[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;

	/*
	 * We don't need page_table_lock: we have mmap_sem exclusively.
	 */
	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 = NULL;
	pte_t *pte;

	/*
	 * We do need page_table_lock: because allocators expect that.
	 */
	spin_lock(&mm->page_table_lock);
	pgd = pgd_offset(mm, addr);
	pud = pud_alloc(mm, pgd, addr);
	if (!pud)
		goto out;

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

	pte = pte_alloc_map(mm, pmd, addr);
	if (!pte) {
		pmd = NULL;
		goto out;
	}
	pte_unmap(pte);
out:
	spin_unlock(&mm->page_table_lock);
	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;

	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;
	}

	spin_lock(&mm->page_table_lock);
	old_pte = pte_offset_map(old_pmd, old_addr);
	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(old_pte - 1);
	spin_unlock(&mm->page_table_lock);
	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;
	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
	 */
	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;
	}

	/* 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 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(current->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(current->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(current->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 = current->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(current->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);

			current->mm->total_vm += pages;
			vm_stat_account(vma->vm_mm, vma->vm_flags,
							vma->vm_file, pages);
			if (vma->vm_flags & VM_LOCKED) {
				current->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
7be7a546 HD	31	/*
	32	* We don't need page_table_lock: we have mmap_sem exclusively.
	33	*/
1da177e4 LT	34	pgd = pgd_offset(mm, addr);
	35	if (pgd_none_or_clear_bad(pgd))
	36	return NULL;
	37
	38	pud = pud_offset(pgd, addr);
	39	if (pud_none_or_clear_bad(pud))
	40	return NULL;
	41
	42	pmd = pmd_offset(pud, addr);
	43	if (pmd_none_or_clear_bad(pmd))
	44	return NULL;
	45
7be7a546	46	return pmd;
1da177e4 LT	47	}
1da177e4 LT	48
7be7a546	49	static pmd_t alloc_new_pmd(struct mm_struct mm, unsigned long addr)
1da177e4 LT	50	{
	51	pgd_t *pgd;
	52	pud_t *pud;
7be7a546 HD	53	pmd_t *pmd = NULL;
7be7a546 HD	54	pte_t *pte;
1da177e4	55
7be7a546 HD	56	/*
	57	* We do need page_table_lock: because allocators expect that.
	58	*/
	59	spin_lock(&mm->page_table_lock);
1da177e4	60	pgd = pgd_offset(mm, addr);
1da177e4 LT	61	pud = pud_alloc(mm, pgd, addr);
1da177e4 LT	62	if (!pud)
7be7a546 HD	63	goto out;
7be7a546 HD	64
1da177e4	65	pmd = pmd_alloc(mm, pud, addr);
7be7a546 HD	66	if (!pmd)
	67	goto out;
	68
	69	pte = pte_alloc_map(mm, pmd, addr);
	70	if (!pte) {
	71	pmd = NULL;
	72	goto out;
	73	}
	74	pte_unmap(pte);
	75	out:
	76	spin_unlock(&mm->page_table_lock);
	77	return pmd;
1da177e4 LT	78	}
1da177e4 LT	79
7be7a546 HD	80	static void move_ptes(struct vm_area_struct vma, pmd_t old_pmd,
	81	unsigned long old_addr, unsigned long old_end,
	82	struct vm_area_struct new_vma, pmd_t new_pmd,
	83	unsigned long new_addr)
1da177e4 LT	84	{
	85	struct address_space *mapping = NULL;
	86	struct mm_struct *mm = vma->vm_mm;
7be7a546	87	pte_t old_pte, new_pte, pte;
1da177e4 LT	88
	89	if (vma->vm_file) {
	90	/*
	91	* Subtle point from Rajesh Venkatasubramanian: before
	92	* moving file-based ptes, we must lock vmtruncate out,
	93	* since it might clean the dst vma before the src vma,
	94	* and we propagate stale pages into the dst afterward.
	95	*/
	96	mapping = vma->vm_file->f_mapping;
	97	spin_lock(&mapping->i_mmap_lock);
	98	if (new_vma->vm_truncate_count &&
	99	new_vma->vm_truncate_count != vma->vm_truncate_count)
	100	new_vma->vm_truncate_count = 0;
	101	}
1da177e4	102
7be7a546 HD	103	spin_lock(&mm->page_table_lock);
	104	old_pte = pte_offset_map(old_pmd, old_addr);
	105	new_pte = pte_offset_map_nested(new_pmd, new_addr);
	106
	107	for (; old_addr < old_end; old_pte++, old_addr += PAGE_SIZE,
	108	new_pte++, new_addr += PAGE_SIZE) {
	109	if (pte_none(*old_pte))
	110	continue;
	111	pte = ptep_clear_flush(vma, old_addr, old_pte);
	112	/* ZERO_PAGE can be dependant on virtual addr */
	113	pte = move_pte(pte, new_vma->vm_page_prot, old_addr, new_addr);
	114	set_pte_at(mm, new_addr, new_pte, pte);
1da177e4	115	}
7be7a546 HD	116
	117	pte_unmap_nested(new_pte - 1);
	118	pte_unmap(old_pte - 1);
1da177e4 LT	119	spin_unlock(&mm->page_table_lock);
	120	if (mapping)
	121	spin_unlock(&mapping->i_mmap_lock);
1da177e4 LT	122	}
1da177e4 LT	123
7be7a546 HD	124	#define LATENCY_LIMIT (64 * PAGE_SIZE)
7be7a546 HD	125
1da177e4 LT	126	static unsigned long move_page_tables(struct vm_area_struct *vma,
	127	unsigned long old_addr, struct vm_area_struct *new_vma,
	128	unsigned long new_addr, unsigned long len)
	129	{
7be7a546 HD	130	unsigned long extent, next, old_end;
7be7a546 HD	131	pmd_t old_pmd, new_pmd;
1da177e4	132
7be7a546 HD	133	old_end = old_addr + len;
7be7a546 HD	134	flush_cache_range(vma, old_addr, old_end);
1da177e4	135
7be7a546	136	for (; old_addr < old_end; old_addr += extent, new_addr += extent) {
1da177e4	137	cond_resched();
7be7a546 HD	138	next = (old_addr + PMD_SIZE) & PMD_MASK;
	139	if (next - 1 > old_end)
	140	next = old_end;
	141	extent = next - old_addr;
	142	old_pmd = get_old_pmd(vma->vm_mm, old_addr);
	143	if (!old_pmd)
	144	continue;
	145	new_pmd = alloc_new_pmd(vma->vm_mm, new_addr);
	146	if (!new_pmd)
	147	break;
	148	next = (new_addr + PMD_SIZE) & PMD_MASK;
	149	if (extent > next - new_addr)
	150	extent = next - new_addr;
	151	if (extent > LATENCY_LIMIT)
	152	extent = LATENCY_LIMIT;
	153	move_ptes(vma, old_pmd, old_addr, old_addr + extent,
	154	new_vma, new_pmd, new_addr);
1da177e4	155	}
7be7a546 HD	156
7be7a546 HD	157	return len + old_addr - old_end; /* how much done */
1da177e4 LT	158	}
	159
	160	static unsigned long move_vma(struct vm_area_struct *vma,
	161	unsigned long old_addr, unsigned long old_len,
	162	unsigned long new_len, unsigned long new_addr)
	163	{
	164	struct mm_struct *mm = vma->vm_mm;
	165	struct vm_area_struct *new_vma;
	166	unsigned long vm_flags = vma->vm_flags;
	167	unsigned long new_pgoff;
	168	unsigned long moved_len;
	169	unsigned long excess = 0;
	170	int split = 0;
	171
	172	/*
	173	* We'd prefer to avoid failure later on in do_munmap:
	174	* which may split one vma into three before unmapping.
	175	*/
	176	if (mm->map_count >= sysctl_max_map_count - 3)
	177	return -ENOMEM;
	178
	179	new_pgoff = vma->vm_pgoff + ((old_addr - vma->vm_start) >> PAGE_SHIFT);
	180	new_vma = copy_vma(&vma, new_addr, new_len, new_pgoff);
	181	if (!new_vma)
	182	return -ENOMEM;
	183
	184	moved_len = move_page_tables(vma, old_addr, new_vma, new_addr, old_len);
	185	if (moved_len < old_len) {
	186	/*
	187	* On error, move entries back from new area to old,
	188	* which will succeed since page tables still there,
	189	* and then proceed to unmap new area instead of old.
	190	*/
	191	move_page_tables(new_vma, new_addr, vma, old_addr, moved_len);
	192	vma = new_vma;
	193	old_len = new_len;
	194	old_addr = new_addr;
	195	new_addr = -ENOMEM;
	196	}
	197
	198	/* Conceal VM_ACCOUNT so old reservation is not undone */
	199	if (vm_flags & VM_ACCOUNT) {
	200	vma->vm_flags &= ~VM_ACCOUNT;
	201	excess = vma->vm_end - vma->vm_start - old_len;
	202	if (old_addr > vma->vm_start &&
	203	old_addr + old_len < vma->vm_end)
	204	split = 1;
	205	}
	206
71799062 KK	207	/*
	208	* if we failed to move page tables we still do total_vm increment
	209	* since do_munmap() will decrement it by old_len == new_len
	210	*/
	211	mm->total_vm += new_len >> PAGE_SHIFT;
ab50b8ed	212	vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
71799062	213
1da177e4 LT	214	if (do_munmap(mm, old_addr, old_len) < 0) {
	215	/* OOM: unable to split vma, just get accounts right */
	216	vm_unacct_memory(excess >> PAGE_SHIFT);
	217	excess = 0;
	218	}
	219
	220	/* Restore VM_ACCOUNT if one or two pieces of vma left */
	221	if (excess) {
	222	vma->vm_flags \|= VM_ACCOUNT;
	223	if (split)
	224	vma->vm_next->vm_flags \|= VM_ACCOUNT;
	225	}
	226
1da177e4 LT	227	if (vm_flags & VM_LOCKED) {
	228	mm->locked_vm += new_len >> PAGE_SHIFT;
	229	if (new_len > old_len)
	230	make_pages_present(new_addr + old_len,
	231	new_addr + new_len);
	232	}
	233
	234	return new_addr;
	235	}
	236
	237	/*
	238	* Expand (or shrink) an existing mapping, potentially moving it at the
	239	* same time (controlled by the MREMAP_MAYMOVE flag and available VM space)
	240	*
	241	* MREMAP_FIXED option added 5-Dec-1999 by Benjamin LaHaise
	242	* This option implies MREMAP_MAYMOVE.
	243	*/
	244	unsigned long do_mremap(unsigned long addr,
	245	unsigned long old_len, unsigned long new_len,
	246	unsigned long flags, unsigned long new_addr)
	247	{
	248	struct vm_area_struct *vma;
	249	unsigned long ret = -EINVAL;
	250	unsigned long charged = 0;
	251
	252	if (flags & ~(MREMAP_FIXED \| MREMAP_MAYMOVE))
	253	goto out;
	254
	255	if (addr & ~PAGE_MASK)
	256	goto out;
	257
	258	old_len = PAGE_ALIGN(old_len);
	259	new_len = PAGE_ALIGN(new_len);
	260
	261	/*
	262	* We allow a zero old-len as a special case
	263	* for DOS-emu "duplicate shm area" thing. But
	264	* a zero new-len is nonsensical.
	265	*/
	266	if (!new_len)
	267	goto out;
	268
	269	/* new_addr is only valid if MREMAP_FIXED is specified */
	270	if (flags & MREMAP_FIXED) {
	271	if (new_addr & ~PAGE_MASK)
	272	goto out;
	273	if (!(flags & MREMAP_MAYMOVE))
	274	goto out;
	275
	276	if (new_len > TASK_SIZE \|\| new_addr > TASK_SIZE - new_len)
	277	goto out;
	278
	279	/* Check if the location we're moving into overlaps the
	280	* old location at all, and fail if it does.
	281	*/
	282	if ((new_addr <= addr) && (new_addr+new_len) > addr)
	283	goto out;
	284
	285	if ((addr <= new_addr) && (addr+old_len) > new_addr)
	286	goto out;
	287
	288	ret = do_munmap(current->mm, new_addr, new_len);
	289	if (ret)
	290	goto out;
291	}
292
293	/*
294	* Always allow a shrinking remap: that just unmaps
295	* the unnecessary pages..
296	* do_munmap does all the needed commit accounting
297	*/
298	if (old_len >= new_len) {
299	ret = do_munmap(current->mm, addr+new_len, old_len - new_len);
300	if (ret && old_len != new_len)
301	goto out;
302	ret = addr;
303	if (!(flags & MREMAP_FIXED) \|\| (new_addr == addr))
304	goto out;
305	old_len = new_len;
306	}
307
308	/*
309	* Ok, we need to grow.. or relocate.
310	*/
311	ret = -EFAULT;
312	vma = find_vma(current->mm, addr);
313	if (!vma \|\| vma->vm_start > addr)
314	goto out;
315	if (is_vm_hugetlb_page(vma)) {
316	ret = -EINVAL;
317	goto out;
318	}
319	/* We can't remap across vm area boundaries */
320	if (old_len > vma->vm_end - addr)
321	goto out;
322	if (vma->vm_flags & VM_DONTEXPAND) {
323	if (new_len > old_len)
324	goto out;
325	}
326	if (vma->vm_flags & VM_LOCKED) {
327	unsigned long locked, lock_limit;
328	locked = current->mm->locked_vm << PAGE_SHIFT;
329	lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
330	locked += new_len - old_len;
331	ret = -EAGAIN;
332	if (locked > lock_limit && !capable(CAP_IPC_LOCK))
333	goto out;
334	}
119f657c AM	335	if (!may_expand_vm(current->mm, (new_len - old_len) >> PAGE_SHIFT)) {
119f657c AM	336	ret = -ENOMEM;
1da177e4	337	goto out;
119f657c	338	}
1da177e4 LT	339
	340	if (vma->vm_flags & VM_ACCOUNT) {
	341	charged = (new_len - old_len) >> PAGE_SHIFT;
	342	if (security_vm_enough_memory(charged))
	343	goto out_nc;
	344	}
	345
	346	/* old_len exactly to the end of the area..
	347	* And we're not relocating the area.
	348	*/
	349	if (old_len == vma->vm_end - addr &&
	350	!((flags & MREMAP_FIXED) && (addr != new_addr)) &&
	351	(old_len != new_len \|\| !(flags & MREMAP_MAYMOVE))) {
	352	unsigned long max_addr = TASK_SIZE;
	353	if (vma->vm_next)
	354	max_addr = vma->vm_next->vm_start;
	355	/* can we just expand the current mapping? */
	356	if (max_addr - addr >= new_len) {
	357	int pages = (new_len - old_len) >> PAGE_SHIFT;
	358
	359	vma_adjust(vma, vma->vm_start,
	360	addr + new_len, vma->vm_pgoff, NULL);
	361
	362	current->mm->total_vm += pages;
ab50b8ed	363	vm_stat_account(vma->vm_mm, vma->vm_flags,
1da177e4 LT	364	vma->vm_file, pages);
	365	if (vma->vm_flags & VM_LOCKED) {
	366	current->mm->locked_vm += pages;
	367	make_pages_present(addr + old_len,
	368	addr + new_len);
	369	}
	370	ret = addr;
	371	goto out;
	372	}
	373	}
	374
	375	/*
	376	* We weren't able to just expand or shrink the area,
	377	* we need to create a new one and move it..
	378	*/
	379	ret = -ENOMEM;
	380	if (flags & MREMAP_MAYMOVE) {
	381	if (!(flags & MREMAP_FIXED)) {
	382	unsigned long map_flags = 0;
	383	if (vma->vm_flags & VM_MAYSHARE)
	384	map_flags \|= MAP_SHARED;
	385
	386	new_addr = get_unmapped_area(vma->vm_file, 0, new_len,
	387	vma->vm_pgoff, map_flags);
	388	ret = new_addr;
	389	if (new_addr & ~PAGE_MASK)
	390	goto out;
	391	}
	392	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	393	}
	394	out:
	395	if (ret & ~PAGE_MASK)
	396	vm_unacct_memory(charged);
	397	out_nc:
	398	return ret;
	399	}
	400
	401	asmlinkage unsigned long sys_mremap(unsigned long addr,
	402	unsigned long old_len, unsigned long new_len,
	403	unsigned long flags, unsigned long new_addr)
	404	{
	405	unsigned long ret;
	406
	407	down_write(&current->mm->mmap_sem);
	408	ret = do_mremap(addr, old_len, new_len, flags, new_addr);
	409	up_write(&current->mm->mmap_sem);
	410	return ret;
	411	}