[linux-2.6-block.git] / mm / fremap.c

/*
 *   linux/mm/fremap.c
 * 
 * Explicit pagetable population and nonlinear (random) mappings support.
 *
 * started by Ingo Molnar, Copyright (C) 2002, 2003
 */
#include <linux/export.h>
#include <linux/backing-dev.h>
#include <linux/mm.h>
#include <linux/swap.h>
#include <linux/file.h>
#include <linux/mman.h>
#include <linux/pagemap.h>
#include <linux/swapops.h>
#include <linux/rmap.h>
#include <linux/syscalls.h>
#include <linux/mmu_notifier.h>

#include <asm/mmu_context.h>
#include <asm/cacheflush.h>
#include <asm/tlbflush.h>

#include "internal.h"

static int mm_counter(struct page *page)
{
	return PageAnon(page) ? MM_ANONPAGES : MM_FILEPAGES;
}

static void zap_pte(struct mm_struct *mm, struct vm_area_struct *vma,
			unsigned long addr, pte_t *ptep)
{
	pte_t pte = *ptep;
	struct page *page;
	swp_entry_t entry;

	if (pte_present(pte)) {
		flush_cache_page(vma, addr, pte_pfn(pte));
		pte = ptep_clear_flush(vma, addr, ptep);
		page = vm_normal_page(vma, addr, pte);
		if (page) {
			if (pte_dirty(pte))
				set_page_dirty(page);
			update_hiwater_rss(mm);
			dec_mm_counter(mm, mm_counter(page));
			page_remove_rmap(page);
			page_cache_release(page);
		}
	} else {	/* zap_pte() is not called when pte_none() */
		if (!pte_file(pte)) {
			update_hiwater_rss(mm);
			entry = pte_to_swp_entry(pte);
			if (non_swap_entry(entry)) {
				if (is_migration_entry(entry)) {
					page = migration_entry_to_page(entry);
					dec_mm_counter(mm, mm_counter(page));
				}
			} else {
				free_swap_and_cache(entry);
				dec_mm_counter(mm, MM_SWAPENTS);
			}
		}
		pte_clear_not_present_full(mm, addr, ptep, 0);
	}
}

/*
 * Install a file pte to a given virtual memory address, release any
 * previously existing mapping.
 */
static int install_file_pte(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long addr, unsigned long pgoff, pgprot_t prot)
{
	int err = -ENOMEM;
	pte_t *pte, ptfile;
	spinlock_t *ptl;

	pte = get_locked_pte(mm, addr, &ptl);
	if (!pte)
		goto out;

	ptfile = pgoff_to_pte(pgoff);

	if (!pte_none(*pte))
		zap_pte(mm, vma, addr, pte);

	set_pte_at(mm, addr, pte, pte_file_mksoft_dirty(ptfile));
	/*
	 * We don't need to run update_mmu_cache() here because the "file pte"
	 * being installed by install_file_pte() is not a real pte - it's a
	 * non-present entry (like a swap entry), noting what file offset should
	 * be mapped there when there's a fault (in a non-linear vma where
	 * that's not obvious).
	 */
	pte_unmap_unlock(pte, ptl);
	err = 0;
out:
	return err;
}

int generic_file_remap_pages(struct vm_area_struct *vma, unsigned long addr,
			     unsigned long size, pgoff_t pgoff)
{
	struct mm_struct *mm = vma->vm_mm;
	int err;

	do {
		err = install_file_pte(mm, vma, addr, pgoff, vma->vm_page_prot);
		if (err)
			return err;

		size -= PAGE_SIZE;
		addr += PAGE_SIZE;
		pgoff++;
	} while (size);

	return 0;
}
EXPORT_SYMBOL(generic_file_remap_pages);

/**
 * sys_remap_file_pages - remap arbitrary pages of an existing VM_SHARED vma
 * @start: start of the remapped virtual memory range
 * @size: size of the remapped virtual memory range
 * @prot: new protection bits of the range (see NOTE)
 * @pgoff: to-be-mapped page of the backing store file
 * @flags: 0 or MAP_NONBLOCKED - the later will cause no IO.
 *
 * sys_remap_file_pages remaps arbitrary pages of an existing VM_SHARED vma
 * (shared backing store file).
 *
 * This syscall works purely via pagetables, so it's the most efficient
 * way to map the same (large) file into a given virtual window. Unlike
 * mmap()/mremap() it does not create any new vmas. The new mappings are
 * also safe across swapout.
 *
 * NOTE: the @prot parameter right now is ignored (but must be zero),
 * and the vma's default protection is used. Arbitrary protections
 * might be implemented in the future.
 */
SYSCALL_DEFINE5(remap_file_pages, unsigned long, start, unsigned long, size,
		unsigned long, prot, unsigned long, pgoff, unsigned long, flags)
{
	struct mm_struct *mm = current->mm;
	struct address_space *mapping;
	struct vm_area_struct *vma;
	int err = -EINVAL;
	int has_write_lock = 0;
	vm_flags_t vm_flags = 0;

	if (prot)
		return err;
	/*
	 * Sanitize the syscall parameters:
	 */
	start = start & PAGE_MASK;
	size = size & PAGE_MASK;

	/* Does the address range wrap, or is the span zero-sized? */
	if (start + size <= start)
		return err;

	/* Does pgoff wrap? */
	if (pgoff + (size >> PAGE_SHIFT) < pgoff)
		return err;

	/* Can we represent this offset inside this architecture's pte's? */
#if PTE_FILE_MAX_BITS < BITS_PER_LONG
	if (pgoff + (size >> PAGE_SHIFT) >= (1UL << PTE_FILE_MAX_BITS))
		return err;
#endif

	/* We need down_write() to change vma->vm_flags. */
	down_read(&mm->mmap_sem);
 retry:
	vma = find_vma(mm, start);

	/*
	 * Make sure the vma is shared, that it supports prefaulting,
	 * and that the remapped range is valid and fully within
	 * the single existing vma.
	 */
	if (!vma || !(vma->vm_flags & VM_SHARED))
		goto out;

	if (!vma->vm_ops || !vma->vm_ops->remap_pages)
		goto out;

	if (start < vma->vm_start || start + size > vma->vm_end)
		goto out;

	/* Must set VM_NONLINEAR before any pages are populated. */
	if (!(vma->vm_flags & VM_NONLINEAR)) {
		/*
		 * vm_private_data is used as a swapout cursor
		 * in a VM_NONLINEAR vma.
		 */
		if (vma->vm_private_data)
			goto out;

		/* Don't need a nonlinear mapping, exit success */
		if (pgoff == linear_page_index(vma, start)) {
			err = 0;
			goto out;
		}

		if (!has_write_lock) {
get_write_lock:
			up_read(&mm->mmap_sem);
			down_write(&mm->mmap_sem);
			has_write_lock = 1;
			goto retry;
		}
		mapping = vma->vm_file->f_mapping;
		/*
		 * page_mkclean doesn't work on nonlinear vmas, so if
		 * dirty pages need to be accounted, emulate with linear
		 * vmas.
		 */
		if (mapping_cap_account_dirty(mapping)) {
			unsigned long addr;
			struct file *file = get_file(vma->vm_file);
			/* mmap_region may free vma; grab the info now */
			vm_flags = vma->vm_flags;

			addr = mmap_region(file, start, size, vm_flags, pgoff);
			fput(file);
			if (IS_ERR_VALUE(addr)) {
				err = addr;
			} else {
				BUG_ON(addr != start);
				err = 0;
			}
			goto out_freed;
		}
		mutex_lock(&mapping->i_mmap_mutex);
		flush_dcache_mmap_lock(mapping);
		vma->vm_flags |= VM_NONLINEAR;
		vma_interval_tree_remove(vma, &mapping->i_mmap);
		vma_nonlinear_insert(vma, &mapping->i_mmap_nonlinear);
		flush_dcache_mmap_unlock(mapping);
		mutex_unlock(&mapping->i_mmap_mutex);
	}

	if (vma->vm_flags & VM_LOCKED) {
		/*
		 * drop PG_Mlocked flag for over-mapped range
		 */
		if (!has_write_lock)
			goto get_write_lock;
		vm_flags = vma->vm_flags;
		munlock_vma_pages_range(vma, start, start + size);
		vma->vm_flags = vm_flags;
	}

	mmu_notifier_invalidate_range_start(mm, start, start + size);
	err = vma->vm_ops->remap_pages(vma, start, size, pgoff);
	mmu_notifier_invalidate_range_end(mm, start, start + size);

	/*
	 * We can't clear VM_NONLINEAR because we'd have to do
	 * it after ->populate completes, and that would prevent
	 * downgrading the lock.  (Locks can't be upgraded).
	 */

out:
	if (vma)
		vm_flags = vma->vm_flags;
out_freed:
	if (likely(!has_write_lock))
		up_read(&mm->mmap_sem);
	else
		up_write(&mm->mmap_sem);
	if (!err && ((vm_flags & VM_LOCKED) || !(flags & MAP_NONBLOCK)))
		mm_populate(start, size);

	return err;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/mm/fremap.c
	3	*
	4	* Explicit pagetable population and nonlinear (random) mappings support.
	5	*
	6	* started by Ingo Molnar, Copyright (C) 2002, 2003
	7	*/
0b173bc4	8	#include <linux/export.h>
4af3c9cc	9	#include <linux/backing-dev.h>
1da177e4 LT	10	#include <linux/mm.h>
	11	#include <linux/swap.h>
	12	#include <linux/file.h>
	13	#include <linux/mman.h>
	14	#include <linux/pagemap.h>
	15	#include <linux/swapops.h>
	16	#include <linux/rmap.h>
1da177e4	17	#include <linux/syscalls.h>
cddb8a5c	18	#include <linux/mmu_notifier.h>
1da177e4 LT	19
	20	#include <asm/mmu_context.h>
	21	#include <asm/cacheflush.h>
	22	#include <asm/tlbflush.h>
	23
ba470de4 RR	24	#include "internal.h"
ba470de4 RR	25
88784396 HD	26	static int mm_counter(struct page *page)
	27	{
	28	return PageAnon(page) ? MM_ANONPAGES : MM_FILEPAGES;
	29	}
	30
d0217ac0	31	static void zap_pte(struct mm_struct mm, struct vm_area_struct vma,
1da177e4 LT	32	unsigned long addr, pte_t *ptep)
	33	{
	34	pte_t pte = *ptep;
88784396 HD	35	struct page *page;
88784396 HD	36	swp_entry_t entry;
1da177e4	37
1da177e4	38	if (pte_present(pte)) {
6aab341e	39	flush_cache_page(vma, addr, pte_pfn(pte));
1da177e4	40	pte = ptep_clear_flush(vma, addr, ptep);
6aab341e LT	41	page = vm_normal_page(vma, addr, pte);
	42	if (page) {
	43	if (pte_dirty(pte))
	44	set_page_dirty(page);
88784396 HD	45	update_hiwater_rss(mm);
88784396 HD	46	dec_mm_counter(mm, mm_counter(page));
edc315fd	47	page_remove_rmap(page);
6aab341e	48	page_cache_release(page);
88784396 HD	49	}
	50	} else { /* zap_pte() is not called when pte_none() */
	51	if (!pte_file(pte)) {
d0217ac0	52	update_hiwater_rss(mm);
88784396 HD	53	entry = pte_to_swp_entry(pte);
	54	if (non_swap_entry(entry)) {
	55	if (is_migration_entry(entry)) {
	56	page = migration_entry_to_page(entry);
	57	dec_mm_counter(mm, mm_counter(page));
	58	}
	59	} else {
	60	free_swap_and_cache(entry);
	61	dec_mm_counter(mm, MM_SWAPENTS);
	62	}
1da177e4	63	}
9888a1ca	64	pte_clear_not_present_full(mm, addr, ptep, 0);
1da177e4 LT	65	}
	66	}
	67
1da177e4 LT	68	/*
	69	* Install a file pte to a given virtual memory address, release any
	70	* previously existing mapping.
	71	*/
d0217ac0	72	static int install_file_pte(struct mm_struct mm, struct vm_area_struct vma,
1da177e4 LT	73	unsigned long addr, unsigned long pgoff, pgprot_t prot)
	74	{
	75	int err = -ENOMEM;
41bb3476	76	pte_t *pte, ptfile;
c74df32c	77	spinlock_t *ptl;
1da177e4	78
c9cfcddf	79	pte = get_locked_pte(mm, addr, &ptl);
1da177e4	80	if (!pte)
c74df32c	81	goto out;
1da177e4	82
41bb3476 CG	83	ptfile = pgoff_to_pte(pgoff);
41bb3476 CG	84
0bf07331	85	if (!pte_none(*pte))
d0217ac0	86	zap_pte(mm, vma, addr, pte);
1da177e4	87
0bf07331	88	set_pte_at(mm, addr, pte, pte_file_mksoft_dirty(ptfile));
668e0d8f HD	89	/*
	90	* We don't need to run update_mmu_cache() here because the "file pte"
	91	* being installed by install_file_pte() is not a real pte - it's a
	92	* non-present entry (like a swap entry), noting what file offset should
	93	* be mapped there when there's a fault (in a non-linear vma where
	94	* that's not obvious).
	95	*/
c74df32c HD	96	pte_unmap_unlock(pte, ptl);
	97	err = 0;
	98	out:
1da177e4 LT	99	return err;
	100	}
	101
0b173bc4 KK	102	int generic_file_remap_pages(struct vm_area_struct *vma, unsigned long addr,
0b173bc4 KK	103	unsigned long size, pgoff_t pgoff)
54cb8821	104	{
0b173bc4	105	struct mm_struct *mm = vma->vm_mm;
54cb8821 NP	106	int err;
	107
	108	do {
	109	err = install_file_pte(mm, vma, addr, pgoff, vma->vm_page_prot);
	110	if (err)
	111	return err;
	112
	113	size -= PAGE_SIZE;
	114	addr += PAGE_SIZE;
	115	pgoff++;
	116	} while (size);
	117
0b173bc4	118	return 0;
54cb8821	119	}
0b173bc4	120	EXPORT_SYMBOL(generic_file_remap_pages);
54cb8821	121
8d63494f RD	122	/**
8d63494f RD	123	* sys_remap_file_pages - remap arbitrary pages of an existing VM_SHARED vma
1da177e4 LT	124	* @start: start of the remapped virtual memory range
1da177e4 LT	125	* @size: size of the remapped virtual memory range
8d63494f RD	126	* @prot: new protection bits of the range (see NOTE)
8d63494f RD	127	* @pgoff: to-be-mapped page of the backing store file
1da177e4 LT	128	* @flags: 0 or MAP_NONBLOCKED - the later will cause no IO.
1da177e4 LT	129	*
8d63494f RD	130	* sys_remap_file_pages remaps arbitrary pages of an existing VM_SHARED vma
	131	* (shared backing store file).
	132	*
	133	* This syscall works purely via pagetables, so it's the most efficient
1da177e4 LT	134	* way to map the same (large) file into a given virtual window. Unlike
	135	* mmap()/mremap() it does not create any new vmas. The new mappings are
	136	* also safe across swapout.
	137	*
7682486b	138	* NOTE: the @prot parameter right now is ignored (but must be zero),
8d63494f RD	139	* and the vma's default protection is used. Arbitrary protections
8d63494f RD	140	* might be implemented in the future.
1da177e4	141	*/
6a6160a7 HC	142	SYSCALL_DEFINE5(remap_file_pages, unsigned long, start, unsigned long, size,
6a6160a7 HC	143	unsigned long, prot, unsigned long, pgoff, unsigned long, flags)
1da177e4 LT	144	{
	145	struct mm_struct *mm = current->mm;
	146	struct address_space *mapping;
1da177e4 LT	147	struct vm_area_struct *vma;
	148	int err = -EINVAL;
	149	int has_write_lock = 0;
a2362d24	150	vm_flags_t vm_flags = 0;
1da177e4	151
8d63494f	152	if (prot)
1da177e4 LT	153	return err;
	154	/*
	155	* Sanitize the syscall parameters:
	156	*/
	157	start = start & PAGE_MASK;
	158	size = size & PAGE_MASK;
	159
	160	/* Does the address range wrap, or is the span zero-sized? */
	161	if (start + size <= start)
	162	return err;
	163
5ec1055a LW	164	/* Does pgoff wrap? */
	165	if (pgoff + (size >> PAGE_SHIFT) < pgoff)
	166	return err;
	167
1da177e4 LT	168	/* Can we represent this offset inside this architecture's pte's? */
	169	#if PTE_FILE_MAX_BITS < BITS_PER_LONG
	170	if (pgoff + (size >> PAGE_SHIFT) >= (1UL << PTE_FILE_MAX_BITS))
	171	return err;
	172	#endif
	173
	174	/* We need down_write() to change vma->vm_flags. */
	175	down_read(&mm->mmap_sem);
	176	retry:
	177	vma = find_vma(mm, start);
	178
	179	/*
	180	* Make sure the vma is shared, that it supports prefaulting,
	181	* and that the remapped range is valid and fully within
940e7da5	182	* the single existing vma.
1da177e4	183	*/
a2362d24	184	if (!vma \|\| !(vma->vm_flags & VM_SHARED))
54cb8821 NP	185	goto out;
54cb8821 NP	186
deb521c4	187	if (!vma->vm_ops \|\| !vma->vm_ops->remap_pages)
54cb8821 NP	188	goto out;
54cb8821 NP	189
e92b05de	190	if (start < vma->vm_start \|\| start + size > vma->vm_end)
54cb8821 NP	191	goto out;
	192
	193	/* Must set VM_NONLINEAR before any pages are populated. */
	194	if (!(vma->vm_flags & VM_NONLINEAR)) {
940e7da5 ML	195	/*
	196	* vm_private_data is used as a swapout cursor
	197	* in a VM_NONLINEAR vma.
	198	*/
	199	if (vma->vm_private_data)
	200	goto out;
	201
54cb8821 NP	202	/* Don't need a nonlinear mapping, exit success */
	203	if (pgoff == linear_page_index(vma, start)) {
	204	err = 0;
	205	goto out;
	206	}
	207
	208	if (!has_write_lock) {
940e7da5	209	get_write_lock:
54cb8821 NP	210	up_read(&mm->mmap_sem);
	211	down_write(&mm->mmap_sem);
	212	has_write_lock = 1;
	213	goto retry;
	214	}
	215	mapping = vma->vm_file->f_mapping;
3ee6dafc MS	216	/*
	217	* page_mkclean doesn't work on nonlinear vmas, so if
	218	* dirty pages need to be accounted, emulate with linear
	219	* vmas.
	220	*/
	221	if (mapping_cap_account_dirty(mapping)) {
	222	unsigned long addr;
cb0942b8	223	struct file *file = get_file(vma->vm_file);
4eb91982 RR	224	/* mmap_region may free vma; grab the info now */
4eb91982 RR	225	vm_flags = vma->vm_flags;
3ee6dafc	226
4eb91982	227	addr = mmap_region(file, start, size, vm_flags, pgoff);
8a459e44	228	fput(file);
3ee6dafc MS	229	if (IS_ERR_VALUE(addr)) {
	230	err = addr;
	231	} else {
	232	BUG_ON(addr != start);
	233	err = 0;
	234	}
4eb91982	235	goto out_freed;
3ee6dafc	236	}
3d48ae45	237	mutex_lock(&mapping->i_mmap_mutex);
54cb8821 NP	238	flush_dcache_mmap_lock(mapping);
54cb8821 NP	239	vma->vm_flags \|= VM_NONLINEAR;
6b2dbba8	240	vma_interval_tree_remove(vma, &mapping->i_mmap);
54cb8821 NP	241	vma_nonlinear_insert(vma, &mapping->i_mmap_nonlinear);
54cb8821 NP	242	flush_dcache_mmap_unlock(mapping);
3d48ae45	243	mutex_unlock(&mapping->i_mmap_mutex);
54cb8821 NP	244	}
54cb8821 NP	245
ba470de4 RR	246	if (vma->vm_flags & VM_LOCKED) {
	247	/*
	248	* drop PG_Mlocked flag for over-mapped range
	249	*/
940e7da5 ML	250	if (!has_write_lock)
940e7da5 ML	251	goto get_write_lock;
a1ea9549	252	vm_flags = vma->vm_flags;
ba470de4	253	munlock_vma_pages_range(vma, start, start + size);
a1ea9549	254	vma->vm_flags = vm_flags;
ba470de4 RR	255	}
ba470de4 RR	256
cddb8a5c	257	mmu_notifier_invalidate_range_start(mm, start, start + size);
0b173bc4	258	err = vma->vm_ops->remap_pages(vma, start, size, pgoff);
cddb8a5c	259	mmu_notifier_invalidate_range_end(mm, start, start + size);
1da177e4	260
54cb8821 NP	261	/*
	262	* We can't clear VM_NONLINEAR because we'd have to do
	263	* it after ->populate completes, and that would prevent
	264	* downgrading the lock. (Locks can't be upgraded).
	265	*/
1da177e4	266
54cb8821	267	out:
6d7825b1 AM	268	if (vma)
6d7825b1 AM	269	vm_flags = vma->vm_flags;
4eb91982	270	out_freed:
1da177e4 LT	271	if (likely(!has_write_lock))
	272	up_read(&mm->mmap_sem);
	273	else
	274	up_write(&mm->mmap_sem);
a1ea9549 ML	275	if (!err && ((vm_flags & VM_LOCKED) \|\| !(flags & MAP_NONBLOCK)))
a1ea9549 ML	276	mm_populate(start, size);
1da177e4 LT	277
	278	return err;
	279	}