hugetlb.c 72.1 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

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

#include <linux/hugetlb.h>
27
#include <linux/node.h>
28
#include "internal.h"
Linus Torvalds's avatar
Linus Torvalds committed
29
30

const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
31
32
static gfp_t htlb_alloc_mask = GFP_HIGHUSER;
unsigned long hugepages_treat_as_movable;
33

34
35
36
37
static int max_hstate;
unsigned int default_hstate_idx;
struct hstate hstates[HUGE_MAX_HSTATE];

38
39
__initdata LIST_HEAD(huge_boot_pages);

40
41
42
/* for command line parsing */
static struct hstate * __initdata parsed_hstate;
static unsigned long __initdata default_hstate_max_huge_pages;
43
static unsigned long __initdata default_hstate_size;
44
45
46

#define for_each_hstate(h) \
	for ((h) = hstates; (h) < &hstates[max_hstate]; (h)++)
47

48
49
50
51
/*
 * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages
 */
static DEFINE_SPINLOCK(hugetlb_lock);
52

53
54
55
/*
 * Region tracking -- allows tracking of reservations and instantiated pages
 *                    across the pages in a mapping.
56
57
58
59
60
61
62
63
64
65
 *
 * 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);
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
 */
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;
}

188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
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;
}

212
213
214
215
/*
 * Convert the address within this vma to the page offset within
 * the mapping, in pagecache page units; huge pages here.
 */
216
217
static pgoff_t vma_hugecache_offset(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
218
{
219
220
	return ((address - vma->vm_start) >> huge_page_shift(h)) +
			(vma->vm_pgoff >> huge_page_order(h));
221
222
}

223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
/*
 * 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);
}
238
EXPORT_SYMBOL_GPL(vma_kernel_pagesize);
239

240
241
242
243
244
245
246
247
248
249
250
251
252
/*
 * 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

253
254
255
256
257
258
259
/*
 * 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)
260
#define HPAGE_RESV_MASK (HPAGE_RESV_OWNER | HPAGE_RESV_UNMAPPED)
261

262
263
264
265
266
267
268
269
270
/*
 * 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.
271
272
273
274
275
276
277
278
279
 *
 * 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.
280
 */
281
282
283
284
285
286
287
288
289
290
291
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;
}

292
293
294
295
296
struct resv_map {
	struct kref refs;
	struct list_head regions;
};

297
static struct resv_map *resv_map_alloc(void)
298
299
300
301
302
303
304
305
306
307
308
{
	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;
}

309
static void resv_map_release(struct kref *ref)
310
311
312
313
314
315
316
317
318
{
	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)
319
320
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
321
	if (!(vma->vm_flags & VM_MAYSHARE))
322
323
		return (struct resv_map *)(get_vma_private_data(vma) &
							~HPAGE_RESV_MASK);
324
	return NULL;
325
326
}

327
static void set_vma_resv_map(struct vm_area_struct *vma, struct resv_map *map)
328
329
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
330
	VM_BUG_ON(vma->vm_flags & VM_MAYSHARE);
331

332
333
	set_vma_private_data(vma, (get_vma_private_data(vma) &
				HPAGE_RESV_MASK) | (unsigned long)map);
334
335
336
337
338
}

static void set_vma_resv_flags(struct vm_area_struct *vma, unsigned long flags)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
339
	VM_BUG_ON(vma->vm_flags & VM_MAYSHARE);
340
341

	set_vma_private_data(vma, get_vma_private_data(vma) | flags);
342
343
344
345
346
}

static int is_vma_resv_set(struct vm_area_struct *vma, unsigned long flag)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
347
348

	return (get_vma_private_data(vma) & flag) != 0;
349
350
351
}

/* Decrement the reserved pages in the hugepage pool by one */
352
353
static void decrement_hugepage_resv_vma(struct hstate *h,
			struct vm_area_struct *vma)
354
{
355
356
357
	if (vma->vm_flags & VM_NORESERVE)
		return;

358
	if (vma->vm_flags & VM_MAYSHARE) {
359
		/* Shared mappings always use reserves */
360
		h->resv_huge_pages--;
361
	} else if (is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
362
363
364
365
		/*
		 * Only the process that called mmap() has reserves for
		 * private mappings.
		 */
366
		h->resv_huge_pages--;
367
368
369
	}
}

370
/* Reset counters to 0 and clear all HPAGE_RESV_* flags */
371
372
373
void reset_vma_resv_huge_pages(struct vm_area_struct *vma)
{
	VM_BUG_ON(!is_vm_hugetlb_page(vma));
374
	if (!(vma->vm_flags & VM_MAYSHARE))
375
376
377
378
		vma->vm_private_data = (void *)0;
}

/* Returns true if the VMA has associated reserve pages */
379
static int vma_has_reserves(struct vm_area_struct *vma)
380
{
381
	if (vma->vm_flags & VM_MAYSHARE)
382
383
384
385
		return 1;
	if (is_vma_resv_set(vma, HPAGE_RESV_OWNER))
		return 1;
	return 0;
386
387
}

388
389
390
391
392
393
394
395
396
397
398
399
static void clear_gigantic_page(struct page *page,
			unsigned long addr, unsigned long sz)
{
	int i;
	struct page *p = page;

	might_sleep();
	for (i = 0; i < sz/PAGE_SIZE; i++, p = mem_map_next(p, page, i)) {
		cond_resched();
		clear_user_highpage(p, addr + i * PAGE_SIZE);
	}
}
400
401
static void clear_huge_page(struct page *page,
			unsigned long addr, unsigned long sz)
402
403
404
{
	int i;

405
	if (unlikely(sz/PAGE_SIZE > MAX_ORDER_NR_PAGES)) {
Hannes Eder's avatar
Hannes Eder committed
406
407
408
		clear_gigantic_page(page, addr, sz);
		return;
	}
409

410
	might_sleep();
411
	for (i = 0; i < sz/PAGE_SIZE; i++) {
412
		cond_resched();
413
		clear_user_highpage(page + i, addr + i * PAGE_SIZE);
414
415
416
	}
}

417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
static void copy_gigantic_page(struct page *dst, struct page *src,
			   unsigned long addr, struct vm_area_struct *vma)
{
	int i;
	struct hstate *h = hstate_vma(vma);
	struct page *dst_base = dst;
	struct page *src_base = src;
	might_sleep();
	for (i = 0; i < pages_per_huge_page(h); ) {
		cond_resched();
		copy_user_highpage(dst, src, addr + i*PAGE_SIZE, vma);

		i++;
		dst = mem_map_next(dst, dst_base, i);
		src = mem_map_next(src, src_base, i);
	}
}
434
static void copy_huge_page(struct page *dst, struct page *src,
435
			   unsigned long addr, struct vm_area_struct *vma)
436
437
{
	int i;
438
	struct hstate *h = hstate_vma(vma);
439

Hannes Eder's avatar
Hannes Eder committed
440
441
442
443
	if (unlikely(pages_per_huge_page(h) > MAX_ORDER_NR_PAGES)) {
		copy_gigantic_page(dst, src, addr, vma);
		return;
	}
444

445
	might_sleep();
446
	for (i = 0; i < pages_per_huge_page(h); i++) {
447
		cond_resched();
448
		copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma);
449
450
451
	}
}

452
static void enqueue_huge_page(struct hstate *h, struct page *page)
Linus Torvalds's avatar
Linus Torvalds committed
453
454
{
	int nid = page_to_nid(page);
455
456
457
	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
458
459
}

460
461
static struct page *dequeue_huge_page_vma(struct hstate *h,
				struct vm_area_struct *vma,
462
				unsigned long address, int avoid_reserve)
Linus Torvalds's avatar
Linus Torvalds committed
463
{
464
	int nid;
Linus Torvalds's avatar
Linus Torvalds committed
465
	struct page *page = NULL;
466
	struct mempolicy *mpol;
467
	nodemask_t *nodemask;
468
	struct zonelist *zonelist = huge_zonelist(vma, address,
469
					htlb_alloc_mask, &mpol, &nodemask);
470
471
	struct zone *zone;
	struct zoneref *z;
Linus Torvalds's avatar
Linus Torvalds committed
472

473
474
475
476
477
	/*
	 * 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
	 */
478
	if (!vma_has_reserves(vma) &&
479
			h->free_huge_pages - h->resv_huge_pages == 0)
480
481
		return NULL;

482
	/* If reserves cannot be used, ensure enough pages are in the pool */
483
	if (avoid_reserve && h->free_huge_pages - h->resv_huge_pages == 0)
484
485
		return NULL;

486
487
	for_each_zone_zonelist_nodemask(zone, z, zonelist,
						MAX_NR_ZONES - 1, nodemask) {
488
489
		nid = zone_to_nid(zone);
		if (cpuset_zone_allowed_softwall(zone, htlb_alloc_mask) &&
490
491
		    !list_empty(&h->hugepage_freelists[nid])) {
			page = list_entry(h->hugepage_freelists[nid].next,
492
493
					  struct page, lru);
			list_del(&page->lru);
494
495
			h->free_huge_pages--;
			h->free_huge_pages_node[nid]--;
496
497

			if (!avoid_reserve)
498
				decrement_hugepage_resv_vma(h, vma);
499

Ken Chen's avatar
Ken Chen committed
500
			break;
501
		}
Linus Torvalds's avatar
Linus Torvalds committed
502
	}
503
	mpol_cond_put(mpol);
Linus Torvalds's avatar
Linus Torvalds committed
504
505
506
	return page;
}

507
static void update_and_free_page(struct hstate *h, struct page *page)
508
509
{
	int i;
510

511
512
	VM_BUG_ON(h->order >= MAX_ORDER);

513
514
515
	h->nr_huge_pages--;
	h->nr_huge_pages_node[page_to_nid(page)]--;
	for (i = 0; i < pages_per_huge_page(h); i++) {
516
517
518
519
520
521
		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);
522
	arch_release_hugepage(page);
523
	__free_pages(page, huge_page_order(h));
524
525
}

526
527
528
529
530
531
532
533
534
535
536
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;
}

537
538
static void free_huge_page(struct page *page)
{
539
540
541
542
	/*
	 * Can't pass hstate in here because it is called from the
	 * compound page destructor.
	 */
543
	struct hstate *h = page_hstate(page);
544
	int nid = page_to_nid(page);
545
	struct address_space *mapping;
546

547
	mapping = (struct address_space *) page_private(page);
548
	set_page_private(page, 0);
549
	page->mapping = NULL;
550
	BUG_ON(page_count(page));
551
552
553
	INIT_LIST_HEAD(&page->lru);

	spin_lock(&hugetlb_lock);
554
	if (h->surplus_huge_pages_node[nid] && huge_page_order(h) < MAX_ORDER) {
555
556
557
		update_and_free_page(h, page);
		h->surplus_huge_pages--;
		h->surplus_huge_pages_node[nid]--;
558
	} else {
559
		enqueue_huge_page(h, page);
560
	}
561
	spin_unlock(&hugetlb_lock);
562
	if (mapping)
563
		hugetlb_put_quota(mapping, 1);
564
565
}

566
static void prep_new_huge_page(struct hstate *h, struct page *page, int nid)
567
568
569
{
	set_compound_page_dtor(page, free_huge_page);
	spin_lock(&hugetlb_lock);
570
571
	h->nr_huge_pages++;
	h->nr_huge_pages_node[nid]++;
572
573
574
575
	spin_unlock(&hugetlb_lock);
	put_page(page); /* free it into the hugepage allocator */
}

576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
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);
		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;
}

604
static struct page *alloc_fresh_huge_page_node(struct hstate *h, int nid)
Linus Torvalds's avatar
Linus Torvalds committed
605
606
{
	struct page *page;
607

608
609
610
	if (h->order >= MAX_ORDER)
		return NULL;

611
	page = alloc_pages_exact_node(nid,
612
613
		htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|
						__GFP_REPEAT|__GFP_NOWARN,
614
		huge_page_order(h));
Linus Torvalds's avatar
Linus Torvalds committed
615
	if (page) {
616
		if (arch_prepare_hugepage(page)) {
617
			__free_pages(page, huge_page_order(h));
618
			return NULL;
619
		}
620
		prep_new_huge_page(h, page, nid);
Linus Torvalds's avatar
Linus Torvalds committed
621
	}
622
623
624
625

	return page;
}

626
/*
627
628
629
630
631
 * 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.
632
 */
633
static int next_node_allowed(int nid, nodemask_t *nodes_allowed)
634
{
635
	nid = next_node(nid, *nodes_allowed);
636
	if (nid == MAX_NUMNODES)
637
		nid = first_node(*nodes_allowed);
638
639
640
641
642
	VM_BUG_ON(nid >= MAX_NUMNODES);

	return nid;
}

643
644
645
646
647
648
649
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;
}

650
/*
651
652
653
654
 * 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.
655
 */
656
657
static int hstate_next_node_to_alloc(struct hstate *h,
					nodemask_t *nodes_allowed)
658
{
659
660
661
662
663
664
	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);
665
666

	return nid;
667
668
}

669
static int alloc_fresh_huge_page(struct hstate *h, nodemask_t *nodes_allowed)
670
671
672
673
674
675
{
	struct page *page;
	int start_nid;
	int next_nid;
	int ret = 0;

676
	start_nid = hstate_next_node_to_alloc(h, nodes_allowed);
677
	next_nid = start_nid;
678
679

	do {
680
		page = alloc_fresh_huge_page_node(h, next_nid);
681
		if (page) {
682
			ret = 1;
683
684
			break;
		}
685
		next_nid = hstate_next_node_to_alloc(h, nodes_allowed);
686
	} while (next_nid != start_nid);
687

688
689
690
691
692
	if (ret)
		count_vm_event(HTLB_BUDDY_PGALLOC);
	else
		count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);

693
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
694
695
}

696
/*
697
698
699
700
 * 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.
701
 */
702
static int hstate_next_node_to_free(struct hstate *h, nodemask_t *nodes_allowed)
703
{
704
705
706
707
708
709
	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);
710
711

	return nid;
712
713
714
715
716
717
718
719
}

/*
 * 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.
 */
720
721
static int free_pool_huge_page(struct hstate *h, nodemask_t *nodes_allowed,
							 bool acct_surplus)
722
723
724
725
726
{
	int start_nid;
	int next_nid;
	int ret = 0;

727
	start_nid = hstate_next_node_to_free(h, nodes_allowed);
728
729
730
	next_nid = start_nid;

	do {
731
732
733
734
735
736
		/*
		 * 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])) {
737
738
739
740
741
742
			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]--;
743
744
745
746
			if (acct_surplus) {
				h->surplus_huge_pages--;
				h->surplus_huge_pages_node[next_nid]--;
			}
747
748
			update_and_free_page(h, page);
			ret = 1;
749
			break;
750
		}
751
		next_nid = hstate_next_node_to_free(h, nodes_allowed);
752
	} while (next_nid != start_nid);
753
754
755
756

	return ret;
}

757
758
static struct page *alloc_buddy_huge_page(struct hstate *h,
			struct vm_area_struct *vma, unsigned long address)
759
760
{
	struct page *page;
761
	unsigned int nid;
762

763
764
765
	if (h->order >= MAX_ORDER)
		return NULL;

766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
	/*
	 * 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);
790
	if (h->surplus_huge_pages >= h->nr_overcommit_huge_pages) {
791
792
793
		spin_unlock(&hugetlb_lock);
		return NULL;
	} else {
794
795
		h->nr_huge_pages++;
		h->surplus_huge_pages++;
796
797
798
	}
	spin_unlock(&hugetlb_lock);

799
800
	page = alloc_pages(htlb_alloc_mask|__GFP_COMP|
					__GFP_REPEAT|__GFP_NOWARN,
801
					huge_page_order(h));
802

803
804
805
806
807
	if (page && arch_prepare_hugepage(page)) {
		__free_pages(page, huge_page_order(h));
		return NULL;
	}

808
	spin_lock(&hugetlb_lock);
809
	if (page) {
810
811
812
813
814
815
		/*
		 * 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));
816
		nid = page_to_nid(page);
817
		set_compound_page_dtor(page, free_huge_page);
818
819
820
		/*
		 * We incremented the global counters already
		 */
821
822
		h->nr_huge_pages_node[nid]++;
		h->surplus_huge_pages_node[nid]++;
823
		__count_vm_event(HTLB_BUDDY_PGALLOC);
824
	} else {
825
826
		h->nr_huge_pages--;
		h->surplus_huge_pages--;
827
		__count_vm_event(HTLB_BUDDY_PGALLOC_FAIL);
828
	}
829
	spin_unlock(&hugetlb_lock);
830
831
832
833

	return page;
}

834
835
836
837
/*
 * Increase the hugetlb pool such that it can accomodate a reservation
 * of size 'delta'.
 */
838
static int gather_surplus_pages(struct hstate *h, int delta)
839
840
841
842
843
844
{
	struct list_head surplus_list;
	struct page *page, *tmp;
	int ret, i;
	int needed, allocated;

845
	needed = (h->resv_huge_pages + delta) - h->free_huge_pages;
846
	if (needed <= 0) {
847
		h->resv_huge_pages += delta;
848
		return 0;
849
	}
850
851
852
853
854
855
856
857

	allocated = 0;
	INIT_LIST_HEAD(&surplus_list);

	ret = -ENOMEM;
retry:
	spin_unlock(&hugetlb_lock);
	for (i = 0; i < needed; i++) {
858
		page = alloc_buddy_huge_page(h, NULL, 0);
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
		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);
879
880
	needed = (h->resv_huge_pages + delta) -
			(h->free_huge_pages + allocated);
881
882
883
884
885
886
887
	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
888
889
890
	 * 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.
891
892
	 */
	needed += allocated;
893
	h->resv_huge_pages += delta;
894
895
	ret = 0;
free:
896
	/* Free the needed pages to the hugetlb pool */
897
	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
898
899
		if ((--needed) < 0)
			break;
900
		list_del(&page->lru);
901
		enqueue_huge_page(h, page);
902
903
904
905
906
907
908
	}

	/* 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);
909
			/*
910
911
912
			 * 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
913
914
915
			 * unlocked which is safe because free_huge_page takes
			 * hugetlb_lock before deciding how to free the page.
			 */
916
			free_huge_page(page);
917
		}
918
		spin_lock(&hugetlb_lock);
919
920
921
922
923
924
925
926
927
	}

	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.
928
 * Called with hugetlb_lock held.
929
 */
930
931
static void return_unused_surplus_pages(struct hstate *h,
					unsigned long unused_resv_pages)
932
933
934
{
	unsigned long nr_pages;

935
	/* Uncommit the reservation */
936
	h->resv_huge_pages -= unused_resv_pages;
937

938
939
940
941
	/* Cannot return gigantic pages currently */
	if (h->order >= MAX_ORDER)
		return;

942
	nr_pages = min(unused_resv_pages, h->surplus_huge_pages);
943

944
945
	/*
	 * We want to release as many surplus pages as possible, spread
946
947
948
949
950
	 * 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.
951
952
	 */
	while (nr_pages--) {
953
		if (!free_pool_huge_page(h, &node_states[N_HIGH_MEMORY], 1))
954
			break;
955
956
957
	}
}

958
959
960
961
962
963
964
965
966
/*
 * 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.
 */
967
static long vma_needs_reservation(struct hstate *h,
968
			struct vm_area_struct *vma, unsigned long addr)
969
970
971
972
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

973
	if (vma->vm_flags & VM_MAYSHARE) {
974
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
975
976
977
		return region_chg(&inode->i_mapping->private_list,
							idx, idx + 1);

978
979
	} else if (!is_vma_resv_set(vma, HPAGE_RESV_OWNER)) {
		return 1;
980

981
	} else  {
982
		long err;
983
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
984
985
986
987
988
989
990
		struct resv_map *reservations = vma_resv_map(vma);

		err = region_chg(&reservations->regions, idx, idx + 1);
		if (err < 0)
			return err;
		return 0;
	}
991
}
992
993
static void vma_commit_reservation(struct hstate *h,
			struct vm_area_struct *vma, unsigned long addr)
994
995
996
997
{
	struct address_space *mapping = vma->vm_file->f_mapping;
	struct inode *inode = mapping->host;

998
	if (vma->vm_flags & VM_MAYSHARE) {
999
		pgoff_t idx = vma_hugecache_offset(h, vma, addr);
1000
		region_add(&inode->i_mapping->private_list, idx, idx + 1);
For faster browsing, not all history is shown. View entire blame