0#10lx\n",
4772 4773
				arch_zone_lowest_possible_pfn[i],
				arch_zone_highest_possible_pfn[i]);
M
Mel Gorman 已提交
4774 4775 4776 4777 4778 4779 4780 4781
	}

	/* Print out the PFNs ZONE_MOVABLE begins at in each node */
	printk("Movable zone start PFN for each node\n");
	for (i = 0; i < MAX_NUMNODES; i++) {
		if (zone_movable_pfn[i])
			printk("  Node %d: %lu\n", i, zone_movable_pfn[i]);
	}
4782 4783

	/* Print out the early_node_map[] */
4784 4785 4786
	printk("Early memory PFN ranges\n");
	for_each_mem_pfn_range(i, MAX_NUMNODES, &start_pfn, &end_pfn, &nid)
		printk("  %3d: %0#10lx -> %0#10lx\n", nid, start_pfn, end_pfn);
4787 4788

	/* Initialise every node */
4789
	mminit_verify_pageflags_layout();
4790
	setup_nr_node_ids();
4791 4792
	for_each_online_node(nid) {
		pg_data_t *pgdat = NODE_DATA(nid);
4793
		free_area_init_node(nid, NULL,
4794
				find_min_pfn_for_node(nid), NULL);
4795 4796 4797 4798 4799

		/* Any memory on that node */
		if (pgdat->node_present_pages)
			node_set_state(nid, N_HIGH_MEMORY);
		check_for_regular_memory(pgdat);
4800 4801
	}
}
M
Mel Gorman 已提交
4802

4803
static int __init cmdline_parse_core(char *p, unsigned long *core)
M
Mel Gorman 已提交
4804 4805 4806 4807 4808 4809
{
	unsigned long long coremem;
	if (!p)
		return -EINVAL;

	coremem = memparse(p, &p);
4810
	*core = coremem >> PAGE_SHIFT;
M
Mel Gorman 已提交
4811

4812
	/* Paranoid check that UL is enough for the coremem value */
M
Mel Gorman 已提交
4813 4814 4815 4816
	WARN_ON((coremem >> PAGE_SHIFT) > ULONG_MAX);

	return 0;
}
M
Mel Gorman 已提交
4817

4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835
/*
 * 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
4836
early_param("kernelcore", cmdline_parse_kernelcore);
4837
early_param("movablecore", cmdline_parse_movablecore);
M
Mel Gorman 已提交
4838

T
Tejun Heo 已提交
4839
#endif /* CONFIG_HAVE_MEMBLOCK_NODE_MAP */
4840

4841
/**
4842 4843
 * set_dma_reserve - set the specified number of pages reserved in the first zone
 * @new_dma_reserve: The number of pages to mark reserved
4844 4845 4846 4847
 *
 * 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
4848 4849 4850
 * 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.
4851 4852 4853 4854 4855 4856
 */
void __init set_dma_reserve(unsigned long new_dma_reserve)
{
	dma_reserve = new_dma_reserve;
}

L
Linus Torvalds 已提交
4857 4858
void __init free_area_init(unsigned long *zones_size)
{
4859
	free_area_init_node(0, zones_size,
L
Linus Torvalds 已提交
4860 4861 4862 4863 4864 4865 4866 4867
			__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;

4868
	if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) {
4869
		lru_add_drain_cpu(cpu);
4870 4871 4872 4873 4874 4875 4876 4877
		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.
		 */
4878
		vm_events_fold_cpu(cpu);
4879 4880 4881 4882 4883 4884 4885 4886

		/*
		 * 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.
		 */
4887
		refresh_cpu_vm_stats(cpu);
L
Linus Torvalds 已提交
4888 4889 4890 4891 4892 4893 4894 4895 4896
	}
	return NOTIFY_OK;
}

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

4897 4898 4899 4900 4901 4902 4903 4904
/*
 * 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;
4905
	enum zone_type i, j;
4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916 4917

	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];
			}

4918 4919
			/* we treat the high watermark as reserved pages. */
			max += high_wmark_pages(zone);
4920 4921 4922 4923

			if (max > zone->present_pages)
				max = zone->present_pages;
			reserve_pages += max;
4924 4925 4926 4927 4928 4929 4930 4931 4932 4933
			/*
			 * 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;
4934 4935
		}
	}
4936
	dirty_balance_reserve = reserve_pages;
4937 4938 4939
	totalreserve_pages = reserve_pages;
}

L
Linus Torvalds 已提交
4940 4941 4942 4943 4944 4945 4946 4947 4948
/*
 * 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;
4949
	enum zone_type j, idx;
L
Linus Torvalds 已提交
4950

4951
	for_each_online_pgdat(pgdat) {
L
Linus Torvalds 已提交
4952 4953 4954 4955 4956 4957
		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;

4958 4959
			idx = j;
			while (idx) {
L
Linus Torvalds 已提交
4960 4961
				struct zone *lower_zone;

4962 4963
				idx--;

L
Linus Torvalds 已提交
4964 4965 4966 4967 4968 4969 4970 4971 4972 4973
				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;
			}
		}
	}
4974 4975 4976

	/* update totalreserve_pages */
	calculate_totalreserve_pages();
L
Linus Torvalds 已提交
4977 4978
}

4979
/**
4980
 * setup_per_zone_wmarks - called when min_free_kbytes changes
4981
 * or when memory is hot-{added|removed}
4982
 *
4983 4984
 * Ensures that the watermark[min,low,high] values for each zone are set
 * correctly with respect to min_free_kbytes.
L
Linus Torvalds 已提交
4985
 */
4986
void setup_per_zone_wmarks(void)
L
Linus Torvalds 已提交
4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999
{
	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) {
5000 5001
		u64 tmp;

5002
		spin_lock_irqsave(&zone->lock, flags);
5003 5004
		tmp = (u64)pages_min * zone->present_pages;
		do_div(tmp, lowmem_pages);
L
Linus Torvalds 已提交
5005 5006
		if (is_highmem(zone)) {
			/*
N
Nick Piggin 已提交
5007 5008 5009 5010
			 * __GFP_HIGH and PF_MEMALLOC allocations usually don't
			 * need highmem pages, so cap pages_min to a small
			 * value here.
			 *
5011
			 * The WMARK_HIGH-WMARK_LOW and (WMARK_LOW-WMARK_MIN)
N
Nick Piggin 已提交
5012 5013
			 * deltas controls asynch page reclaim, and so should
			 * not be capped for highmem.
L
Linus Torvalds 已提交
5014 5015 5016 5017 5018 5019 5020 5021
			 */
			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;
5022
			zone->watermark[WMARK_MIN] = min_pages;
L
Linus Torvalds 已提交
5023
		} else {
N
Nick Piggin 已提交
5024 5025
			/*
			 * If it's a lowmem zone, reserve a number of pages
L
Linus Torvalds 已提交
5026 5027
			 * proportionate to the zone's size.
			 */
5028
			zone->watermark[WMARK_MIN] = tmp;
L
Linus Torvalds 已提交
5029 5030
		}

5031 5032
		zone->watermark[WMARK_LOW]  = min_wmark_pages(zone) + (tmp >> 2);
		zone->watermark[WMARK_HIGH] = min_wmark_pages(zone) + (tmp >> 1);
5033
		setup_zone_migrate_reserve(zone);
5034
		spin_unlock_irqrestore(&zone->lock, flags);
L
Linus Torvalds 已提交
5035
	}
5036 5037 5038

	/* update totalreserve_pages */
	calculate_totalreserve_pages();
L
Linus Torvalds 已提交
5039 5040
}

5041
/*
5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061
 * 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
 */
5062
static void __meminit calculate_zone_inactive_ratio(struct zone *zone)
5063
{
5064
	unsigned int gb, ratio;
5065

5066 5067 5068
	/* Zone size in gigabytes */
	gb = zone->present_pages >> (30 - PAGE_SHIFT);
	if (gb)
5069
		ratio = int_sqrt(10 * gb);
5070 5071
	else
		ratio = 1;
5072

5073 5074
	zone->inactive_ratio = ratio;
}
5075

5076
static void __meminit setup_per_zone_inactive_ratio(void)
5077 5078 5079 5080 5081
{
	struct zone *zone;

	for_each_zone(zone)
		calculate_zone_inactive_ratio(zone);
5082 5083
}

L
Linus Torvalds 已提交
5084 5085 5086 5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107
/*
 * 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
 */
5108
int __meminit init_per_zone_wmark_min(void)
L
Linus Torvalds 已提交
5109 5110 5111 5112 5113 5114 5115 5116 5117 5118
{
	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;
5119
	setup_per_zone_wmarks();
5120
	refresh_zone_stat_thresholds();
L
Linus Torvalds 已提交
5121
	setup_per_zone_lowmem_reserve();
5122
	setup_per_zone_inactive_ratio();
L
Linus Torvalds 已提交
5123 5124
	return 0;
}
5125
module_init(init_per_zone_wmark_min)
L
Linus Torvalds 已提交
5126 5127 5128 5129 5130 5131 5132

/*
 * 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, 
5133
	void __user *buffer, size_t *length, loff_t *ppos)
L
Linus Torvalds 已提交
5134
{
5135
	proc_dointvec(table, write, buffer, length, ppos);
5136
	if (write)
5137
		setup_per_zone_wmarks();
L
Linus Torvalds 已提交
5138 5139 5140
	return 0;
}

5141 5142
#ifdef CONFIG_NUMA
int sysctl_min_unmapped_ratio_sysctl_handler(ctl_table *table, int write,
5143
	void __user *buffer, size_t *length, loff_t *ppos)
5144 5145 5146 5147
{
	struct zone *zone;
	int rc;

5148
	rc = proc_dointvec_minmax(table, write, buffer, length, ppos);
5149 5150 5151 5152
	if (rc)
		return rc;

	for_each_zone(zone)
5153
		zone->min_unmapped_pages = (zone->present_pages *
5154 5155 5156
				sysctl_min_unmapped_ratio) / 100;
	return 0;
}
5157 5158

int sysctl_min_slab_ratio_sysctl_handler(ctl_table *table, int write,
5159
	void __user *buffer, size_t *length, loff_t *ppos)
5160 5161 5162 5163
{
	struct zone *zone;
	int rc;

5164
	rc = proc_dointvec_minmax(table, write, buffer, length, ppos);
5165 5166 5167 5168 5169 5170 5171 5172
	if (rc)
		return rc;

	for_each_zone(zone)
		zone->min_slab_pages = (zone->present_pages *
				sysctl_min_slab_ratio) / 100;
	return 0;
}
5173 5174
#endif

L
Linus Torvalds 已提交
5175 5176 5177 5178 5179 5180
/*
 * 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
5181
 * minimum watermarks. The lowmem reserve ratio can only make sense
L
Linus Torvalds 已提交
5182 5183 5184
 * if in function of the boot time zone sizes.
 */
int lowmem_reserve_ratio_sysctl_handler(ctl_table *table, int write,
5185
	void __user *buffer, size_t *length, loff_t *ppos)
L
Linus Torvalds 已提交
5186
{
5187
	proc_dointvec_minmax(table, write, buffer, length, ppos);
L
Linus Torvalds 已提交
5188 5189 5190 5191
	setup_per_zone_lowmem_reserve();
	return 0;
}

5192 5193 5194 5195 5196 5197 5198
/*
 * 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,
5199
	void __user *buffer, size_t *length, loff_t *ppos)
5200 5201 5202 5203 5204
{
	struct zone *zone;
	unsigned int cpu;
	int ret;

5205
	ret = proc_dointvec_minmax(table, write, buffer, length, ppos);
5206 5207
	if (!write || (ret == -EINVAL))
		return ret;
5208
	for_each_populated_zone(zone) {
5209
		for_each_possible_cpu(cpu) {
5210 5211
			unsigned long  high;
			high = zone->present_pages / percpu_pagelist_fraction;
5212 5213
			setup_pagelist_highmark(
				per_cpu_ptr(zone->pageset, cpu), high);
5214 5215 5216 5217 5218
		}
	}
	return 0;
}

5219
int hashdist = HASHDIST_DEFAULT;
L
Linus Torvalds 已提交
5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253

#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,
				     unsigned long limit)
{
	unsigned long long max = limit;
	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
5254
		numentries = nr_kernel_pages;
L
Linus Torvalds 已提交
5255 5256 5257 5258 5259 5260 5261 5262 5263
		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);
5264 5265

		/* Make sure we've got at least a 0-order allocation.. */
5266 5267 5268 5269 5270 5271 5272 5273
		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))
5274
			numentries = PAGE_SIZE / bucketsize;
L
Linus Torvalds 已提交
5275
	}
5276
	numentries = roundup_pow_of_two(numentries);
L
Linus Torvalds 已提交
5277 5278 5279 5280 5281 5282

	/* 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);
	}
5283
	max = min(max, 0x80000000ULL);
L
Linus Torvalds 已提交
5284 5285 5286 5287

	if (numentries > max)
		numentries = max;

5288
	log2qty = ilog2(numentries);
L
Linus Torvalds 已提交
5289 5290 5291 5292

	do {
		size = bucketsize << log2qty;
		if (flags & HASH_EARLY)
5293
			table = alloc_bootmem_nopanic(size);
L
Linus Torvalds 已提交
5294 5295 5296
		else if (hashdist)
			table = __vmalloc(size, GFP_ATOMIC, PAGE_KERNEL);
		else {
5297 5298
			/*
			 * If bucketsize is not a power-of-two, we may free
5299 5300
			 * some pages at the end of hash table which
			 * alloc_pages_exact() automatically does
5301
			 */
5302
			if (get_order(size) < MAX_ORDER) {
5303
				table = alloc_pages_exact(size, GFP_ATOMIC);
5304 5305
				kmemleak_alloc(table, size, 1, GFP_ATOMIC);
			}
L
Linus Torvalds 已提交
5306 5307 5308 5309 5310 5311
		}
	} while (!table && size > PAGE_SIZE && --log2qty);

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

5312
	printk(KERN_INFO "%s hash table entries: %ld (order: %d, %lu bytes)\n",
L
Linus Torvalds 已提交
5313
	       tablename,
5314
	       (1UL << log2qty),
5315
	       ilog2(size) - PAGE_SHIFT,
L
Linus Torvalds 已提交
5316 5317 5318 5319 5320 5321 5322 5323 5324
	       size);

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

	return table;
}
5325

5326 5327 5328 5329 5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340
/* 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);
5341
	return (pfn >> pageblock_order) * NR_PAGEBLOCK_BITS;
5342 5343
#else
	pfn = pfn - zone->zone_start_pfn;
5344
	return (pfn >> pageblock_order) * NR_PAGEBLOCK_BITS;
5345 5346 5347 5348
#endif /* CONFIG_SPARSEMEM */
}

/**
5349
 * get_pageblock_flags_group - Return the requested group of flags for the pageblock_nr_pages block of pages
5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371
 * @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;
5372

5373 5374 5375 5376
	return flags;
}

/**
5377
 * set_pageblock_flags_group - Set the requested group of flags for a pageblock_nr_pages block of pages
5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394
 * @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);
5395 5396
	VM_BUG_ON(pfn < zone->zone_start_pfn);
	VM_BUG_ON(pfn >= zone->zone_start_pfn + zone->spanned_pages);
5397 5398 5399 5400 5401 5402 5403

	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
5404 5405 5406 5407 5408 5409 5410

/*
 * This is designed as sub function...plz see page_isolation.c also.
 * set/clear page block's type to be ISOLATE.
 * page allocater never alloc memory from ISOLATE block.
 */

5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428
static int
__count_immobile_pages(struct zone *zone, struct page *page, int count)
{
	unsigned long pfn, iter, found;
	/*
	 * For avoiding noise data, lru_add_drain_all() should be called
	 * If ZONE_MOVABLE, the zone never contains immobile pages
	 */
	if (zone_idx(zone) == ZONE_MOVABLE)
		return true;

	if (get_pageblock_migratetype(page) == MIGRATE_MOVABLE)
		return true;

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

5429
		if (!pfn_valid_within(check))
5430
			continue;
5431

5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448 5449 5450 5451 5452 5453 5454 5455 5456 5457 5458 5459 5460
		page = pfn_to_page(check);
		if (!page_count(page)) {
			if (PageBuddy(page))
				iter += (1 << page_order(page)) - 1;
			continue;
		}
		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)
			return false;
	}
	return true;
}

bool is_pageblock_removable_nolock(struct page *page)
{
5461 5462
	struct zone *zone;
	unsigned long pfn;
5463 5464 5465 5466 5467

	/*
	 * 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.
5468 5469
	 * We have to take care about the node as well. If the node is offline
	 * its NODE_DATA will be NULL - see page_zone.
5470
	 */
5471 5472 5473 5474 5475 5476
	if (!node_online(page_to_nid(page)))
		return false;

	zone = page_zone(page);
	pfn = page_to_pfn(page);
	if (zone->zone_start_pfn > pfn ||
5477 5478 5479
			zone->zone_start_pfn + zone->spanned_pages <= pfn)
		return false;

5480 5481 5482
	return __count_immobile_pages(zone, page, 0);
}

K
KAMEZAWA Hiroyuki 已提交
5483 5484 5485
int set_migratetype_isolate(struct page *page)
{
	struct zone *zone;
5486
	unsigned long flags, pfn;
5487 5488
	struct memory_isolate_notify arg;
	int notifier_ret;
K
KAMEZAWA Hiroyuki 已提交
5489 5490 5491
	int ret = -EBUSY;

	zone = page_zone(page);
5492

K
KAMEZAWA Hiroyuki 已提交
5493
	spin_lock_irqsave(&zone->lock, flags);
5494 5495 5496 5497 5498 5499

	pfn = page_to_pfn(page);
	arg.start_pfn = pfn;
	arg.nr_pages = pageblock_nr_pages;
	arg.pages_found = 0;

K
KAMEZAWA Hiroyuki 已提交
5500
	/*
5501 5502 5503 5504 5505 5506 5507 5508 5509
	 * It may be possible to isolate a pageblock even if the
	 * migratetype is not MIGRATE_MOVABLE. The memory isolation
	 * notifier chain is used by balloon drivers to return the
	 * number of pages in a range that are held by the balloon
	 * driver to shrink memory. If all the pages are accounted for
	 * by balloons, are free, or on the LRU, isolation can continue.
	 * Later, for example, when memory hotplug notifier runs, these
	 * pages reported as "can be isolated" should be isolated(freed)
	 * by the balloon driver through the memory notifier chain.
K
KAMEZAWA Hiroyuki 已提交
5510
	 */
5511 5512
	notifier_ret = memory_isolate_notify(MEM_ISOLATE_COUNT, &arg);
	notifier_ret = notifier_to_errno(notifier_ret);
5513
	if (notifier_ret)
K
KAMEZAWA Hiroyuki 已提交
5514
		goto out;
5515 5516 5517 5518 5519
	/*
	 * FIXME: Now, memory hotplug doesn't call shrink_slab() by itself.
	 * We just check MOVABLE pages.
	 */
	if (__count_immobile_pages(zone, page, arg.pages_found))
5520 5521
		ret = 0;

5522 5523 5524 5525 5526
	/*
	 * immobile means "not-on-lru" paes. If immobile is larger than
	 * removable-by-driver pages reported by notifier, we'll fail.
	 */

K
KAMEZAWA Hiroyuki 已提交
5527
out:
5528 5529 5530 5531 5532
	if (!ret) {
		set_pageblock_migratetype(page, MIGRATE_ISOLATE);
		move_freepages_block(zone, page, MIGRATE_ISOLATE);
	}

K
KAMEZAWA Hiroyuki 已提交
5533 5534
	spin_unlock_irqrestore(&zone->lock, flags);
	if (!ret)
5535
		drain_all_pages();
K
KAMEZAWA Hiroyuki 已提交
5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551
	return ret;
}

void unset_migratetype_isolate(struct page *page)
{
	struct zone *zone;
	unsigned long flags;
	zone = page_zone(page);
	spin_lock_irqsave(&zone->lock, flags);
	if (get_pageblock_migratetype(page) != MIGRATE_ISOLATE)
		goto out;
	set_pageblock_migratetype(page, MIGRATE_MOVABLE);
	move_freepages_block(zone, page, MIGRATE_MOVABLE);
out:
	spin_unlock_irqrestore(&zone->lock, flags);
}
K
KAMEZAWA Hiroyuki 已提交
5552 5553 5554 5555 5556 5557 5558 5559 5560 5561 5562 5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575 5576 5577 5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594 5595 5596 5597 5598

#ifdef CONFIG_MEMORY_HOTREMOVE
/*
 * 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
5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619

#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
5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 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 5658 5659 5660 5661 5662 5663 5664 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

static struct trace_print_flags pageflag_names[] = {
	{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"	},
#endif
	{-1UL,				NULL		},
};

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

	printk(KERN_ALERT "page flags: %#lx(", flags);

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

	for (i = 0; pageflag_names[i].name && flags; i++) {

		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",
5692
		page, atomic_read(&page->_count), page_mapcount(page),
5693 5694
		page->mapping, page->index);
	dump_page_flags(page->flags);
5695
	mem_cgroup_print_bad_page(page);
5696
}