hugetlb.c 30.6 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
12
/*
 * Generic hugetlb support.
 * (C) William Irwin, April 2004
 */
#include <linux/gfp.h>
#include <linux/list.h>
#include <linux/init.h>
#include <linux/module.h>
#include <linux/mm.h>
#include <linux/sysctl.h>
#include <linux/highmem.h>
#include <linux/nodemask.h>
David Gibson's avatar
David Gibson committed
13
#include <linux/pagemap.h>
14
#include <linux/mempolicy.h>
15
#include <linux/cpuset.h>
16
#include <linux/mutex.h>
17

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
unsigned long nr_overcommit_huge_pages;
35
static int hugetlb_next_nid;
36

37
38
39
40
/*
 * Protects updates to hugepage_freelists, nr_huge_pages, and free_huge_pages
 */
static DEFINE_SPINLOCK(hugetlb_lock);
41

42
43
44
45
46
47
48
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();
49
		clear_user_highpage(page + i, addr + i * PAGE_SIZE);
50
51
52
53
	}
}

static void copy_huge_page(struct page *dst, struct page *src,
54
			   unsigned long addr, struct vm_area_struct *vma)
55
56
57
58
59
60
{
	int i;

	might_sleep();
	for (i = 0; i < HPAGE_SIZE/PAGE_SIZE; i++) {
		cond_resched();
61
		copy_user_highpage(dst + i, src + i, addr + i*PAGE_SIZE, vma);
62
63
64
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
65
66
67
68
69
70
71
72
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]++;
}

73
74
static struct page *dequeue_huge_page(struct vm_area_struct *vma,
				unsigned long address)
Linus Torvalds's avatar
Linus Torvalds committed
75
{
76
	int nid;
Linus Torvalds's avatar
Linus Torvalds committed
77
	struct page *page = NULL;
78
	struct mempolicy *mpol;
79
	struct zonelist *zonelist = huge_zonelist(vma, address,
80
					htlb_alloc_mask, &mpol);
81
	struct zone **z;
Linus Torvalds's avatar
Linus Torvalds committed
82

83
	for (z = zonelist->zones; *z; z++) {
84
		nid = zone_to_nid(*z);
85
		if (cpuset_zone_allowed_softwall(*z, htlb_alloc_mask) &&
86
87
88
89
90
91
		    !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]--;
92
93
			if (vma && vma->vm_flags & VM_MAYSHARE)
				resv_huge_pages--;
Ken Chen's avatar
Ken Chen committed
94
			break;
95
		}
Linus Torvalds's avatar
Linus Torvalds committed
96
	}
97
	mpol_free(mpol);	/* unref if mpol !NULL */
Linus Torvalds's avatar
Linus Torvalds committed
98
99
100
	return page;
}

101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
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);
}

116
117
static void free_huge_page(struct page *page)
{
118
	int nid = page_to_nid(page);
119
	struct address_space *mapping;
120

121
	mapping = (struct address_space *) page_private(page);
122
	BUG_ON(page_count(page));
123
124
125
	INIT_LIST_HEAD(&page->lru);

	spin_lock(&hugetlb_lock);
126
127
128
129
130
131
132
	if (surplus_huge_pages_node[nid]) {
		update_and_free_page(page);
		surplus_huge_pages--;
		surplus_huge_pages_node[nid]--;
	} else {
		enqueue_huge_page(page);
	}
133
	spin_unlock(&hugetlb_lock);
134
	if (mapping)
135
		hugetlb_put_quota(mapping, 1);
136
	set_page_private(page, 0);
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
/*
 * 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;
}

174
static struct page *alloc_fresh_huge_page_node(int nid)
Linus Torvalds's avatar
Linus Torvalds committed
175
176
{
	struct page *page;
177

178
179
180
	page = alloc_pages_node(nid,
		htlb_alloc_mask|__GFP_COMP|__GFP_THISNODE|__GFP_NOWARN,
		HUGETLB_PAGE_ORDER);
Linus Torvalds's avatar
Linus Torvalds committed
181
	if (page) {
182
		set_compound_page_dtor(page, free_huge_page);
183
		spin_lock(&hugetlb_lock);
Linus Torvalds's avatar
Linus Torvalds committed
184
		nr_huge_pages++;
185
		nr_huge_pages_node[nid]++;
186
		spin_unlock(&hugetlb_lock);
187
		put_page(page); /* free it into the hugepage allocator */
Linus Torvalds's avatar
Linus Torvalds committed
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

	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
224
225
}

226
227
228
229
static struct page *alloc_buddy_huge_page(struct vm_area_struct *vma,
						unsigned long address)
{
	struct page *page;
230
	unsigned int nid;
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
	/*
	 * 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);

265
266
	page = alloc_pages(htlb_alloc_mask|__GFP_COMP|__GFP_NOWARN,
					HUGETLB_PAGE_ORDER);
267
268

	spin_lock(&hugetlb_lock);
269
	if (page) {
270
		nid = page_to_nid(page);
271
		set_compound_page_dtor(page, free_huge_page);
272
273
274
275
276
277
278
279
		/*
		 * We incremented the global counters already
		 */
		nr_huge_pages_node[nid]++;
		surplus_huge_pages_node[nid]++;
	} else {
		nr_huge_pages--;
		surplus_huge_pages--;
280
	}
281
	spin_unlock(&hugetlb_lock);
282
283
284
285

	return page;
}

286
287
288
289
290
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
/*
 * 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);
346
347
348
349
350
351
352
353
354
355
356
		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);
		}
357
358
359
360
361
362
363
364
365
366
	}

	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.
 */
367
static void return_unused_surplus_pages(unsigned long unused_resv_pages)
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
{
	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--;
		}
	}
}

397
398
399

static struct page *alloc_huge_page_shared(struct vm_area_struct *vma,
						unsigned long addr)
Linus Torvalds's avatar
Linus Torvalds committed
400
{
401
	struct page *page;
Linus Torvalds's avatar
Linus Torvalds committed
402
403

	spin_lock(&hugetlb_lock);
404
	page = dequeue_huge_page(vma, addr);
Linus Torvalds's avatar
Linus Torvalds committed
405
	spin_unlock(&hugetlb_lock);
406
	return page ? page : ERR_PTR(-VM_FAULT_OOM);
407
}
408

409
410
411
412
static struct page *alloc_huge_page_private(struct vm_area_struct *vma,
						unsigned long addr)
{
	struct page *page = NULL;
413

414
415
416
	if (hugetlb_get_quota(vma->vm_file->f_mapping, 1))
		return ERR_PTR(-VM_FAULT_SIGBUS);

417
418
419
420
421
	spin_lock(&hugetlb_lock);
	if (free_huge_pages > resv_huge_pages)
		page = dequeue_huge_page(vma, addr);
	spin_unlock(&hugetlb_lock);
	if (!page)
422
		page = alloc_buddy_huge_page(vma, addr);
423
	return page ? page : ERR_PTR(-VM_FAULT_OOM);
424
425
426
427
428
429
}

static struct page *alloc_huge_page(struct vm_area_struct *vma,
				    unsigned long addr)
{
	struct page *page;
430
431
	struct address_space *mapping = vma->vm_file->f_mapping;

432
433
434
435
	if (vma->vm_flags & VM_MAYSHARE)
		page = alloc_huge_page_shared(vma, addr);
	else
		page = alloc_huge_page_private(vma, addr);
436
437

	if (!IS_ERR(page)) {
438
		set_page_refcounted(page);
439
		set_page_private(page, (unsigned long) mapping);
440
441
	}
	return page;
442
443
}

Linus Torvalds's avatar
Linus Torvalds committed
444
445
446
447
static int __init hugetlb_init(void)
{
	unsigned long i;

448
449
450
	if (HPAGE_SHIFT == 0)
		return 0;

Linus Torvalds's avatar
Linus Torvalds committed
451
452
453
	for (i = 0; i < MAX_NUMNODES; ++i)
		INIT_LIST_HEAD(&hugepage_freelists[i]);

454
455
	hugetlb_next_nid = first_node(node_online_map);

Linus Torvalds's avatar
Linus Torvalds committed
456
	for (i = 0; i < max_huge_pages; ++i) {
457
		if (!alloc_fresh_huge_page())
Linus Torvalds's avatar
Linus Torvalds committed
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
			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);

474
475
476
477
478
479
480
481
482
483
484
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
485
486
487
488
#ifdef CONFIG_SYSCTL
#ifdef CONFIG_HIGHMEM
static void try_to_free_low(unsigned long count)
{
489
490
	int i;

Linus Torvalds's avatar
Linus Torvalds committed
491
492
493
	for (i = 0; i < MAX_NUMNODES; ++i) {
		struct page *page, *next;
		list_for_each_entry_safe(page, next, &hugepage_freelists[i], lru) {
494
495
			if (count >= nr_huge_pages)
				return;
Linus Torvalds's avatar
Linus Torvalds committed
496
497
498
499
500
			if (PageHighMem(page))
				continue;
			list_del(&page->lru);
			update_and_free_page(page);
			free_huge_pages--;
501
			free_huge_pages_node[page_to_nid(page)]--;
Linus Torvalds's avatar
Linus Torvalds committed
502
503
504
505
506
507
508
509
510
		}
	}
}
#else
static inline void try_to_free_low(unsigned long count)
{
}
#endif

511
#define persistent_huge_pages (nr_huge_pages - surplus_huge_pages)
Linus Torvalds's avatar
Linus Torvalds committed
512
513
static unsigned long set_max_huge_pages(unsigned long count)
{
514
	unsigned long min_count, ret;
Linus Torvalds's avatar
Linus Torvalds committed
515

516
517
518
519
	/*
	 * Increase the pool size
	 * First take pages out of surplus state.  Then make up the
	 * remaining difference by allocating fresh huge pages.
520
521
522
523
524
525
	 *
	 * 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.
526
	 */
Linus Torvalds's avatar
Linus Torvalds committed
527
	spin_lock(&hugetlb_lock);
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
	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.
554
555
556
557
558
559
560
561
	 *
	 * 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.
562
	 */
563
564
	min_count = resv_huge_pages + nr_huge_pages - free_huge_pages;
	min_count = max(count, min_count);
565
566
	try_to_free_low(min_count);
	while (min_count < persistent_huge_pages) {
567
		struct page *page = dequeue_huge_page(NULL, 0);
Linus Torvalds's avatar
Linus Torvalds committed
568
569
570
571
		if (!page)
			break;
		update_and_free_page(page);
	}
572
573
574
575
576
577
	while (count < persistent_huge_pages) {
		if (!adjust_pool_surplus(1))
			break;
	}
out:
	ret = persistent_huge_pages;
Linus Torvalds's avatar
Linus Torvalds committed
578
	spin_unlock(&hugetlb_lock);
579
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
580
581
582
583
584
585
586
587
588
589
}

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;
}
590
591
592
593
594
595
596
597
598
599
600
601
602

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
603
604
605
606
607
608
609
#endif /* CONFIG_SYSCTL */

int hugetlb_report_meminfo(char *buf)
{
	return sprintf(buf,
			"HugePages_Total: %5lu\n"
			"HugePages_Free:  %5lu\n"
610
			"HugePages_Rsvd:  %5lu\n"
611
			"HugePages_Surp:  %5lu\n"
Linus Torvalds's avatar
Linus Torvalds committed
612
613
614
			"Hugepagesize:    %5lu kB\n",
			nr_huge_pages,
			free_huge_pages,
615
			resv_huge_pages,
616
			surplus_huge_pages,
Linus Torvalds's avatar
Linus Torvalds committed
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
			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
641
static int hugetlb_vm_op_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
Linus Torvalds's avatar
Linus Torvalds committed
642
643
{
	BUG();
Nick Piggin's avatar
Nick Piggin committed
644
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
645
646
647
}

struct vm_operations_struct hugetlb_vm_ops = {
Nick Piggin's avatar
Nick Piggin committed
648
	.fault = hugetlb_vm_op_fault,
Linus Torvalds's avatar
Linus Torvalds committed
649
650
};

651
652
static pte_t make_huge_pte(struct vm_area_struct *vma, struct page *page,
				int writable)
David Gibson's avatar
David Gibson committed
653
654
655
{
	pte_t entry;

656
	if (writable) {
David Gibson's avatar
David Gibson committed
657
658
659
660
661
662
663
664
665
666
667
		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;
}

668
669
670
671
672
673
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));
674
675
676
	if (ptep_set_access_flags(vma, address, ptep, entry, 1)) {
		update_mmu_cache(vma, address, entry);
	}
677
678
679
}


David Gibson's avatar
David Gibson committed
680
681
682
683
684
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;
685
	unsigned long addr;
686
687
688
	int cow;

	cow = (vma->vm_flags & (VM_SHARED | VM_MAYWRITE)) == VM_MAYWRITE;
David Gibson's avatar
David Gibson committed
689

690
	for (addr = vma->vm_start; addr < vma->vm_end; addr += HPAGE_SIZE) {
691
692
693
		src_pte = huge_pte_offset(src, addr);
		if (!src_pte)
			continue;
David Gibson's avatar
David Gibson committed
694
695
696
		dst_pte = huge_pte_alloc(dst, addr);
		if (!dst_pte)
			goto nomem;
697
		spin_lock(&dst->page_table_lock);
698
		spin_lock(&src->page_table_lock);
699
		if (!pte_none(*src_pte)) {
700
701
			if (cow)
				ptep_set_wrprotect(src, addr, src_pte);
702
703
704
705
706
707
			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);
708
		spin_unlock(&dst->page_table_lock);
David Gibson's avatar
David Gibson committed
709
710
711
712
713
714
715
	}
	return 0;

nomem:
	return -ENOMEM;
}

716
717
void __unmap_hugepage_range(struct vm_area_struct *vma, unsigned long start,
			    unsigned long end)
David Gibson's avatar
David Gibson committed
718
719
720
{
	struct mm_struct *mm = vma->vm_mm;
	unsigned long address;
721
	pte_t *ptep;
David Gibson's avatar
David Gibson committed
722
723
	pte_t pte;
	struct page *page;
724
	struct page *tmp;
725
726
727
728
729
	/*
	 * 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.
	 */
730
	LIST_HEAD(page_list);
David Gibson's avatar
David Gibson committed
731
732
733
734
735

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

736
	spin_lock(&mm->page_table_lock);
David Gibson's avatar
David Gibson committed
737
	for (address = start; address < end; address += HPAGE_SIZE) {
738
		ptep = huge_pte_offset(mm, address);
739
		if (!ptep)
740
741
			continue;

742
743
744
		if (huge_pmd_unshare(mm, &address, ptep))
			continue;

745
		pte = huge_ptep_get_and_clear(mm, address, ptep);
David Gibson's avatar
David Gibson committed
746
747
		if (pte_none(pte))
			continue;
748

David Gibson's avatar
David Gibson committed
749
		page = pte_page(pte);
750
751
		if (pte_dirty(pte))
			set_page_dirty(page);
752
		list_add(&page->lru, &page_list);
David Gibson's avatar
David Gibson committed
753
	}
Linus Torvalds's avatar
Linus Torvalds committed
754
	spin_unlock(&mm->page_table_lock);
755
	flush_tlb_range(vma, start, end);
756
757
758
759
	list_for_each_entry_safe(page, tmp, &page_list, lru) {
		list_del(&page->lru);
		put_page(page);
	}
Linus Torvalds's avatar
Linus Torvalds committed
760
}
David Gibson's avatar
David Gibson committed
761

762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
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);
	}
}

780
781
782
783
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;
784
	int avoidcopy;
785
786
787
788
789
790
791
792

	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
793
		return 0;
794
795
796
	}

	page_cache_get(old_page);
797
	new_page = alloc_huge_page(vma, address);
798

799
	if (IS_ERR(new_page)) {
800
		page_cache_release(old_page);
801
		return -PTR_ERR(new_page);
802
803
804
	}

	spin_unlock(&mm->page_table_lock);
805
	copy_huge_page(new_page, old_page, address, vma);
806
807
808
809
810
811
812
813
814
815
816
817
	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
818
	return 0;
819
820
}

821
static int hugetlb_no_page(struct mm_struct *mm, struct vm_area_struct *vma,
822
			unsigned long address, pte_t *ptep, int write_access)
823
824
{
	int ret = VM_FAULT_SIGBUS;
825
826
827
828
	unsigned long idx;
	unsigned long size;
	struct page *page;
	struct address_space *mapping;
829
	pte_t new_pte;
830
831
832
833
834
835
836
837
838

	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.
	 */
839
840
841
retry:
	page = find_lock_page(mapping, idx);
	if (!page) {
842
843
844
		size = i_size_read(mapping->host) >> HPAGE_SHIFT;
		if (idx >= size)
			goto out;
845
		page = alloc_huge_page(vma, address);
846
847
		if (IS_ERR(page)) {
			ret = -PTR_ERR(page);
848
849
			goto out;
		}
850
		clear_huge_page(page, address);
851

852
853
		if (vma->vm_flags & VM_SHARED) {
			int err;
Ken Chen's avatar
Ken Chen committed
854
			struct inode *inode = mapping->host;
855
856
857
858
859
860
861
862

			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
863
864
865
866

			spin_lock(&inode->i_lock);
			inode->i_blocks += BLOCKS_PER_HUGEPAGE;
			spin_unlock(&inode->i_lock);
867
868
869
		} else
			lock_page(page);
	}
870

871
	spin_lock(&mm->page_table_lock);
872
873
874
875
	size = i_size_read(mapping->host) >> HPAGE_SHIFT;
	if (idx >= size)
		goto backout;

Nick Piggin's avatar
Nick Piggin committed
876
	ret = 0;
877
	if (!pte_none(*ptep))
878
879
		goto backout;

880
881
882
883
884
885
886
887
888
	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);
	}

889
	spin_unlock(&mm->page_table_lock);
890
891
	unlock_page(page);
out:
892
	return ret;
893
894
895
896
897
898

backout:
	spin_unlock(&mm->page_table_lock);
	unlock_page(page);
	put_page(page);
	goto out;
899
900
}

901
902
903
904
905
int hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
			unsigned long address, int write_access)
{
	pte_t *ptep;
	pte_t entry;
906
	int ret;
907
	static DEFINE_MUTEX(hugetlb_instantiation_mutex);
908
909
910
911
912

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

913
914
915
916
917
918
	/*
	 * 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);
919
	entry = *ptep;
920
921
922
923
924
	if (pte_none(entry)) {
		ret = hugetlb_no_page(mm, vma, address, ptep, write_access);
		mutex_unlock(&hugetlb_instantiation_mutex);
		return ret;
	}
925

Nick Piggin's avatar
Nick Piggin committed
926
	ret = 0;
927
928
929
930
931
932
933

	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);
934
	mutex_unlock(&hugetlb_instantiation_mutex);
935
936

	return ret;
937
938
}

David Gibson's avatar
David Gibson committed
939
940
int follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct *vma,
			struct page **pages, struct vm_area_struct **vmas,
941
942
			unsigned long *position, int *length, int i,
			int write)
David Gibson's avatar
David Gibson committed
943
{
944
945
	unsigned long pfn_offset;
	unsigned long vaddr = *position;
David Gibson's avatar
David Gibson committed
946
947
	int remainder = *length;

948
	spin_lock(&mm->page_table_lock);
David Gibson's avatar
David Gibson committed
949
	while (vaddr < vma->vm_end && remainder) {
950
951
		pte_t *pte;
		struct page *page;
David Gibson's avatar
David Gibson committed
952

953
954
955
956
957
958
		/*
		 * 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
959

960
		if (!pte || pte_none(*pte) || (write && !pte_write(*pte))) {
961
			int ret;
David Gibson's avatar
David Gibson committed
962

963
			spin_unlock(&mm->page_table_lock);
964
			ret = hugetlb_fault(mm, vma, vaddr, write);
965
			spin_lock(&mm->page_table_lock);
966
			if (!(ret & VM_FAULT_ERROR))
967
				continue;
David Gibson's avatar
David Gibson committed
968

969
970
971
972
973
974
			remainder = 0;
			if (!i)
				i = -EFAULT;
			break;
		}

975
976
977
		pfn_offset = (vaddr & ~HPAGE_MASK) >> PAGE_SHIFT;
		page = pte_page(*pte);
same_page:
978
979
		if (pages) {
			get_page(page);
980
			pages[i] = page + pfn_offset;
981
		}
David Gibson's avatar
David Gibson committed
982
983
984
985
986

		if (vmas)
			vmas[i] = vma;

		vaddr += PAGE_SIZE;
987
		++pfn_offset;
David Gibson's avatar
David Gibson committed
988
989
		--remainder;
		++i;
990
991
992
993
994
995
996
997
		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.
			 */
			goto same_page;
		}
David Gibson's avatar
David Gibson committed
998
	}
999
	spin_unlock(&mm->page_table_lock);
David Gibson's avatar
David Gibson committed
1000
	*length = remainder;
For faster browsing, not all history is shown. View entire blame