[linux.git] / mm / mremap.c

/*
 *	mm/mremap.c
 *
 *	(C) Copyright 1996 Linus Torvalds
 *
 *	Address space accounting code	<[email protected]>
 *	(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, *new_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;
	}

	/*
	 * We don't have to worry about the ordering of src and dst
	 * pte locks because exclusive mmap_sem prevents deadlock.
	 */
	old_pte = pte_offset_map_lock(mm, old_pmd, old_addr, &old_ptl);
 	new_pte = pte_offset_map_nested(new_pmd, new_addr);
	new_ptl = pte_lockptr(mm, new_pmd);
	if (new_ptl != old_ptl)
		spin_lock(new_ptl);

	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);
	}

	if (new_ptl != old_ptl)
		spin_unlock(new_ptl);
	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 <[email protected]>
	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;
4c21e2f2	75	spinlock_t old_ptl, new_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
4c21e2f2 HD	91	/*
	92	* We don't have to worry about the ordering of src and dst
	93	* pte locks because exclusive mmap_sem prevents deadlock.
	94	*/
c74df32c HD	95	old_pte = pte_offset_map_lock(mm, old_pmd, old_addr, &old_ptl);
c74df32c HD	96	new_pte = pte_offset_map_nested(new_pmd, new_addr);
4c21e2f2 HD	97	new_ptl = pte_lockptr(mm, new_pmd);
	98	if (new_ptl != old_ptl)
	99	spin_lock(new_ptl);
7be7a546 HD	100
	101	for (; old_addr < old_end; old_pte++, old_addr += PAGE_SIZE,
	102	new_pte++, new_addr += PAGE_SIZE) {
	103	if (pte_none(*old_pte))
	104	continue;
	105	pte = ptep_clear_flush(vma, old_addr, old_pte);
	106	/* ZERO_PAGE can be dependant on virtual addr */
	107	pte = move_pte(pte, new_vma->vm_page_prot, old_addr, new_addr);
	108	set_pte_at(mm, new_addr, new_pte, pte);
1da177e4	109	}
7be7a546	110
4c21e2f2 HD	111	if (new_ptl != old_ptl)
4c21e2f2 HD	112	spin_unlock(new_ptl);
7be7a546	113	pte_unmap_nested(new_pte - 1);
c74df32c	114	pte_unmap_unlock(old_pte - 1, old_ptl);
1da177e4 LT	115	if (mapping)
1da177e4 LT	116	spin_unlock(&mapping->i_mmap_lock);
1da177e4 LT	117	}
1da177e4 LT	118
7be7a546 HD	119	#define LATENCY_LIMIT (64 * PAGE_SIZE)
7be7a546 HD	120
1da177e4 LT	121	static unsigned long move_page_tables(struct vm_area_struct *vma,
	122	unsigned long old_addr, struct vm_area_struct *new_vma,
	123	unsigned long new_addr, unsigned long len)
	124	{
7be7a546 HD	125	unsigned long extent, next, old_end;
7be7a546 HD	126	pmd_t old_pmd, new_pmd;
1da177e4	127
7be7a546 HD	128	old_end = old_addr + len;
7be7a546 HD	129	flush_cache_range(vma, old_addr, old_end);
1da177e4	130
7be7a546	131	for (; old_addr < old_end; old_addr += extent, new_addr += extent) {
1da177e4	132	cond_resched();
7be7a546 HD	133	next = (old_addr + PMD_SIZE) & PMD_MASK;
	134	if (next - 1 > old_end)
	135	next = old_end;
	136	extent = next - old_addr;
	137	old_pmd = get_old_pmd(vma->vm_mm, old_addr);
	138	if (!old_pmd)
	139	continue;
	140	new_pmd = alloc_new_pmd(vma->vm_mm, new_addr);
	141	if (!new_pmd)
	142	break;
	143	next = (new_addr + PMD_SIZE) & PMD_MASK;
	144	if (extent > next - new_addr)
	145	extent = next - new_addr;
	146	if (extent > LATENCY_LIMIT)
	147	extent = LATENCY_LIMIT;
	148	move_ptes(vma, old_pmd, old_addr, old_addr + extent,
	149	new_vma, new_pmd, new_addr);
1da177e4	150	}
7be7a546 HD	151
7be7a546 HD	152	return len + old_addr - old_end; /* how much done */
1da177e4 LT	153	}
	154
	155	static unsigned long move_vma(struct vm_area_struct *vma,
	156	unsigned long old_addr, unsigned long old_len,
	157	unsigned long new_len, unsigned long new_addr)
	158	{
	159	struct mm_struct *mm = vma->vm_mm;
	160	struct vm_area_struct *new_vma;
	161	unsigned long vm_flags = vma->vm_flags;
	162	unsigned long new_pgoff;
	163	unsigned long moved_len;
	164	unsigned long excess = 0;
365e9c87	165	unsigned long hiwater_vm;
1da177e4 LT	166	int split = 0;
	167
	168	/*
	169	* We'd prefer to avoid failure later on in do_munmap:
	170	* which may split one vma into three before unmapping.
	171	*/
	172	if (mm->map_count >= sysctl_max_map_count - 3)
	173	return -ENOMEM;
	174
	175	new_pgoff = vma->vm_pgoff + ((old_addr - vma->vm_start) >> PAGE_SHIFT);
	176	new_vma = copy_vma(&vma, new_addr, new_len, new_pgoff);
	177	if (!new_vma)
	178	return -ENOMEM;
	179
	180	moved_len = move_page_tables(vma, old_addr, new_vma, new_addr, old_len);
	181	if (moved_len < old_len) {
	182	/*
	183	* On error, move entries back from new area to old,
	184	* which will succeed since page tables still there,
	185	* and then proceed to unmap new area instead of old.
	186	*/
	187	move_page_tables(new_vma, new_addr, vma, old_addr, moved_len);
	188	vma = new_vma;
	189	old_len = new_len;
	190	old_addr = new_addr;
	191	new_addr = -ENOMEM;
	192	}
	193
	194	/* Conceal VM_ACCOUNT so old reservation is not undone */
	195	if (vm_flags & VM_ACCOUNT) {
	196	vma->vm_flags &= ~VM_ACCOUNT;
	197	excess = vma->vm_end - vma->vm_start - old_len;
	198	if (old_addr > vma->vm_start &&
	199	old_addr + old_len < vma->vm_end)
	200	split = 1;
	201	}
	202
71799062	203	/*
365e9c87 HD	204	* If we failed to move page tables we still do total_vm increment
	205	* since do_munmap() will decrement it by old_len == new_len.
	206	*
	207	* Since total_vm is about to be raised artificially high for a
	208	* moment, we need to restore high watermark afterwards: if stats
	209	* are taken meanwhile, total_vm and hiwater_vm appear too high.
	210	* If this were a serious issue, we'd add a flag to do_munmap().
71799062	211	*/
365e9c87	212	hiwater_vm = mm->hiwater_vm;
71799062	213	mm->total_vm += new_len >> PAGE_SHIFT;
ab50b8ed	214	vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
71799062	215
1da177e4 LT	216	if (do_munmap(mm, old_addr, old_len) < 0) {
	217	/* OOM: unable to split vma, just get accounts right */
	218	vm_unacct_memory(excess >> PAGE_SHIFT);
	219	excess = 0;
	220	}
365e9c87	221	mm->hiwater_vm = hiwater_vm;
1da177e4 LT	222
	223	/* Restore VM_ACCOUNT if one or two pieces of vma left */
	224	if (excess) {
	225	vma->vm_flags \|= VM_ACCOUNT;
	226	if (split)
	227	vma->vm_next->vm_flags \|= VM_ACCOUNT;
	228	}
	229
1da177e4 LT	230	if (vm_flags & VM_LOCKED) {
	231	mm->locked_vm += new_len >> PAGE_SHIFT;
	232	if (new_len > old_len)
	233	make_pages_present(new_addr + old_len,
	234	new_addr + new_len);
	235	}
	236
	237	return new_addr;
	238	}
	239
	240	/*
	241	* Expand (or shrink) an existing mapping, potentially moving it at the
	242	* same time (controlled by the MREMAP_MAYMOVE flag and available VM space)
	243	*
	244	* MREMAP_FIXED option added 5-Dec-1999 by Benjamin LaHaise
	245	* This option implies MREMAP_MAYMOVE.
	246	*/
	247	unsigned long do_mremap(unsigned long addr,
	248	unsigned long old_len, unsigned long new_len,
	249	unsigned long flags, unsigned long new_addr)
	250	{
d0de32d9	251	struct mm_struct *mm = current->mm;
1da177e4 LT	252	struct vm_area_struct *vma;
	253	unsigned long ret = -EINVAL;
	254	unsigned long charged = 0;
	255
	256	if (flags & ~(MREMAP_FIXED \| MREMAP_MAYMOVE))
	257	goto out;
	258
	259	if (addr & ~PAGE_MASK)
	260	goto out;
	261
	262	old_len = PAGE_ALIGN(old_len);
	263	new_len = PAGE_ALIGN(new_len);
	264
	265	/*
	266	* We allow a zero old-len as a special case
	267	* for DOS-emu "duplicate shm area" thing. But
	268	* a zero new-len is nonsensical.
	269	*/
	270	if (!new_len)
	271	goto out;
	272
	273	/* new_addr is only valid if MREMAP_FIXED is specified */
	274	if (flags & MREMAP_FIXED) {
	275	if (new_addr & ~PAGE_MASK)
	276	goto out;
	277	if (!(flags & MREMAP_MAYMOVE))
	278	goto out;
	279
	280	if (new_len > TASK_SIZE \|\| new_addr > TASK_SIZE - new_len)
	281	goto out;
	282
	283	/* Check if the location we're moving into overlaps the
	284	* old location at all, and fail if it does.
	285	*/
	286	if ((new_addr <= addr) && (new_addr+new_len) > addr)
	287	goto out;
	288
	289	if ((addr <= new_addr) && (addr+old_len) > new_addr)
	290	goto out;
	291
d0de32d9	292	ret = do_munmap(mm, new_addr, new_len);
1da177e4 LT	293	if (ret)
	294	goto out;
	295	}
	296
	297	/*
	298	* Always allow a shrinking remap: that just unmaps
	299	* the unnecessary pages..
	300	* do_munmap does all the needed commit accounting
	301	*/
	302	if (old_len >= new_len) {
d0de32d9	303	ret = do_munmap(mm, addr+new_len, old_len - new_len);
1da177e4 LT	304	if (ret && old_len != new_len)
	305	goto out;
	306	ret = addr;
	307	if (!(flags & MREMAP_FIXED) \|\| (new_addr == addr))
	308	goto out;
	309	old_len = new_len;
	310	}
	311
	312	/*
	313	* Ok, we need to grow.. or relocate.
	314	*/
	315	ret = -EFAULT;
d0de32d9	316	vma = find_vma(mm, addr);
1da177e4 LT	317	if (!vma \|\| vma->vm_start > addr)
	318	goto out;
	319	if (is_vm_hugetlb_page(vma)) {
	320	ret = -EINVAL;
	321	goto out;
	322	}
	323	/* We can't remap across vm area boundaries */
	324	if (old_len > vma->vm_end - addr)
	325	goto out;
	326	if (vma->vm_flags & VM_DONTEXPAND) {
	327	if (new_len > old_len)
	328	goto out;
	329	}
	330	if (vma->vm_flags & VM_LOCKED) {
	331	unsigned long locked, lock_limit;
d0de32d9	332	locked = mm->locked_vm << PAGE_SHIFT;
1da177e4 LT	333	lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
	334	locked += new_len - old_len;
	335	ret = -EAGAIN;
	336	if (locked > lock_limit && !capable(CAP_IPC_LOCK))
	337	goto out;
	338	}
d0de32d9	339	if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT)) {
119f657c	340	ret = -ENOMEM;
1da177e4	341	goto out;
119f657c	342	}
1da177e4 LT	343
	344	if (vma->vm_flags & VM_ACCOUNT) {
	345	charged = (new_len - old_len) >> PAGE_SHIFT;
	346	if (security_vm_enough_memory(charged))
	347	goto out_nc;
	348	}
	349
	350	/* old_len exactly to the end of the area..
	351	* And we're not relocating the area.
	352	*/
	353	if (old_len == vma->vm_end - addr &&
	354	!((flags & MREMAP_FIXED) && (addr != new_addr)) &&
	355	(old_len != new_len \|\| !(flags & MREMAP_MAYMOVE))) {
	356	unsigned long max_addr = TASK_SIZE;
	357	if (vma->vm_next)
	358	max_addr = vma->vm_next->vm_start;
	359	/* can we just expand the current mapping? */
	360	if (max_addr - addr >= new_len) {
	361	int pages = (new_len - old_len) >> PAGE_SHIFT;
	362
	363	vma_adjust(vma, vma->vm_start,
	364	addr + new_len, vma->vm_pgoff, NULL);
	365
d0de32d9 HD	366	mm->total_vm += pages;
d0de32d9 HD	367	vm_stat_account(mm, vma->vm_flags, vma->vm_file, pages);
1da177e4	368	if (vma->vm_flags & VM_LOCKED) {
d0de32d9	369	mm->locked_vm += pages;
1da177e4 LT	370	make_pages_present(addr + old_len,
	371	addr + new_len);
	372	}
	373	ret = addr;
	374	goto out;
	375	}
	376	}
	377
	378	/*
	379	* We weren't able to just expand or shrink the area,
	380	* we need to create a new one and move it..
	381	*/
	382	ret = -ENOMEM;
	383	if (flags & MREMAP_MAYMOVE) {
	384	if (!(flags & MREMAP_FIXED)) {
	385	unsigned long map_flags = 0;
	386	if (vma->vm_flags & VM_MAYSHARE)
	387	map_flags \|= MAP_SHARED;
	388
	389	new_addr = get_unmapped_area(vma->vm_file, 0, new_len,
	390	vma->vm_pgoff, map_flags);
	391	ret = new_addr;
	392	if (new_addr & ~PAGE_MASK)
	393	goto out;
	394	}
	395	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	396	}
	397	out:
	398	if (ret & ~PAGE_MASK)
	399	vm_unacct_memory(charged);
	400	out_nc:
	401	return ret;
	402	}
	403
	404	asmlinkage unsigned long sys_mremap(unsigned long addr,
	405	unsigned long old_len, unsigned long new_len,
	406	unsigned long flags, unsigned long new_addr)
	407	{
	408	unsigned long ret;
	409
	410	down_write(&current->mm->mmap_sem);
	411	ret = do_mremap(addr, old_len, new_len, flags, new_addr);
	412	up_write(&current->mm->mmap_sem);
	413	return ret;
	414	}