hugetlb.c 79.7 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
/*
 * Generic hugetlb support.
 * (C) William Irwin, April 2004
 */
#include <linux/list.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mm.h>
9
#include <linux/seq_file.h>
Linus Torvalds's avatar
Linus Torvalds committed
10
11
#include <linux/sysctl.h>
#include <linux/highmem.h>
Andrea Arcangeli's avatar
Andrea Arcangeli committed
12
#include <linux/mmu_notifier.h>
Linus Torvalds's avatar
Linus Torvalds committed
13
#include <linux/nodemask.h>
David Gibson's avatar
David Gibson committed
14
#include <linux/pagemap.h>
15
#include <linux/mempolicy.h>
16
#include <linux/cpuset.h>
17
#include <linux/mutex.h>
18
#include <linux/bootmem.h>
19
#include <linux/sysfs.h>
20
#include <linux/slab.h>
21
#include <linux/rmap.h>
22
23
#include <linux/swap.h>
#include <linux/swapops.h>
24

David Gibson's avatar
David Gibson committed
25
26
#include <asm/page.h>
#include <asm/pgtable.h>
27
#include <asm/tlb.h>
David Gibson's avatar
David Gibson committed
28

29
#include <linux/io.h>
David Gibson's avatar
David Gibson committed
30
#include <linux/hugetlb.h>
31
#include <linux/node.h>
32
#include "internal.h"
Linus Torvalds's avatar
Linus Torvalds committed
33
34

const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
35
36
static gfp_t htlb_alloc_mask = GFP_HIGHUSER;
unsigned long hugepages_treat_as_movable;
37

38
static int hugetlb_max_hstate;
39
40
41
unsigned int default_hstate_idx;
struct hstate hstates[HUGE_MAX_HSTATE];

42
43
__initdata LIST_HEAD(huge_boot_pages);

44
45
46
/* for command line parsing */
static struct hstate * __initdata parsed_hstate;
static unsigned long __initdata default_hstate_max_huge_pages;
47
static unsigned long __initdata default_hstate_size;
48
49

#define for_each_hstate(h) \
50
	for ((h) = hstates; (h) < &hstates[hugetlb_max_hstate]; (h)++)
51

52
53
54
55
/*
 * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages
 */
static DEFINE_SPINLOCK(hugetlb_lock);
56

57
58
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
static inline void unlock_or_release_subpool(struct hugepage_subpool *spool)
{
	bool free = (spool->count == 0) && (spool->used_hpages == 0);

	spin_unlock(&spool->lock);

	/* If no pages are used, and no other handles to the subpool
	 * remain, free the subpool the subpool remain */
	if (free)
		kfree(spool);
}

struct hugepage_subpool *hugepage_new_subpool(long nr_blocks)
{
	struct hugepage_subpool *spool;

	spool = kmalloc(sizeof(*spool), GFP_KERNEL);
	if (!spool)
		return NULL;

	spin_lock_init(&spool->lock);
	spool->count = 1;
	spool->max_hpages = nr_blocks;
	spool->used_hpages = 0;

	return spool;
}

void hugepage_put_subpool(struct hugepage_subpool *spool)
{
	spin_lock(&spool->lock);
	BUG_ON(!spool->count);
	spool->count--;
	unlock_or_release_subpool(spool);
}

static int hugepage_subpool_get_pages(struct hugepage_subpool *spool,
				      long delta)
{
	int ret = 0;

	if (!spool)
		return 0;

	spin_lock(&spool->lock);
	if ((spool->used_hpages + delta) <= spool->max_hpages) {
		spool->used_hpages += delta;
	} else {
		ret = -ENOMEM;
	}
	spin_unlock(&spool->lock);

	return ret;
}

static void hugepage_subpool_put_pages(struct hugepage_subpool *spool,
				       long delta)
{
	if (!spool)
		return;

	spin_lock(&spool->lock);
	spool->used_hpages -= delta;
	/* If hugetlbfs_put_super couldn't free spool due to
	* an outstanding quota reference, free it now. */
	unlock_or_release_subpool(spool);
}

static inline struct hugepage_subpool *subpool_inode(struct inode *inode)
{
	return HUGETLBFS_SB(inode->i_sb)->spool;
}

static inline struct hugepage_subpool *subpool_vma(struct vm_area_struct *vma)
{
	return subpool_inode(vma->vm_file->f_dentry->d_inode);
}

135
136
137
/*
 * Region tracking -- allows tracking of reservations and instantiated pages
 *                    across the pages in a mapping.
138
139
140
141
142
143
 *
 * 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:
 *
144
 *	down_write(&mm->mmap_sem);
145
 * or
146
147
 *	down_read(&mm->mmap_sem);
 *	mutex_lock(&hugetlb_instantiation_mutex);
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
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
 */
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;

Lucas De Marchi's avatar
Lucas De Marchi committed
228
		/* We overlap with this area, if it extends further than
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
		 * 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;
}

270
271
272
273
274
275
276
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) {
277
278
		long seg_from;
		long seg_to;
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293

		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;
}

294
295
296
297
/*
 * Convert the address within this vma to the page offset within
 * the mapping, in pagecache page units; huge pages here.
 */
298
299
static pgoff_t vma_hugecache_offset(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
300
{
301
302
	return ((address - vma->vm_start) >> huge_page_shift(h)) +
			(vma->vm_pgoff >> huge_page_order(h));
303
304
}

305
306
307
308
309
310
pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
				     unsigned long address)
{
	return vma_hugecache_offset(hstate_vma(vma), vma, address);
}

311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
/*
 * Return the size of the pages allocated when backing a VMA. In the majority
 * cases this will be same size as used by the page table entries.
 */
unsigned long vma_kernel_pagesize(struct vm_area_struct *vma)
{
	struct hstate *hstate;

	if (!is_vm_hugetlb_page(vma))
		return PAGE_SIZE;

	hstate = hstate_vma(vma);

	return 1UL << (hstate->order + PAGE_SHIFT);
}
326
EXPORT_SYMBOL_GPL(vma_kernel_pagesize);
327

328
329
330
331
332
333
334
335
336
337
338
339
340
/*
 * Return the page size being used by the MMU to back a VMA. In the majority
 * of cases, the page size used by the kernel matches the MMU size. On
 * architectures where it differs, an architecture-specific version of this
 * function is required.
 */
#ifndef vma_mmu_pagesize
unsigned long vma_mmu_pagesize(struct vm_area_struct *vma)
{
	return vma_kernel_pagesize(vma);
}
#endif

341
342
343
344
345
346
347
/*
 * 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)
348
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED)
349

350
351
352
353
354
355
356
357
358
/*
 * 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.
359
360
361
362
363
364
365
366
367
 *
 * 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.
368
 */
369
370
371
372
373
374
375
376
377
378
379
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;
}

380
381
382
383
384
struct resv_map {
	struct kref refs;
	struct list_head regions;
};

385
static struct resv_map *resv_map_alloc(void)
386
387
388
389
390
391
392
393
394
395
396
{
	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;
}

397
static void resv_map_release(struct kref *ref)
398
399
400
401
402
403
404
405
406
{
	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)
407
408
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
409
	if (!(vma->vm_flags & VM_MAYSHARE))
410
411
		return (struct resv_map *)(get_vma_private_data(vma) &
							~HPAGE_RESV_MASK);
412
	return NULL;
413
414
}

415
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map)
416
417
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
418
	VM_BUG_ON(vma->vm_flags & VM_MAYSHARE);
419

420
421
	set_vma_private_data(vma, (get_vma_private_data(vma) &
				HPAGE_RESV_MASK) | (unsigned long)map);
422
423
424
425
426
}

static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
427
	VM_BUG_ON(vma->vm_flags & VM_MAYSHARE);
428
429

	set_vma_private_data(vma, get_vma_private_data(vma) | flags);
430
431
432
433
434
}

static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
435
436

	return (get_vma_private_data(vma) & flag) != 0;
437
438
439
}

/* Decrement the reserved pages in the hugepage pool by one */
440
441
static void decrement_hugepage_resv_vma(struct hstate *h,
			struct vm_area_struct *vma)
442
{
443
444
445
	if (vma->vm_flags & VM_NORESERVE)
		return;

446
	if (vma->vm_flags & VM_MAYSHARE) {
447
		/* Shared mappings always use reserves */
448
		h->resv_huge_pages--;
449
	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
450
451
452
453
		/*
		 * Only the process that called mmap() has reserves for
		 * private mappings.
		 */
454
		h->resv_huge_pages--;
455
456
457
	}
}

458
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */
459
460
461
void reset_vma_resv_huge_pages(struct vm_area_struct *vma)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
462
	if (!(vma->vm_flags & VM_MAYSHARE))
463
464
465
466
		vma->vm_private_data = (void *)0;
}

/* Returns true if the VMA has associated reserve pages */
467
static int vma_has_reserves(struct vm_area_struct *vma)
468
{
469
	if (vma->vm_flags & VM_MAYSHARE)
470
471
472
473
		return 1;
	if (is_vma_resv_set(vma, HPAGE_RESV_OWNER))
		return 1;
	return 0;
474
475
}

476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
static void copy_gigantic_page(struct page *dst, struct page *src)
{
	int i;
	struct hstate *h = page_hstate(src);
	struct page *dst_base = dst;
	struct page *src_base = src;

	for (i = 0; i < pages_per_huge_page(h); ) {
		cond_resched();
		copy_highpage(dst, src);

		i++;
		dst = mem_map_next(dst, dst_base, i);
		src = mem_map_next(src, src_base, i);
	}
}

void copy_huge_page(struct page *dst, struct page *src)
{
	int i;
	struct hstate *h = page_hstate(src);

	if (unlikely(pages_per_huge_page(h) > MAX_ORDER_NR_PAGES)) {
		copy_gigantic_page(dst, src);
		return;
	}

	might_sleep();
	for (i = 0; i < pages_per_huge_page(h); i++) {
		cond_resched();
		copy_highpage(dst + i, src + i);
	}
}

510
static void enqueue_huge_page(struct hstate *h, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
511
512
{
	int nid = page_to_nid(page);
513
	list_move(&page->lru, &h->hugepage_freelists[nid]);
514
515
	h->free_huge_pages++;
	h->free_huge_pages_node[nid]++;
Linus Torvalds's avatar
Linus Torvalds committed
516
517
}

518
519
520
521
522
523
524
static struct page *dequeue_huge_page_node(struct hstate *h, int nid)
{
	struct page *page;

	if (list_empty(&h->hugepage_freelists[nid]))
		return NULL;
	page = list_entry(h->hugepage_freelists[nid].next, struct page, lru);
525
	list_move(&page->lru, &h->hugepage_activelist);
526
	set_page_refcounted(page);
527
528
529
530
531
	h->free_huge_pages--;
	h->free_huge_pages_node[nid]--;
	return page;
}

532
533
static struct page *dequeue_huge_page_vma(struct hstate *h,
				struct vm_area_struct *vma,
534
				unsigned long address, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
535
{
536
	struct page *page = NULL;
537
	struct mempolicy *mpol;
538
	nodemask_t *nodemask;
539
	struct zonelist *zonelist;
540
541
	struct zone *zone;
	struct zoneref *z;
542
	unsigned int cpuset_mems_cookie;
Linus Torvalds's avatar
Linus Torvalds committed
543

544
545
retry_cpuset:
	cpuset_mems_cookie = get_mems_allowed();
546
547
	zonelist = huge_zonelist(vma, address,
					htlb_alloc_mask, &mpol, &nodemask);
548
549
550
551
552
	/*
	 * 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
	 */
553
	if (!vma_has_reserves(vma) &&
554
			h->free_huge_pages - h->resv_huge_pages == 0)
555
		goto err;
556

557
	/* If reserves cannot be used, ensure enough pages are in the pool */
558
	if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0)
559
		goto err;
560

561
562
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						MAX_NR_ZONES - 1, nodemask) {
563
564
565
566
567
568
569
		if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask)) {
			page = dequeue_huge_page_node(h, zone_to_nid(zone));
			if (page) {
				if (!avoid_reserve)
					decrement_hugepage_resv_vma(h, vma);
				break;
			}
570
		}
Linus Torvalds's avatar
Linus Torvalds committed
571
	}
572

573
	mpol_cond_put(mpol);
574
575
	if (unlikely(!put_mems_allowed(cpuset_mems_cookie) && !page))
		goto retry_cpuset;
Linus Torvalds's avatar
Linus Torvalds committed
576
	return page;
577
578
579
580

err:
	mpol_cond_put(mpol);
	return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
581
582
}

583
static void update_and_free_page(struct hstate *h, struct page *page)
584
585
{
	int i;
586

587
588
	VM_BUG_ON(h->order >= MAX_ORDER);

589
590
591
	h->nr_huge_pages--;
	h->nr_huge_pages_node[page_to_nid(page)]--;
	for (i = 0; i < pages_per_huge_page(h); i++) {
592
593
594
595
		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);
596
597
598
	}
	set_compound_page_dtor(page, NULL);
	set_page_refcounted(page);
599
	arch_release_hugepage(page);
600
	__free_pages(page, huge_page_order(h));
601
602
}

603
604
605
606
607
608
609
610
611
612
613
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;
}

614
615
static void free_huge_page(struct page *page)
{
616
617
618
619
	/*
	 * Can't pass hstate in here because it is called from the
	 * compound page destructor.
	 */
620
	struct hstate *h = page_hstate(page);
621
	int nid = page_to_nid(page);
622
623
	struct hugepage_subpool *spool =
		(struct hugepage_subpool *)page_private(page);
624

625
	set_page_private(page, 0);
626
	page->mapping = NULL;
627
	BUG_ON(page_count(page));
628
	BUG_ON(page_mapcount(page));
629
630

	spin_lock(&hugetlb_lock);
631
	if (h->surplus_huge_pages_node[nid] && huge_page_order(h) < MAX_ORDER) {
632
633
		/* remove the page from active list */
		list_del(&page->lru);
634
635
636
		update_and_free_page(h, page);
		h->surplus_huge_pages--;
		h->surplus_huge_pages_node[nid]--;
637
	} else {
638
		enqueue_huge_page(h, page);
639
	}
640
	spin_unlock(&hugetlb_lock);
641
	hugepage_subpool_put_pages(spool, 1);
642
643
}

644
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid)
645
{
646
	INIT_LIST_HEAD(&page->lru);
647
648
	set_compound_page_dtor(page, free_huge_page);
	spin_lock(&hugetlb_lock);
649
650
	h->nr_huge_pages++;
	h->nr_huge_pages_node[nid]++;
651
652
653
654
	spin_unlock(&hugetlb_lock);
	put_page(page); /* free it into the hugepage allocator */
}

655
656
657
658
659
660
661
662
663
664
665
static void prep_compound_gigantic_page(struct page *page, unsigned long order)
{
	int i;
	int nr_pages = 1 << order;
	struct page *p = page + 1;

	/* we rely on prep_new_huge_page to set the destructor */
	set_compound_order(page, order);
	__SetPageHead(page);
	for (i = 1; i < nr_pages; i++, p = mem_map_next(p, page, i)) {
		__SetPageTail(p);
666
		set_page_count(p, 0);
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
		p->first_page = page;
	}
}

int PageHuge(struct page *page)
{
	compound_page_dtor *dtor;

	if (!PageCompound(page))
		return 0;

	page = compound_head(page);
	dtor = get_compound_page_dtor(page);

	return dtor == free_huge_page;
}
683
684
EXPORT_SYMBOL_GPL(PageHuge);

685
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid)
Linus Torvalds's avatar
Linus Torvalds committed
686
687
{
	struct page *page;
688

689
690
691
	if (h->order >= MAX_ORDER)
		return NULL;

692
	page = alloc_pages_exact_node(nid,
693
694
		htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
						__GFP_REPEAT|__GFP_NOWARN,
695
		huge_page_order(h));
Linus Torvalds's avatar
Linus Torvalds committed
696
	if (page) {
697
		if (arch_prepare_hugepage(page)) {
698
			__free_pages(page, huge_page_order(h));
699
			return NULL;
700
		}
701
		prep_new_huge_page(h, page, nid);
Linus Torvalds's avatar
Linus Torvalds committed
702
	}
703
704
705
706

	return page;
}

707
/*
708
709
710
711
712
 * common helper functions for hstate_next_node_to_{alloc|free}.
 * We may have allocated or freed a huge page based on a different
 * nodes_allowed previously, so h->next_node_to_{alloc|free} might
 * be outside of *nodes_allowed.  Ensure that we use an allowed
 * node for alloc or free.
713
 */
714
static int next_node_allowed(int nid, nodemask_t *nodes_allowed)
715
{
716
	nid = next_node(nid, *nodes_allowed);
717
	if (nid == MAX_NUMNODES)
718
		nid = first_node(*nodes_allowed);
719
720
721
722
723
	VM_BUG_ON(nid >= MAX_NUMNODES);

	return nid;
}

724
725
726
727
728
729
730
static int get_valid_node_allowed(int nid, nodemask_t *nodes_allowed)
{
	if (!node_isset(nid, *nodes_allowed))
		nid = next_node_allowed(nid, nodes_allowed);
	return nid;
}

731
/*
732
733
734
735
 * returns the previously saved node ["this node"] from which to
 * allocate a persistent huge page for the pool and advance the
 * next node from which to allocate, handling wrap at end of node
 * mask.
736
 */
737
738
static int hstate_next_node_to_alloc(struct hstate *h,
					nodemask_t *nodes_allowed)
739
{
740
741
742
743
744
745
	int nid;

	VM_BUG_ON(!nodes_allowed);

	nid = get_valid_node_allowed(h->next_nid_to_alloc, nodes_allowed);
	h->next_nid_to_alloc = next_node_allowed(nid, nodes_allowed);
746
747

	return nid;
748
749
}

750
static int alloc_fresh_huge_page(struct hstate *h, nodemask_t *nodes_allowed)
751
752
753
754
755
756
{
	struct page *page;
	int start_nid;
	int next_nid;
	int ret = 0;

757
	start_nid = hstate_next_node_to_alloc(h, nodes_allowed);
758
	next_nid = start_nid;
759
760

	do {
761
		page = alloc_fresh_huge_page_node(h, next_nid);
762
		if (page) {
763
			ret = 1;
764
765
			break;
		}
766
		next_nid = hstate_next_node_to_alloc(h, nodes_allowed);
767
	} while (next_nid != start_nid);
768

769
770
771
772
773
	if (ret)
		count_vm_event(HTLB_BUDDY_PGALLOC);
	else
		count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);

774
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
775
776
}

777
/*
778
779
780
781
 * helper for free_pool_huge_page() - return the previously saved
 * node ["this node"] from which to free a huge page.  Advance the
 * next node id whether or not we find a free huge page to free so
 * that the next attempt to free addresses the next node.
782
 */
783
static int hstate_next_node_to_free(struct hstate *h, nodemask_t *nodes_allowed)
784
{
785
786
787
788
789
790
	int nid;

	VM_BUG_ON(!nodes_allowed);

	nid = get_valid_node_allowed(h->next_nid_to_free, nodes_allowed);
	h->next_nid_to_free = next_node_allowed(nid, nodes_allowed);
791
792

	return nid;
793
794
795
796
797
798
799
800
}

/*
 * Free huge page from pool from next node to free.
 * Attempt to keep persistent huge pages more or less
 * balanced over allowed nodes.
 * Called with hugetlb_lock locked.
 */
801
802
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed,
							 bool acct_surplus)
803
804
805
806
807
{
	int start_nid;
	int next_nid;
	int ret = 0;

808
	start_nid = hstate_next_node_to_free(h, nodes_allowed);
809
810
811
	next_nid = start_nid;

	do {
812
813
814
815
816
817
		/*
		 * If we're returning unused surplus pages, only examine
		 * nodes with surplus pages.
		 */
		if ((!acct_surplus || h->surplus_huge_pages_node[next_nid]) &&
		    !list_empty(&h->hugepage_freelists[next_nid])) {
818
819
820
821
822
823
			struct page *page =
				list_entry(h->hugepage_freelists[next_nid].next,
					  struct page, lru);
			list_del(&page->lru);
			h->free_huge_pages--;
			h->free_huge_pages_node[next_nid]--;
824
825
826
827
			if (acct_surplus) {
				h->surplus_huge_pages--;
				h->surplus_huge_pages_node[next_nid]--;
			}
828
829
			update_and_free_page(h, page);
			ret = 1;
830
			break;
831
		}
832
		next_nid = hstate_next_node_to_free(h, nodes_allowed);
833
	} while (next_nid != start_nid);
834
835
836
837

	return ret;
}

838
static struct page *alloc_buddy_huge_page(struct hstate *h, int nid)
839
840
{
	struct page *page;
841
	unsigned int r_nid;
842

843
844
845
	if (h->order >= MAX_ORDER)
		return NULL;

846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
	/*
	 * 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);
870
	if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) {
871
872
873
		spin_unlock(&hugetlb_lock);
		return NULL;
	} else {
874
875
		h->nr_huge_pages++;
		h->surplus_huge_pages++;
876
877
878
	}
	spin_unlock(&hugetlb_lock);

879
880
881
882
883
884
885
886
	if (nid == NUMA_NO_NODE)
		page = alloc_pages(htlb_alloc_mask|__GFP_COMP|
				   __GFP_REPEAT|__GFP_NOWARN,
				   huge_page_order(h));
	else
		page = alloc_pages_exact_node(nid,
			htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
			__GFP_REPEAT|__GFP_NOWARN, huge_page_order(h));
887

888
889
	if (page && arch_prepare_hugepage(page)) {
		__free_pages(page, huge_page_order(h));
890
		page = NULL;
891
892
	}

893
	spin_lock(&hugetlb_lock);
894
	if (page) {
895
		INIT_LIST_HEAD(&page->lru);
896
		r_nid = page_to_nid(page);
897
		set_compound_page_dtor(page, free_huge_page);
898
899
900
		/*
		 * We incremented the global counters already
		 */
901
902
		h->nr_huge_pages_node[r_nid]++;
		h->surplus_huge_pages_node[r_nid]++;
903
		__count_vm_event(HTLB_BUDDY_PGALLOC);
904
	} else {
905
906
		h->nr_huge_pages--;
		h->surplus_huge_pages--;
907
		__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);
908
	}
909
	spin_unlock(&hugetlb_lock);
910
911
912
913

	return page;
}

914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
/*
 * This allocation function is useful in the context where vma is irrelevant.
 * E.g. soft-offlining uses this function because it only cares physical
 * address of error page.
 */
struct page *alloc_huge_page_node(struct hstate *h, int nid)
{
	struct page *page;

	spin_lock(&hugetlb_lock);
	page = dequeue_huge_page_node(h, nid);
	spin_unlock(&hugetlb_lock);

	if (!page)
		page = alloc_buddy_huge_page(h, nid);

	return page;
}

933
/*
Lucas De Marchi's avatar
Lucas De Marchi committed
934
 * Increase the hugetlb pool such that it can accommodate a reservation
935
936
 * of size 'delta'.
 */
937
static int gather_surplus_pages(struct hstate *h, int delta)
938
939
940
941
942
{
	struct list_head surplus_list;
	struct page *page, *tmp;
	int ret, i;
	int needed, allocated;
943
	bool alloc_ok = true;
944

945
	needed = (h->resv_huge_pages + delta) - h->free_huge_pages;
946
	if (needed <= 0) {
947
		h->resv_huge_pages += delta;
948
		return 0;
949
	}
950
951
952
953
954
955
956
957

	allocated = 0;
	INIT_LIST_HEAD(&surplus_list);

	ret = -ENOMEM;
retry:
	spin_unlock(&hugetlb_lock);
	for (i = 0; i < needed; i++) {
958
		page = alloc_buddy_huge_page(h, NUMA_NO_NODE);
959
960
961
962
		if (!page) {
			alloc_ok = false;
			break;
		}
963
964
		list_add(&page->lru, &surplus_list);
	}
965
	allocated += i;
966
967
968
969
970
971

	/*
	 * 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);
972
973
	needed = (h->resv_huge_pages + delta) -
			(h->free_huge_pages + allocated);
974
975
976
977
978
979
980
981
982
983
	if (needed > 0) {
		if (alloc_ok)
			goto retry;
		/*
		 * We were not able to allocate enough pages to
		 * satisfy the entire reservation so we free what
		 * we've allocated so far.
		 */
		goto free;
	}
984
985
	/*
	 * The surplus_list now contains _at_least_ the number of extra pages
Lucas De Marchi's avatar
Lucas De Marchi committed
986
	 * needed to accommodate the reservation.  Add the appropriate number
987
	 * of pages to the hugetlb pool and free the extras back to the buddy
988
989
990
	 * 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.
991
992
	 */
	needed += allocated;
993
	h->resv_huge_pages += delta;
994
	ret = 0;
995

996
	/* Free the needed pages to the hugetlb pool */
997
	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
998
999
		if ((--needed) < 0)
			break;
1000
1001
1002
1003
1004
1005
		/*
		 * 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));
1006
		enqueue_huge_page(h, page);
1007
	}
1008
free:
1009
	spin_unlock(&hugetlb_lock);
1010
1011
1012
1013

	/* Free unnecessary surplus pages to the buddy allocator */
	if (!list_empty(&surplus_list)) {
		list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
1014
			put_page(page);
1015
		}
1016
	}
1017
	spin_lock(&hugetlb_lock);
1018
1019
1020
1021
1022
1023
1024
1025

	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.
1026
 * Called with hugetlb_lock held.
1027
 */
1028
1029
static void return_unused_surplus_pages(struct hstate *h,
					unsigned long unused_resv_pages)
1030
1031
1032
{
	unsigned long nr_pages;

1033
	/* Uncommit the reservation */
1034
	h->resv_huge_pages -= unused_resv_pages;
1035

1036
1037
1038
1039
	/* Cannot return gigantic pages currently */
	if (h->order >= MAX_ORDER)
		return;

1040
	nr_pages = min(unused_resv_pages, h->surplus_huge_pages);
1041

1042
1043
	/*
	 * We want to release as many surplus pages as possible, spread
1044
1045
1046
1047
1048
	 * evenly across all nodes with memory. Iterate across these nodes
	 * until we can no longer free unreserved surplus pages. This occurs
	 * when the nodes with surplus pages have no free pages.
	 * free_pool_huge_page() will balance the the freed pages across the
	 * on-line nodes with memory and will handle the hstate accounting.
1049
1050
	 */
	while (nr_pages--) {
1051
		if (!free_pool_huge_page(h, &node_states[N_HIGH_MEMORY], 1))
1052
			break;
1053
1054
1055
	}
}

1056
1057
1058
/*
 * Determine if the huge page at addr within the vma has an associated
 * reservation.  Where it does not we will need to logically increase
1059
1060
1061
1062
1063
1064
 * reservation and actually increase subpool usage 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 allocated from the subpool and instantiated the change should
 * be committed via vma_commit_reservation.  No action is required on
 * failure.
1065
 */
1066
static long vma_needs_reservation(struct hstate *h,
1067
			struct vm_area_struct *vma, unsigned long addr)
1068
1069
1070
1071
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

1072
	if (vma->vm_flags & VM_MAYSHARE) {
1073
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
1074
1075
1076
		return region_chg(&inode->i_mapping->private_list,
							idx, idx + 1);

1077
1078
	} else if (!is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
		return 1;
1079

1080
	} else  {
1081
		long err;
1082
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
1083
1084
1085
1086
1087
1088
1089
		struct resv_map *reservations = vma_resv_map(vma);

		err = region_chg(&reservations->regions, idx, idx + 1);
		if (err < 0)
			return err;
		return 0;
	}
1090
}
1091
1092
static void vma_commit_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
1093
1094
1095
1096
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

1097
	if (vma->vm_flags & VM_MAYSHARE) {
1098
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
1099
		region_add(&inode->i_mapping->private_list, idx, idx + 1);
1100
1101

	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
1102
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
1103
1104
1105
1106
		struct resv_map *reservations = vma_resv_map(vma);

		/* Mark this page used in the map. */
		region_add(&reservations->regions, idx, idx + 1);
1107
1108
1109
	}
}

1110
static struct page *alloc_huge_page(struct vm_area_struct *vma,
1111
				    unsigned long addr, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
1112
{
1113
	struct hugepage_subpool *spool = subpool_vma(vma);
1114
	struct hstate *h = hstate_vma(vma);
1115
	struct page *page;
1116
	long chg;
1117
1118

	/*
1119
1120
1121
1122
1123
1124
	 * Processes that did not create the mapping will have no
	 * reserves and will not have accounted against subpool
	 * limit. Check that the subpool limit can be made before
	 * satisfying the allocation MAP_NORESERVE mappings may also
	 * need pages and subpool limit allocated allocated if no reserve
	 * mapping overlaps.
1125
	 */
1126
	chg = vma_needs_reservation(h, vma, addr);
1127
	if (chg < 0)
1128
		return ERR_PTR(-ENOMEM);
1129
	if (chg)
1130
		if (hugepage_subpool_get_pages(spool, chg))
1131
			return ERR_PTR(-ENOSPC);
Linus Torvalds's avatar
Linus Torvalds committed
1132
1133

	spin_lock(&hugetlb_lock);
1134
	page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve);
Linus Torvalds's avatar
Linus Torvalds committed
1135
	spin_unlock(&hugetlb_lock);
1136

Ken Chen's avatar
Ken Chen committed
1137
	if (!page) {
1138
		page = alloc_buddy_huge_page(h, NUMA_NO_NODE);
Ken Chen's avatar
Ken Chen committed
1139
		if (!page) {
1140
			hugepage_subpool_put_pages(spool, chg);
1141
			return ERR_PTR(-ENOSPC);
Ken Chen's avatar
Ken Chen committed
1142
1143
		}
	}
1144

1145
	set_page_private(page, (unsigned long)spool);
1146

1147
	vma_commit_reservation(h, vma, addr);
1148

1149
	return page;
1150
1151
}

1152
int __weak alloc_bootmem_huge_page(struct hstate *h)
1153
1154
{
	struct huge_bootmem_page *m;
1155
	int nr_nodes = nodes_weight(node_states[N_HIGH_MEMORY]);
1156
1157
1158
1159
1160

	while (nr_nodes) {
		void *addr;

		addr = __alloc_bootmem_node_nopanic(
1161
				NODE_DATA(hstate_next_node_to_alloc(h,
1162
						&node_states[N_HIGH_MEMORY])),
1163
1164
1165
1166
1167
1168
1169
1170
1171
				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;
1172
			goto found;
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185