[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/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/module.h>
#include <linux/syscalls.h>

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

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

	if (pte_none(pte))
		return;
	if (pte_present(pte)) {
		unsigned long pfn = pte_pfn(pte);
		struct page *page;

		flush_cache_page(vma, addr, pfn);
		pte = ptep_clear_flush(vma, addr, ptep);
		if (unlikely(!pfn_valid(pfn))) {
			print_bad_pte(vma, pte, addr);
			return;
		}
		page = pfn_to_page(pfn);
		if (pte_dirty(pte))
			set_page_dirty(page);
		page_remove_rmap(page);
		page_cache_release(page);
		dec_mm_counter(mm, file_rss);
	} else {
		if (!pte_file(pte))
			free_swap_and_cache(pte_to_swp_entry(pte));
		pte_clear(mm, addr, ptep);
	}
}

/*
 * Install a file page to a given virtual memory address, release any
 * previously existing mapping.
 */
int install_page(struct mm_struct *mm, struct vm_area_struct *vma,
		unsigned long addr, struct page *page, pgprot_t prot)
{
	struct inode *inode;
	pgoff_t size;
	int err = -ENOMEM;
	pte_t *pte;
	pmd_t *pmd;
	pud_t *pud;
	pgd_t *pgd;
	pte_t pte_val;

	BUG_ON(vma->vm_flags & VM_RESERVED);

	pgd = pgd_offset(mm, addr);
	spin_lock(&mm->page_table_lock);
	
	pud = pud_alloc(mm, pgd, addr);
	if (!pud)
		goto err_unlock;

	pmd = pmd_alloc(mm, pud, addr);
	if (!pmd)
		goto err_unlock;

	pte = pte_alloc_map(mm, pmd, addr);
	if (!pte)
		goto err_unlock;

	/*
	 * This page may have been truncated. Tell the
	 * caller about it.
	 */
	err = -EINVAL;
	inode = vma->vm_file->f_mapping->host;
	size = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
	if (!page->mapping || page->index >= size)
		goto err_unlock;
	err = -ENOMEM;
	if (page_mapcount(page) > INT_MAX/2)
		goto err_unlock;

	zap_pte(mm, vma, addr, pte);

	inc_mm_counter(mm, file_rss);
	flush_icache_page(vma, page);
	set_pte_at(mm, addr, pte, mk_pte(page, prot));
	page_add_file_rmap(page);
	pte_val = *pte;
	pte_unmap(pte);
	update_mmu_cache(vma, addr, pte_val);

	err = 0;
err_unlock:
	spin_unlock(&mm->page_table_lock);
	return err;
}
EXPORT_SYMBOL(install_page);


/*
 * Install a file pte to a given virtual memory address, release any
 * previously existing mapping.
 */
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;
	pmd_t *pmd;
	pud_t *pud;
	pgd_t *pgd;
	pte_t pte_val;

	BUG_ON(vma->vm_flags & VM_RESERVED);

	pgd = pgd_offset(mm, addr);
	spin_lock(&mm->page_table_lock);
	
	pud = pud_alloc(mm, pgd, addr);
	if (!pud)
		goto err_unlock;

	pmd = pmd_alloc(mm, pud, addr);
	if (!pmd)
		goto err_unlock;

	pte = pte_alloc_map(mm, pmd, addr);
	if (!pte)
		goto err_unlock;

	zap_pte(mm, vma, addr, pte);

	set_pte_at(mm, addr, pte, pgoff_to_pte(pgoff));
	pte_val = *pte;
	pte_unmap(pte);
	update_mmu_cache(vma, addr, pte_val);
	spin_unlock(&mm->page_table_lock);
	return 0;

err_unlock:
	spin_unlock(&mm->page_table_lock);
	return err;
}


/***
 * sys_remap_file_pages - remap arbitrary pages of a shared backing store
 *                        file within an existing vma.
 * @start: start of the remapped virtual memory range
 * @size: size of the remapped virtual memory range
 * @prot: new protection bits of the range
 * @pgoff: to be mapped page of the backing store file
 * @flags: 0 or MAP_NONBLOCKED - the later will cause no IO.
 *
 * 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, and the vma's default
 * protection is used. Arbitrary protections might be implemented in the
 * future.
 */
asmlinkage long sys_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;
	unsigned long end = start + size;
	struct vm_area_struct *vma;
	int err = -EINVAL;
	int has_write_lock = 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;

	/* 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.  vm_private_data is used as a
	 * swapout cursor in a VM_NONLINEAR vma (unless VM_RESERVED
	 * or VM_LOCKED, but VM_LOCKED could be revoked later on).
	 */
	if (vma && (vma->vm_flags & VM_SHARED) &&
		(!vma->vm_private_data ||
			(vma->vm_flags & (VM_NONLINEAR|VM_RESERVED))) &&
		vma->vm_ops && vma->vm_ops->populate &&
			end > start && start >= vma->vm_start &&
				end <= vma->vm_end) {

		/* Must set VM_NONLINEAR before any pages are populated. */
		if (pgoff != linear_page_index(vma, start) &&
		    !(vma->vm_flags & VM_NONLINEAR)) {
			if (!has_write_lock) {
				up_read(&mm->mmap_sem);
				down_write(&mm->mmap_sem);
				has_write_lock = 1;
				goto retry;
			}
			mapping = vma->vm_file->f_mapping;
			spin_lock(&mapping->i_mmap_lock);
			flush_dcache_mmap_lock(mapping);
			vma->vm_flags |= VM_NONLINEAR;
			vma_prio_tree_remove(vma, &mapping->i_mmap);
			vma_nonlinear_insert(vma, &mapping->i_mmap_nonlinear);
			flush_dcache_mmap_unlock(mapping);
			spin_unlock(&mapping->i_mmap_lock);
		}

		err = vma->vm_ops->populate(vma, start, size,
					    vma->vm_page_prot,
					    pgoff, flags & MAP_NONBLOCK);

		/*
		 * 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).
		 */
	}
	if (likely(!has_write_lock))
		up_read(&mm->mmap_sem);
	else
		up_write(&mm->mmap_sem);

	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	*/
	8
	9	#include <linux/mm.h>
	10	#include <linux/swap.h>
	11	#include <linux/file.h>
	12	#include <linux/mman.h>
	13	#include <linux/pagemap.h>
	14	#include <linux/swapops.h>
	15	#include <linux/rmap.h>
	16	#include <linux/module.h>
	17	#include <linux/syscalls.h>
	18
	19	#include <asm/mmu_context.h>
	20	#include <asm/cacheflush.h>
	21	#include <asm/tlbflush.h>
	22
	23	static inline void zap_pte(struct mm_struct mm, struct vm_area_struct vma,
	24	unsigned long addr, pte_t *ptep)
	25	{
	26	pte_t pte = *ptep;
	27
	28	if (pte_none(pte))
	29	return;
	30	if (pte_present(pte)) {
	31	unsigned long pfn = pte_pfn(pte);
b5810039	32	struct page *page;
1da177e4 LT	33
	34	flush_cache_page(vma, addr, pfn);
	35	pte = ptep_clear_flush(vma, addr, ptep);
b5810039 NP	36	if (unlikely(!pfn_valid(pfn))) {
	37	print_bad_pte(vma, pte, addr);
	38	return;
1da177e4	39	}
b5810039 NP	40	page = pfn_to_page(pfn);
	41	if (pte_dirty(pte))
	42	set_page_dirty(page);
	43	page_remove_rmap(page);
	44	page_cache_release(page);
	45	dec_mm_counter(mm, file_rss);
1da177e4 LT	46	} else {
	47	if (!pte_file(pte))
	48	free_swap_and_cache(pte_to_swp_entry(pte));
	49	pte_clear(mm, addr, ptep);
	50	}
	51	}
	52
	53	/*
	54	* Install a file page to a given virtual memory address, release any
	55	* previously existing mapping.
	56	*/
	57	int install_page(struct mm_struct mm, struct vm_area_struct vma,
	58	unsigned long addr, struct page *page, pgprot_t prot)
	59	{
	60	struct inode *inode;
	61	pgoff_t size;
	62	int err = -ENOMEM;
	63	pte_t *pte;
	64	pmd_t *pmd;
	65	pud_t *pud;
	66	pgd_t *pgd;
	67	pte_t pte_val;
	68
b5810039 NP	69	BUG_ON(vma->vm_flags & VM_RESERVED);
b5810039 NP	70
1da177e4 LT	71	pgd = pgd_offset(mm, addr);
	72	spin_lock(&mm->page_table_lock);
	73
	74	pud = pud_alloc(mm, pgd, addr);
	75	if (!pud)
	76	goto err_unlock;
	77
	78	pmd = pmd_alloc(mm, pud, addr);
	79	if (!pmd)
	80	goto err_unlock;
	81
	82	pte = pte_alloc_map(mm, pmd, addr);
	83	if (!pte)
	84	goto err_unlock;
	85
	86	/*
	87	* This page may have been truncated. Tell the
	88	* caller about it.
	89	*/
	90	err = -EINVAL;
	91	inode = vma->vm_file->f_mapping->host;
	92	size = (i_size_read(inode) + PAGE_CACHE_SIZE - 1) >> PAGE_CACHE_SHIFT;
	93	if (!page->mapping \|\| page->index >= size)
	94	goto err_unlock;
f5154a98 HD	95	err = -ENOMEM;
	96	if (page_mapcount(page) > INT_MAX/2)
	97	goto err_unlock;
1da177e4 LT	98
	99	zap_pte(mm, vma, addr, pte);
	100
4294621f	101	inc_mm_counter(mm, file_rss);
1da177e4 LT	102	flush_icache_page(vma, page);
	103	set_pte_at(mm, addr, pte, mk_pte(page, prot));
	104	page_add_file_rmap(page);
	105	pte_val = *pte;
	106	pte_unmap(pte);
	107	update_mmu_cache(vma, addr, pte_val);
	108
	109	err = 0;
	110	err_unlock:
	111	spin_unlock(&mm->page_table_lock);
	112	return err;
	113	}
	114	EXPORT_SYMBOL(install_page);
	115
	116
	117	/*
	118	* Install a file pte to a given virtual memory address, release any
	119	* previously existing mapping.
	120	*/
	121	int install_file_pte(struct mm_struct mm, struct vm_area_struct vma,
	122	unsigned long addr, unsigned long pgoff, pgprot_t prot)
	123	{
	124	int err = -ENOMEM;
	125	pte_t *pte;
	126	pmd_t *pmd;
	127	pud_t *pud;
	128	pgd_t *pgd;
	129	pte_t pte_val;
	130
b5810039 NP	131	BUG_ON(vma->vm_flags & VM_RESERVED);
b5810039 NP	132
1da177e4 LT	133	pgd = pgd_offset(mm, addr);
	134	spin_lock(&mm->page_table_lock);
	135
	136	pud = pud_alloc(mm, pgd, addr);
	137	if (!pud)
	138	goto err_unlock;
	139
	140	pmd = pmd_alloc(mm, pud, addr);
	141	if (!pmd)
	142	goto err_unlock;
	143
	144	pte = pte_alloc_map(mm, pmd, addr);
	145	if (!pte)
	146	goto err_unlock;
	147
	148	zap_pte(mm, vma, addr, pte);
	149
	150	set_pte_at(mm, addr, pte, pgoff_to_pte(pgoff));
	151	pte_val = *pte;
	152	pte_unmap(pte);
	153	update_mmu_cache(vma, addr, pte_val);
	154	spin_unlock(&mm->page_table_lock);
	155	return 0;
	156
	157	err_unlock:
	158	spin_unlock(&mm->page_table_lock);
	159	return err;
	160	}
	161
	162
	163	/***
	164	* sys_remap_file_pages - remap arbitrary pages of a shared backing store
	165	* file within an existing vma.
	166	* @start: start of the remapped virtual memory range
	167	* @size: size of the remapped virtual memory range
	168	* @prot: new protection bits of the range
	169	* @pgoff: to be mapped page of the backing store file
	170	* @flags: 0 or MAP_NONBLOCKED - the later will cause no IO.
	171	*
	172	* this syscall works purely via pagetables, so it's the most efficient
	173	* way to map the same (large) file into a given virtual window. Unlike
	174	* mmap()/mremap() it does not create any new vmas. The new mappings are
	175	* also safe across swapout.
	176	*
	177	* NOTE: the 'prot' parameter right now is ignored, and the vma's default
	178	* protection is used. Arbitrary protections might be implemented in the
	179	* future.
	180	*/
	181	asmlinkage long sys_remap_file_pages(unsigned long start, unsigned long size,
	182	unsigned long __prot, unsigned long pgoff, unsigned long flags)
	183	{
	184	struct mm_struct *mm = current->mm;
	185	struct address_space *mapping;
	186	unsigned long end = start + size;
	187	struct vm_area_struct *vma;
	188	int err = -EINVAL;
	189	int has_write_lock = 0;
	190
	191	if (__prot)
	192	return err;
	193	/*
	194	* Sanitize the syscall parameters:
	195	*/
	196	start = start & PAGE_MASK;
197	size = size & PAGE_MASK;
198
199	/* Does the address range wrap, or is the span zero-sized? */
200	if (start + size <= start)
201	return err;
202
203	/* Can we represent this offset inside this architecture's pte's? */
204	#if PTE_FILE_MAX_BITS < BITS_PER_LONG
205	if (pgoff + (size >> PAGE_SHIFT) >= (1UL << PTE_FILE_MAX_BITS))
206	return err;
207	#endif
208
209	/* We need down_write() to change vma->vm_flags. */
210	down_read(&mm->mmap_sem);
211	retry:
212	vma = find_vma(mm, start);
213
214	/*
215	* Make sure the vma is shared, that it supports prefaulting,
216	* and that the remapped range is valid and fully within
217	* the single existing vma. vm_private_data is used as a
218	* swapout cursor in a VM_NONLINEAR vma (unless VM_RESERVED
219	* or VM_LOCKED, but VM_LOCKED could be revoked later on).
220	*/
221	if (vma && (vma->vm_flags & VM_SHARED) &&
222	(!vma->vm_private_data \|\|
223	(vma->vm_flags & (VM_NONLINEAR\|VM_RESERVED))) &&
224	vma->vm_ops && vma->vm_ops->populate &&
225	end > start && start >= vma->vm_start &&
226	end <= vma->vm_end) {
227
228	/* Must set VM_NONLINEAR before any pages are populated. */
229	if (pgoff != linear_page_index(vma, start) &&
230	!(vma->vm_flags & VM_NONLINEAR)) {
231	if (!has_write_lock) {
232	up_read(&mm->mmap_sem);
233	down_write(&mm->mmap_sem);
234	has_write_lock = 1;
235	goto retry;
236	}
237	mapping = vma->vm_file->f_mapping;
238	spin_lock(&mapping->i_mmap_lock);
239	flush_dcache_mmap_lock(mapping);
240	vma->vm_flags \|= VM_NONLINEAR;
241	vma_prio_tree_remove(vma, &mapping->i_mmap);
242	vma_nonlinear_insert(vma, &mapping->i_mmap_nonlinear);
243	flush_dcache_mmap_unlock(mapping);
244	spin_unlock(&mapping->i_mmap_lock);
245	}
246
247	err = vma->vm_ops->populate(vma, start, size,
248	vma->vm_page_prot,
249	pgoff, flags & MAP_NONBLOCK);
250
251	/*
252	* We can't clear VM_NONLINEAR because we'd have to do
253	* it after ->populate completes, and that would prevent
254	* downgrading the lock. (Locks can't be upgraded).
255	*/
256	}
257	if (likely(!has_write_lock))
258	up_read(&mm->mmap_sem);
259	else
260	up_write(&mm->mmap_sem);
261
262	return err;
263	}
264