[linux-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>

/*
 * 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_NODUMP;
		break;
	case MADV_DODUMP:
		new_flags &= ~VM_NODUMP;
		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;
}

/*
 * 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;

	if (!file)
		return -EBADF;

	if (file->f_mapping->a_ops->get_xip_mem) {
		/* 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;

	if (unlikely(vma->vm_flags & VM_NONLINEAR)) {
		struct zap_details details = {
			.nonlinear_vma = vma,
			.last_index = ULONG_MAX,
		};
		zap_page_range(vma, start, end - start, &details);
	} else
		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.
 *
 * NOTE: Currently, only shmfs/tmpfs is supported for this operation.
 * Other filesystems return -ENOSYS.
 */
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;

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

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

	if (!vma->vm_file || !vma->vm_file->f_mapping
		|| !vma->vm_file->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 */
	up_read(&current->mm->mmap_sem);
	error = do_fallocate(vma->vm_file,
				FALLOC_FL_PUNCH_HOLE | FALLOC_FL_KEEP_SIZE,
				offset, end - start);
	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)
{
	int ret = 0;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
	for (; start < end; start += PAGE_SIZE) {
		struct page *p;
		int ret = get_user_pages_fast(start, 1, 0, &p);
		if (ret != 1)
			return ret;
		if (bhv == MADV_SOFT_OFFLINE) {
			printk(KERN_INFO "Soft offlining page %lx at %lx\n",
				page_to_pfn(p), start);
			ret = soft_offline_page(p, MF_COUNT_INCREASED);
			if (ret)
				break;
			continue;
		}
		printk(KERN_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 ret;
}
#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;

#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;

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

	if (start & ~PAGE_MASK)
		goto out;
	len = (len_in + ~PAGE_MASK) & PAGE_MASK;

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

	end = start + len;
	if (end < start)
		goto out;

	error = 0;
	if (end == start)
		goto out;

	/*
	 * 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;

	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:
	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>
1da177e4	18
0a27a14a NP	19	/*
	20	* Any behaviour which results in changes to the vma->vm_flags needs to
	21	* take mmap_sem for writing. Others, which simply traverse vmas, need
	22	* to only take it for reading.
	23	*/
	24	static int madvise_need_mmap_write(int behavior)
	25	{
	26	switch (behavior) {
	27	case MADV_REMOVE:
	28	case MADV_WILLNEED:
	29	case MADV_DONTNEED:
	30	return 0;
	31	default:
	32	/* be safe, default to 1. list exceptions explicitly */
	33	return 1;
	34	}
	35	}
	36
1da177e4 LT	37	/*
	38	* We can potentially split a vm area into separate
	39	* areas, each area with its own behavior.
	40	*/
05b74384 PM	41	static long madvise_behavior(struct vm_area_struct * vma,
	42	struct vm_area_struct **prev,
	43	unsigned long start, unsigned long end, int behavior)
1da177e4 LT	44	{
	45	struct mm_struct * mm = vma->vm_mm;
	46	int error = 0;
05b74384	47	pgoff_t pgoff;
3866ea90	48	unsigned long new_flags = vma->vm_flags;
e798c6e8 PM	49
e798c6e8 PM	50	switch (behavior) {
f8225661 MT	51	case MADV_NORMAL:
	52	new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
	53	break;
e798c6e8	54	case MADV_SEQUENTIAL:
f8225661	55	new_flags = (new_flags & ~VM_RAND_READ) \| VM_SEQ_READ;
e798c6e8 PM	56	break;
e798c6e8 PM	57	case MADV_RANDOM:
f8225661	58	new_flags = (new_flags & ~VM_SEQ_READ) \| VM_RAND_READ;
e798c6e8	59	break;
f8225661 MT	60	case MADV_DONTFORK:
	61	new_flags \|= VM_DONTCOPY;
	62	break;
	63	case MADV_DOFORK:
3866ea90 HD	64	if (vma->vm_flags & VM_IO) {
	65	error = -EINVAL;
	66	goto out;
	67	}
f8225661	68	new_flags &= ~VM_DONTCOPY;
e798c6e8	69	break;
accb61fe JB	70	case MADV_DONTDUMP:
	71	new_flags \|= VM_NODUMP;
	72	break;
	73	case MADV_DODUMP:
	74	new_flags &= ~VM_NODUMP;
	75	break;
f8af4da3 HD	76	case MADV_MERGEABLE:
	77	case MADV_UNMERGEABLE:
	78	error = ksm_madvise(vma, start, end, behavior, &new_flags);
	79	if (error)
	80	goto out;
	81	break;
0af4e98b	82	case MADV_HUGEPAGE:
a664b2d8	83	case MADV_NOHUGEPAGE:
60ab3244	84	error = hugepage_madvise(vma, &new_flags, behavior);
0af4e98b AA	85	if (error)
	86	goto out;
	87	break;
e798c6e8 PM	88	}
e798c6e8 PM	89
05b74384 PM	90	if (new_flags == vma->vm_flags) {
05b74384 PM	91	*prev = vma;
836d5ffd	92	goto out;
05b74384 PM	93	}
	94
	95	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	96	prev = vma_merge(mm, prev, start, end, new_flags, vma->anon_vma,
	97	vma->vm_file, pgoff, vma_policy(vma));
	98	if (*prev) {
	99	vma = *prev;
	100	goto success;
	101	}
	102
	103	*prev = vma;
1da177e4 LT	104
	105	if (start != vma->vm_start) {
	106	error = split_vma(mm, vma, start, 1);
	107	if (error)
	108	goto out;
	109	}
	110
	111	if (end != vma->vm_end) {
	112	error = split_vma(mm, vma, end, 0);
	113	if (error)
	114	goto out;
	115	}
	116
836d5ffd	117	success:
1da177e4 LT	118	/*
	119	* vm_flags is protected by the mmap_sem held in write mode.
	120	*/
e798c6e8	121	vma->vm_flags = new_flags;
1da177e4 LT	122
	123	out:
	124	if (error == -ENOMEM)
	125	error = -EAGAIN;
	126	return error;
	127	}
	128
	129	/*
	130	* Schedule all required I/O operations. Do not wait for completion.
	131	*/
	132	static long madvise_willneed(struct vm_area_struct * vma,
05b74384	133	struct vm_area_struct ** prev,
1da177e4 LT	134	unsigned long start, unsigned long end)
	135	{
	136	struct file *file = vma->vm_file;
	137
1bef4003 S	138	if (!file)
	139	return -EBADF;
	140
70688e4d	141	if (file->f_mapping->a_ops->get_xip_mem) {
fe77ba6f CO	142	/* no bad return value, but ignore advice */
	143	return 0;
	144	}
	145
05b74384	146	*prev = vma;
1da177e4 LT	147	start = ((start - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	148	if (end > vma->vm_end)
	149	end = vma->vm_end;
	150	end = ((end - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	151
f7e839dd	152	force_page_cache_readahead(file->f_mapping, file, start, end - start);
1da177e4 LT	153	return 0;
	154	}
	155
	156	/*
	157	* Application no longer needs these pages. If the pages are dirty,
	158	* it's OK to just throw them away. The app will be more careful about
	159	* data it wants to keep. Be sure to free swap resources too. The
7e6cbea3	160	* zap_page_range call sets things up for shrink_active_list to actually free
1da177e4 LT	161	* these pages later if no one else has touched them in the meantime,
1da177e4 LT	162	* although we could add these pages to a global reuse list for
7e6cbea3	163	* shrink_active_list to pick up before reclaiming other pages.
1da177e4 LT	164	*
	165	* NB: This interface discards data rather than pushes it out to swap,
	166	* as some implementations do. This has performance implications for
	167	* applications like large transactional databases which want to discard
	168	* pages in anonymous maps after committing to backing store the data
	169	* that was kept in them. There is no reason to write this data out to
	170	* the swap area if the application is discarding it.
	171	*
	172	* An interface that causes the system to free clean pages and flush
	173	* dirty pages is already available as msync(MS_INVALIDATE).
	174	*/
	175	static long madvise_dontneed(struct vm_area_struct * vma,
05b74384	176	struct vm_area_struct ** prev,
1da177e4 LT	177	unsigned long start, unsigned long end)
1da177e4 LT	178	{
05b74384	179	*prev = vma;
6aab341e	180	if (vma->vm_flags & (VM_LOCKED\|VM_HUGETLB\|VM_PFNMAP))
1da177e4 LT	181	return -EINVAL;
	182
	183	if (unlikely(vma->vm_flags & VM_NONLINEAR)) {
	184	struct zap_details details = {
	185	.nonlinear_vma = vma,
	186	.last_index = ULONG_MAX,
	187	};
	188	zap_page_range(vma, start, end - start, &details);
	189	} else
	190	zap_page_range(vma, start, end - start, NULL);
	191	return 0;
	192	}
	193
f6b3ec23 BP	194	/*
	195	* Application wants to free up the pages and associated backing store.
	196	* This is effectively punching a hole into the middle of a file.
	197	*
	198	* NOTE: Currently, only shmfs/tmpfs is supported for this operation.
	199	* Other filesystems return -ENOSYS.
	200	*/
	201	static long madvise_remove(struct vm_area_struct *vma,
00e9fa2d	202	struct vm_area_struct **prev,
f6b3ec23 BP	203	unsigned long start, unsigned long end)
f6b3ec23 BP	204	{
3f31d075	205	loff_t offset;
90ed52eb	206	int error;
f6b3ec23	207
90ed52eb	208	prev = NULL; / tell sys_madvise we drop mmap_sem */
00e9fa2d	209
f6b3ec23 BP	210	if (vma->vm_flags & (VM_LOCKED\|VM_NONLINEAR\|VM_HUGETLB))
	211	return -EINVAL;
	212
	213	if (!vma->vm_file \|\| !vma->vm_file->f_mapping
	214	\|\| !vma->vm_file->f_mapping->host) {
	215	return -EINVAL;
	216	}
	217
69cf0fac HD	218	if ((vma->vm_flags & (VM_SHARED\|VM_WRITE)) != (VM_SHARED\|VM_WRITE))
	219	return -EACCES;
	220
f6b3ec23 BP	221	offset = (loff_t)(start - vma->vm_start)
f6b3ec23 BP	222	+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
90ed52eb	223
3f31d075	224	/* filesystem's fallocate may need to take i_mutex */
0a27a14a	225	up_read(&current->mm->mmap_sem);
3f31d075 HD	226	error = do_fallocate(vma->vm_file,
	227	FALLOC_FL_PUNCH_HOLE \| FALLOC_FL_KEEP_SIZE,
	228	offset, end - start);
0a27a14a	229	down_read(&current->mm->mmap_sem);
90ed52eb	230	return error;
f6b3ec23 BP	231	}
f6b3ec23 BP	232
9893e49d AK	233	#ifdef CONFIG_MEMORY_FAILURE
	234	/*
	235	* Error injection support for memory error handling.
	236	*/
afcf938e	237	static int madvise_hwpoison(int bhv, unsigned long start, unsigned long end)
9893e49d AK	238	{
	239	int ret = 0;
	240
	241	if (!capable(CAP_SYS_ADMIN))
	242	return -EPERM;
	243	for (; start < end; start += PAGE_SIZE) {
	244	struct page *p;
d15f107d	245	int ret = get_user_pages_fast(start, 1, 0, &p);
9893e49d AK	246	if (ret != 1)
9893e49d AK	247	return ret;
afcf938e AK	248	if (bhv == MADV_SOFT_OFFLINE) {
	249	printk(KERN_INFO "Soft offlining page %lx at %lx\n",
	250	page_to_pfn(p), start);
	251	ret = soft_offline_page(p, MF_COUNT_INCREASED);
	252	if (ret)
	253	break;
	254	continue;
	255	}
9893e49d AK	256	printk(KERN_INFO "Injecting memory failure for page %lx at %lx\n",
	257	page_to_pfn(p), start);
	258	/* Ignore return value for now */
cd42f4a3	259	memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED);
9893e49d AK	260	}
	261	return ret;
	262	}
	263	#endif
	264
165cd402	265	static long
	266	madvise_vma(struct vm_area_struct vma, struct vm_area_struct *prev,
	267	unsigned long start, unsigned long end, int behavior)
1da177e4	268	{
1da177e4	269	switch (behavior) {
f6b3ec23	270	case MADV_REMOVE:
3866ea90	271	return madvise_remove(vma, prev, start, end);
1da177e4	272	case MADV_WILLNEED:
3866ea90	273	return madvise_willneed(vma, prev, start, end);
1da177e4	274	case MADV_DONTNEED:
3866ea90	275	return madvise_dontneed(vma, prev, start, end);
1da177e4	276	default:
3866ea90	277	return madvise_behavior(vma, prev, start, end, behavior);
1da177e4	278	}
1da177e4 LT	279	}
1da177e4 LT	280
75927af8 NP	281	static int
	282	madvise_behavior_valid(int behavior)
	283	{
	284	switch (behavior) {
	285	case MADV_DOFORK:
	286	case MADV_DONTFORK:
	287	case MADV_NORMAL:
	288	case MADV_SEQUENTIAL:
	289	case MADV_RANDOM:
	290	case MADV_REMOVE:
	291	case MADV_WILLNEED:
	292	case MADV_DONTNEED:
f8af4da3 HD	293	#ifdef CONFIG_KSM
	294	case MADV_MERGEABLE:
	295	case MADV_UNMERGEABLE:
0af4e98b AA	296	#endif
	297	#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	298	case MADV_HUGEPAGE:
a664b2d8	299	case MADV_NOHUGEPAGE:
f8af4da3	300	#endif
accb61fe JB	301	case MADV_DONTDUMP:
accb61fe JB	302	case MADV_DODUMP:
75927af8 NP	303	return 1;
	304
	305	default:
	306	return 0;
	307	}
	308	}
3866ea90	309
1da177e4 LT	310	/*
	311	* The madvise(2) system call.
	312	*
	313	* Applications can use madvise() to advise the kernel how it should
	314	* handle paging I/O in this VM area. The idea is to help the kernel
	315	* use appropriate read-ahead and caching techniques. The information
	316	* provided is advisory only, and can be safely disregarded by the
	317	* kernel without affecting the correct operation of the application.
	318	*
	319	* behavior values:
	320	* MADV_NORMAL - the default behavior is to read clusters. This
	321	* results in some read-ahead and read-behind.
	322	* MADV_RANDOM - the system should read the minimum amount of data
	323	* on any access, since it is unlikely that the appli-
	324	* cation will need more than what it asks for.
	325	* MADV_SEQUENTIAL - pages in the given range will probably be accessed
	326	* once, so they can be aggressively read ahead, and
	327	* can be freed soon after they are accessed.
	328	* MADV_WILLNEED - the application is notifying the system to read
	329	* some pages ahead.
	330	* MADV_DONTNEED - the application is finished with the given range,
	331	* so the kernel can free resources associated with it.
f6b3ec23 BP	332	* MADV_REMOVE - the application wants to free up the given range of
f6b3ec23 BP	333	* pages and associated backing store.
3866ea90 HD	334	* MADV_DONTFORK - omit this area from child's address space when forking:
	335	* typically, to avoid COWing pages pinned by get_user_pages().
	336	* MADV_DOFORK - cancel MADV_DONTFORK: no longer omit this area when forking.
f8af4da3 HD	337	* MADV_MERGEABLE - the application recommends that KSM try to merge pages in
	338	* this area with pages of identical content from other such areas.
	339	* MADV_UNMERGEABLE- cancel MADV_MERGEABLE: no longer merge pages with others.
1da177e4 LT	340	*
	341	* return values:
	342	* zero - success
	343	* -EINVAL - start + len < 0, start is not page-aligned,
	344	* "behavior" is not a valid value, or application
	345	* is attempting to release locked or shared pages.
	346	* -ENOMEM - addresses in the specified range are not currently
	347	* mapped, or are outside the AS of the process.
	348	* -EIO - an I/O error occurred while paging in data.
	349	* -EBADF - map exists, but area maps something that isn't a file.
	350	* -EAGAIN - a kernel resource was temporarily unavailable.
	351	*/
3480b257	352	SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
1da177e4	353	{
05b74384 PM	354	unsigned long end, tmp;
05b74384 PM	355	struct vm_area_struct * vma, *prev;
1da177e4 LT	356	int unmapped_error = 0;
1da177e4 LT	357	int error = -EINVAL;
f7977793	358	int write;
1da177e4 LT	359	size_t len;
1da177e4 LT	360
9893e49d	361	#ifdef CONFIG_MEMORY_FAILURE
afcf938e AK	362	if (behavior == MADV_HWPOISON \|\| behavior == MADV_SOFT_OFFLINE)
afcf938e AK	363	return madvise_hwpoison(behavior, start, start+len_in);
9893e49d	364	#endif
75927af8 NP	365	if (!madvise_behavior_valid(behavior))
	366	return error;
	367
f7977793 JB	368	write = madvise_need_mmap_write(behavior);
f7977793 JB	369	if (write)
0a27a14a NP	370	down_write(&current->mm->mmap_sem);
	371	else
	372	down_read(&current->mm->mmap_sem);
1da177e4 LT	373
	374	if (start & ~PAGE_MASK)
	375	goto out;
	376	len = (len_in + ~PAGE_MASK) & PAGE_MASK;
	377
	378	/* Check to see whether len was rounded up from small -ve to zero */
	379	if (len_in && !len)
	380	goto out;
	381
	382	end = start + len;
	383	if (end < start)
	384	goto out;
	385
	386	error = 0;
	387	if (end == start)
	388	goto out;
	389
	390	/*
	391	* If the interval [start,end) covers some unmapped address
	392	* ranges, just ignore them, but return -ENOMEM at the end.
05b74384	393	* - different from the way of handling in mlock etc.
1da177e4	394	*/
05b74384	395	vma = find_vma_prev(current->mm, start, &prev);
836d5ffd HD	396	if (vma && start > vma->vm_start)
	397	prev = vma;
	398
1da177e4 LT	399	for (;;) {
	400	/* Still start < end. */
	401	error = -ENOMEM;
	402	if (!vma)
	403	goto out;
	404
05b74384	405	/* Here start < (end\|vma->vm_end). */
1da177e4 LT	406	if (start < vma->vm_start) {
	407	unmapped_error = -ENOMEM;
	408	start = vma->vm_start;
05b74384 PM	409	if (start >= end)
05b74384 PM	410	goto out;
1da177e4 LT	411	}
1da177e4 LT	412
05b74384 PM	413	/* Here vma->vm_start <= start < (end\|vma->vm_end) */
	414	tmp = vma->vm_end;
	415	if (end < tmp)
	416	tmp = end;
1da177e4	417
05b74384 PM	418	/* Here vma->vm_start <= start < tmp <= (end\|vma->vm_end). */
05b74384 PM	419	error = madvise_vma(vma, &prev, start, tmp, behavior);
1da177e4 LT	420	if (error)
1da177e4 LT	421	goto out;
05b74384	422	start = tmp;
90ed52eb	423	if (prev && start < prev->vm_end)
05b74384 PM	424	start = prev->vm_end;
	425	error = unmapped_error;
	426	if (start >= end)
	427	goto out;
90ed52eb HD	428	if (prev)
	429	vma = prev->vm_next;
	430	else /* madvise_remove dropped mmap_sem */
	431	vma = find_vma(current->mm, start);
1da177e4	432	}
1da177e4	433	out:
f7977793	434	if (write)
0a27a14a NP	435	up_write(&current->mm->mmap_sem);
	436	else
	437	up_read(&current->mm->mmap_sem);
	438
1da177e4 LT	439	return error;
1da177e4 LT	440	}