hugetlb.c 59.1 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
/*
 * Generic hugetlb support.
 * (C) William Irwin, April 2004
 */
#include <linux/gfp.h>
#include <linux/list.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mm.h>
10
#include <linux/seq_file.h>
Linus Torvalds's avatar
Linus Torvalds committed
11
12
#include <linux/sysctl.h>
#include <linux/highmem.h>
Andrea Arcangeli's avatar
Andrea Arcangeli committed
13
#include <linux/mmu_notifier.h>
Linus Torvalds's avatar
Linus Torvalds committed
14
#include <linux/nodemask.h>
David Gibson's avatar
David Gibson committed
15
#include <linux/pagemap.h>
16
#include <linux/mempolicy.h>
17
#include <linux/cpuset.h>
18
#include <linux/mutex.h>
19
#include <linux/bootmem.h>
20
#include <linux/sysfs.h>
21

David Gibson's avatar
David Gibson committed
22
23
#include <asm/page.h>
#include <asm/pgtable.h>
24
#include <asm/io.h>
David Gibson's avatar
David Gibson committed
25
26

#include <linux/hugetlb.h>
27
#include "internal.h"
Linus Torvalds's avatar
Linus Torvalds committed
28
29

const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
30
31
static gfp_t htlb_alloc_mask = GFP_HIGHUSER;
unsigned long hugepages_treat_as_movable;
32

33
34
35
36
static int max_hstate;
unsigned int default_hstate_idx;
struct hstate hstates[HUGE_MAX_HSTATE];

37
38
__initdata LIST_HEAD(huge_boot_pages);

39
40
41
/* for command line parsing */
static struct hstate * __initdata parsed_hstate;
static unsigned long __initdata default_hstate_max_huge_pages;
42
static unsigned long __initdata default_hstate_size;
43
44
45

#define for_each_hstate(h) \
	for ((h) = hstates; (h) < &hstates[max_hstate]; (h)++)
46

47
48
49
50
/*
 * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages
 */
static DEFINE_SPINLOCK(hugetlb_lock);
51

52
53
54
/*
 * Region tracking -- allows tracking of reservations and instantiated pages
 *                    across the pages in a mapping.
55
56
57
58
59
60
61
62
63
64
 *
 * The region data structures are protected by a combination of the mmap_sem
 * and the hugetlb_instantion_mutex.  To access or modify a region the caller
 * must either hold the mmap_sem for write, or the mmap_sem for read and
 * the hugetlb_instantiation mutex:
 *
 * 	down_write(&mm->mmap_sem);
 * or
 * 	down_read(&mm->mmap_sem);
 * 	mutex_lock(&hugetlb_instantiation_mutex);
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
 */
struct file_region {
	struct list_head link;
	long from;
	long to;
};

static long region_add(struct list_head *head, long f, long t)
{
	struct file_region *rg, *nrg, *trg;

	/* Locate the region we are either in or before. */
	list_for_each_entry(rg, head, link)
		if (f <= rg->to)
			break;

	/* Round our left edge to the current segment if it encloses us. */
	if (f > rg->from)
		f = rg->from;

	/* Check for and consume any regions we now overlap with. */
	nrg = rg;
	list_for_each_entry_safe(rg, trg, rg->link.prev, link) {
		if (&rg->link == head)
			break;
		if (rg->from > t)
			break;

		/* If this area reaches higher then extend our area to
		 * include it completely.  If this is not the first area
		 * which we intend to reuse, free it. */
		if (rg->to > t)
			t = rg->to;
		if (rg != nrg) {
			list_del(&rg->link);
			kfree(rg);
		}
	}
	nrg->from = f;
	nrg->to = t;
	return 0;
}

static long region_chg(struct list_head *head, long f, long t)
{
	struct file_region *rg, *nrg;
	long chg = 0;

	/* Locate the region we are before or in. */
	list_for_each_entry(rg, head, link)
		if (f <= rg->to)
			break;

	/* If we are below the current region then a new region is required.
	 * Subtle, allocate a new region at the position but make it zero
	 * size such that we can guarantee to record the reservation. */
	if (&rg->link == head || t < rg->from) {
		nrg = kmalloc(sizeof(*nrg), GFP_KERNEL);
		if (!nrg)
			return -ENOMEM;
		nrg->from = f;
		nrg->to   = f;
		INIT_LIST_HEAD(&nrg->link);
		list_add(&nrg->link, rg->link.prev);

		return t - f;
	}

	/* Round our left edge to the current segment if it encloses us. */
	if (f > rg->from)
		f = rg->from;
	chg = t - f;

	/* Check for and consume any regions we now overlap with. */
	list_for_each_entry(rg, rg->link.prev, link) {
		if (&rg->link == head)
			break;
		if (rg->from > t)
			return chg;

		/* We overlap with this area, if it extends futher than
		 * us then we must extend ourselves.  Account for its
		 * existing reservation. */
		if (rg->to > t) {
			chg += rg->to - t;
			t = rg->to;
		}
		chg -= rg->to - rg->from;
	}
	return chg;
}

static long region_truncate(struct list_head *head, long end)
{
	struct file_region *rg, *trg;
	long chg = 0;

	/* Locate the region we are either in or before. */
	list_for_each_entry(rg, head, link)
		if (end <= rg->to)
			break;
	if (&rg->link == head)
		return 0;

	/* If we are in the middle of a region then adjust it. */
	if (end > rg->from) {
		chg = rg->to - end;
		rg->to = end;
		rg = list_entry(rg->link.next, typeof(*rg), link);
	}

	/* Drop any remaining regions. */
	list_for_each_entry_safe(rg, trg, rg->link.prev, link) {
		if (&rg->link == head)
			break;
		chg += rg->to - rg->from;
		list_del(&rg->link);
		kfree(rg);
	}
	return chg;
}

187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
static long region_count(struct list_head *head, long f, long t)
{
	struct file_region *rg;
	long chg = 0;

	/* Locate each segment we overlap with, and count that overlap. */
	list_for_each_entry(rg, head, link) {
		int seg_from;
		int seg_to;

		if (rg->to <= f)
			continue;
		if (rg->from >= t)
			break;

		seg_from = max(rg->from, f);
		seg_to = min(rg->to, t);

		chg += seg_to - seg_from;
	}

	return chg;
}

211
212
213
214
/*
 * Convert the address within this vma to the page offset within
 * the mapping, in pagecache page units; huge pages here.
 */
215
216
static pgoff_t vma_hugecache_offset(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
217
{
218
219
	return ((address - vma->vm_start) >> huge_page_shift(h)) +
			(vma->vm_pgoff >> huge_page_order(h));
220
221
}

222
223
224
225
226
227
228
/*
 * Flags for MAP_PRIVATE reservations.  These are stored in the bottom
 * bits of the reservation map pointer, which are always clear due to
 * alignment.
 */
#define HPAGE_RESV_OWNER    (1UL << 0)
#define HPAGE_RESV_UNMAPPED (1UL << 1)
229
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED)
230

231
232
233
234
235
236
237
238
239
/*
 * These helpers are used to track how many pages are reserved for
 * faults in a MAP_PRIVATE mapping. Only the process that called mmap()
 * is guaranteed to have their future faults succeed.
 *
 * With the exception of reset_vma_resv_huge_pages() which is called at fork(),
 * the reserve counters are updated with the hugetlb_lock held. It is safe
 * to reset the VMA at fork() time as it is not in use yet and there is no
 * chance of the global counters getting corrupted as a result of the values.
240
241
242
243
244
245
246
247
248
 *
 * The private mapping reservation is represented in a subtly different
 * manner to a shared mapping.  A shared mapping has a region map associated
 * with the underlying file, this region map represents the backing file
 * pages which have ever had a reservation assigned which this persists even
 * after the page is instantiated.  A private mapping has a region map
 * associated with the original mmap which is attached to all VMAs which
 * reference it, this region map represents those offsets which have consumed
 * reservation ie. where pages have been instantiated.
249
 */
250
251
252
253
254
255
256
257
258
259
260
static unsigned long get_vma_private_data(struct vm_area_struct *vma)
{
	return (unsigned long)vma->vm_private_data;
}

static void set_vma_private_data(struct vm_area_struct *vma,
							unsigned long value)
{
	vma->vm_private_data = (void *)value;
}

261
262
263
264
265
struct resv_map {
	struct kref refs;
	struct list_head regions;
};

266
static struct resv_map *resv_map_alloc(void)
267
268
269
270
271
272
273
274
275
276
277
{
	struct resv_map *resv_map = kmalloc(sizeof(*resv_map), GFP_KERNEL);
	if (!resv_map)
		return NULL;

	kref_init(&resv_map->refs);
	INIT_LIST_HEAD(&resv_map->regions);

	return resv_map;
}

278
static void resv_map_release(struct kref *ref)
279
280
281
282
283
284
285
286
287
{
	struct resv_map *resv_map = container_of(ref, struct resv_map, refs);

	/* Clear out any active regions before we release the map. */
	region_truncate(&resv_map->regions, 0);
	kfree(resv_map);
}

static struct resv_map *vma_resv_map(struct vm_area_struct *vma)
288
289
290
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
	if (!(vma->vm_flags & VM_SHARED))
291
292
		return (struct resv_map *)(get_vma_private_data(vma) &
							~HPAGE_RESV_MASK);
293
	return NULL;
294
295
}

296
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map)
297
298
299
300
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
	VM_BUG_ON(vma->vm_flags & VM_SHARED);

301
302
	set_vma_private_data(vma, (get_vma_private_data(vma) &
				HPAGE_RESV_MASK) | (unsigned long)map);
303
304
305
306
307
}

static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
308
309
310
	VM_BUG_ON(vma->vm_flags & VM_SHARED);

	set_vma_private_data(vma, get_vma_private_data(vma) | flags);
311
312
313
314
315
}

static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
316
317

	return (get_vma_private_data(vma) & flag) != 0;
318
319
320
}

/* Decrement the reserved pages in the hugepage pool by one */
321
322
static void decrement_hugepage_resv_vma(struct hstate *h,
			struct vm_area_struct *vma)
323
{
324
325
326
	if (vma->vm_flags & VM_NORESERVE)
		return;

327
328
	if (vma->vm_flags & VM_SHARED) {
		/* Shared mappings always use reserves */
329
		h->resv_huge_pages--;
330
	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
331
332
333
334
		/*
		 * Only the process that called mmap() has reserves for
		 * private mappings.
		 */
335
		h->resv_huge_pages--;
336
337
338
	}
}

339
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */
340
341
342
343
344
345
346
347
void reset_vma_resv_huge_pages(struct vm_area_struct *vma)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
	if (!(vma->vm_flags & VM_SHARED))
		vma->vm_private_data = (void *)0;
}

/* Returns true if the VMA has associated reserve pages */
348
static int vma_has_reserves(struct vm_area_struct *vma)
349
350
{
	if (vma->vm_flags & VM_SHARED)
351
352
353
354
		return 1;
	if (is_vma_resv_set(vma, HPAGE_RESV_OWNER))
		return 1;
	return 0;
355
356
}

357
358
359
360
361
362
363
364
365
366
367
368
static void clear_gigantic_page(struct page *page,
			unsigned long addr, unsigned long sz)
{
	int i;
	struct page *p = page;

	might_sleep();
	for (i = 0; i < sz/PAGE_SIZE; i++, p = mem_map_next(p, page, i)) {
		cond_resched();
		clear_user_highpage(p, addr + i * PAGE_SIZE);
	}
}
369
370
static void clear_huge_page(struct page *page,
			unsigned long addr, unsigned long sz)
371
372
373
{
	int i;

374
375
376
	if (unlikely(sz > MAX_ORDER_NR_PAGES))
		return clear_gigantic_page(page, addr, sz);

377
	might_sleep();
378
	for (i = 0; i < sz/PAGE_SIZE; i++) {
379
		cond_resched();
380
		clear_user_highpage(page + i, addr + i * PAGE_SIZE);
381
382
383
	}
}

384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
static void copy_gigantic_page(struct page *dst, struct page *src,
			   unsigned long addr, struct vm_area_struct *vma)
{
	int i;
	struct hstate *h = hstate_vma(vma);
	struct page *dst_base = dst;
	struct page *src_base = src;
	might_sleep();
	for (i = 0; i < pages_per_huge_page(h); ) {
		cond_resched();
		copy_user_highpage(dst, src, addr + i*PAGE_SIZE, vma);

		i++;
		dst = mem_map_next(dst, dst_base, i);
		src = mem_map_next(src, src_base, i);
	}
}
401
static void copy_huge_page(struct page *dst, struct page *src,
402
			   unsigned long addr, struct vm_area_struct *vma)
403
404
{
	int i;
405
	struct hstate *h = hstate_vma(vma);
406

407
408
409
	if (unlikely(pages_per_huge_page(h) > MAX_ORDER_NR_PAGES))
		return copy_gigantic_page(dst, src, addr, vma);

410
	might_sleep();
411
	for (i = 0; i < pages_per_huge_page(h); i++) {
412
		cond_resched();
413
		copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma);
414
415
416
	}
}

417
static void enqueue_huge_page(struct hstate *h, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
418
419
{
	int nid = page_to_nid(page);
420
421
422
	list_add(&page->lru, &h->hugepage_freelists[nid]);
	h->free_huge_pages++;
	h->free_huge_pages_node[nid]++;
Linus Torvalds's avatar
Linus Torvalds committed
423
424
}

425
static struct page *dequeue_huge_page(struct hstate *h)
426
427
428
429
430
{
	int nid;
	struct page *page = NULL;

	for (nid = 0; nid < MAX_NUMNODES; ++nid) {
431
432
		if (!list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
433
434
					  struct page, lru);
			list_del(&page->lru);
435
436
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
437
438
439
440
441
442
			break;
		}
	}
	return page;
}

443
444
static struct page *dequeue_huge_page_vma(struct hstate *h,
				struct vm_area_struct *vma,
445
				unsigned long address, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
446
{
447
	int nid;
Linus Torvalds's avatar
Linus Torvalds committed
448
	struct page *page = NULL;
449
	struct mempolicy *mpol;
450
	nodemask_t *nodemask;
451
	struct zonelist *zonelist = huge_zonelist(vma, address,
452
					htlb_alloc_mask, &mpol, &nodemask);
453
454
	struct zone *zone;
	struct zoneref *z;
Linus Torvalds's avatar
Linus Torvalds committed
455

456
457
458
459
460
	/*
	 * A child process with MAP_PRIVATE mappings created by their parent
	 * have no page reserves. This check ensures that reservations are
	 * not "stolen". The child may still get SIGKILLed
	 */
461
	if (!vma_has_reserves(vma) &&
462
			h->free_huge_pages - h->resv_huge_pages == 0)
463
464
		return NULL;

465
	/* If reserves cannot be used, ensure enough pages are in the pool */
466
	if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0)
467
468
		return NULL;

469
470
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						MAX_NR_ZONES - 1, nodemask) {
471
472
		nid = zone_to_nid(zone);
		if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask) &&
473
474
		    !list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
475
476
					  struct page, lru);
			list_del(&page->lru);
477
478
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
479
480

			if (!avoid_reserve)
481
				decrement_hugepage_resv_vma(h, vma);
482

Ken Chen's avatar
Ken Chen committed
483
			break;
484
		}
Linus Torvalds's avatar
Linus Torvalds committed
485
	}
486
	mpol_cond_put(mpol);
Linus Torvalds's avatar
Linus Torvalds committed
487
488
489
	return page;
}

490
static void update_and_free_page(struct hstate *h, struct page *page)
491
492
{
	int i;
493
494
495
496

	h->nr_huge_pages--;
	h->nr_huge_pages_node[page_to_nid(page)]--;
	for (i = 0; i < pages_per_huge_page(h); i++) {
497
498
499
500
501
502
		page[i].flags &= ~(1 << PG_locked | 1 << PG_error | 1 << PG_referenced |
				1 << PG_dirty | 1 << PG_active | 1 << PG_reserved |
				1 << PG_private | 1<< PG_writeback);
	}
	set_compound_page_dtor(page, NULL);
	set_page_refcounted(page);
503
	arch_release_hugepage(page);
504
	__free_pages(page, huge_page_order(h));
505
506
}

507
508
509
510
511
512
513
514
515
516
517
struct hstate *size_to_hstate(unsigned long size)
{
	struct hstate *h;

	for_each_hstate(h) {
		if (huge_page_size(h) == size)
			return h;
	}
	return NULL;
}

518
519
static void free_huge_page(struct page *page)
{
520
521
522
523
	/*
	 * Can't pass hstate in here because it is called from the
	 * compound page destructor.
	 */
524
	struct hstate *h = page_hstate(page);
525
	int nid = page_to_nid(page);
526
	struct address_space *mapping;
527

528
	mapping = (struct address_space *) page_private(page);
529
	set_page_private(page, 0);
530
	BUG_ON(page_count(page));
531
532
533
	INIT_LIST_HEAD(&page->lru);

	spin_lock(&hugetlb_lock);
534
	if (h->surplus_huge_pages_node[nid] && huge_page_order(h) < MAX_ORDER) {
535
536
537
		update_and_free_page(h, page);
		h->surplus_huge_pages--;
		h->surplus_huge_pages_node[nid]--;
538
	} else {
539
		enqueue_huge_page(h, page);
540
	}
541
	spin_unlock(&hugetlb_lock);
542
	if (mapping)
543
		hugetlb_put_quota(mapping, 1);
544
545
}

546
547
548
549
550
/*
 * Increment or decrement surplus_huge_pages.  Keep node-specific counters
 * balanced by operating on them in a round-robin fashion.
 * Returns 1 if an adjustment was made.
 */
551
static int adjust_pool_surplus(struct hstate *h, int delta)
552
553
554
555
556
557
558
559
560
561
562
563
{
	static int prev_nid;
	int nid = prev_nid;
	int ret = 0;

	VM_BUG_ON(delta != -1 && delta != 1);
	do {
		nid = next_node(nid, node_online_map);
		if (nid == MAX_NUMNODES)
			nid = first_node(node_online_map);

		/* To shrink on this node, there must be a surplus page */
564
		if (delta < 0 && !h->surplus_huge_pages_node[nid])
565
566
			continue;
		/* Surplus cannot exceed the total number of pages */
567
568
		if (delta > 0 && h->surplus_huge_pages_node[nid] >=
						h->nr_huge_pages_node[nid])
569
570
			continue;

571
572
		h->surplus_huge_pages += delta;
		h->surplus_huge_pages_node[nid] += delta;
573
574
575
576
577
578
579
580
		ret = 1;
		break;
	} while (nid != prev_nid);

	prev_nid = nid;
	return ret;
}

581
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid)
582
583
584
{
	set_compound_page_dtor(page, free_huge_page);
	spin_lock(&hugetlb_lock);
585
586
	h->nr_huge_pages++;
	h->nr_huge_pages_node[nid]++;
587
588
589
590
	spin_unlock(&hugetlb_lock);
	put_page(page); /* free it into the hugepage allocator */
}

591
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid)
Linus Torvalds's avatar
Linus Torvalds committed
592
593
{
	struct page *page;
594

595
596
597
	if (h->order >= MAX_ORDER)
		return NULL;

598
	page = alloc_pages_node(nid,
599
600
		htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
						__GFP_REPEAT|__GFP_NOWARN,
601
		huge_page_order(h));
Linus Torvalds's avatar
Linus Torvalds committed
602
	if (page) {
603
		if (arch_prepare_hugepage(page)) {
604
			__free_pages(page, huge_page_order(h));
605
			return NULL;
606
		}
607
		prep_new_huge_page(h, page, nid);
Linus Torvalds's avatar
Linus Torvalds committed
608
	}
609
610
611
612

	return page;
}

613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
/*
 * Use a helper variable to find the next node and then
 * copy it back to hugetlb_next_nid afterwards:
 * otherwise there's a window in which a racer might
 * pass invalid nid MAX_NUMNODES to alloc_pages_node.
 * But we don't need to use a spin_lock here: it really
 * doesn't matter if occasionally a racer chooses the
 * same nid as we do.  Move nid forward in the mask even
 * if we just successfully allocated a hugepage so that
 * the next caller gets hugepages on the next node.
 */
static int hstate_next_node(struct hstate *h)
{
	int next_nid;
	next_nid = next_node(h->hugetlb_next_nid, node_online_map);
	if (next_nid == MAX_NUMNODES)
		next_nid = first_node(node_online_map);
	h->hugetlb_next_nid = next_nid;
	return next_nid;
}

634
static int alloc_fresh_huge_page(struct hstate *h)
635
636
637
638
639
640
{
	struct page *page;
	int start_nid;
	int next_nid;
	int ret = 0;

641
	start_nid = h->hugetlb_next_nid;
642
643

	do {
644
		page = alloc_fresh_huge_page_node(h, h->hugetlb_next_nid);
645
646
		if (page)
			ret = 1;
647
		next_nid = hstate_next_node(h);
648
	} while (!page && h->hugetlb_next_nid != start_nid);
649

650
651
652
653
654
	if (ret)
		count_vm_event(HTLB_BUDDY_PGALLOC);
	else
		count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);

655
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
656
657
}

658
659
static struct page *alloc_buddy_huge_page(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
660
661
{
	struct page *page;
662
	unsigned int nid;
663

664
665
666
	if (h->order >= MAX_ORDER)
		return NULL;

667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
	/*
	 * Assume we will successfully allocate the surplus page to
	 * prevent racing processes from causing the surplus to exceed
	 * overcommit
	 *
	 * This however introduces a different race, where a process B
	 * tries to grow the static hugepage pool while alloc_pages() is
	 * called by process A. B will only examine the per-node
	 * counters in determining if surplus huge pages can be
	 * converted to normal huge pages in adjust_pool_surplus(). A
	 * won't be able to increment the per-node counter, until the
	 * lock is dropped by B, but B doesn't drop hugetlb_lock until
	 * no more huge pages can be converted from surplus to normal
	 * state (and doesn't try to convert again). Thus, we have a
	 * case where a surplus huge page exists, the pool is grown, and
	 * the surplus huge page still exists after, even though it
	 * should just have been converted to a normal huge page. This
	 * does not leak memory, though, as the hugepage will be freed
	 * once it is out of use. It also does not allow the counters to
	 * go out of whack in adjust_pool_surplus() as we don't modify
	 * the node values until we've gotten the hugepage and only the
	 * per-node value is checked there.
	 */
	spin_lock(&hugetlb_lock);
691
	if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) {
692
693
694
		spin_unlock(&hugetlb_lock);
		return NULL;
	} else {
695
696
		h->nr_huge_pages++;
		h->surplus_huge_pages++;
697
698
699
	}
	spin_unlock(&hugetlb_lock);

700
701
	page = alloc_pages(htlb_alloc_mask|__GFP_COMP|
					__GFP_REPEAT|__GFP_NOWARN,
702
					huge_page_order(h));
703

704
705
706
707
708
	if (page && arch_prepare_hugepage(page)) {
		__free_pages(page, huge_page_order(h));
		return NULL;
	}

709
	spin_lock(&hugetlb_lock);
710
	if (page) {
711
712
713
714
715
716
		/*
		 * This page is now managed by the hugetlb allocator and has
		 * no users -- drop the buddy allocator's reference.
		 */
		put_page_testzero(page);
		VM_BUG_ON(page_count(page));
717
		nid = page_to_nid(page);
718
		set_compound_page_dtor(page, free_huge_page);
719
720
721
		/*
		 * We incremented the global counters already
		 */
722
723
		h->nr_huge_pages_node[nid]++;
		h->surplus_huge_pages_node[nid]++;
724
		__count_vm_event(HTLB_BUDDY_PGALLOC);
725
	} else {
726
727
		h->nr_huge_pages--;
		h->surplus_huge_pages--;
728
		__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);
729
	}
730
	spin_unlock(&hugetlb_lock);
731
732
733
734

	return page;
}

735
736
737
738
/*
 * Increase the hugetlb pool such that it can accomodate a reservation
 * of size 'delta'.
 */
739
static int gather_surplus_pages(struct hstate *h, int delta)
740
741
742
743
744
745
{
	struct list_head surplus_list;
	struct page *page, *tmp;
	int ret, i;
	int needed, allocated;

746
	needed = (h->resv_huge_pages + delta) - h->free_huge_pages;
747
	if (needed <= 0) {
748
		h->resv_huge_pages += delta;
749
		return 0;
750
	}
751
752
753
754
755
756
757
758

	allocated = 0;
	INIT_LIST_HEAD(&surplus_list);

	ret = -ENOMEM;
retry:
	spin_unlock(&hugetlb_lock);
	for (i = 0; i < needed; i++) {
759
		page = alloc_buddy_huge_page(h, NULL, 0);
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
		if (!page) {
			/*
			 * We were not able to allocate enough pages to
			 * satisfy the entire reservation so we free what
			 * we've allocated so far.
			 */
			spin_lock(&hugetlb_lock);
			needed = 0;
			goto free;
		}

		list_add(&page->lru, &surplus_list);
	}
	allocated += needed;

	/*
	 * After retaking hugetlb_lock, we need to recalculate 'needed'
	 * because either resv_huge_pages or free_huge_pages may have changed.
	 */
	spin_lock(&hugetlb_lock);
780
781
	needed = (h->resv_huge_pages + delta) -
			(h->free_huge_pages + allocated);
782
783
784
785
786
787
788
	if (needed > 0)
		goto retry;

	/*
	 * The surplus_list now contains _at_least_ the number of extra pages
	 * needed to accomodate the reservation.  Add the appropriate number
	 * of pages to the hugetlb pool and free the extras back to the buddy
789
790
791
	 * allocator.  Commit the entire reservation here to prevent another
	 * process from stealing the pages as they are added to the pool but
	 * before they are reserved.
792
793
	 */
	needed += allocated;
794
	h->resv_huge_pages += delta;
795
796
	ret = 0;
free:
797
	/* Free the needed pages to the hugetlb pool */
798
	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
799
800
		if ((--needed) < 0)
			break;
801
		list_del(&page->lru);
802
		enqueue_huge_page(h, page);
803
804
805
806
807
808
809
	}

	/* Free unnecessary surplus pages to the buddy allocator */
	if (!list_empty(&surplus_list)) {
		spin_unlock(&hugetlb_lock);
		list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
			list_del(&page->lru);
810
			/*
811
812
813
			 * The page has a reference count of zero already, so
			 * call free_huge_page directly instead of using
			 * put_page.  This must be done with hugetlb_lock
814
815
816
			 * unlocked which is safe because free_huge_page takes
			 * hugetlb_lock before deciding how to free the page.
			 */
817
			free_huge_page(page);
818
		}
819
		spin_lock(&hugetlb_lock);
820
821
822
823
824
825
826
827
828
829
	}

	return ret;
}

/*
 * When releasing a hugetlb pool reservation, any surplus pages that were
 * allocated to satisfy the reservation must be explicitly freed if they were
 * never used.
 */
830
831
static void return_unused_surplus_pages(struct hstate *h,
					unsigned long unused_resv_pages)
832
833
834
835
836
{
	static int nid = -1;
	struct page *page;
	unsigned long nr_pages;

837
838
839
840
841
842
843
844
	/*
	 * We want to release as many surplus pages as possible, spread
	 * evenly across all nodes. Iterate across all nodes until we
	 * can no longer free unreserved surplus pages. This occurs when
	 * the nodes with surplus pages have no free pages.
	 */
	unsigned long remaining_iterations = num_online_nodes();

845
	/* Uncommit the reservation */
846
	h->resv_huge_pages -= unused_resv_pages;
847

848
849
850
851
	/* Cannot return gigantic pages currently */
	if (h->order >= MAX_ORDER)
		return;

852
	nr_pages = min(unused_resv_pages, h->surplus_huge_pages);
853

854
	while (remaining_iterations-- && nr_pages) {
855
856
857
858
		nid = next_node(nid, node_online_map);
		if (nid == MAX_NUMNODES)
			nid = first_node(node_online_map);

859
		if (!h->surplus_huge_pages_node[nid])
860
861
			continue;

862
863
		if (!list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
864
865
					  struct page, lru);
			list_del(&page->lru);
866
867
868
869
870
			update_and_free_page(h, page);
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
			h->surplus_huge_pages--;
			h->surplus_huge_pages_node[nid]--;
871
			nr_pages--;
872
			remaining_iterations = num_online_nodes();
873
874
875
876
		}
	}
}

877
878
879
880
881
882
883
884
885
/*
 * Determine if the huge page at addr within the vma has an associated
 * reservation.  Where it does not we will need to logically increase
 * reservation and actually increase quota before an allocation can occur.
 * Where any new reservation would be required the reservation change is
 * prepared, but not committed.  Once the page has been quota'd allocated
 * an instantiated the change should be committed via vma_commit_reservation.
 * No action is required on failure.
 */
886
887
static int vma_needs_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
888
889
890
891
892
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

	if (vma->vm_flags & VM_SHARED) {
893
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
894
895
896
		return region_chg(&inode->i_mapping->private_list,
							idx, idx + 1);

897
898
	} else if (!is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
		return 1;
899

900
901
	} else  {
		int err;
902
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
903
904
905
906
907
908
909
		struct resv_map *reservations = vma_resv_map(vma);

		err = region_chg(&reservations->regions, idx, idx + 1);
		if (err < 0)
			return err;
		return 0;
	}
910
}
911
912
static void vma_commit_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
913
914
915
916
917
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

	if (vma->vm_flags & VM_SHARED) {
918
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
919
		region_add(&inode->i_mapping->private_list, idx, idx + 1);
920
921

	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
922
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
923
924
925
926
		struct resv_map *reservations = vma_resv_map(vma);

		/* Mark this page used in the map. */
		region_add(&reservations->regions, idx, idx + 1);
927
928
929
	}
}

930
static struct page *alloc_huge_page(struct vm_area_struct *vma,
931
				    unsigned long addr, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
932
{
933
	struct hstate *h = hstate_vma(vma);
934
	struct page *page;
935
936
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;
937
	unsigned int chg;
938
939
940
941
942

	/*
	 * Processes that did not create the mapping will have no reserves and
	 * will not have accounted against quota. Check that the quota can be
	 * made before satisfying the allocation
943
944
	 * MAP_NORESERVE mappings may also need pages and quota allocated
	 * if no reserve mapping overlaps.
945
	 */
946
	chg = vma_needs_reservation(h, vma, addr);
947
948
949
	if (chg < 0)
		return ERR_PTR(chg);
	if (chg)
950
951
		if (hugetlb_get_quota(inode->i_mapping, chg))
			return ERR_PTR(-ENOSPC);
Linus Torvalds's avatar
Linus Torvalds committed
952
953

	spin_lock(&hugetlb_lock);
954
	page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve);
Linus Torvalds's avatar
Linus Torvalds committed
955
	spin_unlock(&hugetlb_lock);
956

Ken Chen's avatar
Ken Chen committed
957
	if (!page) {
958
		page = alloc_buddy_huge_page(h, vma, addr);
Ken Chen's avatar
Ken Chen committed
959
		if (!page) {
960
			hugetlb_put_quota(inode->i_mapping, chg);
Ken Chen's avatar
Ken Chen committed
961
962
963
			return ERR_PTR(-VM_FAULT_OOM);
		}
	}
964

965
966
	set_page_refcounted(page);
	set_page_private(page, (unsigned long) mapping);
967

968
	vma_commit_reservation(h, vma, addr);
969

970
	return page;
971
972
}

973
__attribute__((weak)) int alloc_bootmem_huge_page(struct hstate *h)
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
{
	struct huge_bootmem_page *m;
	int nr_nodes = nodes_weight(node_online_map);

	while (nr_nodes) {
		void *addr;

		addr = __alloc_bootmem_node_nopanic(
				NODE_DATA(h->hugetlb_next_nid),
				huge_page_size(h), huge_page_size(h), 0);

		if (addr) {
			/*
			 * Use the beginning of the huge page to store the
			 * huge_bootmem_page struct (until gather_bootmem
			 * puts them into the mem_map).
			 */
			m = addr;
			if (m)
				goto found;
		}
		hstate_next_node(h);
		nr_nodes--;
	}
	return 0;

found:
	BUG_ON((unsigned long)virt_to_phys(m) & (huge_page_size(h) - 1));
	/* Put them into a private list first because mem_map is not up yet */
	list_add(&m->list, &huge_boot_pages);
	m->hstate = h;
	return 1;
}

/* Put bootmem huge pages into the standard lists after mem_map is up */
static void __init gather_bootmem_prealloc(void)
{
	struct huge_bootmem_page *m;

	list_for_each_entry(m, &huge_boot_pages, list) {
		struct page *page = virt_to_page(m);
		struct hstate *h = m->hstate;
		__ClearPageReserved(page);
		WARN_ON(page_count(page) != 1);
		prep_compound_page(page, h->order);
		prep_new_huge_page(h, page, page_to_nid(page));
	}
}

1023
static void __init hugetlb_hstate_alloc_pages(struct hstate *h)
Linus Torvalds's avatar
Linus Torvalds committed
1024
1025
{
	unsigned long i;
1026

1027
	for (i = 0; i < h->max_huge_pages; ++i) {
1028
1029
1030
1031
		if (h->order >= MAX_ORDER) {
			if (!alloc_bootmem_huge_page(h))
				break;
		} else if (!alloc_fresh_huge_page(h))
Linus Torvalds's avatar
Linus Torvalds committed
1032
1033
			break;
	}
1034
	h->max_huge_pages = i;
1035
1036
1037
1038
1039
1040
1041
}

static void __init hugetlb_init_hstates(void)
{
	struct hstate *h;

	for_each_hstate(h) {
1042
1043
1044
		/* oversize hugepages were init'ed in early boot */
		if (h->order < MAX_ORDER)
			hugetlb_hstate_alloc_pages(h);
1045
1046
1047
	}
}

Andi Kleen's avatar
Andi Kleen committed
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
static char * __init memfmt(char *buf, unsigned long n)
{
	if (n >= (1UL << 30))
		sprintf(buf, "%lu GB", n >> 30);
	else if (n >= (1UL << 20))
		sprintf(buf, "%lu MB", n >> 20);
	else
		sprintf(buf, "%lu KB", n >> 10);
	return buf;
}

1059
1060
1061
1062
1063
static void __init report_hugepages(void)
{
	struct hstate *h;

	for_each_hstate(h) {
Andi Kleen's avatar
Andi Kleen committed
1064
1065
1066
1067
1068
		char buf[32];
		printk(KERN_INFO "HugeTLB registered %s page size, "
				 "pre-allocated %ld pages\n",
			memfmt(buf, huge_page_size(h)),
			h->free_huge_pages);
1069
1070
1071
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
1072
#ifdef CONFIG_HIGHMEM
1073
static void try_to_free_low(struct hstate *h, unsigned long count)
Linus Torvalds's avatar
Linus Torvalds committed
1074
{
1075
1076
	int i;

1077
1078
1079
	if (h->order >= MAX_ORDER)
		return;

Linus Torvalds's avatar
Linus Torvalds committed
1080
1081
	for (i = 0; i < MAX_NUMNODES; ++i) {
		struct page *page, *next;
1082
1083
1084
		struct list_head *freel = &h->hugepage_freelists[i];
		list_for_each_entry_safe(page, next, freel, lru) {
			if (count >= h->nr_huge_pages)
1085
				return;
Linus Torvalds's avatar
Linus Torvalds committed
1086
1087
1088
			if (PageHighMem(page))
				continue;
			list_del(&page->lru);
1089
			update_and_free_page(h, page);
1090
1091
			h->free_huge_pages--;
			h->free_huge_pages_node[page_to_nid(page)]--;
Linus Torvalds's avatar
Linus Torvalds committed
1092
1093
1094
1095
		}
	}
}
#else
1096
static inline void try_to_free_low(struct hstate *h, unsigned long count)
Linus Torvalds's avatar
Linus Torvalds committed
1097
1098
1099
1100
{
}
#endif

1101
#define persistent_huge_pages(h) (h->nr_huge_pages - h->surplus_huge_pages)
1102
static unsigned long set_max_huge_pages(struct hstate *h, unsigned long count)
Linus Torvalds's avatar
Linus Torvalds committed
1103
{
1104
	unsigned long min_count, ret;
Linus Torvalds's avatar
Linus Torvalds committed
1105

1106
1107
1108
	if (h->order >= MAX_ORDER)
		return h->max_huge_pages;

1109
1110
1111
1112
	/*
	 * Increase the pool size
	 * First take pages out of surplus state.  Then make up the
	 * remaining difference by allocating fresh huge pages.
1113
1114
1115
1116
1117
1118
	 *
	 * We might race with alloc_buddy_huge_page() here and be unable
	 * to convert a surplus huge page to a normal huge page. That is
	 * not critical, though, it just means the overall size of the
	 * pool might be one hugepage larger than it needs to be, but
	 * within all the constraints specified by the sysctls.
1119
	 */
Linus Torvalds's avatar
Linus Torvalds committed
1120
	spin_lock(&hugetlb_lock);
1121
1122
	while (h->surplus_huge_pages && count > persistent_huge_pages(h)) {
		if (!adjust_pool_surplus(h, -1))
1123
1124
1125
			break;
	}

1126
	while (count > persistent_huge_pages(h)) {
1127
1128
1129
1130
1131
1132
		/*
		 * If this allocation races such that we no longer need the
		 * page, free_huge_page will handle it by freeing the page
		 * and reducing the surplus.
		 */
		spin_unlock(&hugetlb_lock);
1133
		ret = alloc_fresh_huge_page(h);
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
		spin_lock(&hugetlb_lock);
		if (!ret)
			goto out;

	}

	/*
	 * Decrease the pool size
	 * First return free pages to the buddy allocator (being careful
	 * to keep enough around to satisfy reservations).  Then place
	 * pages into surplus state as needed so the pool will shrink
	 * to the desired size as pages become free.
1146
1147
1148
1149
1150
1151
1152
1153
	 *
	 * By placing pages into the surplus state independent of the
	 * overcommit value, we are allowing the surplus pool size to
	 * exceed overcommit. There are few sane options here. Since
	 * alloc_buddy_huge_page() is checking the global counter,
	 * though, we'll note that we're not allowed to exceed surplus
	 * and won't grow the pool anywhere else. Not until one of the
	 * sysctls are changed, or the surplus pages go out of use.
1154
	 */
1155
	min_count = h->resv_huge_pages + h->nr_huge_pages - h->free_huge_pages;
1156
	min_count = max(count, min_count);
1157
1158
1159
	try_to_free_low(h, min_count);
	while (min_count < persistent_huge_pages(h)) {
		struct page *page = dequeue_huge_page(h);
Linus Torvalds's avatar
Linus Torvalds committed
1160
1161
		if (!page)
			break;
1162
		update_and_free_page(h, page);
Linus Torvalds's avatar
Linus Torvalds committed
1163
	}
1164
1165
	while (count < persistent_huge_pages(h)) {
		if (!adjust_pool_surplus(h, 1))
1166
1167
1168
			break;
	}
out:
1169
	ret = persistent_huge_pages(h);
Linus Torvalds's avatar
Linus Torvalds committed
1170
	spin_unlock(&hugetlb_lock);
1171
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
1172
1173
}

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
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
#define HSTATE_ATTR_RO(_name) \
	static struct kobj_attribute _name##_attr = __ATTR_RO(_name)

#define HSTATE_ATTR(_name) \
	static struct kobj_attribute _name##_attr = \
		__ATTR(_name, 0644, _name##_show, _name##_store)

static struct kobject *hugepages_kobj;
static struct kobject *hstate_kobjs[HUGE_MAX_HSTATE];

static struct hstate *kobj_to_hstate(struct kobject *kobj)
{
	int i;
	for (i = 0; i < HUGE_MAX_HSTATE; i++)
		if (hstate_kobjs[i] == kobj)
			return &hstates[i];
	BUG();
	return NULL;
}

static ssize_t nr_hugepages_show(struct kobject *kobj,
					struct kobj_attribute *attr, char *buf)
{
	struct hstate *h = kobj_to_hstate(kobj);
	return sprintf(buf, "%lu\n", h->nr_huge_pages);
}
static ssize_t nr_hugepages_store(struct kobject *kobj,
		struct kobj_attribute *attr, const char *buf, size_t count)
{
	int err;
	unsigned long input;
	struct hstate *h = kobj_to_hstate(kobj);

	err = strict_strtoul(buf, 10, &input);
	if (err)
		return 0;

	h->max_huge_pages = set_max_huge_pages(h, input);

	return count;
}
HSTATE_ATTR(nr_hugepages);

static ssize_t nr_overcommit_hugepages_show(struct kobject *kobj,
					struct kobj_attribute *attr, char *buf)
{
	struct hstate *h = kobj_to_hstate(kobj);
	return sprintf(buf, "%lu\n", h->nr_overcommit_huge_pages);
}
static ssize_t nr_overcommit_hugepages_store(struct kobject *kobj,