[linux.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/sched.h>
#include <linux/ksm.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)
{
	struct address_space *mapping;
	loff_t offset, endoff;
	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;

	mapping = vma->vm_file->f_mapping;

	offset = (loff_t)(start - vma->vm_start)
			+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
	endoff = (loff_t)(end - vma->vm_start - 1)
			+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);

	/* vmtruncate_range needs to take i_mutex */
	up_read(&current->mm->mmap_sem);
	error = vmtruncate_range(mapping->host, offset, endoff);
	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>
e8edc6e0	14	#include <linux/sched.h>
f8af4da3	15	#include <linux/ksm.h>
1da177e4	16
0a27a14a NP	17	/*
	18	* Any behaviour which results in changes to the vma->vm_flags needs to
	19	* take mmap_sem for writing. Others, which simply traverse vmas, need
	20	* to only take it for reading.
	21	*/
	22	static int madvise_need_mmap_write(int behavior)
	23	{
	24	switch (behavior) {
	25	case MADV_REMOVE:
	26	case MADV_WILLNEED:
	27	case MADV_DONTNEED:
	28	return 0;
	29	default:
	30	/* be safe, default to 1. list exceptions explicitly */
	31	return 1;
	32	}
	33	}
	34
1da177e4 LT	35	/*
	36	* We can potentially split a vm area into separate
	37	* areas, each area with its own behavior.
	38	*/
05b74384 PM	39	static long madvise_behavior(struct vm_area_struct * vma,
	40	struct vm_area_struct **prev,
	41	unsigned long start, unsigned long end, int behavior)
1da177e4 LT	42	{
	43	struct mm_struct * mm = vma->vm_mm;
	44	int error = 0;
05b74384	45	pgoff_t pgoff;
3866ea90	46	unsigned long new_flags = vma->vm_flags;
e798c6e8 PM	47
e798c6e8 PM	48	switch (behavior) {
f8225661 MT	49	case MADV_NORMAL:
	50	new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
	51	break;
e798c6e8	52	case MADV_SEQUENTIAL:
f8225661	53	new_flags = (new_flags & ~VM_RAND_READ) \| VM_SEQ_READ;
e798c6e8 PM	54	break;
e798c6e8 PM	55	case MADV_RANDOM:
f8225661	56	new_flags = (new_flags & ~VM_SEQ_READ) \| VM_RAND_READ;
e798c6e8	57	break;
f8225661 MT	58	case MADV_DONTFORK:
	59	new_flags \|= VM_DONTCOPY;
	60	break;
	61	case MADV_DOFORK:
3866ea90 HD	62	if (vma->vm_flags & VM_IO) {
	63	error = -EINVAL;
	64	goto out;
	65	}
f8225661	66	new_flags &= ~VM_DONTCOPY;
e798c6e8	67	break;
accb61fe JB	68	case MADV_DONTDUMP:
	69	new_flags \|= VM_NODUMP;
	70	break;
	71	case MADV_DODUMP:
	72	new_flags &= ~VM_NODUMP;
	73	break;
f8af4da3 HD	74	case MADV_MERGEABLE:
	75	case MADV_UNMERGEABLE:
	76	error = ksm_madvise(vma, start, end, behavior, &new_flags);
	77	if (error)
	78	goto out;
	79	break;
0af4e98b	80	case MADV_HUGEPAGE:
a664b2d8	81	case MADV_NOHUGEPAGE:
60ab3244	82	error = hugepage_madvise(vma, &new_flags, behavior);
0af4e98b AA	83	if (error)
	84	goto out;
	85	break;
e798c6e8 PM	86	}
e798c6e8 PM	87
05b74384 PM	88	if (new_flags == vma->vm_flags) {
05b74384 PM	89	*prev = vma;
836d5ffd	90	goto out;
05b74384 PM	91	}
	92
	93	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	94	prev = vma_merge(mm, prev, start, end, new_flags, vma->anon_vma,
	95	vma->vm_file, pgoff, vma_policy(vma));
	96	if (*prev) {
	97	vma = *prev;
	98	goto success;
	99	}
	100
	101	*prev = vma;
1da177e4 LT	102
	103	if (start != vma->vm_start) {
	104	error = split_vma(mm, vma, start, 1);
	105	if (error)
	106	goto out;
	107	}
	108
	109	if (end != vma->vm_end) {
	110	error = split_vma(mm, vma, end, 0);
	111	if (error)
	112	goto out;
	113	}
	114
836d5ffd	115	success:
1da177e4 LT	116	/*
	117	* vm_flags is protected by the mmap_sem held in write mode.
	118	*/
e798c6e8	119	vma->vm_flags = new_flags;
1da177e4 LT	120
	121	out:
	122	if (error == -ENOMEM)
	123	error = -EAGAIN;
	124	return error;
	125	}
	126
	127	/*
	128	* Schedule all required I/O operations. Do not wait for completion.
	129	*/
	130	static long madvise_willneed(struct vm_area_struct * vma,
05b74384	131	struct vm_area_struct ** prev,
1da177e4 LT	132	unsigned long start, unsigned long end)
	133	{
	134	struct file *file = vma->vm_file;
	135
1bef4003 S	136	if (!file)
	137	return -EBADF;
	138
70688e4d	139	if (file->f_mapping->a_ops->get_xip_mem) {
fe77ba6f CO	140	/* no bad return value, but ignore advice */
	141	return 0;
	142	}
	143
05b74384	144	*prev = vma;
1da177e4 LT	145	start = ((start - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	146	if (end > vma->vm_end)
	147	end = vma->vm_end;
	148	end = ((end - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	149
f7e839dd	150	force_page_cache_readahead(file->f_mapping, file, start, end - start);
1da177e4 LT	151	return 0;
	152	}
	153
	154	/*
	155	* Application no longer needs these pages. If the pages are dirty,
	156	* it's OK to just throw them away. The app will be more careful about
	157	* data it wants to keep. Be sure to free swap resources too. The
7e6cbea3	158	* zap_page_range call sets things up for shrink_active_list to actually free
1da177e4 LT	159	* these pages later if no one else has touched them in the meantime,
1da177e4 LT	160	* although we could add these pages to a global reuse list for
7e6cbea3	161	* shrink_active_list to pick up before reclaiming other pages.
1da177e4 LT	162	*
	163	* NB: This interface discards data rather than pushes it out to swap,
	164	* as some implementations do. This has performance implications for
	165	* applications like large transactional databases which want to discard
	166	* pages in anonymous maps after committing to backing store the data
	167	* that was kept in them. There is no reason to write this data out to
	168	* the swap area if the application is discarding it.
	169	*
	170	* An interface that causes the system to free clean pages and flush
	171	* dirty pages is already available as msync(MS_INVALIDATE).
	172	*/
	173	static long madvise_dontneed(struct vm_area_struct * vma,
05b74384	174	struct vm_area_struct ** prev,
1da177e4 LT	175	unsigned long start, unsigned long end)
1da177e4 LT	176	{
05b74384	177	*prev = vma;
6aab341e	178	if (vma->vm_flags & (VM_LOCKED\|VM_HUGETLB\|VM_PFNMAP))
1da177e4 LT	179	return -EINVAL;
	180
	181	if (unlikely(vma->vm_flags & VM_NONLINEAR)) {
	182	struct zap_details details = {
	183	.nonlinear_vma = vma,
	184	.last_index = ULONG_MAX,
	185	};
	186	zap_page_range(vma, start, end - start, &details);
	187	} else
	188	zap_page_range(vma, start, end - start, NULL);
	189	return 0;
	190	}
	191
f6b3ec23 BP	192	/*
	193	* Application wants to free up the pages and associated backing store.
	194	* This is effectively punching a hole into the middle of a file.
	195	*
	196	* NOTE: Currently, only shmfs/tmpfs is supported for this operation.
	197	* Other filesystems return -ENOSYS.
	198	*/
	199	static long madvise_remove(struct vm_area_struct *vma,
00e9fa2d	200	struct vm_area_struct **prev,
f6b3ec23 BP	201	unsigned long start, unsigned long end)
	202	{
	203	struct address_space *mapping;
90ed52eb HD	204	loff_t offset, endoff;
90ed52eb HD	205	int error;
f6b3ec23	206
90ed52eb	207	prev = NULL; / tell sys_madvise we drop mmap_sem */
00e9fa2d	208
f6b3ec23 BP	209	if (vma->vm_flags & (VM_LOCKED\|VM_NONLINEAR\|VM_HUGETLB))
	210	return -EINVAL;
	211
	212	if (!vma->vm_file \|\| !vma->vm_file->f_mapping
	213	\|\| !vma->vm_file->f_mapping->host) {
	214	return -EINVAL;
	215	}
	216
69cf0fac HD	217	if ((vma->vm_flags & (VM_SHARED\|VM_WRITE)) != (VM_SHARED\|VM_WRITE))
	218	return -EACCES;
	219
f6b3ec23 BP	220	mapping = vma->vm_file->f_mapping;
	221
	222	offset = (loff_t)(start - vma->vm_start)
	223	+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
	224	endoff = (loff_t)(end - vma->vm_start - 1)
	225	+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
90ed52eb	226
bd5fe6c5	227	/* vmtruncate_range needs to take i_mutex */
0a27a14a	228	up_read(&current->mm->mmap_sem);
90ed52eb	229	error = vmtruncate_range(mapping->host, offset, endoff);
0a27a14a	230	down_read(&current->mm->mmap_sem);
90ed52eb	231	return error;
f6b3ec23 BP	232	}
f6b3ec23 BP	233
9893e49d AK	234	#ifdef CONFIG_MEMORY_FAILURE
	235	/*
	236	* Error injection support for memory error handling.
	237	*/
afcf938e	238	static int madvise_hwpoison(int bhv, unsigned long start, unsigned long end)
9893e49d AK	239	{
	240	int ret = 0;
	241
	242	if (!capable(CAP_SYS_ADMIN))
	243	return -EPERM;
	244	for (; start < end; start += PAGE_SIZE) {
	245	struct page *p;
d15f107d	246	int ret = get_user_pages_fast(start, 1, 0, &p);
9893e49d AK	247	if (ret != 1)
9893e49d AK	248	return ret;
afcf938e AK	249	if (bhv == MADV_SOFT_OFFLINE) {
	250	printk(KERN_INFO "Soft offlining page %lx at %lx\n",
	251	page_to_pfn(p), start);
	252	ret = soft_offline_page(p, MF_COUNT_INCREASED);
	253	if (ret)
	254	break;
	255	continue;
	256	}
9893e49d AK	257	printk(KERN_INFO "Injecting memory failure for page %lx at %lx\n",
	258	page_to_pfn(p), start);
	259	/* Ignore return value for now */
cd42f4a3	260	memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED);
9893e49d AK	261	}
	262	return ret;
	263	}
	264	#endif
	265
165cd402	266	static long
	267	madvise_vma(struct vm_area_struct vma, struct vm_area_struct *prev,
	268	unsigned long start, unsigned long end, int behavior)
1da177e4	269	{
1da177e4	270	switch (behavior) {
f6b3ec23	271	case MADV_REMOVE:
3866ea90	272	return madvise_remove(vma, prev, start, end);
1da177e4	273	case MADV_WILLNEED:
3866ea90	274	return madvise_willneed(vma, prev, start, end);
1da177e4	275	case MADV_DONTNEED:
3866ea90	276	return madvise_dontneed(vma, prev, start, end);
1da177e4	277	default:
3866ea90	278	return madvise_behavior(vma, prev, start, end, behavior);
1da177e4	279	}
1da177e4 LT	280	}
1da177e4 LT	281
75927af8 NP	282	static int
	283	madvise_behavior_valid(int behavior)
	284	{
	285	switch (behavior) {
	286	case MADV_DOFORK:
	287	case MADV_DONTFORK:
	288	case MADV_NORMAL:
	289	case MADV_SEQUENTIAL:
	290	case MADV_RANDOM:
	291	case MADV_REMOVE:
	292	case MADV_WILLNEED:
	293	case MADV_DONTNEED:
f8af4da3 HD	294	#ifdef CONFIG_KSM
	295	case MADV_MERGEABLE:
	296	case MADV_UNMERGEABLE:
0af4e98b AA	297	#endif
	298	#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	299	case MADV_HUGEPAGE:
a664b2d8	300	case MADV_NOHUGEPAGE:
f8af4da3	301	#endif
accb61fe JB	302	case MADV_DONTDUMP:
accb61fe JB	303	case MADV_DODUMP:
75927af8 NP	304	return 1;
	305
	306	default:
	307	return 0;
	308	}
	309	}
3866ea90	310
1da177e4 LT	311	/*
	312	* The madvise(2) system call.
	313	*
	314	* Applications can use madvise() to advise the kernel how it should
	315	* handle paging I/O in this VM area. The idea is to help the kernel
	316	* use appropriate read-ahead and caching techniques. The information
	317	* provided is advisory only, and can be safely disregarded by the
	318	* kernel without affecting the correct operation of the application.
	319	*
	320	* behavior values:
	321	* MADV_NORMAL - the default behavior is to read clusters. This
	322	* results in some read-ahead and read-behind.
	323	* MADV_RANDOM - the system should read the minimum amount of data
	324	* on any access, since it is unlikely that the appli-
	325	* cation will need more than what it asks for.
	326	* MADV_SEQUENTIAL - pages in the given range will probably be accessed
	327	* once, so they can be aggressively read ahead, and
	328	* can be freed soon after they are accessed.
	329	* MADV_WILLNEED - the application is notifying the system to read
	330	* some pages ahead.
	331	* MADV_DONTNEED - the application is finished with the given range,
	332	* so the kernel can free resources associated with it.
f6b3ec23 BP	333	* MADV_REMOVE - the application wants to free up the given range of
f6b3ec23 BP	334	* pages and associated backing store.
3866ea90 HD	335	* MADV_DONTFORK - omit this area from child's address space when forking:
	336	* typically, to avoid COWing pages pinned by get_user_pages().
	337	* MADV_DOFORK - cancel MADV_DONTFORK: no longer omit this area when forking.
f8af4da3 HD	338	* MADV_MERGEABLE - the application recommends that KSM try to merge pages in
	339	* this area with pages of identical content from other such areas.
	340	* MADV_UNMERGEABLE- cancel MADV_MERGEABLE: no longer merge pages with others.
1da177e4 LT	341	*
	342	* return values:
	343	* zero - success
	344	* -EINVAL - start + len < 0, start is not page-aligned,
	345	* "behavior" is not a valid value, or application
	346	* is attempting to release locked or shared pages.
	347	* -ENOMEM - addresses in the specified range are not currently
	348	* mapped, or are outside the AS of the process.
	349	* -EIO - an I/O error occurred while paging in data.
	350	* -EBADF - map exists, but area maps something that isn't a file.
	351	* -EAGAIN - a kernel resource was temporarily unavailable.
	352	*/
3480b257	353	SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
1da177e4	354	{
05b74384 PM	355	unsigned long end, tmp;
05b74384 PM	356	struct vm_area_struct * vma, *prev;
1da177e4 LT	357	int unmapped_error = 0;
1da177e4 LT	358	int error = -EINVAL;
f7977793	359	int write;
1da177e4 LT	360	size_t len;
1da177e4 LT	361
9893e49d	362	#ifdef CONFIG_MEMORY_FAILURE
afcf938e AK	363	if (behavior == MADV_HWPOISON \|\| behavior == MADV_SOFT_OFFLINE)
afcf938e AK	364	return madvise_hwpoison(behavior, start, start+len_in);
9893e49d	365	#endif
75927af8 NP	366	if (!madvise_behavior_valid(behavior))
	367	return error;
	368
f7977793 JB	369	write = madvise_need_mmap_write(behavior);
f7977793 JB	370	if (write)
0a27a14a NP	371	down_write(&current->mm->mmap_sem);
	372	else
	373	down_read(&current->mm->mmap_sem);
1da177e4 LT	374
	375	if (start & ~PAGE_MASK)
	376	goto out;
	377	len = (len_in + ~PAGE_MASK) & PAGE_MASK;
	378
	379	/* Check to see whether len was rounded up from small -ve to zero */
	380	if (len_in && !len)
	381	goto out;
	382
	383	end = start + len;
	384	if (end < start)
	385	goto out;
	386
	387	error = 0;
	388	if (end == start)
	389	goto out;
	390
	391	/*
	392	* If the interval [start,end) covers some unmapped address
	393	* ranges, just ignore them, but return -ENOMEM at the end.
05b74384	394	* - different from the way of handling in mlock etc.
1da177e4	395	*/
05b74384	396	vma = find_vma_prev(current->mm, start, &prev);
836d5ffd HD	397	if (vma && start > vma->vm_start)
	398	prev = vma;
	399
1da177e4 LT	400	for (;;) {
	401	/* Still start < end. */
	402	error = -ENOMEM;
	403	if (!vma)
	404	goto out;
	405
05b74384	406	/* Here start < (end\|vma->vm_end). */
1da177e4 LT	407	if (start < vma->vm_start) {
	408	unmapped_error = -ENOMEM;
	409	start = vma->vm_start;
05b74384 PM	410	if (start >= end)
05b74384 PM	411	goto out;
1da177e4 LT	412	}
1da177e4 LT	413
05b74384 PM	414	/* Here vma->vm_start <= start < (end\|vma->vm_end) */
	415	tmp = vma->vm_end;
	416	if (end < tmp)
	417	tmp = end;
1da177e4	418
05b74384 PM	419	/* Here vma->vm_start <= start < tmp <= (end\|vma->vm_end). */
05b74384 PM	420	error = madvise_vma(vma, &prev, start, tmp, behavior);
1da177e4 LT	421	if (error)
1da177e4 LT	422	goto out;
05b74384	423	start = tmp;
90ed52eb	424	if (prev && start < prev->vm_end)
05b74384 PM	425	start = prev->vm_end;
	426	error = unmapped_error;
	427	if (start >= end)
	428	goto out;
90ed52eb HD	429	if (prev)
	430	vma = prev->vm_next;
	431	else /* madvise_remove dropped mmap_sem */
	432	vma = find_vma(current->mm, start);
1da177e4	433	}
1da177e4	434	out:
f7977793	435	if (write)
0a27a14a NP	436	up_write(&current->mm->mmap_sem);
	437	else
	438	up_read(&current->mm->mmap_sem);
	439
1da177e4 LT	440	return error;
1da177e4 LT	441	}