hugetlb.c 30.7 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
12
/*
 * Generic hugetlb support.
 * (C) William Irwin, April 2004
 */
#include <linux/gfp.h>
#include <linux/list.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mm.h>
#include <linux/sysctl.h>
#include <linux/highmem.h>
#include <linux/nodemask.h>
David Gibson's avatar
David Gibson committed
13
#include <linux/pagemap.h>
14
#include <linux/mempolicy.h>
15
#include <linux/cpuset.h>
16
#include <linux/mutex.h>
17

David Gibson's avatar
David Gibson committed
18
19
20
21
#include <asm/page.h>
#include <asm/pgtable.h>

#include <linux/hugetlb.h>
22
#include "internal.h"
Linus Torvalds's avatar
Linus Torvalds committed
23
24

const unsigned long hugetlb_zero = 0, hugetlb_infinity = ~0UL;
25
static unsigned long nr_huge_pages, free_huge_pages, resv_huge_pages;
26
static unsigned long surplus_huge_pages;
Linus Torvalds's avatar
Linus Torvalds committed
27
28
29
30
unsigned long max_huge_pages;
static struct list_head hugepage_freelists[MAX_NUMNODES];
static unsigned int nr_huge_pages_node[MAX_NUMNODES];
static unsigned int free_huge_pages_node[MAX_NUMNODES];
31
static unsigned int surplus_huge_pages_node[MAX_NUMNODES];
32
33
static gfp_t htlb_alloc_mask = GFP_HIGHUSER;
unsigned long hugepages_treat_as_movable;
34
int hugetlb_dynamic_pool;
35
unsigned long nr_overcommit_huge_pages;
36
static int hugetlb_next_nid;
37

38
39
40
41
/*
 * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages
 */
static DEFINE_SPINLOCK(hugetlb_lock);
42

43
44
45
46
47
48
49
static void clear_huge_page(struct page *page, unsigned long addr)
{
	int i;

	might_sleep();
	for (i = 0; i < (HPAGE_SIZE/PAGE_SIZE); i++) {
		cond_resched();
50
		clear_user_highpage(page + i, addr + i * PAGE_SIZE);
51
52
53
54
	}
}

static void copy_huge_page(struct page *dst, struct page *src,
55
			   unsigned long addr, struct vm_area_struct *vma)
56
57
58
59
60
61
{
	int i;

	might_sleep();
	for (i = 0; i < HPAGE_SIZE/PAGE_SIZE; i++) {
		cond_resched();
62
		copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma);
63
64
65
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
66
67
68
69
70
71
72
73
static void enqueue_huge_page(struct page *page)
{
	int nid = page_to_nid(page);
	list_add(&page->lru, &hugepage_freelists[nid]);
	free_huge_pages++;
	free_huge_pages_node[nid]++;
}

74
75
static struct page *dequeue_huge_page(struct vm_area_struct *vma,
				unsigned long address)
Linus Torvalds's avatar
Linus Torvalds committed
76
{
77
	int nid;
Linus Torvalds's avatar
Linus Torvalds committed
78
	struct page *page = NULL;
79
	struct mempolicy *mpol;
80
	struct zonelist *zonelist = huge_zonelist(vma, address,
81
					htlb_alloc_mask, &mpol);
82
	struct zone **z;
Linus Torvalds's avatar
Linus Torvalds committed
83

84
	for (z = zonelist->zones; *z; z++) {
85
		nid = zone_to_nid(*z);
86
		if (cpuset_zone_allowed_softwall(*z, htlb_alloc_mask) &&
87
88
89
90
91
92
		    !list_empty(&hugepage_freelists[nid])) {
			page = list_entry(hugepage_freelists[nid].next,
					  struct page, lru);
			list_del(&page->lru);
			free_huge_pages--;
			free_huge_pages_node[nid]--;
93
94
			if (vma && vma->vm_flags & VM_MAYSHARE)
				resv_huge_pages--;
Ken Chen's avatar
Ken Chen committed
95
			break;
96
		}
Linus Torvalds's avatar
Linus Torvalds committed
97
	}
98
	mpol_free(mpol);	/* unref if mpol !NULL */
Linus Torvalds's avatar
Linus Torvalds committed
99
100
101
	return page;
}

102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
static void update_and_free_page(struct page *page)
{
	int i;
	nr_huge_pages--;
	nr_huge_pages_node[page_to_nid(page)]--;
	for (i = 0; i < (HPAGE_SIZE / PAGE_SIZE); i++) {
		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);
	__free_pages(page, HUGETLB_PAGE_ORDER);
}

117
118
static void free_huge_page(struct page *page)
{
119
	int nid = page_to_nid(page);
120
	struct address_space *mapping;
121

122
	mapping = (struct address_space *) page_private(page);
123
	BUG_ON(page_count(page));
124
125
126
	INIT_LIST_HEAD(&page->lru);

	spin_lock(&hugetlb_lock);
127
128
129
130
131
132
133
	if (surplus_huge_pages_node[nid]) {
		update_and_free_page(page);
		surplus_huge_pages--;
		surplus_huge_pages_node[nid]--;
	} else {
		enqueue_huge_page(page);
	}
134
	spin_unlock(&hugetlb_lock);
135
	if (mapping)
136
		hugetlb_put_quota(mapping, 1);
137
	set_page_private(page, 0);
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
/*
 * Increment or decrement surplus_huge_pages.  Keep node-specific counters
 * balanced by operating on them in a round-robin fashion.
 * Returns 1 if an adjustment was made.
 */
static int adjust_pool_surplus(int delta)
{
	static int prev_nid;
	int nid = prev_nid;
	int ret = 0;

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

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

		surplus_huge_pages += delta;
		surplus_huge_pages_node[nid] += delta;
		ret = 1;
		break;
	} while (nid != prev_nid);

	prev_nid = nid;
	return ret;
}

175
static struct page *alloc_fresh_huge_page_node(int nid)
Linus Torvalds's avatar
Linus Torvalds committed
176
177
{
	struct page *page;
178

179
180
181
	page = alloc_pages_node(nid,
		htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|__GFP_NOWARN,
		HUGETLB_PAGE_ORDER);
Linus Torvalds's avatar
Linus Torvalds committed
182
	if (page) {
183
		set_compound_page_dtor(page, free_huge_page);
184
		spin_lock(&hugetlb_lock);
Linus Torvalds's avatar
Linus Torvalds committed
185
		nr_huge_pages++;
186
		nr_huge_pages_node[nid]++;
187
		spin_unlock(&hugetlb_lock);
188
		put_page(page); /* free it into the hugepage allocator */
Linus Torvalds's avatar
Linus Torvalds committed
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

	return page;
}

static int alloc_fresh_huge_page(void)
{
	struct page *page;
	int start_nid;
	int next_nid;
	int ret = 0;

	start_nid = hugetlb_next_nid;

	do {
		page = alloc_fresh_huge_page_node(hugetlb_next_nid);
		if (page)
			ret = 1;
		/*
		 * Use a helper variable to find the next node and then
		 * copy it back to hugetlb_next_nid afterwards:
		 * otherwise there's a window in which a racer might
		 * pass invalid nid MAX_NUMNODES to alloc_pages_node.
		 * But we don't need to use a spin_lock here: it really
		 * doesn't matter if occasionally a racer chooses the
		 * same nid as we do.  Move nid forward in the mask even
		 * if we just successfully allocated a hugepage so that
		 * the next caller gets hugepages on the next node.
		 */
		next_nid = next_node(hugetlb_next_nid, node_online_map);
		if (next_nid == MAX_NUMNODES)
			next_nid = first_node(node_online_map);
		hugetlb_next_nid = next_nid;
	} while (!page && hugetlb_next_nid != start_nid);

	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
225
226
}

227
228
229
230
static struct page *alloc_buddy_huge_page(struct vm_area_struct *vma,
						unsigned long address)
{
	struct page *page;
231
	unsigned int nid;
232

233
234
235
236
	/* Check if the dynamic pool is enabled */
	if (!hugetlb_dynamic_pool)
		return NULL;

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
	/*
	 * 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);
	if (surplus_huge_pages >= nr_overcommit_huge_pages) {
		spin_unlock(&hugetlb_lock);
		return NULL;
	} else {
		nr_huge_pages++;
		surplus_huge_pages++;
	}
	spin_unlock(&hugetlb_lock);

270
271
	page = alloc_pages(htlb_alloc_mask|__GFP_COMP|__GFP_NOWARN,
					HUGETLB_PAGE_ORDER);
272
273

	spin_lock(&hugetlb_lock);
274
	if (page) {
275
		nid = page_to_nid(page);
276
		set_compound_page_dtor(page, free_huge_page);
277
278
279
280
281
282
283
284
		/*
		 * We incremented the global counters already
		 */
		nr_huge_pages_node[nid]++;
		surplus_huge_pages_node[nid]++;
	} else {
		nr_huge_pages--;
		surplus_huge_pages--;
285
	}
286
	spin_unlock(&hugetlb_lock);
287
288
289
290

	return page;
}

291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
/*
 * Increase the hugetlb pool such that it can accomodate a reservation
 * of size 'delta'.
 */
static int gather_surplus_pages(int delta)
{
	struct list_head surplus_list;
	struct page *page, *tmp;
	int ret, i;
	int needed, allocated;

	needed = (resv_huge_pages + delta) - free_huge_pages;
	if (needed <= 0)
		return 0;

	allocated = 0;
	INIT_LIST_HEAD(&surplus_list);

	ret = -ENOMEM;
retry:
	spin_unlock(&hugetlb_lock);
	for (i = 0; i < needed; i++) {
		page = alloc_buddy_huge_page(NULL, 0);
		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);
	needed = (resv_huge_pages + delta) - (free_huge_pages + allocated);
	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
	 * allocator.
	 */
	needed += allocated;
	ret = 0;
free:
	list_for_each_entry_safe(page, tmp, &surplus_list, lru) {
		list_del(&page->lru);
		if ((--needed) >= 0)
			enqueue_huge_page(page);
351
352
353
354
355
356
357
358
359
360
361
		else {
			/*
			 * Decrement the refcount and free the page using its
			 * destructor.  This must be done with hugetlb_lock
			 * unlocked which is safe because free_huge_page takes
			 * hugetlb_lock before deciding how to free the page.
			 */
			spin_unlock(&hugetlb_lock);
			put_page(page);
			spin_lock(&hugetlb_lock);
		}
362
363
364
365
366
367
368
369
370
371
	}

	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.
 */
372
static void return_unused_surplus_pages(unsigned long unused_resv_pages)
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
{
	static int nid = -1;
	struct page *page;
	unsigned long nr_pages;

	nr_pages = min(unused_resv_pages, surplus_huge_pages);

	while (nr_pages) {
		nid = next_node(nid, node_online_map);
		if (nid == MAX_NUMNODES)
			nid = first_node(node_online_map);

		if (!surplus_huge_pages_node[nid])
			continue;

		if (!list_empty(&hugepage_freelists[nid])) {
			page = list_entry(hugepage_freelists[nid].next,
					  struct page, lru);
			list_del(&page->lru);
			update_and_free_page(page);
			free_huge_pages--;
			free_huge_pages_node[nid]--;
			surplus_huge_pages--;
			surplus_huge_pages_node[nid]--;
			nr_pages--;
		}
	}
}

402
403
404

static struct page *alloc_huge_page_shared(struct vm_area_struct *vma,
						unsigned long addr)
Linus Torvalds's avatar
Linus Torvalds committed
405
{
406
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
407
408

	spin_lock(&hugetlb_lock);
409
	page = dequeue_huge_page(vma, addr);
Linus Torvalds's avatar
Linus Torvalds committed
410
	spin_unlock(&hugetlb_lock);
411
	return page ? page : ERR_PTR(-VM_FAULT_OOM);
412
}
413

414
415
416
417
static struct page *alloc_huge_page_private(struct vm_area_struct *vma,
						unsigned long addr)
{
	struct page *page = NULL;
418

419
420
421
	if (hugetlb_get_quota(vma->vm_file->f_mapping, 1))
		return ERR_PTR(-VM_FAULT_SIGBUS);

422
423
424
425
426
	spin_lock(&hugetlb_lock);
	if (free_huge_pages > resv_huge_pages)
		page = dequeue_huge_page(vma, addr);
	spin_unlock(&hugetlb_lock);
	if (!page)
427
		page = alloc_buddy_huge_page(vma, addr);
428
	return page ? page : ERR_PTR(-VM_FAULT_OOM);
429
430
431
432
433
434
}

static struct page *alloc_huge_page(struct vm_area_struct *vma,
				    unsigned long addr)
{
	struct page *page;
435
436
	struct address_space *mapping = vma->vm_file->f_mapping;

437
438
439
440
	if (vma->vm_flags & VM_MAYSHARE)
		page = alloc_huge_page_shared(vma, addr);
	else
		page = alloc_huge_page_private(vma, addr);
441
442

	if (!IS_ERR(page)) {
443
		set_page_refcounted(page);
444
		set_page_private(page, (unsigned long) mapping);
445
446
	}
	return page;
447
448
}

Linus Torvalds's avatar
Linus Torvalds committed
449
450
451
452
static int __init hugetlb_init(void)
{
	unsigned long i;

453
454
455
	if (HPAGE_SHIFT == 0)
		return 0;

Linus Torvalds's avatar
Linus Torvalds committed
456
457
458
	for (i = 0; i < MAX_NUMNODES; ++i)
		INIT_LIST_HEAD(&hugepage_freelists[i]);

459
460
	hugetlb_next_nid = first_node(node_online_map);

Linus Torvalds's avatar
Linus Torvalds committed
461
	for (i = 0; i < max_huge_pages; ++i) {
462
		if (!alloc_fresh_huge_page())
Linus Torvalds's avatar
Linus Torvalds committed
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
			break;
	}
	max_huge_pages = free_huge_pages = nr_huge_pages = i;
	printk("Total HugeTLB memory allocated, %ld\n", free_huge_pages);
	return 0;
}
module_init(hugetlb_init);

static int __init hugetlb_setup(char *s)
{
	if (sscanf(s, "%lu", &max_huge_pages) <= 0)
		max_huge_pages = 0;
	return 1;
}
__setup("hugepages=", hugetlb_setup);

479
480
481
482
483
484
485
486
487
488
489
static unsigned int cpuset_mems_nr(unsigned int *array)
{
	int node;
	unsigned int nr = 0;

	for_each_node_mask(node, cpuset_current_mems_allowed)
		nr += array[node];

	return nr;
}

Linus Torvalds's avatar
Linus Torvalds committed
490
491
492
493
#ifdef CONFIG_SYSCTL
#ifdef CONFIG_HIGHMEM
static void try_to_free_low(unsigned long count)
{
494
495
	int i;

Linus Torvalds's avatar
Linus Torvalds committed
496
497
498
	for (i = 0; i < MAX_NUMNODES; ++i) {
		struct page *page, *next;
		list_for_each_entry_safe(page, next, &hugepage_freelists[i], lru) {
499
500
			if (count >= nr_huge_pages)
				return;
Linus Torvalds's avatar
Linus Torvalds committed
501
502
503
504
505
			if (PageHighMem(page))
				continue;
			list_del(&page->lru);
			update_and_free_page(page);
			free_huge_pages--;
506
			free_huge_pages_node[page_to_nid(page)]--;
Linus Torvalds's avatar
Linus Torvalds committed
507
508
509
510
511
512
513
514
515
		}
	}
}
#else
static inline void try_to_free_low(unsigned long count)
{
}
#endif

516
#define persistent_huge_pages (nr_huge_pages - surplus_huge_pages)
Linus Torvalds's avatar
Linus Torvalds committed
517
518
static unsigned long set_max_huge_pages(unsigned long count)
{
519
	unsigned long min_count, ret;
Linus Torvalds's avatar
Linus Torvalds committed
520

521
522
523
524
	/*
	 * Increase the pool size
	 * First take pages out of surplus state.  Then make up the
	 * remaining difference by allocating fresh huge pages.
525
526
527
528
529
530
	 *
	 * We might race with alloc_buddy_huge_page() here and be unable
	 * to convert a surplus huge page to a normal huge page. That is
	 * not critical, though, it just means the overall size of the
	 * pool might be one hugepage larger than it needs to be, but
	 * within all the constraints specified by the sysctls.
531
	 */
Linus Torvalds's avatar
Linus Torvalds committed
532
	spin_lock(&hugetlb_lock);
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
	while (surplus_huge_pages && count > persistent_huge_pages) {
		if (!adjust_pool_surplus(-1))
			break;
	}

	while (count > persistent_huge_pages) {
		int ret;
		/*
		 * If this allocation races such that we no longer need the
		 * page, free_huge_page will handle it by freeing the page
		 * and reducing the surplus.
		 */
		spin_unlock(&hugetlb_lock);
		ret = alloc_fresh_huge_page();
		spin_lock(&hugetlb_lock);
		if (!ret)
			goto out;

	}

	/*
	 * Decrease the pool size
	 * First return free pages to the buddy allocator (being careful
	 * to keep enough around to satisfy reservations).  Then place
	 * pages into surplus state as needed so the pool will shrink
	 * to the desired size as pages become free.
559
560
561
562
563
564
565
566
	 *
	 * By placing pages into the surplus state independent of the
	 * overcommit value, we are allowing the surplus pool size to
	 * exceed overcommit. There are few sane options here. Since
	 * alloc_buddy_huge_page() is checking the global counter,
	 * though, we'll note that we're not allowed to exceed surplus
	 * and won't grow the pool anywhere else. Not until one of the
	 * sysctls are changed, or the surplus pages go out of use.
567
	 */
568
569
	min_count = resv_huge_pages + nr_huge_pages - free_huge_pages;
	min_count = max(count, min_count);
570
571
	try_to_free_low(min_count);
	while (min_count < persistent_huge_pages) {
572
		struct page *page = dequeue_huge_page(NULL, 0);
Linus Torvalds's avatar
Linus Torvalds committed
573
574
575
576
		if (!page)
			break;
		update_and_free_page(page);
	}
577
578
579
580
581
582
	while (count < persistent_huge_pages) {
		if (!adjust_pool_surplus(1))
			break;
	}
out:
	ret = persistent_huge_pages;
Linus Torvalds's avatar
Linus Torvalds committed
583
	spin_unlock(&hugetlb_lock);
584
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
585
586
587
588
589
590
591
592
593
594
}

int hugetlb_sysctl_handler(struct ctl_table *table, int write,
			   struct file *file, void __user *buffer,
			   size_t *length, loff_t *ppos)
{
	proc_doulongvec_minmax(table, write, file, buffer, length, ppos);
	max_huge_pages = set_max_huge_pages(max_huge_pages);
	return 0;
}
595
596
597
598
599
600
601
602
603
604
605
606
607

int hugetlb_treat_movable_handler(struct ctl_table *table, int write,
			struct file *file, void __user *buffer,
			size_t *length, loff_t *ppos)
{
	proc_dointvec(table, write, file, buffer, length, ppos);
	if (hugepages_treat_as_movable)
		htlb_alloc_mask = GFP_HIGHUSER_MOVABLE;
	else
		htlb_alloc_mask = GFP_HIGHUSER;
	return 0;
}

Linus Torvalds's avatar
Linus Torvalds committed
608
609
610
611
612
613
614
#endif /* CONFIG_SYSCTL */

int hugetlb_report_meminfo(char *buf)
{
	return sprintf(buf,
			"HugePages_Total: %5lu\n"
			"HugePages_Free:  %5lu\n"
615
			"HugePages_Rsvd:  %5lu\n"
616
			"HugePages_Surp:  %5lu\n"
Linus Torvalds's avatar
Linus Torvalds committed
617
618
619
			"Hugepagesize:    %5lu kB\n",
			nr_huge_pages,
			free_huge_pages,
620
			resv_huge_pages,
621
			surplus_huge_pages,
Linus Torvalds's avatar
Linus Torvalds committed
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
			HPAGE_SIZE/1024);
}

int hugetlb_report_node_meminfo(int nid, char *buf)
{
	return sprintf(buf,
		"Node %d HugePages_Total: %5u\n"
		"Node %d HugePages_Free:  %5u\n",
		nid, nr_huge_pages_node[nid],
		nid, free_huge_pages_node[nid]);
}

/* Return the number pages of memory we physically have, in PAGE_SIZE units. */
unsigned long hugetlb_total_pages(void)
{
	return nr_huge_pages * (HPAGE_SIZE / PAGE_SIZE);
}

/*
 * We cannot handle pagefaults against hugetlb pages at all.  They cause
 * handle_mm_fault() to try to instantiate regular-sized pages in the
 * hugegpage VMA.  do_page_fault() is supposed to trap this, so BUG is we get
 * this far.
 */
Nick Piggin's avatar
Nick Piggin committed
646
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
Linus Torvalds's avatar
Linus Torvalds committed
647
648
{
	BUG();
Nick Piggin's avatar
Nick Piggin committed
649
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
650
651
652
}

struct vm_operations_struct hugetlb_vm_ops = {
Nick Piggin's avatar
Nick Piggin committed
653
	.fault = hugetlb_vm_op_fault,
Linus Torvalds's avatar
Linus Torvalds committed
654
655
};

656
657
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page,
				int writable)
David Gibson's avatar
David Gibson committed
658
659
660
{
	pte_t entry;

661
	if (writable) {
David Gibson's avatar
David Gibson committed
662
663
664
665
666
667
668
669
670
671
672
		entry =
		    pte_mkwrite(pte_mkdirty(mk_pte(page, vma->vm_page_prot)));
	} else {
		entry = pte_wrprotect(mk_pte(page, vma->vm_page_prot));
	}
	entry = pte_mkyoung(entry);
	entry = pte_mkhuge(entry);

	return entry;
}

673
674
675
676
677
678
static void set_huge_ptep_writable(struct vm_area_struct *vma,
				   unsigned long address, pte_t *ptep)
{
	pte_t entry;

	entry = pte_mkwrite(pte_mkdirty(*ptep));
679
680
681
	if (ptep_set_access_flags(vma, address, ptep, entry, 1)) {
		update_mmu_cache(vma, address, entry);
	}
682
683
684
}


David Gibson's avatar
David Gibson committed
685
686
687
688
689
int copy_hugetlb_page_range(struct mm_struct *dst, struct mm_struct *src,
			    struct vm_area_struct *vma)
{
	pte_t *src_pte, *dst_pte, entry;
	struct page *ptepage;
690
	unsigned long addr;
691
692
693
	int cow;

	cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
David Gibson's avatar
David Gibson committed
694

695
	for (addr = vma->vm_start; addr < vma->vm_end; addr += HPAGE_SIZE) {
696
697
698
		src_pte = huge_pte_offset(src, addr);
		if (!src_pte)
			continue;
David Gibson's avatar
David Gibson committed
699
700
701
		dst_pte = huge_pte_alloc(dst, addr);
		if (!dst_pte)
			goto nomem;
702
		spin_lock(&dst->page_table_lock);
703
		spin_lock(&src->page_table_lock);
704
		if (!pte_none(*src_pte)) {
705
706
			if (cow)
				ptep_set_wrprotect(src, addr, src_pte);
707
708
709
710
711
712
			entry = *src_pte;
			ptepage = pte_page(entry);
			get_page(ptepage);
			set_huge_pte_at(dst, addr, dst_pte, entry);
		}
		spin_unlock(&src->page_table_lock);
713
		spin_unlock(&dst->page_table_lock);
David Gibson's avatar
David Gibson committed
714
715
716
717
718
719
720
	}
	return 0;

nomem:
	return -ENOMEM;
}

721
722
void __unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
			    unsigned long end)
David Gibson's avatar
David Gibson committed
723
724
725
{
	struct mm_struct *mm = vma->vm_mm;
	unsigned long address;
726
	pte_t *ptep;
David Gibson's avatar
David Gibson committed
727
728
	pte_t pte;
	struct page *page;
729
	struct page *tmp;
730
731
732
733
734
	/*
	 * A page gathering list, protected by per file i_mmap_lock. The
	 * lock is used to avoid list corruption from multiple unmapping
	 * of the same page since we are using page->lru.
	 */
735
	LIST_HEAD(page_list);
David Gibson's avatar
David Gibson committed
736
737
738
739
740

	WARN_ON(!is_vm_hugetlb_page(vma));
	BUG_ON(start & ~HPAGE_MASK);
	BUG_ON(end & ~HPAGE_MASK);

741
	spin_lock(&mm->page_table_lock);
David Gibson's avatar
David Gibson committed
742
	for (address = start; address < end; address += HPAGE_SIZE) {
743
		ptep = huge_pte_offset(mm, address);
744
		if (!ptep)
745
746
			continue;

747
748
749
		if (huge_pmd_unshare(mm, &address, ptep))
			continue;

750
		pte = huge_ptep_get_and_clear(mm, address, ptep);
David Gibson's avatar
David Gibson committed
751
752
		if (pte_none(pte))
			continue;
753

David Gibson's avatar
David Gibson committed
754
		page = pte_page(pte);
755
756
		if (pte_dirty(pte))
			set_page_dirty(page);
757
		list_add(&page->lru, &page_list);
David Gibson's avatar
David Gibson committed
758
	}
Linus Torvalds's avatar
Linus Torvalds committed
759
	spin_unlock(&mm->page_table_lock);
760
	flush_tlb_range(vma, start, end);
761
762
763
764
	list_for_each_entry_safe(page, tmp, &page_list, lru) {
		list_del(&page->lru);
		put_page(page);
	}
Linus Torvalds's avatar
Linus Torvalds committed
765
}
David Gibson's avatar
David Gibson committed
766

767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
void unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
			  unsigned long end)
{
	/*
	 * It is undesirable to test vma->vm_file as it should be non-null
	 * for valid hugetlb area. However, vm_file will be NULL in the error
	 * cleanup path of do_mmap_pgoff. When hugetlbfs ->mmap method fails,
	 * do_mmap_pgoff() nullifies vma->vm_file before calling this function
	 * to clean up. Since no pte has actually been setup, it is safe to
	 * do nothing in this case.
	 */
	if (vma->vm_file) {
		spin_lock(&vma->vm_file->f_mapping->i_mmap_lock);
		__unmap_hugepage_range(vma, start, end);
		spin_unlock(&vma->vm_file->f_mapping->i_mmap_lock);
	}
}

785
786
787
788
static int hugetlb_cow(struct mm_struct *mm, struct vm_area_struct *vma,
			unsigned long address, pte_t *ptep, pte_t pte)
{
	struct page *old_page, *new_page;
789
	int avoidcopy;
790
791
792
793
794
795
796
797

	old_page = pte_page(pte);

	/* If no-one else is actually using this page, avoid the copy
	 * and just make the page writable */
	avoidcopy = (page_count(old_page) == 1);
	if (avoidcopy) {
		set_huge_ptep_writable(vma, address, ptep);
Nick Piggin's avatar
Nick Piggin committed
798
		return 0;
799
800
801
	}

	page_cache_get(old_page);
802
	new_page = alloc_huge_page(vma, address);
803

804
	if (IS_ERR(new_page)) {
805
		page_cache_release(old_page);
806
		return -PTR_ERR(new_page);
807
808
809
	}

	spin_unlock(&mm->page_table_lock);
810
	copy_huge_page(new_page, old_page, address, vma);
811
812
813
814
815
816
817
818
819
820
821
822
	spin_lock(&mm->page_table_lock);

	ptep = huge_pte_offset(mm, address & HPAGE_MASK);
	if (likely(pte_same(*ptep, pte))) {
		/* Break COW */
		set_huge_pte_at(mm, address, ptep,
				make_huge_pte(vma, new_page, 1));
		/* Make the old page be freed below */
		new_page = old_page;
	}
	page_cache_release(new_page);
	page_cache_release(old_page);
Nick Piggin's avatar
Nick Piggin committed
823
	return 0;
824
825
}

826
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma,
827
			unsigned long address, pte_t *ptep, int write_access)
828
829
{
	int ret = VM_FAULT_SIGBUS;
830
831
832
833
	unsigned long idx;
	unsigned long size;
	struct page *page;
	struct address_space *mapping;
834
	pte_t new_pte;
835
836
837
838
839
840
841
842
843

	mapping = vma->vm_file->f_mapping;
	idx = ((address - vma->vm_start) >> HPAGE_SHIFT)
		+ (vma->vm_pgoff >> (HPAGE_SHIFT - PAGE_SHIFT));

	/*
	 * Use page lock to guard against racing truncation
	 * before we get page_table_lock.
	 */
844
845
846
retry:
	page = find_lock_page(mapping, idx);
	if (!page) {
847
848
849
		size = i_size_read(mapping->host) >> HPAGE_SHIFT;
		if (idx >= size)
			goto out;
850
		page = alloc_huge_page(vma, address);
851
852
		if (IS_ERR(page)) {
			ret = -PTR_ERR(page);
853
854
			goto out;
		}
855
		clear_huge_page(page, address);
856

857
858
		if (vma->vm_flags & VM_SHARED) {
			int err;
Ken Chen's avatar
Ken Chen committed
859
			struct inode *inode = mapping->host;
860
861
862
863
864
865
866
867

			err = add_to_page_cache(page, mapping, idx, GFP_KERNEL);
			if (err) {
				put_page(page);
				if (err == -EEXIST)
					goto retry;
				goto out;
			}
Ken Chen's avatar
Ken Chen committed
868
869
870
871

			spin_lock(&inode->i_lock);
			inode->i_blocks += BLOCKS_PER_HUGEPAGE;
			spin_unlock(&inode->i_lock);
872
873
874
		} else
			lock_page(page);
	}
875

876
	spin_lock(&mm->page_table_lock);
877
878
879
880
	size = i_size_read(mapping->host) >> HPAGE_SHIFT;
	if (idx >= size)
		goto backout;

Nick Piggin's avatar
Nick Piggin committed
881
	ret = 0;
882
	if (!pte_none(*ptep))
883
884
		goto backout;

885
886
887
888
889
890
891
892
893
	new_pte = make_huge_pte(vma, page, ((vma->vm_flags & VM_WRITE)
				&& (vma->vm_flags & VM_SHARED)));
	set_huge_pte_at(mm, address, ptep, new_pte);

	if (write_access && !(vma->vm_flags & VM_SHARED)) {
		/* Optimization, do the COW without a second fault */
		ret = hugetlb_cow(mm, vma, address, ptep, new_pte);
	}

894
	spin_unlock(&mm->page_table_lock);
895
896
	unlock_page(page);
out:
897
	return ret;
898
899
900
901
902
903

backout:
	spin_unlock(&mm->page_table_lock);
	unlock_page(page);
	put_page(page);
	goto out;
904
905
}

906
907
908
909
910
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
			unsigned long address, int write_access)
{
	pte_t *ptep;
	pte_t entry;
911
	int ret;
912
	static DEFINE_MUTEX(hugetlb_instantiation_mutex);
913
914
915
916
917

	ptep = huge_pte_alloc(mm, address);
	if (!ptep)
		return VM_FAULT_OOM;

918
919
920
921
922
923
	/*
	 * Serialize hugepage allocation and instantiation, so that we don't
	 * get spurious allocation failures if two CPUs race to instantiate
	 * the same page in the page cache.
	 */
	mutex_lock(&hugetlb_instantiation_mutex);
924
	entry = *ptep;
925
926
927
928
929
	if (pte_none(entry)) {
		ret = hugetlb_no_page(mm, vma, address, ptep, write_access);
		mutex_unlock(&hugetlb_instantiation_mutex);
		return ret;
	}
930

Nick Piggin's avatar
Nick Piggin committed
931
	ret = 0;
932
933
934
935
936
937
938

	spin_lock(&mm->page_table_lock);
	/* Check for a racing update before calling hugetlb_cow */
	if (likely(pte_same(entry, *ptep)))
		if (write_access && !pte_write(entry))
			ret = hugetlb_cow(mm, vma, address, ptep, entry);
	spin_unlock(&mm->page_table_lock);
939
	mutex_unlock(&hugetlb_instantiation_mutex);
940
941

	return ret;
942
943
}

David Gibson's avatar
David Gibson committed
944
945
int follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma,
			struct page **pages, struct vm_area_struct **vmas,
946
947
			unsigned long *position, int *length, int i,
			int write)
David Gibson's avatar
David Gibson committed
948
{
949
950
	unsigned long pfn_offset;
	unsigned long vaddr = *position;
David Gibson's avatar
David Gibson committed
951
952
	int remainder = *length;

953
	spin_lock(&mm->page_table_lock);
David Gibson's avatar
David Gibson committed
954
	while (vaddr < vma->vm_end && remainder) {
955
956
		pte_t *pte;
		struct page *page;
David Gibson's avatar
David Gibson committed
957

958
959
960
961
962
963
		/*
		 * Some archs (sparc64, sh*) have multiple pte_ts to
		 * each hugepage.  We have to make * sure we get the
		 * first, for the page indexing below to work.
		 */
		pte = huge_pte_offset(mm, vaddr & HPAGE_MASK);
David Gibson's avatar
David Gibson committed
964

965
		if (!pte || pte_none(*pte) || (write && !pte_write(*pte))) {
966
			int ret;
David Gibson's avatar
David Gibson committed
967

968
			spin_unlock(&mm->page_table_lock);
969
			ret = hugetlb_fault(mm, vma, vaddr, write);
970
			spin_lock(&mm->page_table_lock);
971
			if (!(ret & VM_FAULT_ERROR))
972
				continue;
David Gibson's avatar
David Gibson committed
973

974
975
976
977
978
979
			remainder = 0;
			if (!i)
				i = -EFAULT;
			break;
		}

980
981
982
		pfn_offset = (vaddr & ~HPAGE_MASK) >> PAGE_SHIFT;
		page = pte_page(*pte);
same_page:
983
984
		if (pages) {
			get_page(page);
985
			pages[i] = page + pfn_offset;
986
		}
David Gibson's avatar
David Gibson committed
987
988
989
990
991

		if (vmas)
			vmas[i] = vma;

		vaddr += PAGE_SIZE;
992
		++pfn_offset;
David Gibson's avatar
David Gibson committed
993
994
		--remainder;
		++i;
995
996
997
998
999
1000
		if (vaddr < vma->vm_end && remainder &&
				pfn_offset < HPAGE_SIZE/PAGE_SIZE) {
			/*
			 * We use pfn_offset to avoid touching the pageframes
			 * of this compound page.
			 */
For faster browsing, not all history is shown. View entire blame