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

/*
 *	mm/mremap.c
 *
 *	(C) Copyright 1996 Linus Torvalds
 *
 *	Address space accounting code	<alan@redhat.com>
 *	(C) Copyright 2002 Red Hat Inc, All Rights Reserved
 */

#include <linux/mm.h>
#include <linux/hugetlb.h>
#include <linux/slab.h>
#include <linux/shm.h>
#include <linux/mman.h>
#include <linux/swap.h>
#include <linux/fs.h>
#include <linux/highmem.h>
#include <linux/security.h>
#include <linux/syscalls.h>

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

static pmd_t *get_old_pmd(struct mm_struct *mm, unsigned long addr)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;

	pgd = pgd_offset(mm, addr);
	if (pgd_none_or_clear_bad(pgd))
		return NULL;

	pud = pud_offset(pgd, addr);
	if (pud_none_or_clear_bad(pud))
		return NULL;

	pmd = pmd_offset(pud, addr);
	if (pmd_none_or_clear_bad(pmd))
		return NULL;

	return pmd;
}

static pmd_t *alloc_new_pmd(struct mm_struct *mm, unsigned long addr)
{
	pgd_t *pgd;
	pud_t *pud;
	pmd_t *pmd;

	pgd = pgd_offset(mm, addr);
	pud = pud_alloc(mm, pgd, addr);
	if (!pud)
		return NULL;

	pmd = pmd_alloc(mm, pud, addr);
	if (!pmd)
		return NULL;

	if (!pmd_present(*pmd) && __pte_alloc(mm, pmd, addr))
		return NULL;

	return pmd;
}

static void move_ptes(struct vm_area_struct *vma, pmd_t *old_pmd,
		unsigned long old_addr, unsigned long old_end,
		struct vm_area_struct *new_vma, pmd_t *new_pmd,
		unsigned long new_addr)
{
	struct address_space *mapping = NULL;
	struct mm_struct *mm = vma->vm_mm;
	pte_t *old_pte, *new_pte, pte;
	spinlock_t *old_ptl;

	if (vma->vm_file) {
		/*
		 * Subtle point from Rajesh Venkatasubramanian: before
		 * moving file-based ptes, we must lock vmtruncate out,
		 * since it might clean the dst vma before the src vma,
		 * and we propagate stale pages into the dst afterward.
		 */
		mapping = vma->vm_file->f_mapping;
		spin_lock(&mapping->i_mmap_lock);
		if (new_vma->vm_truncate_count &&
		    new_vma->vm_truncate_count != vma->vm_truncate_count)
			new_vma->vm_truncate_count = 0;
	}

	old_pte = pte_offset_map_lock(mm, old_pmd, old_addr, &old_ptl);
 	new_pte = pte_offset_map_nested(new_pmd, new_addr);

	for (; old_addr < old_end; old_pte++, old_addr += PAGE_SIZE,
				   new_pte++, new_addr += PAGE_SIZE) {
		if (pte_none(*old_pte))
			continue;
		pte = ptep_clear_flush(vma, old_addr, old_pte);
		/* ZERO_PAGE can be dependant on virtual addr */
		pte = move_pte(pte, new_vma->vm_page_prot, old_addr, new_addr);
		set_pte_at(mm, new_addr, new_pte, pte);
	}

	pte_unmap_nested(new_pte - 1);
	pte_unmap_unlock(old_pte - 1, old_ptl);
	if (mapping)
		spin_unlock(&mapping->i_mmap_lock);
}

#define LATENCY_LIMIT	(64 * PAGE_SIZE)

static unsigned long move_page_tables(struct vm_area_struct *vma,
		unsigned long old_addr, struct vm_area_struct *new_vma,
		unsigned long new_addr, unsigned long len)
{
	unsigned long extent, next, old_end;
	pmd_t *old_pmd, *new_pmd;

	old_end = old_addr + len;
	flush_cache_range(vma, old_addr, old_end);

	for (; old_addr < old_end; old_addr += extent, new_addr += extent) {
		cond_resched();
		next = (old_addr + PMD_SIZE) & PMD_MASK;
		if (next - 1 > old_end)
			next = old_end;
		extent = next - old_addr;
		old_pmd = get_old_pmd(vma->vm_mm, old_addr);
		if (!old_pmd)
			continue;
		new_pmd = alloc_new_pmd(vma->vm_mm, new_addr);
		if (!new_pmd)
			break;
		next = (new_addr + PMD_SIZE) & PMD_MASK;
		if (extent > next - new_addr)
			extent = next - new_addr;
		if (extent > LATENCY_LIMIT)
			extent = LATENCY_LIMIT;
		move_ptes(vma, old_pmd, old_addr, old_addr + extent,
				new_vma, new_pmd, new_addr);
	}

	return len + old_addr - old_end;	/* how much done */
}

static unsigned long move_vma(struct vm_area_struct *vma,
		unsigned long old_addr, unsigned long old_len,
		unsigned long new_len, unsigned long new_addr)
{
	struct mm_struct *mm = vma->vm_mm;
	struct vm_area_struct *new_vma;
	unsigned long vm_flags = vma->vm_flags;
	unsigned long new_pgoff;
	unsigned long moved_len;
	unsigned long excess = 0;
	unsigned long hiwater_vm;
	int split = 0;

	/*
	 * We'd prefer to avoid failure later on in do_munmap:
	 * which may split one vma into three before unmapping.
	 */
	if (mm->map_count >= sysctl_max_map_count - 3)
		return -ENOMEM;

	new_pgoff = vma->vm_pgoff + ((old_addr - vma->vm_start) >> PAGE_SHIFT);
	new_vma = copy_vma(&vma, new_addr, new_len, new_pgoff);
	if (!new_vma)
		return -ENOMEM;

	moved_len = move_page_tables(vma, old_addr, new_vma, new_addr, old_len);
	if (moved_len < old_len) {
		/*
		 * On error, move entries back from new area to old,
		 * which will succeed since page tables still there,
		 * and then proceed to unmap new area instead of old.
		 */
		move_page_tables(new_vma, new_addr, vma, old_addr, moved_len);
		vma = new_vma;
		old_len = new_len;
		old_addr = new_addr;
		new_addr = -ENOMEM;
	}

	/* Conceal VM_ACCOUNT so old reservation is not undone */
	if (vm_flags & VM_ACCOUNT) {
		vma->vm_flags &= ~VM_ACCOUNT;
		excess = vma->vm_end - vma->vm_start - old_len;
		if (old_addr > vma->vm_start &&
		    old_addr + old_len < vma->vm_end)
			split = 1;
	}

	/*
	 * If we failed to move page tables we still do total_vm increment
	 * since do_munmap() will decrement it by old_len == new_len.
	 *
	 * Since total_vm is about to be raised artificially high for a
	 * moment, we need to restore high watermark afterwards: if stats
	 * are taken meanwhile, total_vm and hiwater_vm appear too high.
	 * If this were a serious issue, we'd add a flag to do_munmap().
	 */
	hiwater_vm = mm->hiwater_vm;
	mm->total_vm += new_len >> PAGE_SHIFT;
	vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);

	if (do_munmap(mm, old_addr, old_len) < 0) {
		/* OOM: unable to split vma, just get accounts right */
		vm_unacct_memory(excess >> PAGE_SHIFT);
		excess = 0;
	}
	mm->hiwater_vm = hiwater_vm;

	/* Restore VM_ACCOUNT if one or two pieces of vma left */
	if (excess) {
		vma->vm_flags |= VM_ACCOUNT;
		if (split)
			vma->vm_next->vm_flags |= VM_ACCOUNT;
	}

	if (vm_flags & VM_LOCKED) {
		mm->locked_vm += new_len >> PAGE_SHIFT;
		if (new_len > old_len)
			make_pages_present(new_addr + old_len,
					   new_addr + new_len);
	}

	return new_addr;
}

/*
 * Expand (or shrink) an existing mapping, potentially moving it at the
 * same time (controlled by the MREMAP_MAYMOVE flag and available VM space)
 *
 * MREMAP_FIXED option added 5-Dec-1999 by Benjamin LaHaise
 * This option implies MREMAP_MAYMOVE.
 */
unsigned long do_mremap(unsigned long addr,
	unsigned long old_len, unsigned long new_len,
	unsigned long flags, unsigned long new_addr)
{
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	unsigned long ret = -EINVAL;
	unsigned long charged = 0;

	if (flags & ~(MREMAP_FIXED | MREMAP_MAYMOVE))
		goto out;

	if (addr & ~PAGE_MASK)
		goto out;

	old_len = PAGE_ALIGN(old_len);
	new_len = PAGE_ALIGN(new_len);

	/*
	 * We allow a zero old-len as a special case
	 * for DOS-emu "duplicate shm area" thing. But
	 * a zero new-len is nonsensical.
	 */
	if (!new_len)
		goto out;

	/* new_addr is only valid if MREMAP_FIXED is specified */
	if (flags & MREMAP_FIXED) {
		if (new_addr & ~PAGE_MASK)
			goto out;
		if (!(flags & MREMAP_MAYMOVE))
			goto out;

		if (new_len > TASK_SIZE || new_addr > TASK_SIZE - new_len)
			goto out;

		/* Check if the location we're moving into overlaps the
		 * old location at all, and fail if it does.
		 */
		if ((new_addr <= addr) && (new_addr+new_len) > addr)
			goto out;

		if ((addr <= new_addr) && (addr+old_len) > new_addr)
			goto out;

		ret = do_munmap(mm, new_addr, new_len);
		if (ret)
			goto out;
	}

	/*
	 * Always allow a shrinking remap: that just unmaps
	 * the unnecessary pages..
	 * do_munmap does all the needed commit accounting
	 */
	if (old_len >= new_len) {
		ret = do_munmap(mm, addr+new_len, old_len - new_len);
		if (ret && old_len != new_len)
			goto out;
		ret = addr;
		if (!(flags & MREMAP_FIXED) || (new_addr == addr))
			goto out;
		old_len = new_len;
	}

	/*
	 * Ok, we need to grow..  or relocate.
	 */
	ret = -EFAULT;
	vma = find_vma(mm, addr);
	if (!vma || vma->vm_start > addr)
		goto out;
	if (is_vm_hugetlb_page(vma)) {
		ret = -EINVAL;
		goto out;
	}
	/* We can't remap across vm area boundaries */
	if (old_len > vma->vm_end - addr)
		goto out;
	if (vma->vm_flags & VM_DONTEXPAND) {
		if (new_len > old_len)
			goto out;
	}
	if (vma->vm_flags & VM_LOCKED) {
		unsigned long locked, lock_limit;
		locked = mm->locked_vm << PAGE_SHIFT;
		lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
		locked += new_len - old_len;
		ret = -EAGAIN;
		if (locked > lock_limit && !capable(CAP_IPC_LOCK))
			goto out;
	}
	if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT)) {
		ret = -ENOMEM;
		goto out;
	}

	if (vma->vm_flags & VM_ACCOUNT) {
		charged = (new_len - old_len) >> PAGE_SHIFT;
		if (security_vm_enough_memory(charged))
			goto out_nc;
	}

	/* old_len exactly to the end of the area..
	 * And we're not relocating the area.
	 */
	if (old_len == vma->vm_end - addr &&
	    !((flags & MREMAP_FIXED) && (addr != new_addr)) &&
	    (old_len != new_len || !(flags & MREMAP_MAYMOVE))) {
		unsigned long max_addr = TASK_SIZE;
		if (vma->vm_next)
			max_addr = vma->vm_next->vm_start;
		/* can we just expand the current mapping? */
		if (max_addr - addr >= new_len) {
			int pages = (new_len - old_len) >> PAGE_SHIFT;

			vma_adjust(vma, vma->vm_start,
				addr + new_len, vma->vm_pgoff, NULL);

			mm->total_vm += pages;
			vm_stat_account(mm, vma->vm_flags, vma->vm_file, pages);
			if (vma->vm_flags & VM_LOCKED) {
				mm->locked_vm += pages;
				make_pages_present(addr + old_len,
						   addr + new_len);
			}
			ret = addr;
			goto out;
		}
	}

	/*
	 * We weren't able to just expand or shrink the area,
	 * we need to create a new one and move it..
	 */
	ret = -ENOMEM;
	if (flags & MREMAP_MAYMOVE) {
		if (!(flags & MREMAP_FIXED)) {
			unsigned long map_flags = 0;
			if (vma->vm_flags & VM_MAYSHARE)
				map_flags |= MAP_SHARED;

			new_addr = get_unmapped_area(vma->vm_file, 0, new_len,
						vma->vm_pgoff, map_flags);
			ret = new_addr;
			if (new_addr & ~PAGE_MASK)
				goto out;
		}
		ret = move_vma(vma, addr, old_len, new_len, new_addr);
	}
out:
	if (ret & ~PAGE_MASK)
		vm_unacct_memory(charged);
out_nc:
	return ret;
}

asmlinkage unsigned long sys_mremap(unsigned long addr,
	unsigned long old_len, unsigned long new_len,
	unsigned long flags, unsigned long new_addr)
{
	unsigned long ret;

	down_write(&current->mm->mmap_sem);
	ret = do_mremap(addr, old_len, new_len, flags, new_addr);
	up_write(&current->mm->mmap_sem);
	return ret;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* mm/mremap.c
	3	*
	4	* (C) Copyright 1996 Linus Torvalds
	5	*
	6	* Address space accounting code <alan@redhat.com>
	7	* (C) Copyright 2002 Red Hat Inc, All Rights Reserved
	8	*/
	9
	10	#include <linux/mm.h>
	11	#include <linux/hugetlb.h>
	12	#include <linux/slab.h>
	13	#include <linux/shm.h>
	14	#include <linux/mman.h>
	15	#include <linux/swap.h>
	16	#include <linux/fs.h>
	17	#include <linux/highmem.h>
	18	#include <linux/security.h>
	19	#include <linux/syscalls.h>
	20
	21	#include <asm/uaccess.h>
	22	#include <asm/cacheflush.h>
	23	#include <asm/tlbflush.h>
	24
7be7a546	25	static pmd_t get_old_pmd(struct mm_struct mm, unsigned long addr)
1da177e4 LT	26	{
	27	pgd_t *pgd;
	28	pud_t *pud;
	29	pmd_t *pmd;
	30
	31	pgd = pgd_offset(mm, addr);
	32	if (pgd_none_or_clear_bad(pgd))
	33	return NULL;
	34
	35	pud = pud_offset(pgd, addr);
	36	if (pud_none_or_clear_bad(pud))
	37	return NULL;
	38
	39	pmd = pmd_offset(pud, addr);
	40	if (pmd_none_or_clear_bad(pmd))
	41	return NULL;
	42
7be7a546	43	return pmd;
1da177e4 LT	44	}
1da177e4 LT	45
7be7a546	46	static pmd_t alloc_new_pmd(struct mm_struct mm, unsigned long addr)
1da177e4 LT	47	{
	48	pgd_t *pgd;
	49	pud_t *pud;
c74df32c	50	pmd_t *pmd;
1da177e4 LT	51
1da177e4 LT	52	pgd = pgd_offset(mm, addr);
1da177e4 LT	53	pud = pud_alloc(mm, pgd, addr);
1da177e4 LT	54	if (!pud)
c74df32c	55	return NULL;
7be7a546	56
1da177e4	57	pmd = pmd_alloc(mm, pud, addr);
7be7a546	58	if (!pmd)
c74df32c	59	return NULL;
7be7a546	60
1bb3630e	61	if (!pmd_present(*pmd) && __pte_alloc(mm, pmd, addr))
c74df32c HD	62	return NULL;
c74df32c HD	63
7be7a546	64	return pmd;
1da177e4 LT	65	}
1da177e4 LT	66
7be7a546 HD	67	static void move_ptes(struct vm_area_struct vma, pmd_t old_pmd,
	68	unsigned long old_addr, unsigned long old_end,
	69	struct vm_area_struct new_vma, pmd_t new_pmd,
	70	unsigned long new_addr)
1da177e4 LT	71	{
	72	struct address_space *mapping = NULL;
	73	struct mm_struct *mm = vma->vm_mm;
7be7a546	74	pte_t old_pte, new_pte, pte;
c74df32c	75	spinlock_t *old_ptl;
1da177e4 LT	76
	77	if (vma->vm_file) {
	78	/*
	79	* Subtle point from Rajesh Venkatasubramanian: before
	80	* moving file-based ptes, we must lock vmtruncate out,
	81	* since it might clean the dst vma before the src vma,
	82	* and we propagate stale pages into the dst afterward.
	83	*/
	84	mapping = vma->vm_file->f_mapping;
	85	spin_lock(&mapping->i_mmap_lock);
	86	if (new_vma->vm_truncate_count &&
	87	new_vma->vm_truncate_count != vma->vm_truncate_count)
	88	new_vma->vm_truncate_count = 0;
	89	}
1da177e4	90
c74df32c HD	91	old_pte = pte_offset_map_lock(mm, old_pmd, old_addr, &old_ptl);
c74df32c HD	92	new_pte = pte_offset_map_nested(new_pmd, new_addr);
7be7a546 HD	93
	94	for (; old_addr < old_end; old_pte++, old_addr += PAGE_SIZE,
	95	new_pte++, new_addr += PAGE_SIZE) {
	96	if (pte_none(*old_pte))
	97	continue;
	98	pte = ptep_clear_flush(vma, old_addr, old_pte);
	99	/* ZERO_PAGE can be dependant on virtual addr */
	100	pte = move_pte(pte, new_vma->vm_page_prot, old_addr, new_addr);
	101	set_pte_at(mm, new_addr, new_pte, pte);
1da177e4	102	}
7be7a546 HD	103
7be7a546 HD	104	pte_unmap_nested(new_pte - 1);
c74df32c	105	pte_unmap_unlock(old_pte - 1, old_ptl);
1da177e4 LT	106	if (mapping)
1da177e4 LT	107	spin_unlock(&mapping->i_mmap_lock);
1da177e4 LT	108	}
1da177e4 LT	109
7be7a546 HD	110	#define LATENCY_LIMIT (64 * PAGE_SIZE)
7be7a546 HD	111
1da177e4 LT	112	static unsigned long move_page_tables(struct vm_area_struct *vma,
	113	unsigned long old_addr, struct vm_area_struct *new_vma,
	114	unsigned long new_addr, unsigned long len)
	115	{
7be7a546 HD	116	unsigned long extent, next, old_end;
7be7a546 HD	117	pmd_t old_pmd, new_pmd;
1da177e4	118
7be7a546 HD	119	old_end = old_addr + len;
7be7a546 HD	120	flush_cache_range(vma, old_addr, old_end);
1da177e4	121
7be7a546	122	for (; old_addr < old_end; old_addr += extent, new_addr += extent) {
1da177e4	123	cond_resched();
7be7a546 HD	124	next = (old_addr + PMD_SIZE) & PMD_MASK;
	125	if (next - 1 > old_end)
	126	next = old_end;
	127	extent = next - old_addr;
	128	old_pmd = get_old_pmd(vma->vm_mm, old_addr);
	129	if (!old_pmd)
	130	continue;
	131	new_pmd = alloc_new_pmd(vma->vm_mm, new_addr);
	132	if (!new_pmd)
	133	break;
	134	next = (new_addr + PMD_SIZE) & PMD_MASK;
	135	if (extent > next - new_addr)
	136	extent = next - new_addr;
	137	if (extent > LATENCY_LIMIT)
	138	extent = LATENCY_LIMIT;
	139	move_ptes(vma, old_pmd, old_addr, old_addr + extent,
	140	new_vma, new_pmd, new_addr);
1da177e4	141	}
7be7a546 HD	142
7be7a546 HD	143	return len + old_addr - old_end; /* how much done */
1da177e4 LT	144	}
	145
	146	static unsigned long move_vma(struct vm_area_struct *vma,
	147	unsigned long old_addr, unsigned long old_len,
	148	unsigned long new_len, unsigned long new_addr)
	149	{
	150	struct mm_struct *mm = vma->vm_mm;
	151	struct vm_area_struct *new_vma;
	152	unsigned long vm_flags = vma->vm_flags;
	153	unsigned long new_pgoff;
	154	unsigned long moved_len;
	155	unsigned long excess = 0;
365e9c87	156	unsigned long hiwater_vm;
1da177e4 LT	157	int split = 0;
	158
	159	/*
	160	* We'd prefer to avoid failure later on in do_munmap:
	161	* which may split one vma into three before unmapping.
	162	*/
	163	if (mm->map_count >= sysctl_max_map_count - 3)
	164	return -ENOMEM;
	165
	166	new_pgoff = vma->vm_pgoff + ((old_addr - vma->vm_start) >> PAGE_SHIFT);
	167	new_vma = copy_vma(&vma, new_addr, new_len, new_pgoff);
	168	if (!new_vma)
	169	return -ENOMEM;
	170
	171	moved_len = move_page_tables(vma, old_addr, new_vma, new_addr, old_len);
	172	if (moved_len < old_len) {
	173	/*
	174	* On error, move entries back from new area to old,
	175	* which will succeed since page tables still there,
	176	* and then proceed to unmap new area instead of old.
	177	*/
	178	move_page_tables(new_vma, new_addr, vma, old_addr, moved_len);
	179	vma = new_vma;
	180	old_len = new_len;
	181	old_addr = new_addr;
	182	new_addr = -ENOMEM;
	183	}
	184
	185	/* Conceal VM_ACCOUNT so old reservation is not undone */
	186	if (vm_flags & VM_ACCOUNT) {
	187	vma->vm_flags &= ~VM_ACCOUNT;
	188	excess = vma->vm_end - vma->vm_start - old_len;
	189	if (old_addr > vma->vm_start &&
	190	old_addr + old_len < vma->vm_end)
	191	split = 1;
	192	}
	193
71799062	194	/*
365e9c87 HD	195	* If we failed to move page tables we still do total_vm increment
	196	* since do_munmap() will decrement it by old_len == new_len.
	197	*
	198	* Since total_vm is about to be raised artificially high for a
	199	* moment, we need to restore high watermark afterwards: if stats
	200	* are taken meanwhile, total_vm and hiwater_vm appear too high.
	201	* If this were a serious issue, we'd add a flag to do_munmap().
71799062	202	*/
365e9c87	203	hiwater_vm = mm->hiwater_vm;
71799062	204	mm->total_vm += new_len >> PAGE_SHIFT;
ab50b8ed	205	vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
71799062	206
1da177e4 LT	207	if (do_munmap(mm, old_addr, old_len) < 0) {
	208	/* OOM: unable to split vma, just get accounts right */
	209	vm_unacct_memory(excess >> PAGE_SHIFT);
	210	excess = 0;
	211	}
365e9c87	212	mm->hiwater_vm = hiwater_vm;
1da177e4 LT	213
	214	/* Restore VM_ACCOUNT if one or two pieces of vma left */
	215	if (excess) {
	216	vma->vm_flags \|= VM_ACCOUNT;
	217	if (split)
	218	vma->vm_next->vm_flags \|= VM_ACCOUNT;
	219	}
	220
1da177e4 LT	221	if (vm_flags & VM_LOCKED) {
	222	mm->locked_vm += new_len >> PAGE_SHIFT;
	223	if (new_len > old_len)
	224	make_pages_present(new_addr + old_len,
	225	new_addr + new_len);
	226	}
	227
	228	return new_addr;
	229	}
	230
	231	/*
	232	* Expand (or shrink) an existing mapping, potentially moving it at the
	233	* same time (controlled by the MREMAP_MAYMOVE flag and available VM space)
	234	*
	235	* MREMAP_FIXED option added 5-Dec-1999 by Benjamin LaHaise
	236	* This option implies MREMAP_MAYMOVE.
	237	*/
	238	unsigned long do_mremap(unsigned long addr,
	239	unsigned long old_len, unsigned long new_len,
	240	unsigned long flags, unsigned long new_addr)
	241	{
d0de32d9	242	struct mm_struct *mm = current->mm;
1da177e4 LT	243	struct vm_area_struct *vma;
	244	unsigned long ret = -EINVAL;
	245	unsigned long charged = 0;
	246
	247	if (flags & ~(MREMAP_FIXED \| MREMAP_MAYMOVE))
	248	goto out;
	249
	250	if (addr & ~PAGE_MASK)
	251	goto out;
	252
	253	old_len = PAGE_ALIGN(old_len);
	254	new_len = PAGE_ALIGN(new_len);
	255
	256	/*
	257	* We allow a zero old-len as a special case
	258	* for DOS-emu "duplicate shm area" thing. But
	259	* a zero new-len is nonsensical.
	260	*/
	261	if (!new_len)
	262	goto out;
	263
	264	/* new_addr is only valid if MREMAP_FIXED is specified */
	265	if (flags & MREMAP_FIXED) {
	266	if (new_addr & ~PAGE_MASK)
	267	goto out;
	268	if (!(flags & MREMAP_MAYMOVE))
	269	goto out;
	270
	271	if (new_len > TASK_SIZE \|\| new_addr > TASK_SIZE - new_len)
	272	goto out;
	273
	274	/* Check if the location we're moving into overlaps the
	275	* old location at all, and fail if it does.
	276	*/
	277	if ((new_addr <= addr) && (new_addr+new_len) > addr)
	278	goto out;
	279
	280	if ((addr <= new_addr) && (addr+old_len) > new_addr)
	281	goto out;
	282
d0de32d9	283	ret = do_munmap(mm, new_addr, new_len);
1da177e4 LT	284	if (ret)
	285	goto out;
	286	}
	287
	288	/*
	289	* Always allow a shrinking remap: that just unmaps
	290	* the unnecessary pages..
	291	* do_munmap does all the needed commit accounting
	292	*/
	293	if (old_len >= new_len) {
d0de32d9	294	ret = do_munmap(mm, addr+new_len, old_len - new_len);
1da177e4 LT	295	if (ret && old_len != new_len)
	296	goto out;
	297	ret = addr;
	298	if (!(flags & MREMAP_FIXED) \|\| (new_addr == addr))
	299	goto out;
	300	old_len = new_len;
	301	}
	302
	303	/*
	304	* Ok, we need to grow.. or relocate.
	305	*/
	306	ret = -EFAULT;
d0de32d9	307	vma = find_vma(mm, addr);
1da177e4 LT	308	if (!vma \|\| vma->vm_start > addr)
	309	goto out;
	310	if (is_vm_hugetlb_page(vma)) {
	311	ret = -EINVAL;
	312	goto out;
	313	}
	314	/* We can't remap across vm area boundaries */
	315	if (old_len > vma->vm_end - addr)
	316	goto out;
	317	if (vma->vm_flags & VM_DONTEXPAND) {
	318	if (new_len > old_len)
	319	goto out;
	320	}
	321	if (vma->vm_flags & VM_LOCKED) {
	322	unsigned long locked, lock_limit;
d0de32d9	323	locked = mm->locked_vm << PAGE_SHIFT;
1da177e4 LT	324	lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
	325	locked += new_len - old_len;
	326	ret = -EAGAIN;
	327	if (locked > lock_limit && !capable(CAP_IPC_LOCK))
	328	goto out;
	329	}
d0de32d9	330	if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT)) {
119f657c	331	ret = -ENOMEM;
1da177e4	332	goto out;
119f657c	333	}
1da177e4 LT	334
	335	if (vma->vm_flags & VM_ACCOUNT) {
	336	charged = (new_len - old_len) >> PAGE_SHIFT;
	337	if (security_vm_enough_memory(charged))
	338	goto out_nc;
	339	}
	340
	341	/* old_len exactly to the end of the area..
	342	* And we're not relocating the area.
	343	*/
	344	if (old_len == vma->vm_end - addr &&
	345	!((flags & MREMAP_FIXED) && (addr != new_addr)) &&
	346	(old_len != new_len \|\| !(flags & MREMAP_MAYMOVE))) {
	347	unsigned long max_addr = TASK_SIZE;
	348	if (vma->vm_next)
	349	max_addr = vma->vm_next->vm_start;
	350	/* can we just expand the current mapping? */
	351	if (max_addr - addr >= new_len) {
	352	int pages = (new_len - old_len) >> PAGE_SHIFT;
	353
	354	vma_adjust(vma, vma->vm_start,
	355	addr + new_len, vma->vm_pgoff, NULL);
	356
d0de32d9 HD	357	mm->total_vm += pages;
d0de32d9 HD	358	vm_stat_account(mm, vma->vm_flags, vma->vm_file, pages);
1da177e4	359	if (vma->vm_flags & VM_LOCKED) {
d0de32d9	360	mm->locked_vm += pages;
1da177e4 LT	361	make_pages_present(addr + old_len,
	362	addr + new_len);
	363	}
	364	ret = addr;
	365	goto out;
	366	}
	367	}
	368
	369	/*
	370	* We weren't able to just expand or shrink the area,
	371	* we need to create a new one and move it..
	372	*/
	373	ret = -ENOMEM;
	374	if (flags & MREMAP_MAYMOVE) {
	375	if (!(flags & MREMAP_FIXED)) {
	376	unsigned long map_flags = 0;
	377	if (vma->vm_flags & VM_MAYSHARE)
	378	map_flags \|= MAP_SHARED;
	379
	380	new_addr = get_unmapped_area(vma->vm_file, 0, new_len,
	381	vma->vm_pgoff, map_flags);
	382	ret = new_addr;
	383	if (new_addr & ~PAGE_MASK)
	384	goto out;
	385	}
	386	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	387	}
	388	out:
	389	if (ret & ~PAGE_MASK)
	390	vm_unacct_memory(charged);
	391	out_nc:
	392	return ret;
	393	}
	394
	395	asmlinkage unsigned long sys_mremap(unsigned long addr,
	396	unsigned long old_len, unsigned long new_len,
	397	unsigned long flags, unsigned long new_addr)
	398	{
	399	unsigned long ret;
	400
	401	down_write(&current->mm->mmap_sem);
	402	ret = do_mremap(addr, old_len, new_len, flags, new_addr);
	403	up_write(&current->mm->mmap_sem);
	404	return ret;
	405	}