[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/ksm.h>
#include <linux/mman.h>
#include <linux/swap.h>
#include <linux/capability.h>
#include <linux/fs.h>
#include <linux/highmem.h>
#include <linux/security.h>
#include <linux/syscalls.h>
#include <linux/mmu_notifier.h>

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

#include "internal.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;
	unsigned long old_start;

	old_start = old_addr;
	mmu_notifier_invalidate_range_start(vma->vm_mm,
					    old_start, old_end);
	if (vma->vm_file) {
		/*
		 * Subtle point from Rajesh Venkatasubramanian: before
		 * moving file-based ptes, we must lock truncate_pagecache
		 * 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_nested(new_ptl, SINGLE_DEPTH_NESTING);
	arch_enter_lazy_mmu_mode();

	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);
		pte = move_pte(pte, new_vma->vm_page_prot, old_addr, new_addr);
		set_pte_at(mm, new_addr, new_pte, pte);
	}

	arch_leave_lazy_mmu_mode();
	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);
	mmu_notifier_invalidate_range_end(vma->vm_mm, old_start, old_end);
}

#define LATENCY_LIMIT	(64 * PAGE_SIZE)

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;
	int err;

	/*
	 * 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;

	/*
	 * Advise KSM to break any KSM pages in the area to be moved:
	 * it would be confusing if they were to turn up at the new
	 * location, where they happen to coincide with different KSM
	 * pages recently unmapped.  But leave vma->vm_flags as it was,
	 * so KSM can come around to merge on vma and new_vma afterwards.
	 */
	err = ksm_madvise(vma, old_addr, old_addr + old_len,
						MADV_UNMERGEABLE, &vm_flags);
	if (err)
		return err;

	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)
			mlock_vma_pages_range(new_vma, new_addr + old_len,
						       new_addr + new_len);
	}

	return new_addr;
}

static struct vm_area_struct *vma_to_resize(unsigned long addr,
	unsigned long old_len, unsigned long new_len, unsigned long *p)
{
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma = find_vma(mm, addr);

	if (!vma || vma->vm_start > addr)
		goto Efault;

	if (is_vm_hugetlb_page(vma))
		goto Einval;

	/* We can't remap across vm area boundaries */
	if (old_len > vma->vm_end - addr)
		goto Efault;

	if (vma->vm_flags & (VM_DONTEXPAND | VM_PFNMAP)) {
		if (new_len > old_len)
			goto Efault;
	}

	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;
		if (locked > lock_limit && !capable(CAP_IPC_LOCK))
			goto Eagain;
	}

	if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT))
		goto Enomem;

	if (vma->vm_flags & VM_ACCOUNT) {
		unsigned long charged = (new_len - old_len) >> PAGE_SHIFT;
		if (security_vm_enough_memory(charged))
			goto Efault;
		*p = charged;
	}

	return vma;

Efault:	/* very odd choice for most of the cases, but... */
	return ERR_PTR(-EFAULT);
Einval:
	return ERR_PTR(-EINVAL);
Enomem:
	return ERR_PTR(-ENOMEM);
Eagain:
	return ERR_PTR(-EAGAIN);
}

static unsigned long mremap_to(unsigned long addr,
	unsigned long old_len, unsigned long new_addr,
	unsigned long new_len)
{
	struct mm_struct *mm = current->mm;
	struct vm_area_struct *vma;
	unsigned long ret = -EINVAL;
	unsigned long charged = 0;

	if (new_addr & ~PAGE_MASK)
		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 = security_file_mmap(NULL, 0, 0, 0, new_addr, 1);
	if (ret)
		goto out;

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

	if (old_len >= new_len) {
		ret = do_munmap(mm, addr+new_len, old_len - new_len);
		if (ret && old_len != new_len)
			goto out;
		old_len = new_len;
	}

	vma = vma_to_resize(addr, old_len, new_len, &charged);
	if (IS_ERR(vma)) {
		ret = PTR_ERR(vma);
		goto out;
	}

	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	if (ret & ~PAGE_MASK)
		vm_unacct_memory(charged);

out:
	return ret;
}

/*
 * 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;

	if (flags & MREMAP_FIXED) {
		if (flags & MREMAP_MAYMOVE)
			ret = mremap_to(addr, old_len, new_addr, new_len);
		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;
		goto out;
	}

	/*
	 * Ok, we need to grow..
	 */
	vma = vma_to_resize(addr, old_len, new_len, &charged);
	if (IS_ERR(vma)) {
		ret = PTR_ERR(vma);
		goto out;
	}

	/* old_len exactly to the end of the area..
	 */
	if (old_len == vma->vm_end - addr) {
		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;
				mlock_vma_pages_range(vma, 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) {
		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);
		if (new_addr & ~PAGE_MASK) {
			ret = new_addr;
			goto out;
		}

		ret = security_file_mmap(NULL, 0, 0, 0, new_addr, 1);
		if (ret)
			goto out;
		ret = move_vma(vma, addr, old_len, new_len, new_addr);
	}
out:
	if (ret & ~PAGE_MASK)
		vm_unacct_memory(charged);
	return ret;
}

SYSCALL_DEFINE5(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	*
046c6884	6	* Address space accounting code <[email protected]>
1da177e4 LT	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>
1ff82995	14	#include <linux/ksm.h>
1da177e4 LT	15	#include <linux/mman.h>
1da177e4 LT	16	#include <linux/swap.h>
c59ede7b	17	#include <linux/capability.h>
1da177e4 LT	18	#include <linux/fs.h>
	19	#include <linux/highmem.h>
	20	#include <linux/security.h>
	21	#include <linux/syscalls.h>
cddb8a5c	22	#include <linux/mmu_notifier.h>
1da177e4 LT	23
	24	#include <asm/uaccess.h>
	25	#include <asm/cacheflush.h>
	26	#include <asm/tlbflush.h>
	27
ba470de4 RR	28	#include "internal.h"
ba470de4 RR	29
7be7a546	30	static pmd_t get_old_pmd(struct mm_struct mm, unsigned long addr)
1da177e4 LT	31	{
	32	pgd_t *pgd;
	33	pud_t *pud;
	34	pmd_t *pmd;
	35
	36	pgd = pgd_offset(mm, addr);
	37	if (pgd_none_or_clear_bad(pgd))
	38	return NULL;
	39
	40	pud = pud_offset(pgd, addr);
	41	if (pud_none_or_clear_bad(pud))
	42	return NULL;
	43
	44	pmd = pmd_offset(pud, addr);
	45	if (pmd_none_or_clear_bad(pmd))
	46	return NULL;
	47
7be7a546	48	return pmd;
1da177e4 LT	49	}
1da177e4 LT	50
7be7a546	51	static pmd_t alloc_new_pmd(struct mm_struct mm, unsigned long addr)
1da177e4 LT	52	{
	53	pgd_t *pgd;
	54	pud_t *pud;
c74df32c	55	pmd_t *pmd;
1da177e4 LT	56
1da177e4 LT	57	pgd = pgd_offset(mm, addr);
1da177e4 LT	58	pud = pud_alloc(mm, pgd, addr);
1da177e4 LT	59	if (!pud)
c74df32c	60	return NULL;
7be7a546	61
1da177e4	62	pmd = pmd_alloc(mm, pud, addr);
7be7a546	63	if (!pmd)
c74df32c	64	return NULL;
7be7a546	65
1bb3630e	66	if (!pmd_present(*pmd) && __pte_alloc(mm, pmd, addr))
c74df32c HD	67	return NULL;
c74df32c HD	68
7be7a546	69	return pmd;
1da177e4 LT	70	}
1da177e4 LT	71
7be7a546 HD	72	static void move_ptes(struct vm_area_struct vma, pmd_t old_pmd,
	73	unsigned long old_addr, unsigned long old_end,
	74	struct vm_area_struct new_vma, pmd_t new_pmd,
	75	unsigned long new_addr)
1da177e4 LT	76	{
	77	struct address_space *mapping = NULL;
	78	struct mm_struct *mm = vma->vm_mm;
7be7a546	79	pte_t old_pte, new_pte, pte;
4c21e2f2	80	spinlock_t old_ptl, new_ptl;
cddb8a5c	81	unsigned long old_start;
1da177e4	82
cddb8a5c AA	83	old_start = old_addr;
	84	mmu_notifier_invalidate_range_start(vma->vm_mm,
	85	old_start, old_end);
1da177e4 LT	86	if (vma->vm_file) {
	87	/*
	88	* Subtle point from Rajesh Venkatasubramanian: before
25d9e2d1 NP	89	* moving file-based ptes, we must lock truncate_pagecache
25d9e2d1 NP	90	* out, since it might clean the dst vma before the src vma,
1da177e4 LT	91	* and we propagate stale pages into the dst afterward.
	92	*/
	93	mapping = vma->vm_file->f_mapping;
	94	spin_lock(&mapping->i_mmap_lock);
	95	if (new_vma->vm_truncate_count &&
	96	new_vma->vm_truncate_count != vma->vm_truncate_count)
	97	new_vma->vm_truncate_count = 0;
	98	}
1da177e4	99
4c21e2f2 HD	100	/*
	101	* We don't have to worry about the ordering of src and dst
	102	* pte locks because exclusive mmap_sem prevents deadlock.
	103	*/
c74df32c HD	104	old_pte = pte_offset_map_lock(mm, old_pmd, old_addr, &old_ptl);
c74df32c HD	105	new_pte = pte_offset_map_nested(new_pmd, new_addr);
4c21e2f2 HD	106	new_ptl = pte_lockptr(mm, new_pmd);
4c21e2f2 HD	107	if (new_ptl != old_ptl)
f20dc5f7	108	spin_lock_nested(new_ptl, SINGLE_DEPTH_NESTING);
6606c3e0	109	arch_enter_lazy_mmu_mode();
7be7a546 HD	110
	111	for (; old_addr < old_end; old_pte++, old_addr += PAGE_SIZE,
	112	new_pte++, new_addr += PAGE_SIZE) {
	113	if (pte_none(*old_pte))
	114	continue;
	115	pte = ptep_clear_flush(vma, old_addr, old_pte);
7be7a546 HD	116	pte = move_pte(pte, new_vma->vm_page_prot, old_addr, new_addr);
7be7a546 HD	117	set_pte_at(mm, new_addr, new_pte, pte);
1da177e4	118	}
7be7a546	119
6606c3e0	120	arch_leave_lazy_mmu_mode();
4c21e2f2 HD	121	if (new_ptl != old_ptl)
4c21e2f2 HD	122	spin_unlock(new_ptl);
7be7a546	123	pte_unmap_nested(new_pte - 1);
c74df32c	124	pte_unmap_unlock(old_pte - 1, old_ptl);
1da177e4 LT	125	if (mapping)
1da177e4 LT	126	spin_unlock(&mapping->i_mmap_lock);
cddb8a5c	127	mmu_notifier_invalidate_range_end(vma->vm_mm, old_start, old_end);
1da177e4 LT	128	}
1da177e4 LT	129
7be7a546 HD	130	#define LATENCY_LIMIT (64 * PAGE_SIZE)
7be7a546 HD	131
b6a2fea3	132	unsigned long move_page_tables(struct vm_area_struct *vma,
1da177e4 LT	133	unsigned long old_addr, struct vm_area_struct *new_vma,
	134	unsigned long new_addr, unsigned long len)
	135	{
7be7a546 HD	136	unsigned long extent, next, old_end;
7be7a546 HD	137	pmd_t old_pmd, new_pmd;
1da177e4	138
7be7a546 HD	139	old_end = old_addr + len;
7be7a546 HD	140	flush_cache_range(vma, old_addr, old_end);
1da177e4	141
7be7a546	142	for (; old_addr < old_end; old_addr += extent, new_addr += extent) {
1da177e4	143	cond_resched();
7be7a546 HD	144	next = (old_addr + PMD_SIZE) & PMD_MASK;
	145	if (next - 1 > old_end)
	146	next = old_end;
	147	extent = next - old_addr;
	148	old_pmd = get_old_pmd(vma->vm_mm, old_addr);
	149	if (!old_pmd)
	150	continue;
	151	new_pmd = alloc_new_pmd(vma->vm_mm, new_addr);
	152	if (!new_pmd)
	153	break;
	154	next = (new_addr + PMD_SIZE) & PMD_MASK;
	155	if (extent > next - new_addr)
	156	extent = next - new_addr;
	157	if (extent > LATENCY_LIMIT)
	158	extent = LATENCY_LIMIT;
	159	move_ptes(vma, old_pmd, old_addr, old_addr + extent,
	160	new_vma, new_pmd, new_addr);
1da177e4	161	}
7be7a546 HD	162
7be7a546 HD	163	return len + old_addr - old_end; /* how much done */
1da177e4 LT	164	}
	165
	166	static unsigned long move_vma(struct vm_area_struct *vma,
	167	unsigned long old_addr, unsigned long old_len,
	168	unsigned long new_len, unsigned long new_addr)
	169	{
	170	struct mm_struct *mm = vma->vm_mm;
	171	struct vm_area_struct *new_vma;
	172	unsigned long vm_flags = vma->vm_flags;
	173	unsigned long new_pgoff;
	174	unsigned long moved_len;
	175	unsigned long excess = 0;
365e9c87	176	unsigned long hiwater_vm;
1da177e4	177	int split = 0;
7103ad32	178	int err;
1da177e4 LT	179
	180	/*
	181	* We'd prefer to avoid failure later on in do_munmap:
	182	* which may split one vma into three before unmapping.
	183	*/
	184	if (mm->map_count >= sysctl_max_map_count - 3)
	185	return -ENOMEM;
	186
1ff82995 HD	187	/*
	188	* Advise KSM to break any KSM pages in the area to be moved:
	189	* it would be confusing if they were to turn up at the new
	190	* location, where they happen to coincide with different KSM
	191	* pages recently unmapped. But leave vma->vm_flags as it was,
	192	* so KSM can come around to merge on vma and new_vma afterwards.
	193	*/
7103ad32 HD	194	err = ksm_madvise(vma, old_addr, old_addr + old_len,
	195	MADV_UNMERGEABLE, &vm_flags);
	196	if (err)
	197	return err;
1ff82995	198
1da177e4 LT	199	new_pgoff = vma->vm_pgoff + ((old_addr - vma->vm_start) >> PAGE_SHIFT);
	200	new_vma = copy_vma(&vma, new_addr, new_len, new_pgoff);
	201	if (!new_vma)
	202	return -ENOMEM;
	203
	204	moved_len = move_page_tables(vma, old_addr, new_vma, new_addr, old_len);
	205	if (moved_len < old_len) {
	206	/*
	207	* On error, move entries back from new area to old,
	208	* which will succeed since page tables still there,
	209	* and then proceed to unmap new area instead of old.
	210	*/
	211	move_page_tables(new_vma, new_addr, vma, old_addr, moved_len);
	212	vma = new_vma;
	213	old_len = new_len;
	214	old_addr = new_addr;
	215	new_addr = -ENOMEM;
	216	}
	217
	218	/* Conceal VM_ACCOUNT so old reservation is not undone */
	219	if (vm_flags & VM_ACCOUNT) {
	220	vma->vm_flags &= ~VM_ACCOUNT;
	221	excess = vma->vm_end - vma->vm_start - old_len;
	222	if (old_addr > vma->vm_start &&
	223	old_addr + old_len < vma->vm_end)
	224	split = 1;
	225	}
	226
71799062	227	/*
365e9c87 HD	228	* If we failed to move page tables we still do total_vm increment
	229	* since do_munmap() will decrement it by old_len == new_len.
	230	*
	231	* Since total_vm is about to be raised artificially high for a
	232	* moment, we need to restore high watermark afterwards: if stats
	233	* are taken meanwhile, total_vm and hiwater_vm appear too high.
	234	* If this were a serious issue, we'd add a flag to do_munmap().
71799062	235	*/
365e9c87	236	hiwater_vm = mm->hiwater_vm;
71799062	237	mm->total_vm += new_len >> PAGE_SHIFT;
ab50b8ed	238	vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
71799062	239
1da177e4 LT	240	if (do_munmap(mm, old_addr, old_len) < 0) {
	241	/* OOM: unable to split vma, just get accounts right */
	242	vm_unacct_memory(excess >> PAGE_SHIFT);
	243	excess = 0;
	244	}
365e9c87	245	mm->hiwater_vm = hiwater_vm;
1da177e4 LT	246
	247	/* Restore VM_ACCOUNT if one or two pieces of vma left */
	248	if (excess) {
	249	vma->vm_flags \|= VM_ACCOUNT;
	250	if (split)
	251	vma->vm_next->vm_flags \|= VM_ACCOUNT;
	252	}
	253
1da177e4 LT	254	if (vm_flags & VM_LOCKED) {
	255	mm->locked_vm += new_len >> PAGE_SHIFT;
	256	if (new_len > old_len)
ba470de4 RR	257	mlock_vma_pages_range(new_vma, new_addr + old_len,
ba470de4 RR	258	new_addr + new_len);
1da177e4 LT	259	}
	260
	261	return new_addr;
	262	}
	263
54f5de70 AV	264	static struct vm_area_struct *vma_to_resize(unsigned long addr,
	265	unsigned long old_len, unsigned long new_len, unsigned long *p)
	266	{
	267	struct mm_struct *mm = current->mm;
	268	struct vm_area_struct *vma = find_vma(mm, addr);
	269
	270	if (!vma \|\| vma->vm_start > addr)
	271	goto Efault;
	272
	273	if (is_vm_hugetlb_page(vma))
	274	goto Einval;
	275
	276	/* We can't remap across vm area boundaries */
	277	if (old_len > vma->vm_end - addr)
	278	goto Efault;
	279
	280	if (vma->vm_flags & (VM_DONTEXPAND \| VM_PFNMAP)) {
	281	if (new_len > old_len)
	282	goto Efault;
	283	}
	284
	285	if (vma->vm_flags & VM_LOCKED) {
	286	unsigned long locked, lock_limit;
	287	locked = mm->locked_vm << PAGE_SHIFT;
	288	lock_limit = current->signal->rlim[RLIMIT_MEMLOCK].rlim_cur;
	289	locked += new_len - old_len;
	290	if (locked > lock_limit && !capable(CAP_IPC_LOCK))
	291	goto Eagain;
	292	}
	293
	294	if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT))
	295	goto Enomem;
	296
	297	if (vma->vm_flags & VM_ACCOUNT) {
	298	unsigned long charged = (new_len - old_len) >> PAGE_SHIFT;
	299	if (security_vm_enough_memory(charged))
	300	goto Efault;
	301	*p = charged;
	302	}
	303
	304	return vma;
	305
	306	Efault: /* very odd choice for most of the cases, but... */
	307	return ERR_PTR(-EFAULT);
	308	Einval:
	309	return ERR_PTR(-EINVAL);
	310	Enomem:
	311	return ERR_PTR(-ENOMEM);
	312	Eagain:
	313	return ERR_PTR(-EAGAIN);
	314	}
	315
ecc1a899 AV	316	static unsigned long mremap_to(unsigned long addr,
	317	unsigned long old_len, unsigned long new_addr,
	318	unsigned long new_len)
	319	{
	320	struct mm_struct *mm = current->mm;
	321	struct vm_area_struct *vma;
	322	unsigned long ret = -EINVAL;
	323	unsigned long charged = 0;
	324
	325	if (new_addr & ~PAGE_MASK)
	326	goto out;
	327
	328	if (new_len > TASK_SIZE \|\| new_addr > TASK_SIZE - new_len)
	329	goto out;
	330
	331	/* Check if the location we're moving into overlaps the
	332	* old location at all, and fail if it does.
	333	*/
	334	if ((new_addr <= addr) && (new_addr+new_len) > addr)
	335	goto out;
	336
	337	if ((addr <= new_addr) && (addr+old_len) > new_addr)
	338	goto out;
	339
	340	ret = security_file_mmap(NULL, 0, 0, 0, new_addr, 1);
	341	if (ret)
	342	goto out;
	343
	344	ret = do_munmap(mm, new_addr, new_len);
	345	if (ret)
	346	goto out;
	347
	348	if (old_len >= new_len) {
	349	ret = do_munmap(mm, addr+new_len, old_len - new_len);
	350	if (ret && old_len != new_len)
	351	goto out;
	352	old_len = new_len;
	353	}
	354
	355	vma = vma_to_resize(addr, old_len, new_len, &charged);
	356	if (IS_ERR(vma)) {
	357	ret = PTR_ERR(vma);
	358	goto out;
	359	}
	360
	361	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	362	if (ret & ~PAGE_MASK)
	363	vm_unacct_memory(charged);
	364
	365	out:
	366	return ret;
	367	}
	368
1da177e4 LT	369	/*
	370	* Expand (or shrink) an existing mapping, potentially moving it at the
	371	* same time (controlled by the MREMAP_MAYMOVE flag and available VM space)
	372	*
	373	* MREMAP_FIXED option added 5-Dec-1999 by Benjamin LaHaise
	374	* This option implies MREMAP_MAYMOVE.
	375	*/
	376	unsigned long do_mremap(unsigned long addr,
	377	unsigned long old_len, unsigned long new_len,
	378	unsigned long flags, unsigned long new_addr)
	379	{
d0de32d9	380	struct mm_struct *mm = current->mm;
1da177e4 LT	381	struct vm_area_struct *vma;
	382	unsigned long ret = -EINVAL;
	383	unsigned long charged = 0;
	384
	385	if (flags & ~(MREMAP_FIXED \| MREMAP_MAYMOVE))
	386	goto out;
	387
	388	if (addr & ~PAGE_MASK)
	389	goto out;
	390
	391	old_len = PAGE_ALIGN(old_len);
	392	new_len = PAGE_ALIGN(new_len);
	393
	394	/*
	395	* We allow a zero old-len as a special case
	396	* for DOS-emu "duplicate shm area" thing. But
	397	* a zero new-len is nonsensical.
	398	*/
	399	if (!new_len)
	400	goto out;
	401
1da177e4	402	if (flags & MREMAP_FIXED) {
ecc1a899 AV	403	if (flags & MREMAP_MAYMOVE)
	404	ret = mremap_to(addr, old_len, new_addr, new_len);
	405	goto out;
1da177e4 LT	406	}
	407
	408	/*
	409	* Always allow a shrinking remap: that just unmaps
	410	* the unnecessary pages..
	411	* do_munmap does all the needed commit accounting
	412	*/
	413	if (old_len >= new_len) {
d0de32d9	414	ret = do_munmap(mm, addr+new_len, old_len - new_len);
1da177e4 LT	415	if (ret && old_len != new_len)
	416	goto out;
	417	ret = addr;
ecc1a899	418	goto out;
1da177e4 LT	419	}
	420
	421	/*
ecc1a899	422	* Ok, we need to grow..
1da177e4	423	*/
54f5de70 AV	424	vma = vma_to_resize(addr, old_len, new_len, &charged);
	425	if (IS_ERR(vma)) {
	426	ret = PTR_ERR(vma);
1da177e4	427	goto out;
119f657c	428	}
1da177e4	429
1da177e4	430	/* old_len exactly to the end of the area..
1da177e4	431	*/
ecc1a899	432	if (old_len == vma->vm_end - addr) {
1da177e4 LT	433	unsigned long max_addr = TASK_SIZE;
	434	if (vma->vm_next)
	435	max_addr = vma->vm_next->vm_start;
	436	/* can we just expand the current mapping? */
	437	if (max_addr - addr >= new_len) {
	438	int pages = (new_len - old_len) >> PAGE_SHIFT;
	439
	440	vma_adjust(vma, vma->vm_start,
	441	addr + new_len, vma->vm_pgoff, NULL);
	442
d0de32d9 HD	443	mm->total_vm += pages;
d0de32d9 HD	444	vm_stat_account(mm, vma->vm_flags, vma->vm_file, pages);
1da177e4	445	if (vma->vm_flags & VM_LOCKED) {
d0de32d9	446	mm->locked_vm += pages;
ba470de4	447	mlock_vma_pages_range(vma, addr + old_len,
1da177e4 LT	448	addr + new_len);
	449	}
	450	ret = addr;
	451	goto out;
	452	}
	453	}
	454
	455	/*
	456	* We weren't able to just expand or shrink the area,
	457	* we need to create a new one and move it..
	458	*/
	459	ret = -ENOMEM;
	460	if (flags & MREMAP_MAYMOVE) {
ecc1a899 AV	461	unsigned long map_flags = 0;
	462	if (vma->vm_flags & VM_MAYSHARE)
	463	map_flags \|= MAP_SHARED;
	464
	465	new_addr = get_unmapped_area(vma->vm_file, 0, new_len,
	466	vma->vm_pgoff, map_flags);
	467	if (new_addr & ~PAGE_MASK) {
	468	ret = new_addr;
	469	goto out;
1da177e4	470	}
ecc1a899 AV	471
	472	ret = security_file_mmap(NULL, 0, 0, 0, new_addr, 1);
	473	if (ret)
	474	goto out;
1da177e4 LT	475	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	476	}
	477	out:
	478	if (ret & ~PAGE_MASK)
	479	vm_unacct_memory(charged);
1da177e4 LT	480	return ret;
	481	}
	482
6a6160a7 HC	483	SYSCALL_DEFINE5(mremap, unsigned long, addr, unsigned long, old_len,
	484	unsigned long, new_len, unsigned long, flags,
	485	unsigned long, new_addr)
1da177e4 LT	486	{
	487	unsigned long ret;
	488
	489	down_write(&current->mm->mmap_sem);
	490	ret = do_mremap(addr, old_len, new_len, flags, new_addr);
	491	up_write(&current->mm->mmap_sem);
	492	return ret;
	493	}