[net-next-2.6.git] / mm / msync.c

/*
 *	linux/mm/msync.c
 *
 * Copyright (C) 1994-1999  Linus Torvalds
 */

/*
 * The msync() system call.
 */
#include <linux/slab.h>
#include <linux/pagemap.h>
#include <linux/fs.h>
#include <linux/mm.h>
#include <linux/mman.h>
#include <linux/hugetlb.h>
#include <linux/writeback.h>
#include <linux/file.h>
#include <linux/syscalls.h>

#include <asm/pgtable.h>
#include <asm/tlbflush.h>

static unsigned long msync_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
				unsigned long addr, unsigned long end)
{
	pte_t *pte;
	spinlock_t *ptl;
	int progress = 0;
	unsigned long ret = 0;

again:
	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
	do {
		struct page *page;

		if (progress >= 64) {
			progress = 0;
			if (need_resched() || need_lockbreak(ptl))
				break;
		}
		progress++;
		if (!pte_present(*pte))
			continue;
		if (!pte_maybe_dirty(*pte))
			continue;
		page = vm_normal_page(vma, addr, *pte);
		if (!page)
			continue;
		if (ptep_clear_flush_dirty(vma, addr, pte) ||
				page_test_and_clear_dirty(page))
			ret += set_page_dirty(page);
		progress += 3;
	} while (pte++, addr += PAGE_SIZE, addr != end);
	pte_unmap_unlock(pte - 1, ptl);
	cond_resched();
	if (addr != end)
		goto again;
	return ret;
}

static inline unsigned long msync_pmd_range(struct vm_area_struct *vma,
			pud_t *pud, unsigned long addr, unsigned long end)
{
	pmd_t *pmd;
	unsigned long next;
	unsigned long ret = 0;

	pmd = pmd_offset(pud, addr);
	do {
		next = pmd_addr_end(addr, end);
		if (pmd_none_or_clear_bad(pmd))
			continue;
		ret += msync_pte_range(vma, pmd, addr, next);
	} while (pmd++, addr = next, addr != end);
	return ret;
}

static inline unsigned long msync_pud_range(struct vm_area_struct *vma,
			pgd_t *pgd, unsigned long addr, unsigned long end)
{
	pud_t *pud;
	unsigned long next;
	unsigned long ret = 0;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
		ret += msync_pmd_range(vma, pud, addr, next);
	} while (pud++, addr = next, addr != end);
	return ret;
}

static unsigned long msync_page_range(struct vm_area_struct *vma,
				unsigned long addr, unsigned long end)
{
	pgd_t *pgd;
	unsigned long next;
	unsigned long ret = 0;

	/* For hugepages we can't go walking the page table normally,
	 * but that's ok, hugetlbfs is memory based, so we don't need
	 * to do anything more on an msync().
	 */
	if (vma->vm_flags & VM_HUGETLB)
		return 0;

	BUG_ON(addr >= end);
	pgd = pgd_offset(vma->vm_mm, addr);
	flush_cache_range(vma, addr, end);
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
		ret += msync_pud_range(vma, pgd, addr, next);
	} while (pgd++, addr = next, addr != end);
	return ret;
}

/*
 * MS_SYNC syncs the entire file - including mappings.
 *
 * MS_ASYNC does not start I/O (it used to, up to 2.5.67).  Instead, it just
 * marks the relevant pages dirty.  The application may now run fsync() to
 * write out the dirty pages and wait on the writeout and check the result.
 * Or the application may run fadvise(FADV_DONTNEED) against the fd to start
 * async writeout immediately.
 * So my _not_ starting I/O in MS_ASYNC we provide complete flexibility to
 * applications.
 */
static int msync_interval(struct vm_area_struct *vma, unsigned long addr,
			unsigned long end, int flags,
			unsigned long *nr_pages_dirtied)
{
	struct file *file = vma->vm_file;

	if ((flags & MS_INVALIDATE) && (vma->vm_flags & VM_LOCKED))
		return -EBUSY;

	if (file && (vma->vm_flags & VM_SHARED))
		*nr_pages_dirtied = msync_page_range(vma, addr, end);
	return 0;
}

asmlinkage long sys_msync(unsigned long start, size_t len, int flags)
{
	unsigned long end;
	struct vm_area_struct *vma;
	int unmapped_error = 0;
	int error = -EINVAL;
	int done = 0;

	if (flags & ~(MS_ASYNC | MS_INVALIDATE | MS_SYNC))
		goto out;
	if (start & ~PAGE_MASK)
		goto out;
	if ((flags & MS_ASYNC) && (flags & MS_SYNC))
		goto out;
	error = -ENOMEM;
	len = (len + ~PAGE_MASK) & PAGE_MASK;
	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.
	 */
	down_read(&current->mm->mmap_sem);
	if (flags & MS_SYNC)
		current->flags |= PF_SYNCWRITE;
	vma = find_vma(current->mm, start);
	if (!vma) {
		error = -ENOMEM;
		goto out_unlock;
	}
	do {
		unsigned long nr_pages_dirtied = 0;
		struct file *file;

		/* Here start < vma->vm_end. */
		if (start < vma->vm_start) {
			unmapped_error = -ENOMEM;
			start = vma->vm_start;
		}
		/* Here vma->vm_start <= start < vma->vm_end. */
		if (end <= vma->vm_end) {
			if (start < end) {
				error = msync_interval(vma, start, end, flags,
							&nr_pages_dirtied);
				if (error)
					goto out_unlock;
			}
			error = unmapped_error;
			done = 1;
		} else {
			/* Here vma->vm_start <= start < vma->vm_end < end. */
			error = msync_interval(vma, start, vma->vm_end, flags,
						&nr_pages_dirtied);
			if (error)
				goto out_unlock;
		}
		file = vma->vm_file;
		start = vma->vm_end;
		if ((flags & MS_ASYNC) && file && nr_pages_dirtied) {
			get_file(file);
			up_read(&current->mm->mmap_sem);
			balance_dirty_pages_ratelimited_nr(file->f_mapping,
							nr_pages_dirtied);
			fput(file);
			down_read(&current->mm->mmap_sem);
			vma = find_vma(current->mm, start);
		} else if ((flags & MS_SYNC) && file &&
				(vma->vm_flags & VM_SHARED)) {
			get_file(file);
			up_read(&current->mm->mmap_sem);
			error = do_fsync(file, 0);
			fput(file);
			down_read(&current->mm->mmap_sem);
			if (error)
				goto out_unlock;
			vma = find_vma(current->mm, start);
		} else {
			vma = vma->vm_next;
		}
	} while (vma && !done);
out_unlock:
	current->flags &= ~PF_SYNCWRITE;
	up_read(&current->mm->mmap_sem);
out:
	return error;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/mm/msync.c
	3	*
	4	* Copyright (C) 1994-1999 Linus Torvalds
	5	*/
	6
	7	/*
	8	* The msync() system call.
	9	*/
	10	#include <linux/slab.h>
	11	#include <linux/pagemap.h>
8f2e9f15	12	#include <linux/fs.h>
1da177e4 LT	13	#include <linux/mm.h>
	14	#include <linux/mman.h>
	15	#include <linux/hugetlb.h>
9c50823e AM	16	#include <linux/writeback.h>
9c50823e AM	17	#include <linux/file.h>
1da177e4 LT	18	#include <linux/syscalls.h>
	19
	20	#include <asm/pgtable.h>
	21	#include <asm/tlbflush.h>
	22
9c50823e	23	static unsigned long msync_pte_range(struct vm_area_struct vma, pmd_t pmd,
1da177e4 LT	24	unsigned long addr, unsigned long end)
	25	{
	26	pte_t *pte;
705e87c0	27	spinlock_t *ptl;
0c942a45	28	int progress = 0;
9c50823e	29	unsigned long ret = 0;
1da177e4	30
0c942a45	31	again:
705e87c0	32	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl);
1da177e4	33	do {
1da177e4 LT	34	struct page *page;
1da177e4 LT	35
0c942a45 HD	36	if (progress >= 64) {
0c942a45 HD	37	progress = 0;
705e87c0	38	if (need_resched() \|\| need_lockbreak(ptl))
0c942a45 HD	39	break;
	40	}
	41	progress++;
1da177e4 LT	42	if (!pte_present(*pte))
1da177e4 LT	43	continue;
b4955ce3 AK	44	if (!pte_maybe_dirty(*pte))
b4955ce3 AK	45	continue;
6aab341e LT	46	page = vm_normal_page(vma, addr, *pte);
6aab341e LT	47	if (!page)
1da177e4	48	continue;
1da177e4	49	if (ptep_clear_flush_dirty(vma, addr, pte) \|\|
9c50823e AM	50	page_test_and_clear_dirty(page))
9c50823e AM	51	ret += set_page_dirty(page);
0c942a45	52	progress += 3;
1da177e4	53	} while (pte++, addr += PAGE_SIZE, addr != end);
705e87c0 HD	54	pte_unmap_unlock(pte - 1, ptl);
705e87c0 HD	55	cond_resched();
0c942a45 HD	56	if (addr != end)
0c942a45 HD	57	goto again;
9c50823e	58	return ret;
1da177e4 LT	59	}
1da177e4 LT	60
9c50823e AM	61	static inline unsigned long msync_pmd_range(struct vm_area_struct *vma,
9c50823e AM	62	pud_t *pud, unsigned long addr, unsigned long end)
1da177e4 LT	63	{
	64	pmd_t *pmd;
	65	unsigned long next;
9c50823e	66	unsigned long ret = 0;
1da177e4 LT	67
	68	pmd = pmd_offset(pud, addr);
	69	do {
	70	next = pmd_addr_end(addr, end);
	71	if (pmd_none_or_clear_bad(pmd))
	72	continue;
9c50823e	73	ret += msync_pte_range(vma, pmd, addr, next);
1da177e4	74	} while (pmd++, addr = next, addr != end);
9c50823e	75	return ret;
1da177e4 LT	76	}
1da177e4 LT	77
9c50823e AM	78	static inline unsigned long msync_pud_range(struct vm_area_struct *vma,
9c50823e AM	79	pgd_t *pgd, unsigned long addr, unsigned long end)
1da177e4 LT	80	{
	81	pud_t *pud;
	82	unsigned long next;
9c50823e	83	unsigned long ret = 0;
1da177e4 LT	84
	85	pud = pud_offset(pgd, addr);
	86	do {
	87	next = pud_addr_end(addr, end);
	88	if (pud_none_or_clear_bad(pud))
	89	continue;
9c50823e	90	ret += msync_pmd_range(vma, pud, addr, next);
1da177e4	91	} while (pud++, addr = next, addr != end);
9c50823e	92	return ret;
1da177e4 LT	93	}
1da177e4 LT	94
9c50823e	95	static unsigned long msync_page_range(struct vm_area_struct *vma,
1da177e4 LT	96	unsigned long addr, unsigned long end)
1da177e4 LT	97	{
1da177e4 LT	98	pgd_t *pgd;
1da177e4 LT	99	unsigned long next;
9c50823e	100	unsigned long ret = 0;
1da177e4 LT	101
	102	/* For hugepages we can't go walking the page table normally,
	103	* but that's ok, hugetlbfs is memory based, so we don't need
b5810039	104	* to do anything more on an msync().
b5810039	105	*/
6aab341e	106	if (vma->vm_flags & VM_HUGETLB)
9c50823e	107	return 0;
1da177e4 LT	108
1da177e4 LT	109	BUG_ON(addr >= end);
705e87c0	110	pgd = pgd_offset(vma->vm_mm, addr);
1da177e4	111	flush_cache_range(vma, addr, end);
1da177e4 LT	112	do {
	113	next = pgd_addr_end(addr, end);
	114	if (pgd_none_or_clear_bad(pgd))
	115	continue;
9c50823e	116	ret += msync_pud_range(vma, pgd, addr, next);
1da177e4	117	} while (pgd++, addr = next, addr != end);
9c50823e	118	return ret;
1da177e4 LT	119	}
1da177e4 LT	120
1da177e4 LT	121	/*
	122	* MS_SYNC syncs the entire file - including mappings.
	123	*
	124	* MS_ASYNC does not start I/O (it used to, up to 2.5.67). Instead, it just
	125	* marks the relevant pages dirty. The application may now run fsync() to
	126	* write out the dirty pages and wait on the writeout and check the result.
	127	* Or the application may run fadvise(FADV_DONTNEED) against the fd to start
	128	* async writeout immediately.
	129	* So my _not_ starting I/O in MS_ASYNC we provide complete flexibility to
	130	* applications.
	131	*/
9c50823e AM	132	static int msync_interval(struct vm_area_struct *vma, unsigned long addr,
	133	unsigned long end, int flags,
	134	unsigned long *nr_pages_dirtied)
1da177e4	135	{
1da177e4 LT	136	struct file *file = vma->vm_file;
	137
	138	if ((flags & MS_INVALIDATE) && (vma->vm_flags & VM_LOCKED))
	139	return -EBUSY;
	140
707c21c8	141	if (file && (vma->vm_flags & VM_SHARED))
9c50823e	142	*nr_pages_dirtied = msync_page_range(vma, addr, end);
707c21c8	143	return 0;
1da177e4 LT	144	}
	145
	146	asmlinkage long sys_msync(unsigned long start, size_t len, int flags)
	147	{
	148	unsigned long end;
	149	struct vm_area_struct *vma;
676758bd AM	150	int unmapped_error = 0;
676758bd AM	151	int error = -EINVAL;
9c50823e	152	int done = 0;
1da177e4	153
1da177e4 LT	154	if (flags & ~(MS_ASYNC \| MS_INVALIDATE \| MS_SYNC))
	155	goto out;
	156	if (start & ~PAGE_MASK)
	157	goto out;
	158	if ((flags & MS_ASYNC) && (flags & MS_SYNC))
	159	goto out;
	160	error = -ENOMEM;
	161	len = (len + ~PAGE_MASK) & PAGE_MASK;
	162	end = start + len;
	163	if (end < start)
	164	goto out;
	165	error = 0;
	166	if (end == start)
	167	goto out;
	168	/*
	169	* If the interval [start,end) covers some unmapped address ranges,
	170	* just ignore them, but return -ENOMEM at the end.
	171	*/
9c50823e AM	172	down_read(&current->mm->mmap_sem);
	173	if (flags & MS_SYNC)
	174	current->flags \|= PF_SYNCWRITE;
1da177e4	175	vma = find_vma(current->mm, start);
676758bd AM	176	if (!vma) {
	177	error = -ENOMEM;
	178	goto out_unlock;
	179	}
9c50823e AM	180	do {
	181	unsigned long nr_pages_dirtied = 0;
	182	struct file *file;
	183
1da177e4 LT	184	/* Here start < vma->vm_end. */
	185	if (start < vma->vm_start) {
	186	unmapped_error = -ENOMEM;
	187	start = vma->vm_start;
	188	}
	189	/* Here vma->vm_start <= start < vma->vm_end. */
	190	if (end <= vma->vm_end) {
	191	if (start < end) {
9c50823e AM	192	error = msync_interval(vma, start, end, flags,
9c50823e AM	193	&nr_pages_dirtied);
1da177e4	194	if (error)
9c50823e	195	goto out_unlock;
1da177e4 LT	196	}
1da177e4 LT	197	error = unmapped_error;
9c50823e AM	198	done = 1;
	199	} else {
	200	/* Here vma->vm_start <= start < vma->vm_end < end. */
	201	error = msync_interval(vma, start, vma->vm_end, flags,
	202	&nr_pages_dirtied);
	203	if (error)
	204	goto out_unlock;
1da177e4	205	}
9c50823e	206	file = vma->vm_file;
1da177e4	207	start = vma->vm_end;
9c50823e AM	208	if ((flags & MS_ASYNC) && file && nr_pages_dirtied) {
	209	get_file(file);
	210	up_read(&current->mm->mmap_sem);
	211	balance_dirty_pages_ratelimited_nr(file->f_mapping,
	212	nr_pages_dirtied);
	213	fput(file);
	214	down_read(&current->mm->mmap_sem);
	215	vma = find_vma(current->mm, start);
707c21c8 AM	216	} else if ((flags & MS_SYNC) && file &&
707c21c8 AM	217	(vma->vm_flags & VM_SHARED)) {
707c21c8 AM	218	get_file(file);
707c21c8 AM	219	up_read(&current->mm->mmap_sem);
8f2e9f15	220	error = do_fsync(file, 0);
707c21c8 AM	221	fput(file);
	222	down_read(&current->mm->mmap_sem);
	223	if (error)
	224	goto out_unlock;
	225	vma = find_vma(current->mm, start);
9c50823e AM	226	} else {
	227	vma = vma->vm_next;
	228	}
676758bd	229	} while (vma && !done);
9c50823e	230	out_unlock:
1da177e4	231	current->flags &= ~PF_SYNCWRITE;
9c50823e AM	232	up_read(&current->mm->mmap_sem);
9c50823e AM	233	out:
1da177e4 LT	234	return error;
1da177e4 LT	235	}