madvise.c 21.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10
/*
 *	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>
P
Prasanna Meda 已提交
11
#include <linux/mempolicy.h>
12
#include <linux/page-isolation.h>
13
#include <linux/userfaultfd_k.h>
L
Linus Torvalds 已提交
14
#include <linux/hugetlb.h>
15
#include <linux/falloc.h>
A
Alexey Dobriyan 已提交
16
#include <linux/sched.h>
H
Hugh Dickins 已提交
17
#include <linux/ksm.h>
18
#include <linux/fs.h>
19
#include <linux/file.h>
20
#include <linux/blkdev.h>
21
#include <linux/backing-dev.h>
22 23
#include <linux/swap.h>
#include <linux/swapops.h>
24
#include <linux/shmem_fs.h>
M
Minchan Kim 已提交
25 26 27
#include <linux/mmu_notifier.h>

#include <asm/tlb.h>
L
Linus Torvalds 已提交
28

29 30
#include "internal.h"

31 32 33 34 35 36 37 38 39 40 41
/*
 * 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:
M
Minchan Kim 已提交
42
	case MADV_FREE:
43 44 45 46 47 48 49
		return 0;
	default:
		/* be safe, default to 1. list exceptions explicitly */
		return 1;
	}
}

L
Linus Torvalds 已提交
50 51 52 53
/*
 * We can potentially split a vm area into separate
 * areas, each area with its own behavior.
 */
54
static long madvise_behavior(struct vm_area_struct *vma,
P
Prasanna Meda 已提交
55 56
		     struct vm_area_struct **prev,
		     unsigned long start, unsigned long end, int behavior)
L
Linus Torvalds 已提交
57
{
58
	struct mm_struct *mm = vma->vm_mm;
L
Linus Torvalds 已提交
59
	int error = 0;
P
Prasanna Meda 已提交
60
	pgoff_t pgoff;
H
Hugh Dickins 已提交
61
	unsigned long new_flags = vma->vm_flags;
62 63

	switch (behavior) {
64 65 66
	case MADV_NORMAL:
		new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
		break;
67
	case MADV_SEQUENTIAL:
68
		new_flags = (new_flags & ~VM_RAND_READ) | VM_SEQ_READ;
69 70
		break;
	case MADV_RANDOM:
71
		new_flags = (new_flags & ~VM_SEQ_READ) | VM_RAND_READ;
72
		break;
73 74 75 76
	case MADV_DONTFORK:
		new_flags |= VM_DONTCOPY;
		break;
	case MADV_DOFORK:
H
Hugh Dickins 已提交
77 78 79 80
		if (vma->vm_flags & VM_IO) {
			error = -EINVAL;
			goto out;
		}
81
		new_flags &= ~VM_DONTCOPY;
82
		break;
83
	case MADV_DONTDUMP:
84
		new_flags |= VM_DONTDUMP;
85 86
		break;
	case MADV_DODUMP:
87 88 89 90 91
		if (new_flags & VM_SPECIAL) {
			error = -EINVAL;
			goto out;
		}
		new_flags &= ~VM_DONTDUMP;
92
		break;
H
Hugh Dickins 已提交
93 94 95
	case MADV_MERGEABLE:
	case MADV_UNMERGEABLE:
		error = ksm_madvise(vma, start, end, behavior, &new_flags);
96 97 98 99 100 101 102
		if (error) {
			/*
			 * madvise() returns EAGAIN if kernel resources, such as
			 * slab, are temporarily unavailable.
			 */
			if (error == -ENOMEM)
				error = -EAGAIN;
H
Hugh Dickins 已提交
103
			goto out;
104
		}
H
Hugh Dickins 已提交
105
		break;
A
Andrea Arcangeli 已提交
106
	case MADV_HUGEPAGE:
A
Andrea Arcangeli 已提交
107
	case MADV_NOHUGEPAGE:
108
		error = hugepage_madvise(vma, &new_flags, behavior);
109 110 111 112 113 114 115
		if (error) {
			/*
			 * madvise() returns EAGAIN if kernel resources, such as
			 * slab, are temporarily unavailable.
			 */
			if (error == -ENOMEM)
				error = -EAGAIN;
A
Andrea Arcangeli 已提交
116
			goto out;
117
		}
A
Andrea Arcangeli 已提交
118
		break;
119 120
	}

P
Prasanna Meda 已提交
121 122
	if (new_flags == vma->vm_flags) {
		*prev = vma;
123
		goto out;
P
Prasanna Meda 已提交
124 125 126 127
	}

	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	*prev = vma_merge(mm, *prev, start, end, new_flags, vma->anon_vma,
128 129
			  vma->vm_file, pgoff, vma_policy(vma),
			  vma->vm_userfaultfd_ctx);
P
Prasanna Meda 已提交
130 131 132 133 134 135
	if (*prev) {
		vma = *prev;
		goto success;
	}

	*prev = vma;
L
Linus Torvalds 已提交
136 137

	if (start != vma->vm_start) {
138 139
		if (unlikely(mm->map_count >= sysctl_max_map_count)) {
			error = -ENOMEM;
L
Linus Torvalds 已提交
140
			goto out;
141 142 143 144 145 146 147 148 149 150 151
		}
		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;
		}
L
Linus Torvalds 已提交
152 153 154
	}

	if (end != vma->vm_end) {
155 156
		if (unlikely(mm->map_count >= sysctl_max_map_count)) {
			error = -ENOMEM;
L
Linus Torvalds 已提交
157
			goto out;
158 159 160 161 162 163 164 165 166 167 168
		}
		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;
		}
L
Linus Torvalds 已提交
169 170
	}

171
success:
L
Linus Torvalds 已提交
172 173 174
	/*
	 * vm_flags is protected by the mmap_sem held in write mode.
	 */
175
	vma->vm_flags = new_flags;
L
Linus Torvalds 已提交
176 177 178 179
out:
	return error;
}

180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200
#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);

201
		if (pte_present(pte) || pte_none(pte))
202 203 204 205 206 207 208 209
			continue;
		entry = pte_to_swp_entry(pte);
		if (unlikely(non_swap_entry(entry)))
			continue;

		page = read_swap_cache_async(entry, GFP_HIGHUSER_MOVABLE,
								vma, index);
		if (page)
210
			put_page(page);
211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240
	}

	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;

241
		page = find_get_entry(mapping, index);
242 243
		if (!radix_tree_exceptional_entry(page)) {
			if (page)
244
				put_page(page);
245 246 247 248 249 250
			continue;
		}
		swap = radix_to_swp_entry(page);
		page = read_swap_cache_async(swap, GFP_HIGHUSER_MOVABLE,
								NULL, 0);
		if (page)
251
			put_page(page);
252 253 254 255 256 257
	}

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

L
Linus Torvalds 已提交
258 259 260
/*
 * Schedule all required I/O operations.  Do not wait for completion.
 */
261 262
static long madvise_willneed(struct vm_area_struct *vma,
			     struct vm_area_struct **prev,
L
Linus Torvalds 已提交
263 264 265 266
			     unsigned long start, unsigned long end)
{
	struct file *file = vma->vm_file;

267
#ifdef CONFIG_SWAP
C
Christoph Hellwig 已提交
268
	if (!file) {
269
		*prev = vma;
C
Christoph Hellwig 已提交
270
		force_swapin_readahead(vma, start, end);
271 272 273
		return 0;
	}

C
Christoph Hellwig 已提交
274 275 276 277 278 279 280
	if (shmem_mapping(file->f_mapping)) {
		*prev = vma;
		force_shm_swapin_readahead(vma, start, end,
					file->f_mapping);
		return 0;
	}
#else
281 282
	if (!file)
		return -EBADF;
C
Christoph Hellwig 已提交
283
#endif
284

M
Matthew Wilcox 已提交
285
	if (IS_DAX(file_inode(file))) {
286 287 288 289
		/* no bad return value, but ignore advice */
		return 0;
	}

P
Prasanna Meda 已提交
290
	*prev = vma;
L
Linus Torvalds 已提交
291 292 293 294 295
	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;

296
	force_page_cache_readahead(file->f_mapping, file, start, end - start);
L
Linus Torvalds 已提交
297 298 299
	return 0;
}

M
Minchan Kim 已提交
300 301 302 303 304 305 306 307 308 309
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;
310
	int nr_swap = 0;
311 312 313 314 315 316
	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;
M
Minchan Kim 已提交
317 318 319 320

	if (pmd_trans_unstable(pmd))
		return 0;

321
	tlb_remove_check_page_size_change(tlb, PAGE_SIZE);
M
Minchan Kim 已提交
322 323 324 325 326
	orig_pte = pte = pte_offset_map_lock(mm, pmd, addr, &ptl);
	arch_enter_lazy_mmu_mode();
	for (; addr != end; pte++, addr += PAGE_SIZE) {
		ptent = *pte;

327
		if (pte_none(ptent))
M
Minchan Kim 已提交
328
			continue;
329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
		/*
		 * 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;
		}
M
Minchan Kim 已提交
345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415

		page = vm_normal_page(vma, addr, ptent);
		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;
			}
			put_page(page);
			unlock_page(page);
			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);
		}
S
Shaohua Li 已提交
416
		mark_page_lazyfree(page);
M
Minchan Kim 已提交
417 418
	}
out:
419 420 421 422 423 424
	if (nr_swap) {
		if (current->mm == mm)
			sync_mm_rss(mm);

		add_mm_counter(mm, MM_SWAPENTS, nr_swap);
	}
M
Minchan Kim 已提交
425 426 427
	arch_leave_lazy_mmu_mode();
	pte_unmap_unlock(orig_pte, ptl);
	cond_resched();
428
next:
M
Minchan Kim 已提交
429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487
	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;

	if (vma->vm_flags & (VM_LOCKED|VM_HUGETLB|VM_PFNMAP))
		return -EINVAL;

	/* 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;
}

static long madvise_free(struct vm_area_struct *vma,
			     struct vm_area_struct **prev,
			     unsigned long start, unsigned long end)
{
	*prev = vma;
	return madvise_free_single_vma(vma, start, end);
}

L
Linus Torvalds 已提交
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
L
Linus Torvalds 已提交
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.
L
Linus Torvalds 已提交
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
static long madvise_dontneed(struct vm_area_struct *vma,
			     struct vm_area_struct **prev,
L
Linus Torvalds 已提交
509 510
			     unsigned long start, unsigned long end)
{
P
Prasanna Meda 已提交
511
	*prev = vma;
512
	if (!can_madv_dontneed_vma(vma))
L
Linus Torvalds 已提交
513 514
		return -EINVAL;

515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551
	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
			 * virtual range passed to MADV_DONTNEED.
			 */
			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
			 * operation may not cause MADV_DONTNEED to
			 * 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);
	}
552
	zap_page_range(vma, start, end - start);
L
Linus Torvalds 已提交
553 554 555
	return 0;
}

556 557 558 559 560
/*
 * 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,
561
				struct vm_area_struct **prev,
562 563
				unsigned long start, unsigned long end)
{
564
	loff_t offset;
565
	int error;
566
	struct file *f;
567

568
	*prev = NULL;	/* tell sys_madvise we drop mmap_sem */
569

570
	if (vma->vm_flags & VM_LOCKED)
571 572
		return -EINVAL;

573 574 575
	f = vma->vm_file;

	if (!f || !f->f_mapping || !f->f_mapping->host) {
576 577 578
			return -EINVAL;
	}

579 580 581
	if ((vma->vm_flags & (VM_SHARED|VM_WRITE)) != (VM_SHARED|VM_WRITE))
		return -EACCES;

582 583
	offset = (loff_t)(start - vma->vm_start)
			+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
584

585 586 587 588 589 590 591
	/*
	 * 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);
592 593 594 595
	if (userfaultfd_remove(vma, start, end)) {
		/* mmap_sem was not released by userfaultfd_remove() */
		up_read(&current->mm->mmap_sem);
	}
596
	error = vfs_fallocate(f,
597 598
				FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
				offset, end - start);
599
	fput(f);
600
	down_read(&current->mm->mmap_sem);
601
	return error;
602 603
}

604 605 606 607
#ifdef CONFIG_MEMORY_FAILURE
/*
 * Error injection support for memory error handling.
 */
608 609
static int madvise_inject_error(int behavior,
		unsigned long start, unsigned long end)
610
{
611 612
	struct page *page;

613 614
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
615

616
	for (; start < end; start += PAGE_SIZE <<
617
				compound_order(compound_head(page))) {
618 619
		int ret;

620
		ret = get_user_pages_fast(start, 1, 0, &page);
621 622
		if (ret != 1)
			return ret;
623

624 625
		if (PageHWPoison(page)) {
			put_page(page);
626 627
			continue;
		}
628 629 630 631 632 633

		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);
634
			if (ret)
635
				return ret;
636 637
			continue;
		}
638 639 640 641
		pr_info("Injecting memory failure for pfn %#lx at process virtual address %#lx\n",
						page_to_pfn(page), start);

		ret = memory_failure(page_to_pfn(page), 0, MF_COUNT_INCREASED);
642 643
		if (ret)
			return ret;
644
	}
645
	return 0;
646 647 648
}
#endif

649 650 651
static long
madvise_vma(struct vm_area_struct *vma, struct vm_area_struct **prev,
		unsigned long start, unsigned long end, int behavior)
L
Linus Torvalds 已提交
652 653
{
	switch (behavior) {
654
	case MADV_REMOVE:
H
Hugh Dickins 已提交
655
		return madvise_remove(vma, prev, start, end);
L
Linus Torvalds 已提交
656
	case MADV_WILLNEED:
H
Hugh Dickins 已提交
657
		return madvise_willneed(vma, prev, start, end);
M
Minchan Kim 已提交
658
	case MADV_FREE:
659
		return madvise_free(vma, prev, start, end);
L
Linus Torvalds 已提交
660
	case MADV_DONTNEED:
H
Hugh Dickins 已提交
661
		return madvise_dontneed(vma, prev, start, end);
L
Linus Torvalds 已提交
662
	default:
H
Hugh Dickins 已提交
663
		return madvise_behavior(vma, prev, start, end, behavior);
L
Linus Torvalds 已提交
664 665 666
	}
}

667
static bool
N
Nick Piggin 已提交
668 669 670 671 672 673 674 675 676 677 678
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:
M
Minchan Kim 已提交
679
	case MADV_FREE:
H
Hugh Dickins 已提交
680 681 682
#ifdef CONFIG_KSM
	case MADV_MERGEABLE:
	case MADV_UNMERGEABLE:
A
Andrea Arcangeli 已提交
683 684 685
#endif
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	case MADV_HUGEPAGE:
A
Andrea Arcangeli 已提交
686
	case MADV_NOHUGEPAGE:
H
Hugh Dickins 已提交
687
#endif
688 689
	case MADV_DONTDUMP:
	case MADV_DODUMP:
690
		return true;
N
Nick Piggin 已提交
691 692

	default:
693
		return false;
N
Nick Piggin 已提交
694 695
	}
}
H
Hugh Dickins 已提交
696

L
Linus Torvalds 已提交
697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718
/*
 * 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.
719 720
 *  MADV_FREE - the application marks pages in the given range as lazy free,
 *		where actual purges are postponed until memory pressure happens.
721 722
 *  MADV_REMOVE - the application wants to free up the given range of
 *		pages and associated backing store.
H
Hugh Dickins 已提交
723 724 725
 *  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.
726 727 728
 *  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.
H
Hugh Dickins 已提交
729 730 731
 *  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.
732 733 734 735 736 737 738 739 740
 *  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.
L
Linus Torvalds 已提交
741 742 743 744 745 746 747 748 749 750 751 752
 *
 * return values:
 *  zero    - success
 *  -EINVAL - start + len < 0, start is not page-aligned,
 *		"behavior" is not a valid value, or application
 *		is attempting to release locked or shared pages.
 *  -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.
 */
753
SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
L
Linus Torvalds 已提交
754
{
P
Prasanna Meda 已提交
755
	unsigned long end, tmp;
756
	struct vm_area_struct *vma, *prev;
L
Linus Torvalds 已提交
757 758
	int unmapped_error = 0;
	int error = -EINVAL;
759
	int write;
L
Linus Torvalds 已提交
760
	size_t len;
761
	struct blk_plug plug;
L
Linus Torvalds 已提交
762

763
#ifdef CONFIG_MEMORY_FAILURE
764
	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
765
		return madvise_inject_error(behavior, start, start + len_in);
766
#endif
N
Nick Piggin 已提交
767 768 769
	if (!madvise_behavior_valid(behavior))
		return error;

L
Linus Torvalds 已提交
770
	if (start & ~PAGE_MASK)
771
		return error;
L
Linus Torvalds 已提交
772 773 774 775
	len = (len_in + ~PAGE_MASK) & PAGE_MASK;

	/* Check to see whether len was rounded up from small -ve to zero */
	if (len_in && !len)
776
		return error;
L
Linus Torvalds 已提交
777 778 779

	end = start + len;
	if (end < start)
780
		return error;
L
Linus Torvalds 已提交
781 782 783

	error = 0;
	if (end == start)
784 785 786
		return error;

	write = madvise_need_mmap_write(behavior);
787 788 789 790
	if (write) {
		if (down_write_killable(&current->mm->mmap_sem))
			return -EINTR;
	} else {
791
		down_read(&current->mm->mmap_sem);
792
	}
L
Linus Torvalds 已提交
793 794 795 796

	/*
	 * If the interval [start,end) covers some unmapped address
	 * ranges, just ignore them, but return -ENOMEM at the end.
P
Prasanna Meda 已提交
797
	 * - different from the way of handling in mlock etc.
L
Linus Torvalds 已提交
798
	 */
P
Prasanna Meda 已提交
799
	vma = find_vma_prev(current->mm, start, &prev);
800 801 802
	if (vma && start > vma->vm_start)
		prev = vma;

803
	blk_start_plug(&plug);
L
Linus Torvalds 已提交
804 805 806 807
	for (;;) {
		/* Still start < end. */
		error = -ENOMEM;
		if (!vma)
808
			goto out;
L
Linus Torvalds 已提交
809

P
Prasanna Meda 已提交
810
		/* Here start < (end|vma->vm_end). */
L
Linus Torvalds 已提交
811 812 813
		if (start < vma->vm_start) {
			unmapped_error = -ENOMEM;
			start = vma->vm_start;
P
Prasanna Meda 已提交
814
			if (start >= end)
815
				goto out;
L
Linus Torvalds 已提交
816 817
		}

P
Prasanna Meda 已提交
818 819 820 821
		/* Here vma->vm_start <= start < (end|vma->vm_end) */
		tmp = vma->vm_end;
		if (end < tmp)
			tmp = end;
L
Linus Torvalds 已提交
822

P
Prasanna Meda 已提交
823 824
		/* Here vma->vm_start <= start < tmp <= (end|vma->vm_end). */
		error = madvise_vma(vma, &prev, start, tmp, behavior);
L
Linus Torvalds 已提交
825
		if (error)
826
			goto out;
P
Prasanna Meda 已提交
827
		start = tmp;
828
		if (prev && start < prev->vm_end)
P
Prasanna Meda 已提交
829 830 831
			start = prev->vm_end;
		error = unmapped_error;
		if (start >= end)
832
			goto out;
833 834 835 836
		if (prev)
			vma = prev->vm_next;
		else	/* madvise_remove dropped mmap_sem */
			vma = find_vma(current->mm, start);
L
Linus Torvalds 已提交
837 838
	}
out:
839
	blk_finish_plug(&plug);
840
	if (write)
841 842 843 844
		up_write(&current->mm->mmap_sem);
	else
		up_read(&current->mm->mmap_sem);

L
Linus Torvalds 已提交
845 846
	return error;
}