[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/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_MERGEABLE:
	case MADV_UNMERGEABLE:
		error = ksm_madvise(vma, start, end, behavior, &new_flags);
		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 and i_alloc_sem */
	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
		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;
f8af4da3 HD	68	case MADV_MERGEABLE:
	69	case MADV_UNMERGEABLE:
	70	error = ksm_madvise(vma, start, end, behavior, &new_flags);
	71	if (error)
	72	goto out;
	73	break;
e798c6e8 PM	74	}
e798c6e8 PM	75
05b74384 PM	76	if (new_flags == vma->vm_flags) {
05b74384 PM	77	*prev = vma;
836d5ffd	78	goto out;
05b74384 PM	79	}
	80
	81	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	82	prev = vma_merge(mm, prev, start, end, new_flags, vma->anon_vma,
	83	vma->vm_file, pgoff, vma_policy(vma));
	84	if (*prev) {
	85	vma = *prev;
	86	goto success;
	87	}
	88
	89	*prev = vma;
1da177e4 LT	90
	91	if (start != vma->vm_start) {
	92	error = split_vma(mm, vma, start, 1);
	93	if (error)
	94	goto out;
	95	}
	96
	97	if (end != vma->vm_end) {
	98	error = split_vma(mm, vma, end, 0);
	99	if (error)
	100	goto out;
	101	}
	102
836d5ffd	103	success:
1da177e4 LT	104	/*
	105	* vm_flags is protected by the mmap_sem held in write mode.
	106	*/
e798c6e8	107	vma->vm_flags = new_flags;
1da177e4 LT	108
	109	out:
	110	if (error == -ENOMEM)
	111	error = -EAGAIN;
	112	return error;
	113	}
	114
	115	/*
	116	* Schedule all required I/O operations. Do not wait for completion.
	117	*/
	118	static long madvise_willneed(struct vm_area_struct * vma,
05b74384	119	struct vm_area_struct ** prev,
1da177e4 LT	120	unsigned long start, unsigned long end)
	121	{
	122	struct file *file = vma->vm_file;
	123
1bef4003 S	124	if (!file)
	125	return -EBADF;
	126
70688e4d	127	if (file->f_mapping->a_ops->get_xip_mem) {
fe77ba6f CO	128	/* no bad return value, but ignore advice */
	129	return 0;
	130	}
	131
05b74384	132	*prev = vma;
1da177e4 LT	133	start = ((start - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	134	if (end > vma->vm_end)
	135	end = vma->vm_end;
	136	end = ((end - vma->vm_start) >> PAGE_SHIFT) + vma->vm_pgoff;
	137
f7e839dd	138	force_page_cache_readahead(file->f_mapping, file, start, end - start);
1da177e4 LT	139	return 0;
	140	}
	141
	142	/*
	143	* Application no longer needs these pages. If the pages are dirty,
	144	* it's OK to just throw them away. The app will be more careful about
	145	* data it wants to keep. Be sure to free swap resources too. The
7e6cbea3	146	* zap_page_range call sets things up for shrink_active_list to actually free
1da177e4 LT	147	* these pages later if no one else has touched them in the meantime,
1da177e4 LT	148	* although we could add these pages to a global reuse list for
7e6cbea3	149	* shrink_active_list to pick up before reclaiming other pages.
1da177e4 LT	150	*
	151	* NB: This interface discards data rather than pushes it out to swap,
	152	* as some implementations do. This has performance implications for
	153	* applications like large transactional databases which want to discard
	154	* pages in anonymous maps after committing to backing store the data
	155	* that was kept in them. There is no reason to write this data out to
	156	* the swap area if the application is discarding it.
	157	*
	158	* An interface that causes the system to free clean pages and flush
	159	* dirty pages is already available as msync(MS_INVALIDATE).
	160	*/
	161	static long madvise_dontneed(struct vm_area_struct * vma,
05b74384	162	struct vm_area_struct ** prev,
1da177e4 LT	163	unsigned long start, unsigned long end)
1da177e4 LT	164	{
05b74384	165	*prev = vma;
6aab341e	166	if (vma->vm_flags & (VM_LOCKED\|VM_HUGETLB\|VM_PFNMAP))
1da177e4 LT	167	return -EINVAL;
	168
	169	if (unlikely(vma->vm_flags & VM_NONLINEAR)) {
	170	struct zap_details details = {
	171	.nonlinear_vma = vma,
	172	.last_index = ULONG_MAX,
	173	};
	174	zap_page_range(vma, start, end - start, &details);
	175	} else
	176	zap_page_range(vma, start, end - start, NULL);
	177	return 0;
	178	}
	179
f6b3ec23 BP	180	/*
	181	* Application wants to free up the pages and associated backing store.
	182	* This is effectively punching a hole into the middle of a file.
	183	*
	184	* NOTE: Currently, only shmfs/tmpfs is supported for this operation.
	185	* Other filesystems return -ENOSYS.
	186	*/
	187	static long madvise_remove(struct vm_area_struct *vma,
00e9fa2d	188	struct vm_area_struct **prev,
f6b3ec23 BP	189	unsigned long start, unsigned long end)
	190	{
	191	struct address_space *mapping;
90ed52eb HD	192	loff_t offset, endoff;
90ed52eb HD	193	int error;
f6b3ec23	194
90ed52eb	195	prev = NULL; / tell sys_madvise we drop mmap_sem */
00e9fa2d	196
f6b3ec23 BP	197	if (vma->vm_flags & (VM_LOCKED\|VM_NONLINEAR\|VM_HUGETLB))
	198	return -EINVAL;
	199
	200	if (!vma->vm_file \|\| !vma->vm_file->f_mapping
	201	\|\| !vma->vm_file->f_mapping->host) {
	202	return -EINVAL;
	203	}
	204
69cf0fac HD	205	if ((vma->vm_flags & (VM_SHARED\|VM_WRITE)) != (VM_SHARED\|VM_WRITE))
	206	return -EACCES;
	207
f6b3ec23 BP	208	mapping = vma->vm_file->f_mapping;
	209
	210	offset = (loff_t)(start - vma->vm_start)
	211	+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
	212	endoff = (loff_t)(end - vma->vm_start - 1)
	213	+ ((loff_t)vma->vm_pgoff << PAGE_SHIFT);
90ed52eb HD	214
90ed52eb HD	215	/* vmtruncate_range needs to take i_mutex and i_alloc_sem */
0a27a14a	216	up_read(&current->mm->mmap_sem);
90ed52eb	217	error = vmtruncate_range(mapping->host, offset, endoff);
0a27a14a	218	down_read(&current->mm->mmap_sem);
90ed52eb	219	return error;
f6b3ec23 BP	220	}
f6b3ec23 BP	221
9893e49d AK	222	#ifdef CONFIG_MEMORY_FAILURE
	223	/*
	224	* Error injection support for memory error handling.
	225	*/
afcf938e	226	static int madvise_hwpoison(int bhv, unsigned long start, unsigned long end)
9893e49d AK	227	{
	228	int ret = 0;
	229
	230	if (!capable(CAP_SYS_ADMIN))
	231	return -EPERM;
	232	for (; start < end; start += PAGE_SIZE) {
	233	struct page *p;
d15f107d	234	int ret = get_user_pages_fast(start, 1, 0, &p);
9893e49d AK	235	if (ret != 1)
9893e49d AK	236	return ret;
afcf938e AK	237	if (bhv == MADV_SOFT_OFFLINE) {
	238	printk(KERN_INFO "Soft offlining page %lx at %lx\n",
	239	page_to_pfn(p), start);
	240	ret = soft_offline_page(p, MF_COUNT_INCREASED);
	241	if (ret)
	242	break;
	243	continue;
	244	}
9893e49d AK	245	printk(KERN_INFO "Injecting memory failure for page %lx at %lx\n",
	246	page_to_pfn(p), start);
	247	/* Ignore return value for now */
82ba011b	248	__memory_failure(page_to_pfn(p), 0, MF_COUNT_INCREASED);
9893e49d AK	249	}
	250	return ret;
	251	}
	252	#endif
	253
165cd402	254	static long
	255	madvise_vma(struct vm_area_struct vma, struct vm_area_struct *prev,
	256	unsigned long start, unsigned long end, int behavior)
1da177e4	257	{
1da177e4	258	switch (behavior) {
f6b3ec23	259	case MADV_REMOVE:
3866ea90	260	return madvise_remove(vma, prev, start, end);
1da177e4	261	case MADV_WILLNEED:
3866ea90	262	return madvise_willneed(vma, prev, start, end);
1da177e4	263	case MADV_DONTNEED:
3866ea90	264	return madvise_dontneed(vma, prev, start, end);
1da177e4	265	default:
3866ea90	266	return madvise_behavior(vma, prev, start, end, behavior);
1da177e4	267	}
1da177e4 LT	268	}
1da177e4 LT	269
75927af8 NP	270	static int
	271	madvise_behavior_valid(int behavior)
	272	{
	273	switch (behavior) {
	274	case MADV_DOFORK:
	275	case MADV_DONTFORK:
	276	case MADV_NORMAL:
	277	case MADV_SEQUENTIAL:
	278	case MADV_RANDOM:
	279	case MADV_REMOVE:
	280	case MADV_WILLNEED:
	281	case MADV_DONTNEED:
f8af4da3 HD	282	#ifdef CONFIG_KSM
	283	case MADV_MERGEABLE:
	284	case MADV_UNMERGEABLE:
	285	#endif
75927af8 NP	286	return 1;
	287
	288	default:
	289	return 0;
	290	}
	291	}
3866ea90	292
1da177e4 LT	293	/*
	294	* The madvise(2) system call.
	295	*
	296	* Applications can use madvise() to advise the kernel how it should
	297	* handle paging I/O in this VM area. The idea is to help the kernel
	298	* use appropriate read-ahead and caching techniques. The information
	299	* provided is advisory only, and can be safely disregarded by the
	300	* kernel without affecting the correct operation of the application.
	301	*
	302	* behavior values:
	303	* MADV_NORMAL - the default behavior is to read clusters. This
	304	* results in some read-ahead and read-behind.
	305	* MADV_RANDOM - the system should read the minimum amount of data
	306	* on any access, since it is unlikely that the appli-
	307	* cation will need more than what it asks for.
	308	* MADV_SEQUENTIAL - pages in the given range will probably be accessed
	309	* once, so they can be aggressively read ahead, and
	310	* can be freed soon after they are accessed.
	311	* MADV_WILLNEED - the application is notifying the system to read
	312	* some pages ahead.
	313	* MADV_DONTNEED - the application is finished with the given range,
	314	* so the kernel can free resources associated with it.
f6b3ec23 BP	315	* MADV_REMOVE - the application wants to free up the given range of
f6b3ec23 BP	316	* pages and associated backing store.
3866ea90 HD	317	* MADV_DONTFORK - omit this area from child's address space when forking:
	318	* typically, to avoid COWing pages pinned by get_user_pages().
	319	* MADV_DOFORK - cancel MADV_DONTFORK: no longer omit this area when forking.
f8af4da3 HD	320	* MADV_MERGEABLE - the application recommends that KSM try to merge pages in
	321	* this area with pages of identical content from other such areas.
	322	* MADV_UNMERGEABLE- cancel MADV_MERGEABLE: no longer merge pages with others.
1da177e4 LT	323	*
	324	* return values:
	325	* zero - success
	326	* -EINVAL - start + len < 0, start is not page-aligned,
	327	* "behavior" is not a valid value, or application
	328	* is attempting to release locked or shared pages.
	329	* -ENOMEM - addresses in the specified range are not currently
	330	* mapped, or are outside the AS of the process.
	331	* -EIO - an I/O error occurred while paging in data.
	332	* -EBADF - map exists, but area maps something that isn't a file.
	333	* -EAGAIN - a kernel resource was temporarily unavailable.
	334	*/
3480b257	335	SYSCALL_DEFINE3(madvise, unsigned long, start, size_t, len_in, int, behavior)
1da177e4	336	{
05b74384 PM	337	unsigned long end, tmp;
05b74384 PM	338	struct vm_area_struct * vma, *prev;
1da177e4 LT	339	int unmapped_error = 0;
1da177e4 LT	340	int error = -EINVAL;
f7977793	341	int write;
1da177e4 LT	342	size_t len;
1da177e4 LT	343
9893e49d	344	#ifdef CONFIG_MEMORY_FAILURE
afcf938e AK	345	if (behavior == MADV_HWPOISON \|\| behavior == MADV_SOFT_OFFLINE)
afcf938e AK	346	return madvise_hwpoison(behavior, start, start+len_in);
9893e49d	347	#endif
75927af8 NP	348	if (!madvise_behavior_valid(behavior))
	349	return error;
	350
f7977793 JB	351	write = madvise_need_mmap_write(behavior);
f7977793 JB	352	if (write)
0a27a14a NP	353	down_write(&current->mm->mmap_sem);
	354	else
	355	down_read(&current->mm->mmap_sem);
1da177e4 LT	356
	357	if (start & ~PAGE_MASK)
	358	goto out;
	359	len = (len_in + ~PAGE_MASK) & PAGE_MASK;
	360
	361	/* Check to see whether len was rounded up from small -ve to zero */
	362	if (len_in && !len)
	363	goto out;
	364
	365	end = start + len;
	366	if (end < start)
	367	goto out;
	368
	369	error = 0;
	370	if (end == start)
	371	goto out;
	372
	373	/*
	374	* If the interval [start,end) covers some unmapped address
	375	* ranges, just ignore them, but return -ENOMEM at the end.
05b74384	376	* - different from the way of handling in mlock etc.
1da177e4	377	*/
05b74384	378	vma = find_vma_prev(current->mm, start, &prev);
836d5ffd HD	379	if (vma && start > vma->vm_start)
	380	prev = vma;
	381
1da177e4 LT	382	for (;;) {
	383	/* Still start < end. */
	384	error = -ENOMEM;
	385	if (!vma)
	386	goto out;
	387
05b74384	388	/* Here start < (end\|vma->vm_end). */
1da177e4 LT	389	if (start < vma->vm_start) {
	390	unmapped_error = -ENOMEM;
	391	start = vma->vm_start;
05b74384 PM	392	if (start >= end)
05b74384 PM	393	goto out;
1da177e4 LT	394	}
1da177e4 LT	395
05b74384 PM	396	/* Here vma->vm_start <= start < (end\|vma->vm_end) */
	397	tmp = vma->vm_end;
	398	if (end < tmp)
	399	tmp = end;
1da177e4	400
05b74384 PM	401	/* Here vma->vm_start <= start < tmp <= (end\|vma->vm_end). */
05b74384 PM	402	error = madvise_vma(vma, &prev, start, tmp, behavior);
1da177e4 LT	403	if (error)
1da177e4 LT	404	goto out;
05b74384	405	start = tmp;
90ed52eb	406	if (prev && start < prev->vm_end)
05b74384 PM	407	start = prev->vm_end;
	408	error = unmapped_error;
	409	if (start >= end)
	410	goto out;
90ed52eb HD	411	if (prev)
	412	vma = prev->vm_next;
	413	else /* madvise_remove dropped mmap_sem */
	414	vma = find_vma(current->mm, start);
1da177e4	415	}
1da177e4	416	out:
f7977793	417	if (write)
0a27a14a NP	418	up_write(&current->mm->mmap_sem);
	419	else
	420	up_read(&current->mm->mmap_sem);
	421
1da177e4 LT	422	return error;
1da177e4 LT	423	}