[deliverable/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/falloc.h>
#include <linux/sched.h>
#include <linux/ksm.h>
#include <linux/fs.h>
#include <linux/file.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;
}

/*
 * 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;
	struct file *f;

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

	if (vma->vm_flags & (VM_LOCKED|VM_NONLINEAR|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 = do_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)
{
	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>
9ab4233d	18	#include <linux/file.h>
1da177e4	19
0a27a14a NP	20	/*
	21	* Any behaviour which results in changes to the vma->vm_flags needs to
	22	* take mmap_sem for writing. Others, which simply traverse vmas, need
	23	* to only take it for reading.
	24	*/
	25	static int madvise_need_mmap_write(int behavior)
	26	{
	27	switch (behavior) {
	28	case MADV_REMOVE:
	29	case MADV_WILLNEED:
	30	case MADV_DONTNEED:
	31	return 0;
	32	default:
	33	/* be safe, default to 1. list exceptions explicitly */
	34	return 1;
	35	}
	36	}
	37
1da177e4 LT	38	/*
	39	* We can potentially split a vm area into separate
	40	* areas, each area with its own behavior.
	41	*/
05b74384 PM	42	static long madvise_behavior(struct vm_area_struct * vma,
	43	struct vm_area_struct **prev,
	44	unsigned long start, unsigned long end, int behavior)
1da177e4 LT	45	{
	46	struct mm_struct * mm = vma->vm_mm;
	47	int error = 0;
05b74384	48	pgoff_t pgoff;
3866ea90	49	unsigned long new_flags = vma->vm_flags;
e798c6e8 PM	50
e798c6e8 PM	51	switch (behavior) {
f8225661 MT	52	case MADV_NORMAL:
	53	new_flags = new_flags & ~VM_RAND_READ & ~VM_SEQ_READ;
	54	break;
e798c6e8	55	case MADV_SEQUENTIAL:
f8225661	56	new_flags = (new_flags & ~VM_RAND_READ) \| VM_SEQ_READ;
e798c6e8 PM	57	break;
e798c6e8 PM	58	case MADV_RANDOM:
f8225661	59	new_flags = (new_flags & ~VM_SEQ_READ) \| VM_RAND_READ;
e798c6e8	60	break;
f8225661 MT	61	case MADV_DONTFORK:
	62	new_flags \|= VM_DONTCOPY;
	63	break;
	64	case MADV_DOFORK:
3866ea90 HD	65	if (vma->vm_flags & VM_IO) {
	66	error = -EINVAL;
	67	goto out;
	68	}
f8225661	69	new_flags &= ~VM_DONTCOPY;
e798c6e8	70	break;
accb61fe	71	case MADV_DONTDUMP:
0103bd16	72	new_flags \|= VM_DONTDUMP;
accb61fe JB	73	break;
accb61fe JB	74	case MADV_DODUMP:
0103bd16 KK	75	if (new_flags & VM_SPECIAL) {
	76	error = -EINVAL;
	77	goto out;
	78	}
	79	new_flags &= ~VM_DONTDUMP;
accb61fe	80	break;
f8af4da3 HD	81	case MADV_MERGEABLE:
	82	case MADV_UNMERGEABLE:
	83	error = ksm_madvise(vma, start, end, behavior, &new_flags);
	84	if (error)
	85	goto out;
	86	break;
0af4e98b	87	case MADV_HUGEPAGE:
a664b2d8	88	case MADV_NOHUGEPAGE:
60ab3244	89	error = hugepage_madvise(vma, &new_flags, behavior);
0af4e98b AA	90	if (error)
	91	goto out;
	92	break;
e798c6e8 PM	93	}
e798c6e8 PM	94
05b74384 PM	95	if (new_flags == vma->vm_flags) {
05b74384 PM	96	*prev = vma;
836d5ffd	97	goto out;
05b74384 PM	98	}
	99
	100	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	101	prev = vma_merge(mm, prev, start, end, new_flags, vma->anon_vma,
	102	vma->vm_file, pgoff, vma_policy(vma));
	103	if (*prev) {
	104	vma = *prev;
	105	goto success;
	106	}
	107
	108	*prev = vma;
1da177e4 LT	109
	110	if (start != vma->vm_start) {
	111	error = split_vma(mm, vma, start, 1);
	112	if (error)
	113	goto out;
	114	}
	115
	116	if (end != vma->vm_end) {
	117	error = split_vma(mm, vma, end, 0);
	118	if (error)
	119	goto out;
	120	}
	121
836d5ffd	122	success:
1da177e4 LT	123	/*
	124	* vm_flags is protected by the mmap_sem held in write mode.
	125	*/
e798c6e8	126	vma->vm_flags = new_flags;
1da177e4 LT	127
	128	out:
	129	if (error == -ENOMEM)
	130	error = -EAGAIN;
	131	return error;
	132	}
	133
	134	/*
	135	* Schedule all required I/O operations. Do not wait for completion.
	136	*/
	137	static long madvise_willneed(struct vm_area_struct * vma,
05b74384	138	struct vm_area_struct ** prev,
1da177e4 LT	139	unsigned long start, unsigned long end)
	140	{
	141	struct file *file = vma->vm_file;
	142
1bef4003 S	143	if (!file)
	144	return -EBADF;
	145
70688e4d	146	if (file->f_mapping->a_ops->get_xip_mem) {
fe77ba6f CO	147	/* no bad return value, but ignore advice */
	148	return 0;
	149	}
	150
05b74384	151	*prev = vma;
1da177e4 LT	152	start = ((start - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	153	if (end > vma->vm_end)
	154	end = vma->vm_end;
	155	end = ((end - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	156
f7e839dd	157	force_page_cache_readahead(file->f_mapping, file, start, end - start);
1da177e4 LT	158	return 0;
	159	}
	160
	161	/*
	162	* Application no longer needs these pages. If the pages are dirty,
	163	* it's OK to just throw them away. The app will be more careful about
	164	* data it wants to keep. Be sure to free swap resources too. The
7e6cbea3	165	* zap_page_range call sets things up for shrink_active_list to actually free
1da177e4 LT	166	* these pages later if no one else has touched them in the meantime,
1da177e4 LT	167	* although we could add these pages to a global reuse list for
7e6cbea3	168	* shrink_active_list to pick up before reclaiming other pages.
1da177e4 LT	169	*
	170	* NB: This interface discards data rather than pushes it out to swap,
	171	* as some implementations do. This has performance implications for
	172	* applications like large transactional databases which want to discard
	173	* pages in anonymous maps after committing to backing store the data
	174	* that was kept in them. There is no reason to write this data out to
	175	* the swap area if the application is discarding it.
	176	*
	177	* An interface that causes the system to free clean pages and flush
	178	* dirty pages is already available as msync(MS_INVALIDATE).
	179	*/
	180	static long madvise_dontneed(struct vm_area_struct * vma,
05b74384	181	struct vm_area_struct ** prev,
1da177e4 LT	182	unsigned long start, unsigned long end)
1da177e4 LT	183	{
05b74384	184	*prev = vma;
6aab341e	185	if (vma->vm_flags & (VM_LOCKED\|VM_HUGETLB\|VM_PFNMAP))
1da177e4 LT	186	return -EINVAL;
	187
	188	if (unlikely(vma->vm_flags & VM_NONLINEAR)) {
	189	struct zap_details details = {
	190	.nonlinear_vma = vma,
	191	.last_index = ULONG_MAX,
	192	};
	193	zap_page_range(vma, start, end - start, &details);
	194	} else
	195	zap_page_range(vma, start, end - start, NULL);
	196	return 0;
	197	}
	198
f6b3ec23 BP	199	/*
	200	* Application wants to free up the pages and associated backing store.
	201	* This is effectively punching a hole into the middle of a file.
	202	*
	203	* NOTE: Currently, only shmfs/tmpfs is supported for this operation.
	204	* Other filesystems return -ENOSYS.
	205	*/
	206	static long madvise_remove(struct vm_area_struct *vma,
00e9fa2d	207	struct vm_area_struct **prev,
f6b3ec23 BP	208	unsigned long start, unsigned long end)
f6b3ec23 BP	209	{
3f31d075	210	loff_t offset;
90ed52eb	211	int error;
9ab4233d	212	struct file *f;
f6b3ec23	213
90ed52eb	214	prev = NULL; / tell sys_madvise we drop mmap_sem */
00e9fa2d	215
f6b3ec23 BP	216	if (vma->vm_flags & (VM_LOCKED\|VM_NONLINEAR\|VM_HUGETLB))
	217	return -EINVAL;
	218
9ab4233d AL	219	f = vma->vm_file;
	220
	221	if (!f \|\| !f->f_mapping \|\| !f->f_mapping->host) {
f6b3ec23 BP	222	return -EINVAL;
	223	}
	224
69cf0fac HD	225	if ((vma->vm_flags & (VM_SHARED\|VM_WRITE)) != (VM_SHARED\|VM_WRITE))
	226	return -EACCES;
	227
f6b3ec23 BP	228	offset = (loff_t)(start - vma->vm_start)
f6b3ec23 BP	229	+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
90ed52eb	230
9ab4233d AL	231	/*
	232	* Filesystem's fallocate may need to take i_mutex. We need to
	233	* explicitly grab a reference because the vma (and hence the
	234	* vma's reference to the file) can go away as soon as we drop
	235	* mmap_sem.
	236	*/
	237	get_file(f);
0a27a14a	238	up_read(&current->mm->mmap_sem);
9ab4233d	239	error = do_fallocate(f,
3f31d075 HD	240	FALLOC_FL_PUNCH_HOLE \| FALLOC_FL_KEEP_SIZE,
3f31d075 HD	241	offset, end - start);
9ab4233d	242	fput(f);
0a27a14a	243	down_read(&current->mm->mmap_sem);
90ed52eb	244	return error;
f6b3ec23 BP	245	}
f6b3ec23 BP	246
9893e49d AK	247	#ifdef CONFIG_MEMORY_FAILURE
	248	/*
	249	* Error injection support for memory error handling.
	250	*/
afcf938e	251	static int madvise_hwpoison(int bhv, unsigned long start, unsigned long end)
9893e49d AK	252	{
	253	int ret = 0;
	254
	255	if (!capable(CAP_SYS_ADMIN))
	256	return -EPERM;
	257	for (; start < end; start += PAGE_SIZE) {
	258	struct page *p;
d15f107d	259	int ret = get_user_pages_fast(start, 1, 0, &p);
9893e49d AK	260	if (ret != 1)
9893e49d AK	261	return ret;
afcf938e AK	262	if (bhv == MADV_SOFT_OFFLINE) {
	263	printk(KERN_INFO "Soft offlining page %lx at %lx\n",
	264	page_to_pfn(p), start);
	265	ret = soft_offline_page(p, MF_COUNT_INCREASED);
	266	if (ret)
	267	break;
	268	continue;
	269	}
9893e49d AK	270	printk(KERN_INFO "Injecting memory failure for page %lx at %lx\n",
	271	page_to_pfn(p), start);
	272	/* Ignore return value for now */
cd42f4a3	273	memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED);
9893e49d AK	274	}
	275	return ret;
	276	}
	277	#endif
	278
165cd402	279	static long
	280	madvise_vma(struct vm_area_struct vma, struct vm_area_struct *prev,
	281	unsigned long start, unsigned long end, int behavior)
1da177e4	282	{
1da177e4	283	switch (behavior) {
f6b3ec23	284	case MADV_REMOVE:
3866ea90	285	return madvise_remove(vma, prev, start, end);
1da177e4	286	case MADV_WILLNEED:
3866ea90	287	return madvise_willneed(vma, prev, start, end);
1da177e4	288	case MADV_DONTNEED:
3866ea90	289	return madvise_dontneed(vma, prev, start, end);
1da177e4	290	default:
3866ea90	291	return madvise_behavior(vma, prev, start, end, behavior);
1da177e4	292	}
1da177e4 LT	293	}
1da177e4 LT	294
75927af8 NP	295	static int
	296	madvise_behavior_valid(int behavior)
	297	{
	298	switch (behavior) {
	299	case MADV_DOFORK:
	300	case MADV_DONTFORK:
	301	case MADV_NORMAL:
	302	case MADV_SEQUENTIAL:
	303	case MADV_RANDOM:
	304	case MADV_REMOVE:
	305	case MADV_WILLNEED:
	306	case MADV_DONTNEED:
f8af4da3 HD	307	#ifdef CONFIG_KSM
	308	case MADV_MERGEABLE:
	309	case MADV_UNMERGEABLE:
0af4e98b AA	310	#endif
	311	#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	312	case MADV_HUGEPAGE:
a664b2d8	313	case MADV_NOHUGEPAGE:
f8af4da3	314	#endif
accb61fe JB	315	case MADV_DONTDUMP:
accb61fe JB	316	case MADV_DODUMP:
75927af8 NP	317	return 1;
	318
	319	default:
	320	return 0;
	321	}
	322	}
3866ea90	323
1da177e4 LT	324	/*
	325	* The madvise(2) system call.
	326	*
	327	* Applications can use madvise() to advise the kernel how it should
	328	* handle paging I/O in this VM area. The idea is to help the kernel
	329	* use appropriate read-ahead and caching techniques. The information
	330	* provided is advisory only, and can be safely disregarded by the
	331	* kernel without affecting the correct operation of the application.
	332	*
	333	* behavior values:
	334	* MADV_NORMAL - the default behavior is to read clusters. This
	335	* results in some read-ahead and read-behind.
	336	* MADV_RANDOM - the system should read the minimum amount of data
	337	* on any access, since it is unlikely that the appli-
	338	* cation will need more than what it asks for.
	339	* MADV_SEQUENTIAL - pages in the given range will probably be accessed
	340	* once, so they can be aggressively read ahead, and
	341	* can be freed soon after they are accessed.
	342	* MADV_WILLNEED - the application is notifying the system to read
	343	* some pages ahead.
	344	* MADV_DONTNEED - the application is finished with the given range,
	345	* so the kernel can free resources associated with it.
f6b3ec23 BP	346	* MADV_REMOVE - the application wants to free up the given range of
f6b3ec23 BP	347	* pages and associated backing store.
3866ea90 HD	348	* MADV_DONTFORK - omit this area from child's address space when forking:
	349	* typically, to avoid COWing pages pinned by get_user_pages().
	350	* MADV_DOFORK - cancel MADV_DONTFORK: no longer omit this area when forking.
f8af4da3 HD	351	* MADV_MERGEABLE - the application recommends that KSM try to merge pages in
	352	* this area with pages of identical content from other such areas.
	353	* MADV_UNMERGEABLE- cancel MADV_MERGEABLE: no longer merge pages with others.
1da177e4 LT	354	*
	355	* return values:
	356	* zero - success
	357	* -EINVAL - start + len < 0, start is not page-aligned,
	358	* "behavior" is not a valid value, or application
	359	* is attempting to release locked or shared pages.
	360	* -ENOMEM - addresses in the specified range are not currently
	361	* mapped, or are outside the AS of the process.
	362	* -EIO - an I/O error occurred while paging in data.
	363	* -EBADF - map exists, but area maps something that isn't a file.
	364	* -EAGAIN - a kernel resource was temporarily unavailable.
	365	*/
3480b257	366	SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
1da177e4	367	{
05b74384 PM	368	unsigned long end, tmp;
05b74384 PM	369	struct vm_area_struct * vma, *prev;
1da177e4 LT	370	int unmapped_error = 0;
1da177e4 LT	371	int error = -EINVAL;
f7977793	372	int write;
1da177e4 LT	373	size_t len;
1da177e4 LT	374
9893e49d	375	#ifdef CONFIG_MEMORY_FAILURE
afcf938e AK	376	if (behavior == MADV_HWPOISON \|\| behavior == MADV_SOFT_OFFLINE)
afcf938e AK	377	return madvise_hwpoison(behavior, start, start+len_in);
9893e49d	378	#endif
75927af8 NP	379	if (!madvise_behavior_valid(behavior))
	380	return error;
	381
f7977793 JB	382	write = madvise_need_mmap_write(behavior);
f7977793 JB	383	if (write)
0a27a14a NP	384	down_write(&current->mm->mmap_sem);
	385	else
	386	down_read(&current->mm->mmap_sem);
1da177e4 LT	387
	388	if (start & ~PAGE_MASK)
	389	goto out;
	390	len = (len_in + ~PAGE_MASK) & PAGE_MASK;
	391
	392	/* Check to see whether len was rounded up from small -ve to zero */
	393	if (len_in && !len)
	394	goto out;
	395
	396	end = start + len;
	397	if (end < start)
	398	goto out;
	399
	400	error = 0;
	401	if (end == start)
	402	goto out;
	403
	404	/*
	405	* If the interval [start,end) covers some unmapped address
	406	* ranges, just ignore them, but return -ENOMEM at the end.
05b74384	407	* - different from the way of handling in mlock etc.
1da177e4	408	*/
05b74384	409	vma = find_vma_prev(current->mm, start, &prev);
836d5ffd HD	410	if (vma && start > vma->vm_start)
	411	prev = vma;
	412
1da177e4 LT	413	for (;;) {
	414	/* Still start < end. */
	415	error = -ENOMEM;
	416	if (!vma)
	417	goto out;
	418
05b74384	419	/* Here start < (end\|vma->vm_end). */
1da177e4 LT	420	if (start < vma->vm_start) {
	421	unmapped_error = -ENOMEM;
	422	start = vma->vm_start;
05b74384 PM	423	if (start >= end)
05b74384 PM	424	goto out;
1da177e4 LT	425	}
1da177e4 LT	426
05b74384 PM	427	/* Here vma->vm_start <= start < (end\|vma->vm_end) */
	428	tmp = vma->vm_end;
	429	if (end < tmp)
	430	tmp = end;
1da177e4	431
05b74384 PM	432	/* Here vma->vm_start <= start < tmp <= (end\|vma->vm_end). */
05b74384 PM	433	error = madvise_vma(vma, &prev, start, tmp, behavior);
1da177e4 LT	434	if (error)
1da177e4 LT	435	goto out;
05b74384	436	start = tmp;
90ed52eb	437	if (prev && start < prev->vm_end)
05b74384 PM	438	start = prev->vm_end;
	439	error = unmapped_error;
	440	if (start >= end)
	441	goto out;
90ed52eb HD	442	if (prev)
	443	vma = prev->vm_next;
	444	else /* madvise_remove dropped mmap_sem */
	445	vma = find_vma(current->mm, start);
1da177e4	446	}
1da177e4	447	out:
f7977793	448	if (write)
0a27a14a NP	449	up_write(&current->mm->mmap_sem);
	450	else
	451	up_read(&current->mm->mmap_sem);
	452
1da177e4 LT	453	return error;
1da177e4 LT	454	}