madvise.c 22.6 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
Linus Torvalds's avatar
Linus Torvalds committed
2
3
4
5
6
7
8
9
10
11
/*
 *	linux/mm/madvise.c
 *
 * Copyright (C) 1999  Linus Torvalds
 * Copyright (C) 2002  Christoph Hellwig
 */

#include <linux/mman.h>
#include <linux/pagemap.h>
#include <linux/syscalls.h>
12
#include <linux/mempolicy.h>
13
#include <linux/page-isolation.h>
14
#include <linux/userfaultfd_k.h>
Linus Torvalds's avatar
Linus Torvalds committed
15
#include <linux/hugetlb.h>
16
#include <linux/falloc.h>
Alexey Dobriyan's avatar
Alexey Dobriyan committed
17
#include <linux/sched.h>
Hugh Dickins's avatar
Hugh Dickins committed
18
#include <linux/ksm.h>
19
#include <linux/fs.h>
20
#include <linux/file.h>
21
#include <linux/blkdev.h>
22
#include <linux/backing-dev.h>
23
24
#include <linux/swap.h>
#include <linux/swapops.h>
25
#include <linux/shmem_fs.h>
Minchan Kim's avatar
Minchan Kim committed
26
27
28
#include <linux/mmu_notifier.h>

#include <asm/tlb.h>
Linus Torvalds's avatar
Linus Torvalds committed
29

30
31
#include "internal.h"

32
33
34
35
36
37
38
39
40
41
42
/*
 * Any behaviour which results in changes to the vma->vm_flags needs to
 * take mmap_sem for writing. Others, which simply traverse vmas, need
 * to only take it for reading.
 */
static int madvise_need_mmap_write(int behavior)
{
	switch (behavior) {
	case MADV_REMOVE:
	case MADV_WILLNEED:
	case MADV_DONTNEED:
Minchan Kim's avatar
Minchan Kim committed
43
	case MADV_FREE:
44
45
46
47
48
49
50
		return 0;
	default:
		/* be safe, default to 1. list exceptions explicitly */
		return 1;
	}
}

Linus Torvalds's avatar
Linus Torvalds committed
51
52
53
54
/*
 * We can potentially split a vm area into separate
 * areas, each area with its own behavior.
 */
55
static long madvise_behavior(struct vm_area_struct *vma,
56
57
		     struct vm_area_struct **prev,
		     unsigned long start, unsigned long end, int behavior)
Linus Torvalds's avatar
Linus Torvalds committed
58
{
59
	struct mm_struct *mm = vma->vm_mm;
Linus Torvalds's avatar
Linus Torvalds committed
60
	int error = 0;
61
	pgoff_t pgoff;
62
	unsigned long new_flags = vma->vm_flags;
63
64

	switch (behavior) {
65
66
67
	case MADV_NORMAL:
		new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
		break;
68
	case MADV_SEQUENTIAL:
69
		new_flags = (new_flags & ~VM_RAND_READ) | VM_SEQ_READ;
70
71
		break;
	case MADV_RANDOM:
72
		new_flags = (new_flags & ~VM_SEQ_READ) | VM_RAND_READ;
73
		break;
74
75
76
77
	case MADV_DONTFORK:
		new_flags |= VM_DONTCOPY;
		break;
	case MADV_DOFORK:
78
79
80
81
		if (vma->vm_flags & VM_IO) {
			error = -EINVAL;
			goto out;
		}
82
		new_flags &= ~VM_DONTCOPY;
83
		break;
84
85
86
87
88
89
90
91
92
93
94
	case MADV_WIPEONFORK:
		/* MADV_WIPEONFORK is only supported on anonymous memory. */
		if (vma->vm_file || vma->vm_flags & VM_SHARED) {
			error = -EINVAL;
			goto out;
		}
		new_flags |= VM_WIPEONFORK;
		break;
	case MADV_KEEPONFORK:
		new_flags &= ~VM_WIPEONFORK;
		break;
95
	case MADV_DONTDUMP:
96
		new_flags |= VM_DONTDUMP;
97
98
		break;
	case MADV_DODUMP:
99
100
101
102
103
		if (new_flags & VM_SPECIAL) {
			error = -EINVAL;
			goto out;
		}
		new_flags &= ~VM_DONTDUMP;
104
		break;
Hugh Dickins's avatar
Hugh Dickins committed
105
106
107
	case MADV_MERGEABLE:
	case MADV_UNMERGEABLE:
		error = ksm_madvise(vma, start, end, behavior, &new_flags);
108
109
110
111
112
113
114
		if (error) {
			/*
			 * madvise() returns EAGAIN if kernel resources, such as
			 * slab, are temporarily unavailable.
			 */
			if (error == -ENOMEM)
				error = -EAGAIN;
Hugh Dickins's avatar
Hugh Dickins committed
115
			goto out;
116
		}
Hugh Dickins's avatar
Hugh Dickins committed
117
		break;
Andrea Arcangeli's avatar
Andrea Arcangeli committed
118
	case MADV_HUGEPAGE:
119
	case MADV_NOHUGEPAGE:
120
		error = hugepage_madvise(vma, &new_flags, behavior);
121
122
123
124
125
126
127
		if (error) {
			/*
			 * madvise() returns EAGAIN if kernel resources, such as
			 * slab, are temporarily unavailable.
			 */
			if (error == -ENOMEM)
				error = -EAGAIN;
Andrea Arcangeli's avatar
Andrea Arcangeli committed
128
			goto out;
129
		}
Andrea Arcangeli's avatar
Andrea Arcangeli committed
130
		break;
131
132
	}

133
134
	if (new_flags == vma->vm_flags) {
		*prev = vma;
135
		goto out;
136
137
138
139
	}

	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	*prev = vma_merge(mm, *prev, start, end, new_flags, vma->anon_vma,
140
141
			  vma->vm_file, pgoff, vma_policy(vma),
			  vma->vm_userfaultfd_ctx);
142
143
144
145
146
147
	if (*prev) {
		vma = *prev;
		goto success;
	}

	*prev = vma;
Linus Torvalds's avatar
Linus Torvalds committed
148
149

	if (start != vma->vm_start) {
150
151
		if (unlikely(mm->map_count >= sysctl_max_map_count)) {
			error = -ENOMEM;
Linus Torvalds's avatar
Linus Torvalds committed
152
			goto out;
153
154
155
156
157
158
159
160
161
162
163
		}
		error = __split_vma(mm, vma, start, 1);
		if (error) {
			/*
			 * madvise() returns EAGAIN if kernel resources, such as
			 * slab, are temporarily unavailable.
			 */
			if (error == -ENOMEM)
				error = -EAGAIN;
			goto out;
		}
Linus Torvalds's avatar
Linus Torvalds committed
164
165
166
	}

	if (end != vma->vm_end) {
167
168
		if (unlikely(mm->map_count >= sysctl_max_map_count)) {
			error = -ENOMEM;
Linus Torvalds's avatar
Linus Torvalds committed
169
			goto out;
170
171
172
173
174
175
176
177
178
179
180
		}
		error = __split_vma(mm, vma, end, 0);
		if (error) {
			/*
			 * madvise() returns EAGAIN if kernel resources, such as
			 * slab, are temporarily unavailable.
			 */
			if (error == -ENOMEM)
				error = -EAGAIN;
			goto out;
		}
Linus Torvalds's avatar
Linus Torvalds committed
181
182
	}

183
success:
Linus Torvalds's avatar
Linus Torvalds committed
184
185
186
	/*
	 * vm_flags is protected by the mmap_sem held in write mode.
	 */
187
	vma->vm_flags = new_flags;
Linus Torvalds's avatar
Linus Torvalds committed
188
189
190
191
out:
	return error;
}

192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
#ifdef CONFIG_SWAP
static int swapin_walk_pmd_entry(pmd_t *pmd, unsigned long start,
	unsigned long end, struct mm_walk *walk)
{
	pte_t *orig_pte;
	struct vm_area_struct *vma = walk->private;
	unsigned long index;

	if (pmd_none_or_trans_huge_or_clear_bad(pmd))
		return 0;

	for (index = start; index != end; index += PAGE_SIZE) {
		pte_t pte;
		swp_entry_t entry;
		struct page *page;
		spinlock_t *ptl;

		orig_pte = pte_offset_map_lock(vma->vm_mm, pmd, start, &ptl);
		pte = *(orig_pte + ((index - start) / PAGE_SIZE));
		pte_unmap_unlock(orig_pte, ptl);

213
		if (pte_present(pte) || pte_none(pte))
214
215
216
217
218
219
			continue;
		entry = pte_to_swp_entry(pte);
		if (unlikely(non_swap_entry(entry)))
			continue;

		page = read_swap_cache_async(entry, GFP_HIGHUSER_MOVABLE,
220
							vma, index, false);
221
		if (page)
222
			put_page(page);
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
	}

	return 0;
}

static void force_swapin_readahead(struct vm_area_struct *vma,
		unsigned long start, unsigned long end)
{
	struct mm_walk walk = {
		.mm = vma->vm_mm,
		.pmd_entry = swapin_walk_pmd_entry,
		.private = vma,
	};

	walk_page_range(start, end, &walk);

	lru_add_drain();	/* Push any new pages onto the LRU now */
}

static void force_shm_swapin_readahead(struct vm_area_struct *vma,
		unsigned long start, unsigned long end,
		struct address_space *mapping)
{
	pgoff_t index;
	struct page *page;
	swp_entry_t swap;

	for (; start < end; start += PAGE_SIZE) {
		index = ((start - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;

253
		page = find_get_entry(mapping, index);
254
255
		if (!radix_tree_exceptional_entry(page)) {
			if (page)
256
				put_page(page);
257
258
259
260
			continue;
		}
		swap = radix_to_swp_entry(page);
		page = read_swap_cache_async(swap, GFP_HIGHUSER_MOVABLE,
261
							NULL, 0, false);
262
		if (page)
263
			put_page(page);
264
265
266
267
268
269
	}

	lru_add_drain();	/* Push any new pages onto the LRU now */
}
#endif		/* CONFIG_SWAP */

Linus Torvalds's avatar
Linus Torvalds committed
270
271
272
/*
 * Schedule all required I/O operations.  Do not wait for completion.
 */
273
274
static long madvise_willneed(struct vm_area_struct *vma,
			     struct vm_area_struct **prev,
Linus Torvalds's avatar
Linus Torvalds committed
275
276
277
278
			     unsigned long start, unsigned long end)
{
	struct file *file = vma->vm_file;

279
	*prev = vma;
280
#ifdef CONFIG_SWAP
281
282
	if (!file) {
		force_swapin_readahead(vma, start, end);
283
284
285
		return 0;
	}

286
287
288
289
290
291
	if (shmem_mapping(file->f_mapping)) {
		force_shm_swapin_readahead(vma, start, end,
					file->f_mapping);
		return 0;
	}
#else
292
293
	if (!file)
		return -EBADF;
294
#endif
295

Matthew Wilcox's avatar
Matthew Wilcox committed
296
	if (IS_DAX(file_inode(file))) {
297
298
299
300
		/* no bad return value, but ignore advice */
		return 0;
	}

Linus Torvalds's avatar
Linus Torvalds committed
301
302
303
304
305
	start = ((start - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	if (end > vma->vm_end)
		end = vma->vm_end;
	end = ((end - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;

306
	force_page_cache_readahead(file->f_mapping, file, start, end - start);
Linus Torvalds's avatar
Linus Torvalds committed
307
308
309
	return 0;
}

Minchan Kim's avatar
Minchan Kim committed
310
311
312
313
314
315
316
317
318
319
static int madvise_free_pte_range(pmd_t *pmd, unsigned long addr,
				unsigned long end, struct mm_walk *walk)

{
	struct mmu_gather *tlb = walk->private;
	struct mm_struct *mm = tlb->mm;
	struct vm_area_struct *vma = walk->vma;
	spinlock_t *ptl;
	pte_t *orig_pte, *pte, ptent;
	struct page *page;
320
	int nr_swap = 0;
321
322
323
324
325
326
	unsigned long next;

	next = pmd_addr_end(addr, end);
	if (pmd_trans_huge(*pmd))
		if (madvise_free_huge_pmd(tlb, vma, pmd, addr, next))
			goto next;
Minchan Kim's avatar
Minchan Kim committed
327
328
329
330

	if (pmd_trans_unstable(pmd))
		return 0;

331
	tlb_remove_check_page_size_change(tlb, PAGE_SIZE);
Minchan Kim's avatar
Minchan Kim committed
332
	orig_pte = pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
333
	flush_tlb_batched_pending(mm);
Minchan Kim's avatar
Minchan Kim committed
334
335
336
337
	arch_enter_lazy_mmu_mode();
	for (; addr != end; pte++, addr += PAGE_SIZE) {
		ptent = *pte;

338
		if (pte_none(ptent))
Minchan Kim's avatar
Minchan Kim committed
339
			continue;
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
		/*
		 * If the pte has swp_entry, just clear page table to
		 * prevent swap-in which is more expensive rather than
		 * (page allocation + zeroing).
		 */
		if (!pte_present(ptent)) {
			swp_entry_t entry;

			entry = pte_to_swp_entry(ptent);
			if (non_swap_entry(entry))
				continue;
			nr_swap--;
			free_swap_and_cache(entry);
			pte_clear_not_present_full(mm, addr, pte, tlb->fullmm);
			continue;
		}
Minchan Kim's avatar
Minchan Kim committed
356

357
		page = _vm_normal_page(vma, addr, ptent, true);
Minchan Kim's avatar
Minchan Kim committed
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
		if (!page)
			continue;

		/*
		 * If pmd isn't transhuge but the page is THP and
		 * is owned by only this process, split it and
		 * deactivate all pages.
		 */
		if (PageTransCompound(page)) {
			if (page_mapcount(page) != 1)
				goto out;
			get_page(page);
			if (!trylock_page(page)) {
				put_page(page);
				goto out;
			}
			pte_unmap_unlock(orig_pte, ptl);
			if (split_huge_page(page)) {
				unlock_page(page);
				put_page(page);
				pte_offset_map_lock(mm, pmd, addr, &ptl);
				goto out;
			}
			unlock_page(page);
382
			put_page(page);
Minchan Kim's avatar
Minchan Kim committed
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
			pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
			pte--;
			addr -= PAGE_SIZE;
			continue;
		}

		VM_BUG_ON_PAGE(PageTransCompound(page), page);

		if (PageSwapCache(page) || PageDirty(page)) {
			if (!trylock_page(page))
				continue;
			/*
			 * If page is shared with others, we couldn't clear
			 * PG_dirty of the page.
			 */
			if (page_mapcount(page) != 1) {
				unlock_page(page);
				continue;
			}

			if (PageSwapCache(page) && !try_to_free_swap(page)) {
				unlock_page(page);
				continue;
			}

			ClearPageDirty(page);
			unlock_page(page);
		}

		if (pte_young(ptent) || pte_dirty(ptent)) {
			/*
			 * Some of architecture(ex, PPC) don't update TLB
			 * with set_pte_at and tlb_remove_tlb_entry so for
			 * the portability, remap the pte with old|clean
			 * after pte clearing.
			 */
			ptent = ptep_get_and_clear_full(mm, addr, pte,
							tlb->fullmm);

			ptent = pte_mkold(ptent);
			ptent = pte_mkclean(ptent);
			set_pte_at(mm, addr, pte, ptent);
			tlb_remove_tlb_entry(tlb, pte, addr);
		}
Shaohua Li's avatar
Shaohua Li committed
427
		mark_page_lazyfree(page);
Minchan Kim's avatar
Minchan Kim committed
428
429
	}
out:
430
431
432
433
434
435
	if (nr_swap) {
		if (current->mm == mm)
			sync_mm_rss(mm);

		add_mm_counter(mm, MM_SWAPENTS, nr_swap);
	}
Minchan Kim's avatar
Minchan Kim committed
436
437
438
	arch_leave_lazy_mmu_mode();
	pte_unmap_unlock(orig_pte, ptl);
	cond_resched();
439
next:
Minchan Kim's avatar
Minchan Kim committed
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
	return 0;
}

static void madvise_free_page_range(struct mmu_gather *tlb,
			     struct vm_area_struct *vma,
			     unsigned long addr, unsigned long end)
{
	struct mm_walk free_walk = {
		.pmd_entry = madvise_free_pte_range,
		.mm = vma->vm_mm,
		.private = tlb,
	};

	tlb_start_vma(tlb, vma);
	walk_page_range(addr, end, &free_walk);
	tlb_end_vma(tlb, vma);
}

static int madvise_free_single_vma(struct vm_area_struct *vma,
			unsigned long start_addr, unsigned long end_addr)
{
	unsigned long start, end;
	struct mm_struct *mm = vma->vm_mm;
	struct mmu_gather tlb;

	/* MADV_FREE works for only anon vma at the moment */
	if (!vma_is_anonymous(vma))
		return -EINVAL;

	start = max(vma->vm_start, start_addr);
	if (start >= vma->vm_end)
		return -EINVAL;
	end = min(vma->vm_end, end_addr);
	if (end <= vma->vm_start)
		return -EINVAL;

	lru_add_drain();
	tlb_gather_mmu(&tlb, mm, start, end);
	update_hiwater_rss(mm);

	mmu_notifier_invalidate_range_start(mm, start, end);
	madvise_free_page_range(&tlb, vma, start, end);
	mmu_notifier_invalidate_range_end(mm, start, end);
	tlb_finish_mmu(&tlb, start, end);

	return 0;
}

Linus Torvalds's avatar
Linus Torvalds committed
488
489
490
491
/*
 * Application no longer needs these pages.  If the pages are dirty,
 * it's OK to just throw them away.  The app will be more careful about
 * data it wants to keep.  Be sure to free swap resources too.  The
492
 * zap_page_range call sets things up for shrink_active_list to actually free
Linus Torvalds's avatar
Linus Torvalds committed
493
494
 * these pages later if no one else has touched them in the meantime,
 * although we could add these pages to a global reuse list for
495
 * shrink_active_list to pick up before reclaiming other pages.
Linus Torvalds's avatar
Linus Torvalds committed
496
497
498
499
500
501
502
503
504
505
506
 *
 * NB: This interface discards data rather than pushes it out to swap,
 * as some implementations do.  This has performance implications for
 * applications like large transactional databases which want to discard
 * pages in anonymous maps after committing to backing store the data
 * that was kept in them.  There is no reason to write this data out to
 * the swap area if the application is discarding it.
 *
 * An interface that causes the system to free clean pages and flush
 * dirty pages is already available as msync(MS_INVALIDATE).
 */
507
508
509
510
511
512
513
514
515
516
517
static long madvise_dontneed_single_vma(struct vm_area_struct *vma,
					unsigned long start, unsigned long end)
{
	zap_page_range(vma, start, end - start);
	return 0;
}

static long madvise_dontneed_free(struct vm_area_struct *vma,
				  struct vm_area_struct **prev,
				  unsigned long start, unsigned long end,
				  int behavior)
Linus Torvalds's avatar
Linus Torvalds committed
518
{
519
	*prev = vma;
520
	if (!can_madv_dontneed_vma(vma))
Linus Torvalds's avatar
Linus Torvalds committed
521
522
		return -EINVAL;

523
524
525
526
527
528
529
530
531
532
533
534
535
536
	if (!userfaultfd_remove(vma, start, end)) {
		*prev = NULL; /* mmap_sem has been dropped, prev is stale */

		down_read(&current->mm->mmap_sem);
		vma = find_vma(current->mm, start);
		if (!vma)
			return -ENOMEM;
		if (start < vma->vm_start) {
			/*
			 * This "vma" under revalidation is the one
			 * with the lowest vma->vm_start where start
			 * is also < vma->vm_end. If start <
			 * vma->vm_start it means an hole materialized
			 * in the user address space within the
537
538
			 * virtual range passed to MADV_DONTNEED
			 * or MADV_FREE.
539
540
541
542
543
544
545
546
547
548
			 */
			return -ENOMEM;
		}
		if (!can_madv_dontneed_vma(vma))
			return -EINVAL;
		if (end > vma->vm_end) {
			/*
			 * Don't fail if end > vma->vm_end. If the old
			 * vma was splitted while the mmap_sem was
			 * released the effect of the concurrent
549
			 * operation may not cause madvise() to
550
551
552
553
554
555
556
557
558
559
560
			 * have an undefined result. There may be an
			 * adjacent next vma that we'll walk
			 * next. userfaultfd_remove() will generate an
			 * UFFD_EVENT_REMOVE repetition on the
			 * end-vma->vm_end range, but the manager can
			 * handle a repetition fine.
			 */
			end = vma->vm_end;
		}
		VM_WARN_ON(start >= end);
	}
561
562
563
564
565
566
567

	if (behavior == MADV_DONTNEED)
		return madvise_dontneed_single_vma(vma, start, end);
	else if (behavior == MADV_FREE)
		return madvise_free_single_vma(vma, start, end);
	else
		return -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
568
569
}

570
571
572
573
574
/*
 * Application wants to free up the pages and associated backing store.
 * This is effectively punching a hole into the middle of a file.
 */
static long madvise_remove(struct vm_area_struct *vma,
575
				struct vm_area_struct **prev,
576
577
				unsigned long start, unsigned long end)
{
578
	loff_t offset;
579
	int error;
580
	struct file *f;
581

582
	*prev = NULL;	/* tell sys_madvise we drop mmap_sem */
583

584
	if (vma->vm_flags & VM_LOCKED)
585
586
		return -EINVAL;

587
588
589
	f = vma->vm_file;

	if (!f || !f->f_mapping || !f->f_mapping->host) {
590
591
592
			return -EINVAL;
	}

593
594
595
	if ((vma->vm_flags & (VM_SHARED|VM_WRITE)) != (VM_SHARED|VM_WRITE))
		return -EACCES;

596
597
	offset = (loff_t)(start - vma->vm_start)
			+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
598

599
600
601
602
603
604
605
	/*
	 * Filesystem's fallocate may need to take i_mutex.  We need to
	 * explicitly grab a reference because the vma (and hence the
	 * vma's reference to the file) can go away as soon as we drop
	 * mmap_sem.
	 */
	get_file(f);
606
607
608
609
	if (userfaultfd_remove(vma, start, end)) {
		/* mmap_sem was not released by userfaultfd_remove() */
		up_read(&current->mm->mmap_sem);
	}
610
	error = vfs_fallocate(f,
611
612
				FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
				offset, end - start);
613
	fput(f);
614
	down_read(&current->mm->mmap_sem);
615
	return error;
616
617
}

618
619
620
621
#ifdef CONFIG_MEMORY_FAILURE
/*
 * Error injection support for memory error handling.
 */
622
623
static int madvise_inject_error(int behavior,
		unsigned long start, unsigned long end)
624
{
625
	struct page *page;
626
	struct zone *zone;
627
	unsigned int order;
628

629
630
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
631

632
633

	for (; start < end; start += PAGE_SIZE << order) {
634
635
		int ret;

636
		ret = get_user_pages_fast(start, 1, 0, &page);
637
638
		if (ret != 1)
			return ret;
639

640
641
642
643
644
645
646
		/*
		 * When soft offlining hugepages, after migrating the page
		 * we dissolve it, therefore in the second loop "page" will
		 * no longer be a compound page, and order will be 0.
		 */
		order = compound_order(compound_head(page));

647
648
		if (PageHWPoison(page)) {
			put_page(page);
649
650
			continue;
		}
651
652
653
654
655
656

		if (behavior == MADV_SOFT_OFFLINE) {
			pr_info("Soft offlining pfn %#lx at process virtual address %#lx\n",
						page_to_pfn(page), start);

			ret = soft_offline_page(page, MF_COUNT_INCREASED);
657
			if (ret)
658
				return ret;
659
660
			continue;
		}
661
662
663
		pr_info("Injecting memory failure for pfn %#lx at process virtual address %#lx\n",
						page_to_pfn(page), start);

664
		ret = memory_failure(page_to_pfn(page), MF_COUNT_INCREASED);
665
666
		if (ret)
			return ret;
667
	}
668
669
670
671
672

	/* Ensure that all poisoned pages are removed from per-cpu lists */
	for_each_populated_zone(zone)
		drain_all_pages(zone);

673
	return 0;
674
675
676
}
#endif

677
678
679
static long
madvise_vma(struct vm_area_struct *vma, struct vm_area_struct **prev,
		unsigned long start, unsigned long end, int behavior)
Linus Torvalds's avatar
Linus Torvalds committed
680
681
{
	switch (behavior) {
682
	case MADV_REMOVE:
683
		return madvise_remove(vma, prev, start, end);
Linus Torvalds's avatar
Linus Torvalds committed
684
	case MADV_WILLNEED:
685
		return madvise_willneed(vma, prev, start, end);
Minchan Kim's avatar
Minchan Kim committed
686
	case MADV_FREE:
Linus Torvalds's avatar
Linus Torvalds committed
687
	case MADV_DONTNEED:
688
		return madvise_dontneed_free(vma, prev, start, end, behavior);
Linus Torvalds's avatar
Linus Torvalds committed
689
	default:
690
		return madvise_behavior(vma, prev, start, end, behavior);
Linus Torvalds's avatar
Linus Torvalds committed
691
692
693
	}
}

694
static bool
695
696
697
698
699
700
701
702
703
704
705
madvise_behavior_valid(int behavior)
{
	switch (behavior) {
	case MADV_DOFORK:
	case MADV_DONTFORK:
	case MADV_NORMAL:
	case MADV_SEQUENTIAL:
	case MADV_RANDOM:
	case MADV_REMOVE:
	case MADV_WILLNEED:
	case MADV_DONTNEED:
Minchan Kim's avatar
Minchan Kim committed
706
	case MADV_FREE:
Hugh Dickins's avatar
Hugh Dickins committed
707
708
709
#ifdef CONFIG_KSM
	case MADV_MERGEABLE:
	case MADV_UNMERGEABLE:
Andrea Arcangeli's avatar
Andrea Arcangeli committed
710
711
712
#endif
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	case MADV_HUGEPAGE:
713
	case MADV_NOHUGEPAGE:
Hugh Dickins's avatar
Hugh Dickins committed
714
#endif
715
716
	case MADV_DONTDUMP:
	case MADV_DODUMP:
717
718
	case MADV_WIPEONFORK:
	case MADV_KEEPONFORK:
719
720
721
722
#ifdef CONFIG_MEMORY_FAILURE
	case MADV_SOFT_OFFLINE:
	case MADV_HWPOISON:
#endif
723
		return true;
724
725

	default:
726
		return false;
727
728
	}
}
729

Linus Torvalds's avatar
Linus Torvalds committed
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
/*
 * The madvise(2) system call.
 *
 * Applications can use madvise() to advise the kernel how it should
 * handle paging I/O in this VM area.  The idea is to help the kernel
 * use appropriate read-ahead and caching techniques.  The information
 * provided is advisory only, and can be safely disregarded by the
 * kernel without affecting the correct operation of the application.
 *
 * behavior values:
 *  MADV_NORMAL - the default behavior is to read clusters.  This
 *		results in some read-ahead and read-behind.
 *  MADV_RANDOM - the system should read the minimum amount of data
 *		on any access, since it is unlikely that the appli-
 *		cation will need more than what it asks for.
 *  MADV_SEQUENTIAL - pages in the given range will probably be accessed
 *		once, so they can be aggressively read ahead, and
 *		can be freed soon after they are accessed.
 *  MADV_WILLNEED - the application is notifying the system to read
 *		some pages ahead.
 *  MADV_DONTNEED - the application is finished with the given range,
 *		so the kernel can free resources associated with it.
752
753
 *  MADV_FREE - the application marks pages in the given range as lazy free,
 *		where actual purges are postponed until memory pressure happens.
754
755
 *  MADV_REMOVE - the application wants to free up the given range of
 *		pages and associated backing store.
756
757
758
 *  MADV_DONTFORK - omit this area from child's address space when forking:
 *		typically, to avoid COWing pages pinned by get_user_pages().
 *  MADV_DOFORK - cancel MADV_DONTFORK: no longer omit this area when forking.
759
760
761
 *  MADV_WIPEONFORK - present the child process with zero-filled memory in this
 *              range after a fork.
 *  MADV_KEEPONFORK - undo the effect of MADV_WIPEONFORK
762
763
764
 *  MADV_HWPOISON - trigger memory error handler as if the given memory range
 *		were corrupted by unrecoverable hardware memory failure.
 *  MADV_SOFT_OFFLINE - try to soft-offline the given range of memory.
Hugh Dickins's avatar
Hugh Dickins committed
765
766
767
 *  MADV_MERGEABLE - the application recommends that KSM try to merge pages in
 *		this area with pages of identical content from other such areas.
 *  MADV_UNMERGEABLE- cancel MADV_MERGEABLE: no longer merge pages with others.
768
769
770
771
772
773
774
775
776
 *  MADV_HUGEPAGE - the application wants to back the given range by transparent
 *		huge pages in the future. Existing pages might be coalesced and
 *		new pages might be allocated as THP.
 *  MADV_NOHUGEPAGE - mark the given range as not worth being backed by
 *		transparent huge pages so the existing pages will not be
 *		coalesced into THP and new pages will not be allocated as THP.
 *  MADV_DONTDUMP - the application wants to prevent pages in the given range
 *		from being included in its core dump.
 *  MADV_DODUMP - cancel MADV_DONTDUMP: no longer exclude from core dump.
Linus Torvalds's avatar
Linus Torvalds committed
777
778
779
780
781
 *
 * return values:
 *  zero    - success
 *  -EINVAL - start + len < 0, start is not page-aligned,
 *		"behavior" is not a valid value, or application
782
783
784
 *		is attempting to release locked or shared pages,
 *		or the specified address range includes file, Huge TLB,
 *		MAP_SHARED or VMPFNMAP range.
Linus Torvalds's avatar
Linus Torvalds committed
785
786
787
788
789
790
 *  -ENOMEM - addresses in the specified range are not currently
 *		mapped, or are outside the AS of the process.
 *  -EIO    - an I/O error occurred while paging in data.
 *  -EBADF  - map exists, but area maps something that isn't a file.
 *  -EAGAIN - a kernel resource was temporarily unavailable.
 */
791
SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
Linus Torvalds's avatar
Linus Torvalds committed
792
{
793
	unsigned long end, tmp;
794
	struct vm_area_struct *vma, *prev;
Linus Torvalds's avatar
Linus Torvalds committed
795
796
	int unmapped_error = 0;
	int error = -EINVAL;
797
	int write;
Linus Torvalds's avatar
Linus Torvalds committed
798
	size_t len;
799
	struct blk_plug plug;
Linus Torvalds's avatar
Linus Torvalds committed
800

801
802
803
	if (!madvise_behavior_valid(behavior))
		return error;

Linus Torvalds's avatar
Linus Torvalds committed
804
	if (start & ~PAGE_MASK)
805
		return error;
Linus Torvalds's avatar
Linus Torvalds committed
806
807
808
809
	len = (len_in + ~PAGE_MASK) & PAGE_MASK;

	/* Check to see whether len was rounded up from small -ve to zero */
	if (len_in && !len)
810
		return error;
Linus Torvalds's avatar
Linus Torvalds committed
811
812
813

	end = start + len;
	if (end < start)
814
		return error;
Linus Torvalds's avatar
Linus Torvalds committed
815
816
817

	error = 0;
	if (end == start)
818
819
		return error;

820
821
822
823
824
#ifdef CONFIG_MEMORY_FAILURE
	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
		return madvise_inject_error(behavior, start, start + len_in);
#endif

825
	write = madvise_need_mmap_write(behavior);
826
827
828
829
	if (write) {
		if (down_write_killable(&current->mm->mmap_sem))
			return -EINTR;
	} else {
830
		down_read(&current->mm->mmap_sem);
831
	}
Linus Torvalds's avatar
Linus Torvalds committed
832
833
834
835

	/*
	 * If the interval [start,end) covers some unmapped address
	 * ranges, just ignore them, but return -ENOMEM at the end.
836
	 * - different from the way of handling in mlock etc.
Linus Torvalds's avatar
Linus Torvalds committed
837
	 */
838
	vma = find_vma_prev(current->mm, start, &prev);
839
840
841
	if (vma && start > vma->vm_start)
		prev = vma;

842
	blk_start_plug(&plug);
Linus Torvalds's avatar
Linus Torvalds committed
843
844
845
846
	for (;;) {
		/* Still start < end. */
		error = -ENOMEM;
		if (!vma)
847
			goto out;
Linus Torvalds's avatar
Linus Torvalds committed
848

849
		/* Here start < (end|vma->vm_end). */
Linus Torvalds's avatar
Linus Torvalds committed
850
851
852
		if (start < vma->vm_start) {
			unmapped_error = -ENOMEM;
			start = vma->vm_start;
853
			if (start >= end)
854
				goto out;
Linus Torvalds's avatar
Linus Torvalds committed
855
856
		}

857
858
859
860
		/* Here vma->vm_start <= start < (end|vma->vm_end) */
		tmp = vma->vm_end;
		if (end < tmp)
			tmp = end;
Linus Torvalds's avatar
Linus Torvalds committed
861

862
863
		/* Here vma->vm_start <= start < tmp <= (end|vma->vm_end). */
		error = madvise_vma(vma, &prev, start, tmp, behavior);
Linus Torvalds's avatar
Linus Torvalds committed
864
		if (error)
865
			goto out;
866
		start = tmp;
867
		if (prev && start < prev->vm_end)
868
869
870
			start = prev->vm_end;
		error = unmapped_error;
		if (start >= end)
871
			goto out;
872
873
874
875
		if (prev)
			vma = prev->vm_next;
		else	/* madvise_remove dropped mmap_sem */
			vma = find_vma(current->mm, start);
Linus Torvalds's avatar
Linus Torvalds committed
876
877
	}
out:
878
	blk_finish_plug(&plug);
879
	if (write)
880
881
882
883
		up_write(&current->mm->mmap_sem);
	else
		up_read(&current->mm->mmap_sem);

Linus Torvalds's avatar
Linus Torvalds committed
884
885
	return error;
}