hugetlb.c 55.6 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
12
/*
 * 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>
#include <linux/sysctl.h>
#include <linux/highmem.h>
#include <linux/nodemask.h>
David Gibson's avatar
David Gibson committed
13
#include <linux/pagemap.h>
14
#include <linux/mempolicy.h>
15
#include <linux/cpuset.h>
16
#include <linux/mutex.h>
17
#include <linux/bootmem.h>
18
#include <linux/sysfs.h>
19

David Gibson's avatar
David Gibson committed
20
21
22
23
#include <asm/page.h>
#include <asm/pgtable.h>

#include <linux/hugetlb.h>
24
#include "internal.h"
Linus Torvalds's avatar
Linus Torvalds committed
25
26

const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
27
28
static gfp_t htlb_alloc_mask = GFP_HIGHUSER;
unsigned long hugepages_treat_as_movable;
29

30
31
32
33
34
35
36
37
38
39
static int max_hstate;
unsigned int default_hstate_idx;
struct hstate hstates[HUGE_MAX_HSTATE];

/* for command line parsing */
static struct hstate * __initdata parsed_hstate;
static unsigned long __initdata default_hstate_max_huge_pages;

#define for_each_hstate(h) \
	for ((h) = hstates; (h) < &hstates[max_hstate]; (h)++)
40

41
42
43
44
/*
 * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages
 */
static DEFINE_SPINLOCK(hugetlb_lock);
45

46
47
48
/*
 * Region tracking -- allows tracking of reservations and instantiated pages
 *                    across the pages in a mapping.
49
50
51
52
53
54
55
56
57
58
 *
 * 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);
59
60
61
62
63
64
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
 */
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;
}

181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
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;
}

205
206
207
208
/*
 * Convert the address within this vma to the page offset within
 * the mapping, in pagecache page units; huge pages here.
 */
209
210
static pgoff_t vma_hugecache_offset(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
211
{
212
213
	return ((address - vma->vm_start) >> huge_page_shift(h)) +
			(vma->vm_pgoff >> huge_page_order(h));
214
215
}

216
217
218
219
220
221
222
/*
 * 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)
223
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED)
224

225
226
227
228
229
230
231
232
233
/*
 * 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.
234
235
236
237
238
239
240
241
242
 *
 * 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.
243
 */
244
245
246
247
248
249
250
251
252
253
254
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;
}

255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
struct resv_map {
	struct kref refs;
	struct list_head regions;
};

struct resv_map *resv_map_alloc(void)
{
	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;
}

void resv_map_release(struct kref *ref)
{
	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)
282
283
284
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
	if (!(vma->vm_flags & VM_SHARED))
285
286
		return (struct resv_map *)(get_vma_private_data(vma) &
							~HPAGE_RESV_MASK);
287
288
289
	return 0;
}

290
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map)
291
292
293
294
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
	VM_BUG_ON(vma->vm_flags & VM_SHARED);

295
296
	set_vma_private_data(vma, (get_vma_private_data(vma) &
				HPAGE_RESV_MASK) | (unsigned long)map);
297
298
299
300
301
}

static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
302
303
304
	VM_BUG_ON(vma->vm_flags & VM_SHARED);

	set_vma_private_data(vma, get_vma_private_data(vma) | flags);
305
306
307
308
309
}

static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
310
311

	return (get_vma_private_data(vma) & flag) != 0;
312
313
314
}

/* Decrement the reserved pages in the hugepage pool by one */
315
316
static void decrement_hugepage_resv_vma(struct hstate *h,
			struct vm_area_struct *vma)
317
{
318
319
320
	if (vma->vm_flags & VM_NORESERVE)
		return;

321
322
	if (vma->vm_flags & VM_SHARED) {
		/* Shared mappings always use reserves */
323
		h->resv_huge_pages--;
324
	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
325
326
327
328
		/*
		 * Only the process that called mmap() has reserves for
		 * private mappings.
		 */
329
		h->resv_huge_pages--;
330
331
332
	}
}

333
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */
334
335
336
337
338
339
340
341
342
343
344
345
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 */
static int vma_has_private_reserves(struct vm_area_struct *vma)
{
	if (vma->vm_flags & VM_SHARED)
		return 0;
346
	if (!is_vma_resv_set(vma, HPAGE_RESV_OWNER))
347
348
349
350
		return 0;
	return 1;
}

351
352
static void clear_huge_page(struct page *page,
			unsigned long addr, unsigned long sz)
353
354
355
356
{
	int i;

	might_sleep();
357
	for (i = 0; i < sz/PAGE_SIZE; i++) {
358
		cond_resched();
359
		clear_user_highpage(page + i, addr + i * PAGE_SIZE);
360
361
362
363
	}
}

static void copy_huge_page(struct page *dst, struct page *src,
364
			   unsigned long addr, struct vm_area_struct *vma)
365
366
{
	int i;
367
	struct hstate *h = hstate_vma(vma);
368
369

	might_sleep();
370
	for (i = 0; i < pages_per_huge_page(h); i++) {
371
		cond_resched();
372
		copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma);
373
374
375
	}
}

376
static void enqueue_huge_page(struct hstate *h, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
377
378
{
	int nid = page_to_nid(page);
379
380
381
	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
382
383
}

384
static struct page *dequeue_huge_page(struct hstate *h)
385
386
387
388
389
{
	int nid;
	struct page *page = NULL;

	for (nid = 0; nid < MAX_NUMNODES; ++nid) {
390
391
		if (!list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
392
393
					  struct page, lru);
			list_del(&page->lru);
394
395
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
396
397
398
399
400
401
			break;
		}
	}
	return page;
}

402
403
static struct page *dequeue_huge_page_vma(struct hstate *h,
				struct vm_area_struct *vma,
404
				unsigned long address, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
405
{
406
	int nid;
Linus Torvalds's avatar
Linus Torvalds committed
407
	struct page *page = NULL;
408
	struct mempolicy *mpol;
409
	nodemask_t *nodemask;
410
	struct zonelist *zonelist = huge_zonelist(vma, address,
411
					htlb_alloc_mask, &mpol, &nodemask);
412
413
	struct zone *zone;
	struct zoneref *z;
Linus Torvalds's avatar
Linus Torvalds committed
414

415
416
417
418
419
420
	/*
	 * 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
	 */
	if (!vma_has_private_reserves(vma) &&
421
			h->free_huge_pages - h->resv_huge_pages == 0)
422
423
		return NULL;

424
	/* If reserves cannot be used, ensure enough pages are in the pool */
425
	if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0)
426
427
		return NULL;

428
429
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						MAX_NR_ZONES - 1, nodemask) {
430
431
		nid = zone_to_nid(zone);
		if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask) &&
432
433
		    !list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
434
435
					  struct page, lru);
			list_del(&page->lru);
436
437
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
438
439

			if (!avoid_reserve)
440
				decrement_hugepage_resv_vma(h, vma);
441

Ken Chen's avatar
Ken Chen committed
442
			break;
443
		}
Linus Torvalds's avatar
Linus Torvalds committed
444
	}
445
	mpol_cond_put(mpol);
Linus Torvalds's avatar
Linus Torvalds committed
446
447
448
	return page;
}

449
static void update_and_free_page(struct hstate *h, struct page *page)
450
451
{
	int i;
452
453
454
455

	h->nr_huge_pages--;
	h->nr_huge_pages_node[page_to_nid(page)]--;
	for (i = 0; i < pages_per_huge_page(h); i++) {
456
457
458
459
460
461
		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);
462
	arch_release_hugepage(page);
463
	__free_pages(page, huge_page_order(h));
464
465
}

466
467
468
469
470
471
472
473
474
475
476
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;
}

477
478
static void free_huge_page(struct page *page)
{
479
480
481
482
	/*
	 * Can't pass hstate in here because it is called from the
	 * compound page destructor.
	 */
483
	struct hstate *h = page_hstate(page);
484
	int nid = page_to_nid(page);
485
	struct address_space *mapping;
486

487
	mapping = (struct address_space *) page_private(page);
488
	set_page_private(page, 0);
489
	BUG_ON(page_count(page));
490
491
492
	INIT_LIST_HEAD(&page->lru);

	spin_lock(&hugetlb_lock);
493
	if (h->surplus_huge_pages_node[nid] && huge_page_order(h) < MAX_ORDER) {
494
495
496
		update_and_free_page(h, page);
		h->surplus_huge_pages--;
		h->surplus_huge_pages_node[nid]--;
497
	} else {
498
		enqueue_huge_page(h, page);
499
	}
500
	spin_unlock(&hugetlb_lock);
501
	if (mapping)
502
		hugetlb_put_quota(mapping, 1);
503
504
}

505
506
507
508
509
/*
 * 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.
 */
510
static int adjust_pool_surplus(struct hstate *h, int delta)
511
512
513
514
515
516
517
518
519
520
521
522
{
	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 */
523
		if (delta < 0 && !h->surplus_huge_pages_node[nid])
524
525
			continue;
		/* Surplus cannot exceed the total number of pages */
526
527
		if (delta > 0 && h->surplus_huge_pages_node[nid] >=
						h->nr_huge_pages_node[nid])
528
529
			continue;

530
531
		h->surplus_huge_pages += delta;
		h->surplus_huge_pages_node[nid] += delta;
532
533
534
535
536
537
538
539
		ret = 1;
		break;
	} while (nid != prev_nid);

	prev_nid = nid;
	return ret;
}

540
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid)
541
542
543
{
	set_compound_page_dtor(page, free_huge_page);
	spin_lock(&hugetlb_lock);
544
545
	h->nr_huge_pages++;
	h->nr_huge_pages_node[nid]++;
546
547
548
549
	spin_unlock(&hugetlb_lock);
	put_page(page); /* free it into the hugepage allocator */
}

550
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid)
Linus Torvalds's avatar
Linus Torvalds committed
551
552
{
	struct page *page;
553

554
555
556
	if (h->order >= MAX_ORDER)
		return NULL;

557
	page = alloc_pages_node(nid,
558
559
		htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
						__GFP_REPEAT|__GFP_NOWARN,
560
		huge_page_order(h));
Linus Torvalds's avatar
Linus Torvalds committed
561
	if (page) {
562
563
		if (arch_prepare_hugepage(page)) {
			__free_pages(page, HUGETLB_PAGE_ORDER);
564
			return NULL;
565
		}
566
		prep_new_huge_page(h, page, nid);
Linus Torvalds's avatar
Linus Torvalds committed
567
	}
568
569
570
571

	return page;
}

572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
/*
 * 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;
}

593
static int alloc_fresh_huge_page(struct hstate *h)
594
595
596
597
598
599
{
	struct page *page;
	int start_nid;
	int next_nid;
	int ret = 0;

600
	start_nid = h->hugetlb_next_nid;
601
602

	do {
603
		page = alloc_fresh_huge_page_node(h, h->hugetlb_next_nid);
604
605
		if (page)
			ret = 1;
606
		next_nid = hstate_next_node(h);
607
	} while (!page && h->hugetlb_next_nid != start_nid);
608

609
610
611
612
613
	if (ret)
		count_vm_event(HTLB_BUDDY_PGALLOC);
	else
		count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);

614
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
615
616
}

617
618
static struct page *alloc_buddy_huge_page(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
619
620
{
	struct page *page;
621
	unsigned int nid;
622

623
624
625
	if (h->order >= MAX_ORDER)
		return NULL;

626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
	/*
	 * 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);
650
	if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) {
651
652
653
		spin_unlock(&hugetlb_lock);
		return NULL;
	} else {
654
655
		h->nr_huge_pages++;
		h->surplus_huge_pages++;
656
657
658
	}
	spin_unlock(&hugetlb_lock);

659
660
	page = alloc_pages(htlb_alloc_mask|__GFP_COMP|
					__GFP_REPEAT|__GFP_NOWARN,
661
					huge_page_order(h));
662
663

	spin_lock(&hugetlb_lock);
664
	if (page) {
665
666
667
668
669
670
		/*
		 * 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));
671
		nid = page_to_nid(page);
672
		set_compound_page_dtor(page, free_huge_page);
673
674
675
		/*
		 * We incremented the global counters already
		 */
676
677
		h->nr_huge_pages_node[nid]++;
		h->surplus_huge_pages_node[nid]++;
678
		__count_vm_event(HTLB_BUDDY_PGALLOC);
679
	} else {
680
681
		h->nr_huge_pages--;
		h->surplus_huge_pages--;
682
		__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);
683
	}
684
	spin_unlock(&hugetlb_lock);
685
686
687
688

	return page;
}

689
690
691
692
/*
 * Increase the hugetlb pool such that it can accomodate a reservation
 * of size 'delta'.
 */
693
static int gather_surplus_pages(struct hstate *h, int delta)
694
695
696
697
698
699
{
	struct list_head surplus_list;
	struct page *page, *tmp;
	int ret, i;
	int needed, allocated;

700
	needed = (h->resv_huge_pages + delta) - h->free_huge_pages;
701
	if (needed <= 0) {
702
		h->resv_huge_pages += delta;
703
		return 0;
704
	}
705
706
707
708
709
710
711
712

	allocated = 0;
	INIT_LIST_HEAD(&surplus_list);

	ret = -ENOMEM;
retry:
	spin_unlock(&hugetlb_lock);
	for (i = 0; i < needed; i++) {
713
		page = alloc_buddy_huge_page(h, NULL, 0);
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
		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);
734
735
	needed = (h->resv_huge_pages + delta) -
			(h->free_huge_pages + allocated);
736
737
738
739
740
741
742
	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
743
744
745
	 * 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.
746
747
	 */
	needed += allocated;
748
	h->resv_huge_pages += delta;
749
750
	ret = 0;
free:
751
	/* Free the needed pages to the hugetlb pool */
752
	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
753
754
		if ((--needed) < 0)
			break;
755
		list_del(&page->lru);
756
		enqueue_huge_page(h, page);
757
758
759
760
761
762
763
	}

	/* 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);
764
			/*
765
766
767
			 * 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
768
769
770
			 * unlocked which is safe because free_huge_page takes
			 * hugetlb_lock before deciding how to free the page.
			 */
771
			free_huge_page(page);
772
		}
773
		spin_lock(&hugetlb_lock);
774
775
776
777
778
779
780
781
782
783
	}

	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.
 */
784
785
static void return_unused_surplus_pages(struct hstate *h,
					unsigned long unused_resv_pages)
786
787
788
789
790
{
	static int nid = -1;
	struct page *page;
	unsigned long nr_pages;

791
792
793
794
795
796
797
798
	/*
	 * 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();

799
	/* Uncommit the reservation */
800
	h->resv_huge_pages -= unused_resv_pages;
801

802
803
804
805
	/* Cannot return gigantic pages currently */
	if (h->order >= MAX_ORDER)
		return;

806
	nr_pages = min(unused_resv_pages, h->surplus_huge_pages);
807

808
	while (remaining_iterations-- && nr_pages) {
809
810
811
812
		nid = next_node(nid, node_online_map);
		if (nid == MAX_NUMNODES)
			nid = first_node(node_online_map);

813
		if (!h->surplus_huge_pages_node[nid])
814
815
			continue;

816
817
		if (!list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
818
819
					  struct page, lru);
			list_del(&page->lru);
820
821
822
823
824
			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]--;
825
			nr_pages--;
826
			remaining_iterations = num_online_nodes();
827
828
829
830
		}
	}
}

831
832
833
834
835
836
837
838
839
/*
 * 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.
 */
840
841
static int vma_needs_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
842
843
844
845
846
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

	if (vma->vm_flags & VM_SHARED) {
847
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
848
849
850
		return region_chg(&inode->i_mapping->private_list,
							idx, idx + 1);

851
852
	} else if (!is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
		return 1;
853

854
855
	} else  {
		int err;
856
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
857
858
859
860
861
862
863
		struct resv_map *reservations = vma_resv_map(vma);

		err = region_chg(&reservations->regions, idx, idx + 1);
		if (err < 0)
			return err;
		return 0;
	}
864
}
865
866
static void vma_commit_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
867
868
869
870
871
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

	if (vma->vm_flags & VM_SHARED) {
872
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
873
		region_add(&inode->i_mapping->private_list, idx, idx + 1);
874
875

	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
876
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
877
878
879
880
		struct resv_map *reservations = vma_resv_map(vma);

		/* Mark this page used in the map. */
		region_add(&reservations->regions, idx, idx + 1);
881
882
883
	}
}

884
static struct page *alloc_huge_page(struct vm_area_struct *vma,
885
				    unsigned long addr, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
886
{
887
	struct hstate *h = hstate_vma(vma);
888
	struct page *page;
889
890
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;
891
	unsigned int chg;
892
893
894
895
896

	/*
	 * 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
897
898
	 * MAP_NORESERVE mappings may also need pages and quota allocated
	 * if no reserve mapping overlaps.
899
	 */
900
	chg = vma_needs_reservation(h, vma, addr);
901
902
903
	if (chg < 0)
		return ERR_PTR(chg);
	if (chg)
904
905
		if (hugetlb_get_quota(inode->i_mapping, chg))
			return ERR_PTR(-ENOSPC);
Linus Torvalds's avatar
Linus Torvalds committed
906
907

	spin_lock(&hugetlb_lock);
908
	page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve);
Linus Torvalds's avatar
Linus Torvalds committed
909
	spin_unlock(&hugetlb_lock);
910

Ken Chen's avatar
Ken Chen committed
911
	if (!page) {
912
		page = alloc_buddy_huge_page(h, vma, addr);
Ken Chen's avatar
Ken Chen committed
913
		if (!page) {
914
			hugetlb_put_quota(inode->i_mapping, chg);
Ken Chen's avatar
Ken Chen committed
915
916
917
			return ERR_PTR(-VM_FAULT_OOM);
		}
	}
918

919
920
	set_page_refcounted(page);
	set_page_private(page, (unsigned long) mapping);
921

922
	vma_commit_reservation(h, vma, addr);
923

924
	return page;
925
926
}

927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
static __initdata LIST_HEAD(huge_boot_pages);

struct huge_bootmem_page {
	struct list_head list;
	struct hstate *hstate;
};

static int __init alloc_bootmem_huge_page(struct hstate *h)
{
	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));
	}
}

984
static void __init hugetlb_hstate_alloc_pages(struct hstate *h)
Linus Torvalds's avatar
Linus Torvalds committed
985
986
{
	unsigned long i;
987

988
	for (i = 0; i < h->max_huge_pages; ++i) {
989
990
991
992
		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
993
994
			break;
	}
995
	h->max_huge_pages = i;
996
997
998
999
1000
}

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