hugetlb.c 86.9 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
/*
 * Generic hugetlb support.
3
 * (C) Nadia Yvette Chambers, April 2004
Linus Torvalds's avatar
Linus Torvalds committed
4
5
6
7
8
 */
#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/hugetlb_cgroup.h>
32
#include <linux/node.h>
33
#include "internal.h"
Linus Torvalds's avatar
Linus Torvalds committed
34
35

const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
36
37
static gfp_t htlb_alloc_mask = GFP_HIGHUSER;
unsigned long hugepages_treat_as_movable;
38

39
int hugetlb_max_hstate __read_mostly;
40
41
42
unsigned int default_hstate_idx;
struct hstate hstates[HUGE_MAX_HSTATE];

43
44
__initdata LIST_HEAD(huge_boot_pages);

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

50
51
52
/*
 * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages
 */
53
DEFINE_SPINLOCK(hugetlb_lock);
54

55
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
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)
{
Al Viro's avatar
Al Viro committed
130
	return subpool_inode(file_inode(vma->vm_file));
131
132
}

133
134
135
/*
 * Region tracking -- allows tracking of reservations and instantiated pages
 *                    across the pages in a mapping.
136
137
138
139
140
141
 *
 * 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:
 *
142
 *	down_write(&mm->mmap_sem);
143
 * or
144
145
 *	down_read(&mm->mmap_sem);
 *	mutex_lock(&hugetlb_instantiation_mutex);
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
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
 */
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
226
		/* We overlap with this area, if it extends further than
227
228
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
		 * 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;
}

268
269
270
271
272
273
274
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) {
275
276
		long seg_from;
		long seg_to;
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291

		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;
}

292
293
294
295
/*
 * Convert the address within this vma to the page offset within
 * the mapping, in pagecache page units; huge pages here.
 */
296
297
static pgoff_t vma_hugecache_offset(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
298
{
299
300
	return ((address - vma->vm_start) >> huge_page_shift(h)) +
			(vma->vm_pgoff >> huge_page_order(h));
301
302
}

303
304
305
306
307
308
pgoff_t linear_hugepage_index(struct vm_area_struct *vma,
				     unsigned long address)
{
	return vma_hugecache_offset(hstate_vma(vma), vma, address);
}

309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
/*
 * 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);
}
324
EXPORT_SYMBOL_GPL(vma_kernel_pagesize);
325

326
327
328
329
330
331
332
333
334
335
336
337
338
/*
 * 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

339
340
341
342
343
344
345
/*
 * 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)
346
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED)
347

348
349
350
351
352
353
354
355
356
/*
 * 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.
357
358
359
360
361
362
363
364
365
 *
 * 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.
366
 */
367
368
369
370
371
372
373
374
375
376
377
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;
}

378
379
380
381
382
struct resv_map {
	struct kref refs;
	struct list_head regions;
};

383
static struct resv_map *resv_map_alloc(void)
384
385
386
387
388
389
390
391
392
393
394
{
	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;
}

395
static void resv_map_release(struct kref *ref)
396
397
398
399
400
401
402
403
404
{
	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)
405
406
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
407
	if (!(vma->vm_flags & VM_MAYSHARE))
408
409
		return (struct resv_map *)(get_vma_private_data(vma) &
							~HPAGE_RESV_MASK);
410
	return NULL;
411
412
}

413
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map)
414
415
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
416
	VM_BUG_ON(vma->vm_flags & VM_MAYSHARE);
417

418
419
	set_vma_private_data(vma, (get_vma_private_data(vma) &
				HPAGE_RESV_MASK) | (unsigned long)map);
420
421
422
423
424
}

static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
425
	VM_BUG_ON(vma->vm_flags & VM_MAYSHARE);
426
427

	set_vma_private_data(vma, get_vma_private_data(vma) | flags);
428
429
430
431
432
}

static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
433
434

	return (get_vma_private_data(vma) & flag) != 0;
435
436
437
}

/* Decrement the reserved pages in the hugepage pool by one */
438
439
static void decrement_hugepage_resv_vma(struct hstate *h,
			struct vm_area_struct *vma)
440
{
441
442
443
	if (vma->vm_flags & VM_NORESERVE)
		return;

444
	if (vma->vm_flags & VM_MAYSHARE) {
445
		/* Shared mappings always use reserves */
446
		h->resv_huge_pages--;
447
	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
448
449
450
451
		/*
		 * Only the process that called mmap() has reserves for
		 * private mappings.
		 */
452
		h->resv_huge_pages--;
453
454
455
	}
}

456
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */
457
458
459
void reset_vma_resv_huge_pages(struct vm_area_struct *vma)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
460
	if (!(vma->vm_flags & VM_MAYSHARE))
461
462
463
464
		vma->vm_private_data = (void *)0;
}

/* Returns true if the VMA has associated reserve pages */
465
static int vma_has_reserves(struct vm_area_struct *vma)
466
{
467
	if (vma->vm_flags & VM_MAYSHARE)
468
469
470
471
		return 1;
	if (is_vma_resv_set(vma, HPAGE_RESV_OWNER))
		return 1;
	return 0;
472
473
}

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
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);
	}
}

508
static void enqueue_huge_page(struct hstate *h, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
509
510
{
	int nid = page_to_nid(page);
511
	list_move(&page->lru, &h->hugepage_freelists[nid]);
512
513
	h->free_huge_pages++;
	h->free_huge_pages_node[nid]++;
Linus Torvalds's avatar
Linus Torvalds committed
514
515
}

516
517
518
519
520
521
522
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);
523
	list_move(&page->lru, &h->hugepage_activelist);
524
	set_page_refcounted(page);
525
526
527
528
529
	h->free_huge_pages--;
	h->free_huge_pages_node[nid]--;
	return page;
}

530
531
static struct page *dequeue_huge_page_vma(struct hstate *h,
				struct vm_area_struct *vma,
532
				unsigned long address, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
533
{
534
	struct page *page = NULL;
535
	struct mempolicy *mpol;
536
	nodemask_t *nodemask;
537
	struct zonelist *zonelist;
538
539
	struct zone *zone;
	struct zoneref *z;
540
	unsigned int cpuset_mems_cookie;
Linus Torvalds's avatar
Linus Torvalds committed
541

542
543
retry_cpuset:
	cpuset_mems_cookie = get_mems_allowed();
544
545
	zonelist = huge_zonelist(vma, address,
					htlb_alloc_mask, &mpol, &nodemask);
546
547
548
549
550
	/*
	 * 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
	 */
551
	if (!vma_has_reserves(vma) &&
552
			h->free_huge_pages - h->resv_huge_pages == 0)
553
		goto err;
554

555
	/* If reserves cannot be used, ensure enough pages are in the pool */
556
	if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0)
557
		goto err;
558

559
560
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						MAX_NR_ZONES - 1, nodemask) {
561
562
563
564
565
566
567
		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;
			}
568
		}
Linus Torvalds's avatar
Linus Torvalds committed
569
	}
570

571
	mpol_cond_put(mpol);
572
573
	if (unlikely(!put_mems_allowed(cpuset_mems_cookie) && !page))
		goto retry_cpuset;
Linus Torvalds's avatar
Linus Torvalds committed
574
	return page;
575
576
577
578

err:
	mpol_cond_put(mpol);
	return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
579
580
}

581
static void update_and_free_page(struct hstate *h, struct page *page)
582
583
{
	int i;
584

585
586
	VM_BUG_ON(h->order >= MAX_ORDER);

587
588
589
	h->nr_huge_pages--;
	h->nr_huge_pages_node[page_to_nid(page)]--;
	for (i = 0; i < pages_per_huge_page(h); i++) {
590
591
592
593
		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);
594
	}
595
	VM_BUG_ON(hugetlb_cgroup_from_page(page));
596
597
	set_compound_page_dtor(page, NULL);
	set_page_refcounted(page);
598
	arch_release_hugepage(page);
599
	__free_pages(page, huge_page_order(h));
600
601
}

602
603
604
605
606
607
608
609
610
611
612
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;
}

613
614
static void free_huge_page(struct page *page)
{
615
616
617
618
	/*
	 * Can't pass hstate in here because it is called from the
	 * compound page destructor.
	 */
619
	struct hstate *h = page_hstate(page);
620
	int nid = page_to_nid(page);
621
622
	struct hugepage_subpool *spool =
		(struct hugepage_subpool *)page_private(page);
623

624
	set_page_private(page, 0);
625
	page->mapping = NULL;
626
	BUG_ON(page_count(page));
627
	BUG_ON(page_mapcount(page));
628
629

	spin_lock(&hugetlb_lock);
630
631
	hugetlb_cgroup_uncharge_page(hstate_index(h),
				     pages_per_huge_page(h), page);
632
	if (h->surplus_huge_pages_node[nid] && huge_page_order(h) < MAX_ORDER) {
633
634
		/* remove the page from active list */
		list_del(&page->lru);
635
636
637
		update_and_free_page(h, page);
		h->surplus_huge_pages--;
		h->surplus_huge_pages_node[nid]--;
638
	} else {
639
		arch_clear_hugepage_flags(page);
640
		enqueue_huge_page(h, page);
641
	}
642
	spin_unlock(&hugetlb_lock);
643
	hugepage_subpool_put_pages(spool, 1);
644
645
}

646
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid)
647
{
648
	INIT_LIST_HEAD(&page->lru);
649
650
	set_compound_page_dtor(page, free_huge_page);
	spin_lock(&hugetlb_lock);
651
	set_hugetlb_cgroup(page, NULL);
652
653
	h->nr_huge_pages++;
	h->nr_huge_pages_node[nid]++;
654
655
656
657
	spin_unlock(&hugetlb_lock);
	put_page(page); /* free it into the hugepage allocator */
}

658
659
660
661
662
663
664
665
666
667
668
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);
669
		set_page_count(p, 0);
670
671
672
673
		p->first_page = page;
	}
}

Andrew Morton's avatar
Andrew Morton committed
674
675
676
677
678
/*
 * PageHuge() only returns true for hugetlbfs pages, but not for normal or
 * transparent huge pages.  See the PageTransHuge() documentation for more
 * details.
 */
679
680
681
682
683
684
685
686
687
688
689
690
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;
}
691
692
EXPORT_SYMBOL_GPL(PageHuge);

693
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid)
Linus Torvalds's avatar
Linus Torvalds committed
694
695
{
	struct page *page;
696

697
698
699
	if (h->order >= MAX_ORDER)
		return NULL;

700
	page = alloc_pages_exact_node(nid,
701
702
		htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
						__GFP_REPEAT|__GFP_NOWARN,
703
		huge_page_order(h));
Linus Torvalds's avatar
Linus Torvalds committed
704
	if (page) {
705
		if (arch_prepare_hugepage(page)) {
706
			__free_pages(page, huge_page_order(h));
707
			return NULL;
708
		}
709
		prep_new_huge_page(h, page, nid);
Linus Torvalds's avatar
Linus Torvalds committed
710
	}
711
712
713
714

	return page;
}

715
/*
716
717
718
719
720
 * 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.
721
 */
722
static int next_node_allowed(int nid, nodemask_t *nodes_allowed)
723
{
724
	nid = next_node(nid, *nodes_allowed);
725
	if (nid == MAX_NUMNODES)
726
		nid = first_node(*nodes_allowed);
727
728
729
730
731
	VM_BUG_ON(nid >= MAX_NUMNODES);

	return nid;
}

732
733
734
735
736
737
738
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;
}

739
/*
740
741
742
743
 * 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.
744
 */
745
746
static int hstate_next_node_to_alloc(struct hstate *h,
					nodemask_t *nodes_allowed)
747
{
748
749
750
751
752
753
	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);
754
755

	return nid;
756
757
}

758
static int alloc_fresh_huge_page(struct hstate *h, nodemask_t *nodes_allowed)
759
760
761
762
763
764
{
	struct page *page;
	int start_nid;
	int next_nid;
	int ret = 0;

765
	start_nid = hstate_next_node_to_alloc(h, nodes_allowed);
766
	next_nid = start_nid;
767
768

	do {
769
		page = alloc_fresh_huge_page_node(h, next_nid);
770
		if (page) {
771
			ret = 1;
772
773
			break;
		}
774
		next_nid = hstate_next_node_to_alloc(h, nodes_allowed);
775
	} while (next_nid != start_nid);
776

777
778
779
780
781
	if (ret)
		count_vm_event(HTLB_BUDDY_PGALLOC);
	else
		count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);

782
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
783
784
}

785
/*
786
787
788
789
 * 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.
790
 */
791
static int hstate_next_node_to_free(struct hstate *h, nodemask_t *nodes_allowed)
792
{
793
794
795
796
797
798
	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);
799
800

	return nid;
801
802
803
804
805
806
807
808
}

/*
 * 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.
 */
809
810
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed,
							 bool acct_surplus)
811
812
813
814
815
{
	int start_nid;
	int next_nid;
	int ret = 0;

816
	start_nid = hstate_next_node_to_free(h, nodes_allowed);
817
818
819
	next_nid = start_nid;

	do {
820
821
822
823
824
825
		/*
		 * 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])) {
826
827
828
829
830
831
			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]--;
832
833
834
835
			if (acct_surplus) {
				h->surplus_huge_pages--;
				h->surplus_huge_pages_node[next_nid]--;
			}
836
837
			update_and_free_page(h, page);
			ret = 1;
838
			break;
839
		}
840
		next_nid = hstate_next_node_to_free(h, nodes_allowed);
841
	} while (next_nid != start_nid);
842
843
844
845

	return ret;
}

846
static struct page *alloc_buddy_huge_page(struct hstate *h, int nid)
847
848
{
	struct page *page;
849
	unsigned int r_nid;
850

851
852
853
	if (h->order >= MAX_ORDER)
		return NULL;

854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
	/*
	 * 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);
878
	if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) {
879
880
881
		spin_unlock(&hugetlb_lock);
		return NULL;
	} else {
882
883
		h->nr_huge_pages++;
		h->surplus_huge_pages++;
884
885
886
	}
	spin_unlock(&hugetlb_lock);

887
888
889
890
891
892
893
894
	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));
895

896
897
	if (page && arch_prepare_hugepage(page)) {
		__free_pages(page, huge_page_order(h));
898
		page = NULL;
899
900
	}

901
	spin_lock(&hugetlb_lock);
902
	if (page) {
903
		INIT_LIST_HEAD(&page->lru);
904
		r_nid = page_to_nid(page);
905
		set_compound_page_dtor(page, free_huge_page);
906
		set_hugetlb_cgroup(page, NULL);
907
908
909
		/*
		 * We incremented the global counters already
		 */
910
911
		h->nr_huge_pages_node[r_nid]++;
		h->surplus_huge_pages_node[r_nid]++;
912
		__count_vm_event(HTLB_BUDDY_PGALLOC);
913
	} else {
914
915
		h->nr_huge_pages--;
		h->surplus_huge_pages--;
916
		__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);
917
	}
918
	spin_unlock(&hugetlb_lock);
919
920
921
922

	return page;
}

923
924
925
926
927
928
929
930
931
932
933
934
935
/*
 * 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);

936
	if (!page)
937
938
939
940
941
		page = alloc_buddy_huge_page(h, nid);

	return page;
}

942
/*
Lucas De Marchi's avatar
Lucas De Marchi committed
943
 * Increase the hugetlb pool such that it can accommodate a reservation
944
945
 * of size 'delta'.
 */
946
static int gather_surplus_pages(struct hstate *h, int delta)
947
948
949
950
951
{
	struct list_head surplus_list;
	struct page *page, *tmp;
	int ret, i;
	int needed, allocated;
952
	bool alloc_ok = true;
953

954
	needed = (h->resv_huge_pages + delta) - h->free_huge_pages;
955
	if (needed <= 0) {
956
		h->resv_huge_pages += delta;
957
		return 0;
958
	}
959
960
961
962
963
964
965
966

	allocated = 0;
	INIT_LIST_HEAD(&surplus_list);

	ret = -ENOMEM;
retry:
	spin_unlock(&hugetlb_lock);
	for (i = 0; i < needed; i++) {
967
		page = alloc_buddy_huge_page(h, NUMA_NO_NODE);
968
969
970
971
		if (!page) {
			alloc_ok = false;
			break;
		}
972
973
		list_add(&page->lru, &surplus_list);
	}
974
	allocated += i;
975
976
977
978
979
980

	/*
	 * 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);
981
982
	needed = (h->resv_huge_pages + delta) -
			(h->free_huge_pages + allocated);
983
984
985
986
987
988
989
990
991
992
	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;
	}
993
994
	/*
	 * The surplus_list now contains _at_least_ the number of extra pages
Lucas De Marchi's avatar
Lucas De Marchi committed
995
	 * needed to accommodate the reservation.  Add the appropriate number
996
	 * of pages to the hugetlb pool and free the extras back to the buddy
997
998
999
	 * 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.
1000
	 */
For faster browsing, not all history is shown. View entire blame