[linux-2.6-block.git] / mm / madvise.c

/*
 *	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>
#include <linux/mempolicy.h>
#include <linux/page-isolation.h>
#include <linux/hugetlb.h>
#include <linux/falloc.h>
#include <linux/sched.h>
#include <linux/ksm.h>
#include <linux/fs.h>
#include <linux/file.h>
#include <linux/blkdev.h>
#include <linux/swap.h>
#include <linux/swapops.h>

/*
 * 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:
		return 0;
	default:
		/* be safe, default to 1. list exceptions explicitly */
		return 1;
	}
}

/*
 * We can potentially split a vm area into separate
 * areas, each area with its own behavior.
 */
static long madvise_behavior(struct vm_area_struct *vma,
		     struct vm_area_struct **prev,
		     unsigned long start, unsigned long end, int behavior)
{
	struct mm_struct *mm = vma->vm_mm;
	int error = 0;
	pgoff_t pgoff;
	unsigned long new_flags = vma->vm_flags;

	switch (behavior) {
	case MADV_NORMAL:
		new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
		break;
	case MADV_SEQUENTIAL:
		new_flags = (new_flags & ~VM_RAND_READ) | VM_SEQ_READ;
		break;
	case MADV_RANDOM:
		new_flags = (new_flags & ~VM_SEQ_READ) | VM_RAND_READ;
		break;
	case MADV_DONTFORK:
		new_flags |= VM_DONTCOPY;
		break;
	case MADV_DOFORK:
		if (vma->vm_flags & VM_IO) {
			error = -EINVAL;
			goto out;
		}
		new_flags &= ~VM_DONTCOPY;
		break;
	case MADV_DONTDUMP:
		new_flags |= VM_DONTDUMP;
		break;
	case MADV_DODUMP:
		if (new_flags & VM_SPECIAL) {
			error = -EINVAL;
			goto out;
		}
		new_flags &= ~VM_DONTDUMP;
		break;
	case MADV_MERGEABLE:
	case MADV_UNMERGEABLE:
		error = ksm_madvise(vma, start, end, behavior, &new_flags);
		if (error)
			goto out;
		break;
	case MADV_HUGEPAGE:
	case MADV_NOHUGEPAGE:
		error = hugepage_madvise(vma, &new_flags, behavior);
		if (error)
			goto out;
		break;
	}

	if (new_flags == vma->vm_flags) {
		*prev = vma;
		goto out;
	}

	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	*prev = vma_merge(mm, *prev, start, end, new_flags, vma->anon_vma,
				vma->vm_file, pgoff, vma_policy(vma));
	if (*prev) {
		vma = *prev;
		goto success;
	}

	*prev = vma;

	if (start != vma->vm_start) {
		error = split_vma(mm, vma, start, 1);
		if (error)
			goto out;
	}

	if (end != vma->vm_end) {
		error = split_vma(mm, vma, end, 0);
		if (error)
			goto out;
	}

success:
	/*
	 * vm_flags is protected by the mmap_sem held in write mode.
	 */
	vma->vm_flags = new_flags;

out:
	if (error == -ENOMEM)
		error = -EAGAIN;
	return error;
}

#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);

		if (pte_present(pte) || pte_none(pte))
			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)
			page_cache_release(page);
	}

	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;

		page = find_get_entry(mapping, index);
		if (!radix_tree_exceptional_entry(page)) {
			if (page)
				page_cache_release(page);
			continue;
		}
		swap = radix_to_swp_entry(page);
		page = read_swap_cache_async(swap, GFP_HIGHUSER_MOVABLE,
								NULL, 0);
		if (page)
			page_cache_release(page);
	}

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

/*
 * Schedule all required I/O operations.  Do not wait for completion.
 */
static long madvise_willneed(struct vm_area_struct *vma,
			     struct vm_area_struct **prev,
			     unsigned long start, unsigned long end)
{
	struct file *file = vma->vm_file;

#ifdef CONFIG_SWAP
	if (!file) {
		*prev = vma;
		force_swapin_readahead(vma, start, end);
		return 0;
	}

	if (shmem_mapping(file->f_mapping)) {
		*prev = vma;
		force_shm_swapin_readahead(vma, start, end,
					file->f_mapping);
		return 0;
	}
#else
	if (!file)
		return -EBADF;
#endif

	if (IS_DAX(file_inode(file))) {
		/* no bad return value, but ignore advice */
		return 0;
	}

	*prev = vma;
	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;

	force_page_cache_readahead(file->f_mapping, file, start, end - start);
	return 0;
}

/*
 * 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
 * zap_page_range call sets things up for shrink_active_list to actually free
 * 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
 * shrink_active_list to pick up before reclaiming other pages.
 *
 * 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).
 */
static long madvise_dontneed(struct vm_area_struct *vma,
			     struct vm_area_struct **prev,
			     unsigned long start, unsigned long end)
{
	*prev = vma;
	if (vma->vm_flags & (VM_LOCKED|VM_HUGETLB|VM_PFNMAP))
		return -EINVAL;

	zap_page_range(vma, start, end - start, NULL);
	return 0;
}

/*
 * 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,
				struct vm_area_struct **prev,
				unsigned long start, unsigned long end)
{
	loff_t offset;
	int error;
	struct file *f;

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

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

	f = vma->vm_file;

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

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

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

	/*
	 * 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);
	up_read(&current->mm->mmap_sem);
	error = vfs_fallocate(f,
				FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
				offset, end - start);
	fput(f);
	down_read(&current->mm->mmap_sem);
	return error;
}

#ifdef CONFIG_MEMORY_FAILURE
/*
 * Error injection support for memory error handling.
 */
static int madvise_hwpoison(int bhv, unsigned long start, unsigned long end)
{
	struct page *p;
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
	for (; start < end; start += PAGE_SIZE <<
				compound_order(compound_head(p))) {
		int ret;

		ret = get_user_pages_fast(start, 1, 0, &p);
		if (ret != 1)
			return ret;

		if (PageHWPoison(p)) {
			put_page(p);
			continue;
		}
		if (bhv == MADV_SOFT_OFFLINE) {
			pr_info("Soft offlining page %#lx at %#lx\n",
				page_to_pfn(p), start);
			ret = soft_offline_page(p, MF_COUNT_INCREASED);
			if (ret)
				return ret;
			continue;
		}
		pr_info("Injecting memory failure for page %#lx at %#lx\n",
		       page_to_pfn(p), start);
		/* Ignore return value for now */
		memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED);
	}
	return 0;
}
#endif

static long
madvise_vma(struct vm_area_struct *vma, struct vm_area_struct **prev,
		unsigned long start, unsigned long end, int behavior)
{
	switch (behavior) {
	case MADV_REMOVE:
		return madvise_remove(vma, prev, start, end);
	case MADV_WILLNEED:
		return madvise_willneed(vma, prev, start, end);
	case MADV_DONTNEED:
		return madvise_dontneed(vma, prev, start, end);
	default:
		return madvise_behavior(vma, prev, start, end, behavior);
	}
}

static int
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:
#ifdef CONFIG_KSM
	case MADV_MERGEABLE:
	case MADV_UNMERGEABLE:
#endif
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	case MADV_HUGEPAGE:
	case MADV_NOHUGEPAGE:
#endif
	case MADV_DONTDUMP:
	case MADV_DODUMP:
		return 1;

	default:
		return 0;
	}
}

/*
 * 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.
 *  MADV_REMOVE - the application wants to free up the given range of
 *		pages and associated backing store.
 *  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.
 *  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.
 *
 * 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.
 */
SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
{
	unsigned long end, tmp;
	struct vm_area_struct *vma, *prev;
	int unmapped_error = 0;
	int error = -EINVAL;
	int write;
	size_t len;
	struct blk_plug plug;

#ifdef CONFIG_MEMORY_FAILURE
	if (behavior == MADV_HWPOISON || behavior == MADV_SOFT_OFFLINE)
		return madvise_hwpoison(behavior, start, start+len_in);
#endif
	if (!madvise_behavior_valid(behavior))
		return error;

	if (start & ~PAGE_MASK)
		return error;
	len = (len_in + ~PAGE_MASK) & PAGE_MASK;

	/* Check to see whether len was rounded up from small -ve to zero */
	if (len_in && !len)
		return error;

	end = start + len;
	if (end < start)
		return error;

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

	write = madvise_need_mmap_write(behavior);
	if (write)
		down_write(&current->mm->mmap_sem);
	else
		down_read(&current->mm->mmap_sem);

	/*
	 * If the interval [start,end) covers some unmapped address
	 * ranges, just ignore them, but return -ENOMEM at the end.
	 * - different from the way of handling in mlock etc.
	 */
	vma = find_vma_prev(current->mm, start, &prev);
	if (vma && start > vma->vm_start)
		prev = vma;

	blk_start_plug(&plug);
	for (;;) {
		/* Still start < end. */
		error = -ENOMEM;
		if (!vma)
			goto out;

		/* Here start < (end|vma->vm_end). */
		if (start < vma->vm_start) {
			unmapped_error = -ENOMEM;
			start = vma->vm_start;
			if (start >= end)
				goto out;
		}

		/* Here vma->vm_start <= start < (end|vma->vm_end) */
		tmp = vma->vm_end;
		if (end < tmp)
			tmp = end;

		/* Here vma->vm_start <= start < tmp <= (end|vma->vm_end). */
		error = madvise_vma(vma, &prev, start, tmp, behavior);
		if (error)
			goto out;
		start = tmp;
		if (prev && start < prev->vm_end)
			start = prev->vm_end;
		error = unmapped_error;
		if (start >= end)
			goto out;
		if (prev)
			vma = prev->vm_next;
		else	/* madvise_remove dropped mmap_sem */
			vma = find_vma(current->mm, start);
	}
out:
	blk_finish_plug(&plug);
	if (write)
		up_write(&current->mm->mmap_sem);
	else
		up_read(&current->mm->mmap_sem);

	return error;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/mm/madvise.c
	3	*
	4	* Copyright (C) 1999 Linus Torvalds
	5	* Copyright (C) 2002 Christoph Hellwig
	6	*/
	7
	8	#include <linux/mman.h>
	9	#include <linux/pagemap.h>
	10	#include <linux/syscalls.h>
05b74384	11	#include <linux/mempolicy.h>
afcf938e	12	#include <linux/page-isolation.h>
1da177e4	13	#include <linux/hugetlb.h>
3f31d075	14	#include <linux/falloc.h>
e8edc6e0	15	#include <linux/sched.h>
f8af4da3	16	#include <linux/ksm.h>
3f31d075	17	#include <linux/fs.h>
9ab4233d	18	#include <linux/file.h>
1998cc04 SL	19	#include <linux/blkdev.h>
	20	#include <linux/swap.h>
	21	#include <linux/swapops.h>
1da177e4	22
0a27a14a NP	23	/*
	24	* Any behaviour which results in changes to the vma->vm_flags needs to
	25	* take mmap_sem for writing. Others, which simply traverse vmas, need
	26	* to only take it for reading.
	27	*/
	28	static int madvise_need_mmap_write(int behavior)
	29	{
	30	switch (behavior) {
	31	case MADV_REMOVE:
	32	case MADV_WILLNEED:
	33	case MADV_DONTNEED:
	34	return 0;
	35	default:
	36	/* be safe, default to 1. list exceptions explicitly */
	37	return 1;
	38	}
	39	}
	40
1da177e4 LT	41	/*
	42	* We can potentially split a vm area into separate
	43	* areas, each area with its own behavior.
	44	*/
ec9bed9d	45	static long madvise_behavior(struct vm_area_struct *vma,
05b74384 PM	46	struct vm_area_struct **prev,
05b74384 PM	47	unsigned long start, unsigned long end, int behavior)
1da177e4	48	{
ec9bed9d	49	struct mm_struct *mm = vma->vm_mm;
1da177e4	50	int error = 0;
05b74384	51	pgoff_t pgoff;
3866ea90	52	unsigned long new_flags = vma->vm_flags;
e798c6e8 PM	53
e798c6e8 PM	54	switch (behavior) {
f8225661 MT	55	case MADV_NORMAL:
	56	new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
	57	break;
e798c6e8	58	case MADV_SEQUENTIAL:
f8225661	59	new_flags = (new_flags & ~VM_RAND_READ) \| VM_SEQ_READ;
e798c6e8 PM	60	break;
e798c6e8 PM	61	case MADV_RANDOM:
f8225661	62	new_flags = (new_flags & ~VM_SEQ_READ) \| VM_RAND_READ;
e798c6e8	63	break;
f8225661 MT	64	case MADV_DONTFORK:
	65	new_flags \|= VM_DONTCOPY;
	66	break;
	67	case MADV_DOFORK:
3866ea90 HD	68	if (vma->vm_flags & VM_IO) {
	69	error = -EINVAL;
	70	goto out;
	71	}
f8225661	72	new_flags &= ~VM_DONTCOPY;
e798c6e8	73	break;
accb61fe	74	case MADV_DONTDUMP:
0103bd16	75	new_flags \|= VM_DONTDUMP;
accb61fe JB	76	break;
accb61fe JB	77	case MADV_DODUMP:
0103bd16 KK	78	if (new_flags & VM_SPECIAL) {
	79	error = -EINVAL;
	80	goto out;
	81	}
	82	new_flags &= ~VM_DONTDUMP;
accb61fe	83	break;
f8af4da3 HD	84	case MADV_MERGEABLE:
	85	case MADV_UNMERGEABLE:
	86	error = ksm_madvise(vma, start, end, behavior, &new_flags);
	87	if (error)
	88	goto out;
	89	break;
0af4e98b	90	case MADV_HUGEPAGE:
a664b2d8	91	case MADV_NOHUGEPAGE:
60ab3244	92	error = hugepage_madvise(vma, &new_flags, behavior);
0af4e98b AA	93	if (error)
	94	goto out;
	95	break;
e798c6e8 PM	96	}
e798c6e8 PM	97
05b74384 PM	98	if (new_flags == vma->vm_flags) {
05b74384 PM	99	*prev = vma;
836d5ffd	100	goto out;
05b74384 PM	101	}
	102
	103	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	104	prev = vma_merge(mm, prev, start, end, new_flags, vma->anon_vma,
	105	vma->vm_file, pgoff, vma_policy(vma));
	106	if (*prev) {
	107	vma = *prev;
	108	goto success;
	109	}
	110
	111	*prev = vma;
1da177e4 LT	112
	113	if (start != vma->vm_start) {
	114	error = split_vma(mm, vma, start, 1);
	115	if (error)
	116	goto out;
	117	}
	118
	119	if (end != vma->vm_end) {
	120	error = split_vma(mm, vma, end, 0);
	121	if (error)
	122	goto out;
	123	}
	124
836d5ffd	125	success:
1da177e4 LT	126	/*
	127	* vm_flags is protected by the mmap_sem held in write mode.
	128	*/
e798c6e8	129	vma->vm_flags = new_flags;
1da177e4 LT	130
	131	out:
	132	if (error == -ENOMEM)
	133	error = -EAGAIN;
	134	return error;
	135	}
	136
1998cc04 SL	137	#ifdef CONFIG_SWAP
	138	static int swapin_walk_pmd_entry(pmd_t *pmd, unsigned long start,
	139	unsigned long end, struct mm_walk *walk)
	140	{
	141	pte_t *orig_pte;
	142	struct vm_area_struct *vma = walk->private;
	143	unsigned long index;
	144
	145	if (pmd_none_or_trans_huge_or_clear_bad(pmd))
	146	return 0;
	147
	148	for (index = start; index != end; index += PAGE_SIZE) {
	149	pte_t pte;
	150	swp_entry_t entry;
	151	struct page *page;
	152	spinlock_t *ptl;
	153
	154	orig_pte = pte_offset_map_lock(vma->vm_mm, pmd, start, &ptl);
	155	pte = *(orig_pte + ((index - start) / PAGE_SIZE));
	156	pte_unmap_unlock(orig_pte, ptl);
	157
0661a336	158	if (pte_present(pte) \|\| pte_none(pte))
1998cc04 SL	159	continue;
	160	entry = pte_to_swp_entry(pte);
	161	if (unlikely(non_swap_entry(entry)))
	162	continue;
	163
	164	page = read_swap_cache_async(entry, GFP_HIGHUSER_MOVABLE,
	165	vma, index);
	166	if (page)
	167	page_cache_release(page);
	168	}
	169
	170	return 0;
	171	}
	172
	173	static void force_swapin_readahead(struct vm_area_struct *vma,
	174	unsigned long start, unsigned long end)
	175	{
	176	struct mm_walk walk = {
	177	.mm = vma->vm_mm,
	178	.pmd_entry = swapin_walk_pmd_entry,
	179	.private = vma,
	180	};
	181
	182	walk_page_range(start, end, &walk);
	183
	184	lru_add_drain(); /* Push any new pages onto the LRU now */
	185	}
	186
	187	static void force_shm_swapin_readahead(struct vm_area_struct *vma,
	188	unsigned long start, unsigned long end,
	189	struct address_space *mapping)
	190	{
	191	pgoff_t index;
	192	struct page *page;
	193	swp_entry_t swap;
	194
	195	for (; start < end; start += PAGE_SIZE) {
	196	index = ((start - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	197
55231e5c	198	page = find_get_entry(mapping, index);
1998cc04 SL	199	if (!radix_tree_exceptional_entry(page)) {
	200	if (page)
	201	page_cache_release(page);
	202	continue;
	203	}
	204	swap = radix_to_swp_entry(page);
	205	page = read_swap_cache_async(swap, GFP_HIGHUSER_MOVABLE,
	206	NULL, 0);
	207	if (page)
	208	page_cache_release(page);
	209	}
	210
	211	lru_add_drain(); /* Push any new pages onto the LRU now */
	212	}
	213	#endif /* CONFIG_SWAP */
	214
1da177e4 LT	215	/*
	216	* Schedule all required I/O operations. Do not wait for completion.
	217	*/
ec9bed9d VC	218	static long madvise_willneed(struct vm_area_struct *vma,
ec9bed9d VC	219	struct vm_area_struct **prev,
1da177e4 LT	220	unsigned long start, unsigned long end)
	221	{
	222	struct file *file = vma->vm_file;
	223
1998cc04	224	#ifdef CONFIG_SWAP
97b713ba	225	if (!file) {
1998cc04	226	*prev = vma;
97b713ba	227	force_swapin_readahead(vma, start, end);
1998cc04 SL	228	return 0;
1998cc04 SL	229	}
1998cc04	230
97b713ba CH	231	if (shmem_mapping(file->f_mapping)) {
	232	*prev = vma;
	233	force_shm_swapin_readahead(vma, start, end,
	234	file->f_mapping);
	235	return 0;
	236	}
	237	#else
1bef4003 S	238	if (!file)
1bef4003 S	239	return -EBADF;
97b713ba	240	#endif
1bef4003	241
e748dcd0	242	if (IS_DAX(file_inode(file))) {
fe77ba6f CO	243	/* no bad return value, but ignore advice */
	244	return 0;
	245	}
	246
05b74384	247	*prev = vma;
1da177e4 LT	248	start = ((start - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	249	if (end > vma->vm_end)
	250	end = vma->vm_end;
	251	end = ((end - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	252
f7e839dd	253	force_page_cache_readahead(file->f_mapping, file, start, end - start);
1da177e4 LT	254	return 0;
	255	}
	256
	257	/*
	258	* Application no longer needs these pages. If the pages are dirty,
	259	* it's OK to just throw them away. The app will be more careful about
	260	* data it wants to keep. Be sure to free swap resources too. The
7e6cbea3	261	* zap_page_range call sets things up for shrink_active_list to actually free
1da177e4 LT	262	* these pages later if no one else has touched them in the meantime,
1da177e4 LT	263	* although we could add these pages to a global reuse list for
7e6cbea3	264	* shrink_active_list to pick up before reclaiming other pages.
1da177e4 LT	265	*
	266	* NB: This interface discards data rather than pushes it out to swap,
	267	* as some implementations do. This has performance implications for
	268	* applications like large transactional databases which want to discard
	269	* pages in anonymous maps after committing to backing store the data
	270	* that was kept in them. There is no reason to write this data out to
	271	* the swap area if the application is discarding it.
	272	*
	273	* An interface that causes the system to free clean pages and flush
	274	* dirty pages is already available as msync(MS_INVALIDATE).
	275	*/
ec9bed9d VC	276	static long madvise_dontneed(struct vm_area_struct *vma,
ec9bed9d VC	277	struct vm_area_struct **prev,
1da177e4 LT	278	unsigned long start, unsigned long end)
1da177e4 LT	279	{
05b74384	280	*prev = vma;
6aab341e	281	if (vma->vm_flags & (VM_LOCKED\|VM_HUGETLB\|VM_PFNMAP))
1da177e4 LT	282	return -EINVAL;
1da177e4 LT	283
8a5f14a2	284	zap_page_range(vma, start, end - start, NULL);
1da177e4 LT	285	return 0;
	286	}
	287
f6b3ec23 BP	288	/*
	289	* Application wants to free up the pages and associated backing store.
	290	* This is effectively punching a hole into the middle of a file.
f6b3ec23 BP	291	*/
f6b3ec23 BP	292	static long madvise_remove(struct vm_area_struct *vma,
00e9fa2d	293	struct vm_area_struct **prev,
f6b3ec23 BP	294	unsigned long start, unsigned long end)
f6b3ec23 BP	295	{
3f31d075	296	loff_t offset;
90ed52eb	297	int error;
9ab4233d	298	struct file *f;
f6b3ec23	299
90ed52eb	300	prev = NULL; / tell sys_madvise we drop mmap_sem */
00e9fa2d	301
0661a336	302	if (vma->vm_flags & (VM_LOCKED \| VM_HUGETLB))
f6b3ec23 BP	303	return -EINVAL;
f6b3ec23 BP	304
9ab4233d AL	305	f = vma->vm_file;
	306
	307	if (!f \|\| !f->f_mapping \|\| !f->f_mapping->host) {
f6b3ec23 BP	308	return -EINVAL;
	309	}
	310
69cf0fac HD	311	if ((vma->vm_flags & (VM_SHARED\|VM_WRITE)) != (VM_SHARED\|VM_WRITE))
	312	return -EACCES;
	313
f6b3ec23 BP	314	offset = (loff_t)(start - vma->vm_start)
f6b3ec23 BP	315	+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
90ed52eb	316
9ab4233d AL	317	/*
	318	* Filesystem's fallocate may need to take i_mutex. We need to
	319	* explicitly grab a reference because the vma (and hence the
	320	* vma's reference to the file) can go away as soon as we drop
	321	* mmap_sem.
	322	*/
	323	get_file(f);
0a27a14a	324	up_read(&current->mm->mmap_sem);
72c72bdf	325	error = vfs_fallocate(f,
3f31d075 HD	326	FALLOC_FL_PUNCH_HOLE \| FALLOC_FL_KEEP_SIZE,
3f31d075 HD	327	offset, end - start);
9ab4233d	328	fput(f);
0a27a14a	329	down_read(&current->mm->mmap_sem);
90ed52eb	330	return error;
f6b3ec23 BP	331	}
f6b3ec23 BP	332
9893e49d AK	333	#ifdef CONFIG_MEMORY_FAILURE
	334	/*
	335	* Error injection support for memory error handling.
	336	*/
afcf938e	337	static int madvise_hwpoison(int bhv, unsigned long start, unsigned long end)
9893e49d	338	{
20cb6cab	339	struct page *p;
9893e49d AK	340	if (!capable(CAP_SYS_ADMIN))
9893e49d AK	341	return -EPERM;
20cb6cab WL	342	for (; start < end; start += PAGE_SIZE <<
20cb6cab WL	343	compound_order(compound_head(p))) {
325c4ef5 AM	344	int ret;
	345
	346	ret = get_user_pages_fast(start, 1, 0, &p);
9893e49d AK	347	if (ret != 1)
9893e49d AK	348	return ret;
325c4ef5	349
29b4eede WL	350	if (PageHWPoison(p)) {
	351	put_page(p);
	352	continue;
	353	}
afcf938e	354	if (bhv == MADV_SOFT_OFFLINE) {
b194b8cd	355	pr_info("Soft offlining page %#lx at %#lx\n",
afcf938e AK	356	page_to_pfn(p), start);
	357	ret = soft_offline_page(p, MF_COUNT_INCREASED);
	358	if (ret)
8302423b	359	return ret;
afcf938e AK	360	continue;
afcf938e AK	361	}
b194b8cd	362	pr_info("Injecting memory failure for page %#lx at %#lx\n",
9893e49d AK	363	page_to_pfn(p), start);
9893e49d AK	364	/* Ignore return value for now */
cd42f4a3	365	memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED);
9893e49d	366	}
325c4ef5	367	return 0;
9893e49d AK	368	}
	369	#endif
	370
165cd402	371	static long
	372	madvise_vma(struct vm_area_struct vma, struct vm_area_struct *prev,
	373	unsigned long start, unsigned long end, int behavior)
1da177e4	374	{
1da177e4	375	switch (behavior) {
f6b3ec23	376	case MADV_REMOVE:
3866ea90	377	return madvise_remove(vma, prev, start, end);
1da177e4	378	case MADV_WILLNEED:
3866ea90	379	return madvise_willneed(vma, prev, start, end);
1da177e4	380	case MADV_DONTNEED:
3866ea90	381	return madvise_dontneed(vma, prev, start, end);
1da177e4	382	default:
3866ea90	383	return madvise_behavior(vma, prev, start, end, behavior);
1da177e4	384	}
1da177e4 LT	385	}
1da177e4 LT	386
75927af8 NP	387	static int
	388	madvise_behavior_valid(int behavior)
	389	{
	390	switch (behavior) {
	391	case MADV_DOFORK:
	392	case MADV_DONTFORK:
	393	case MADV_NORMAL:
	394	case MADV_SEQUENTIAL:
	395	case MADV_RANDOM:
	396	case MADV_REMOVE:
	397	case MADV_WILLNEED:
	398	case MADV_DONTNEED:
f8af4da3 HD	399	#ifdef CONFIG_KSM
	400	case MADV_MERGEABLE:
	401	case MADV_UNMERGEABLE:
0af4e98b AA	402	#endif
	403	#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	404	case MADV_HUGEPAGE:
a664b2d8	405	case MADV_NOHUGEPAGE:
f8af4da3	406	#endif
accb61fe JB	407	case MADV_DONTDUMP:
accb61fe JB	408	case MADV_DODUMP:
75927af8 NP	409	return 1;
	410
	411	default:
	412	return 0;
	413	}
	414	}
3866ea90	415
1da177e4 LT	416	/*
	417	* The madvise(2) system call.
	418	*
	419	* Applications can use madvise() to advise the kernel how it should
	420	* handle paging I/O in this VM area. The idea is to help the kernel
	421	* use appropriate read-ahead and caching techniques. The information
	422	* provided is advisory only, and can be safely disregarded by the
	423	* kernel without affecting the correct operation of the application.
	424	*
	425	* behavior values:
	426	* MADV_NORMAL - the default behavior is to read clusters. This
	427	* results in some read-ahead and read-behind.
	428	* MADV_RANDOM - the system should read the minimum amount of data
	429	* on any access, since it is unlikely that the appli-
	430	* cation will need more than what it asks for.
	431	* MADV_SEQUENTIAL - pages in the given range will probably be accessed
	432	* once, so they can be aggressively read ahead, and
	433	* can be freed soon after they are accessed.
	434	* MADV_WILLNEED - the application is notifying the system to read
	435	* some pages ahead.
	436	* MADV_DONTNEED - the application is finished with the given range,
	437	* so the kernel can free resources associated with it.
f6b3ec23 BP	438	* MADV_REMOVE - the application wants to free up the given range of
f6b3ec23 BP	439	* pages and associated backing store.
3866ea90 HD	440	* MADV_DONTFORK - omit this area from child's address space when forking:
	441	* typically, to avoid COWing pages pinned by get_user_pages().
	442	* MADV_DOFORK - cancel MADV_DONTFORK: no longer omit this area when forking.
f8af4da3 HD	443	* MADV_MERGEABLE - the application recommends that KSM try to merge pages in
	444	* this area with pages of identical content from other such areas.
	445	* MADV_UNMERGEABLE- cancel MADV_MERGEABLE: no longer merge pages with others.
1da177e4 LT	446	*
	447	* return values:
	448	* zero - success
	449	* -EINVAL - start + len < 0, start is not page-aligned,
	450	* "behavior" is not a valid value, or application
	451	* is attempting to release locked or shared pages.
	452	* -ENOMEM - addresses in the specified range are not currently
	453	* mapped, or are outside the AS of the process.
	454	* -EIO - an I/O error occurred while paging in data.
	455	* -EBADF - map exists, but area maps something that isn't a file.
	456	* -EAGAIN - a kernel resource was temporarily unavailable.
	457	*/
3480b257	458	SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
1da177e4	459	{
05b74384	460	unsigned long end, tmp;
ec9bed9d	461	struct vm_area_struct vma, prev;
1da177e4 LT	462	int unmapped_error = 0;
1da177e4 LT	463	int error = -EINVAL;
f7977793	464	int write;
1da177e4	465	size_t len;
1998cc04	466	struct blk_plug plug;
1da177e4	467
9893e49d	468	#ifdef CONFIG_MEMORY_FAILURE
afcf938e AK	469	if (behavior == MADV_HWPOISON \|\| behavior == MADV_SOFT_OFFLINE)
afcf938e AK	470	return madvise_hwpoison(behavior, start, start+len_in);
9893e49d	471	#endif
75927af8 NP	472	if (!madvise_behavior_valid(behavior))
	473	return error;
	474
1da177e4	475	if (start & ~PAGE_MASK)
84d96d89	476	return error;
1da177e4 LT	477	len = (len_in + ~PAGE_MASK) & PAGE_MASK;
	478
	479	/* Check to see whether len was rounded up from small -ve to zero */
	480	if (len_in && !len)
84d96d89	481	return error;
1da177e4 LT	482
	483	end = start + len;
	484	if (end < start)
84d96d89	485	return error;
1da177e4 LT	486
	487	error = 0;
	488	if (end == start)
84d96d89 RV	489	return error;
	490
	491	write = madvise_need_mmap_write(behavior);
	492	if (write)
	493	down_write(&current->mm->mmap_sem);
	494	else
	495	down_read(&current->mm->mmap_sem);
1da177e4 LT	496
	497	/*
	498	* If the interval [start,end) covers some unmapped address
	499	* ranges, just ignore them, but return -ENOMEM at the end.
05b74384	500	* - different from the way of handling in mlock etc.
1da177e4	501	*/
05b74384	502	vma = find_vma_prev(current->mm, start, &prev);
836d5ffd HD	503	if (vma && start > vma->vm_start)
	504	prev = vma;
	505
1998cc04	506	blk_start_plug(&plug);
1da177e4 LT	507	for (;;) {
	508	/* Still start < end. */
	509	error = -ENOMEM;
	510	if (!vma)
84d96d89	511	goto out;
1da177e4	512
05b74384	513	/* Here start < (end\|vma->vm_end). */
1da177e4 LT	514	if (start < vma->vm_start) {
	515	unmapped_error = -ENOMEM;
	516	start = vma->vm_start;
05b74384	517	if (start >= end)
84d96d89	518	goto out;
1da177e4 LT	519	}
1da177e4 LT	520
05b74384 PM	521	/* Here vma->vm_start <= start < (end\|vma->vm_end) */
	522	tmp = vma->vm_end;
	523	if (end < tmp)
	524	tmp = end;
1da177e4	525
05b74384 PM	526	/* Here vma->vm_start <= start < tmp <= (end\|vma->vm_end). */
05b74384 PM	527	error = madvise_vma(vma, &prev, start, tmp, behavior);
1da177e4	528	if (error)
84d96d89	529	goto out;
05b74384	530	start = tmp;
90ed52eb	531	if (prev && start < prev->vm_end)
05b74384 PM	532	start = prev->vm_end;
	533	error = unmapped_error;
	534	if (start >= end)
84d96d89	535	goto out;
90ed52eb HD	536	if (prev)
	537	vma = prev->vm_next;
	538	else /* madvise_remove dropped mmap_sem */
	539	vma = find_vma(current->mm, start);
1da177e4	540	}
1da177e4	541	out:
84d96d89	542	blk_finish_plug(&plug);
f7977793	543	if (write)
0a27a14a NP	544	up_write(&current->mm->mmap_sem);
	545	else
	546	up_read(&current->mm->mmap_sem);
	547
1da177e4 LT	548	return error;
1da177e4 LT	549	}