[linux-block.git] / mm / mremap.c

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

#include <linux/mm.h>
#include <linux/hugetlb.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(*pmd))
		return NULL;

	return pmd;
}

static pmd_t *alloc_new_pmd(struct mm_struct *mm, struct vm_area_struct *vma,
			    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;

	VM_BUG_ON(pmd_trans_huge(*pmd));

	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 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;
		mutex_lock(&mapping->i_mmap_mutex);
	}

	/*
	 * 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(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_get_and_clear(mm, 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(new_pte - 1);
	pte_unmap_unlock(old_pte - 1, old_ptl);
	if (mapping)
		mutex_unlock(&mapping->i_mmap_mutex);
}

#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;
	bool need_flush = false;

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

	mmu_notifier_invalidate_range_start(vma->vm_mm, old_addr, old_end);

	for (; old_addr < old_end; old_addr += extent, new_addr += extent) {
		cond_resched();
		next = (old_addr + PMD_SIZE) & PMD_MASK;
		/* even if next overflowed, extent below will be ok */
		extent = next - old_addr;
		if (extent > old_end - old_addr)
			extent = old_end - old_addr;
		old_pmd = get_old_pmd(vma->vm_mm, old_addr);
		if (!old_pmd)
			continue;
		new_pmd = alloc_new_pmd(vma->vm_mm, vma, new_addr);
		if (!new_pmd)
			break;
		if (pmd_trans_huge(*old_pmd)) {
			int err = 0;
			if (extent == HPAGE_PMD_SIZE)
				err = move_huge_pmd(vma, new_vma, old_addr,
						    new_addr, old_end,
						    old_pmd, new_pmd);
			if (err > 0) {
				need_flush = true;
				continue;
			} else if (!err) {
				split_huge_page_pmd(vma->vm_mm, old_pmd);
			}
			VM_BUG_ON(pmd_trans_huge(*old_pmd));
		}
		if (pmd_none(*new_pmd) && __pte_alloc(new_vma->vm_mm, new_vma,
						      new_pmd, new_addr))
			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);
		need_flush = true;
	}
	if (likely(need_flush))
		flush_tlb_range(vma, old_end-len, old_addr);

	mmu_notifier_invalidate_range_end(vma->vm_mm, old_end-len, old_end);

	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) {
		/*
		 * Before moving the page tables from the new vma to
		 * the old vma, we need to be sure the old vma is
		 * queued after new vma in the same_anon_vma list to
		 * prevent SMP races with rmap_walk (that could lead
		 * rmap_walk to miss some page table).
		 */
		anon_vma_moveto_tail(vma);

		/*
		 * 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;
	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;

	/* Need to be careful about a growing mapping */
	if (new_len > old_len) {
		unsigned long pgoff;

		if (vma->vm_flags & (VM_DONTEXPAND | VM_PFNMAP))
			goto Efault;
		pgoff = (addr - vma->vm_start) >> PAGE_SHIFT;
		pgoff += vma->vm_pgoff;
		if (pgoff + (new_len >> PAGE_SHIFT) < pgoff)
			goto Einval;
	}

	if (vma->vm_flags & VM_LOCKED) {
		unsigned long locked, lock_limit;
		locked = mm->locked_vm << PAGE_SHIFT;
		lock_limit = rlimit(RLIMIT_MEMLOCK);
		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_mm(mm, 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;
	unsigned long map_flags;

	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 = 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;
	}

	map_flags = MAP_FIXED;
	if (vma->vm_flags & VM_MAYSHARE)
		map_flags |= MAP_SHARED;

	ret = get_unmapped_area(vma->vm_file, new_addr, new_len, vma->vm_pgoff +
				((addr - vma->vm_start) >> PAGE_SHIFT),
				map_flags);
	if (ret & ~PAGE_MASK)
		goto out1;

	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	if (!(ret & ~PAGE_MASK))
		goto out;
out1:
	vm_unacct_memory(charged);

out:
	return ret;
}

static int vma_expandable(struct vm_area_struct *vma, unsigned long delta)
{
	unsigned long end = vma->vm_end + delta;
	if (end < vma->vm_end) /* overflow */
		return 0;
	if (vma->vm_next && vma->vm_next->vm_start < end) /* intersection */
		return 0;
	if (get_unmapped_area(NULL, vma->vm_start, end - vma->vm_start,
			      0, MAP_FIXED) & ~PAGE_MASK)
		return 0;
	return 1;
}

/*
 * 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.
 */
SYSCALL_DEFINE5(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;

	down_write(&current->mm->mmap_sem);

	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) {
		/* can we just expand the current mapping? */
		if (vma_expandable(vma, new_len - old_len)) {
			int pages = (new_len - old_len) >> PAGE_SHIFT;

			if (vma_adjust(vma, vma->vm_start, addr + new_len,
				       vma->vm_pgoff, NULL)) {
				ret = -ENOMEM;
				goto out;
			}

			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 +
					((addr - vma->vm_start) >> PAGE_SHIFT),
					map_flags);
		if (new_addr & ~PAGE_MASK) {
			ret = new_addr;
			goto out;
		}

		ret = move_vma(vma, addr, old_len, new_len, new_addr);
	}
out:
	if (ret & ~PAGE_MASK)
		vm_unacct_memory(charged);
	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 <alan@lxorguk.ukuu.org.uk>
1da177e4 LT	7	* (C) Copyright 2002 Red Hat Inc, All Rights Reserved
	8	*/
	9
	10	#include <linux/mm.h>
	11	#include <linux/hugetlb.h>
1da177e4	12	#include <linux/shm.h>
1ff82995	13	#include <linux/ksm.h>
1da177e4 LT	14	#include <linux/mman.h>
1da177e4 LT	15	#include <linux/swap.h>
c59ede7b	16	#include <linux/capability.h>
1da177e4 LT	17	#include <linux/fs.h>
	18	#include <linux/highmem.h>
	19	#include <linux/security.h>
	20	#include <linux/syscalls.h>
cddb8a5c	21	#include <linux/mmu_notifier.h>
1da177e4 LT	22
	23	#include <asm/uaccess.h>
	24	#include <asm/cacheflush.h>
	25	#include <asm/tlbflush.h>
	26
ba470de4 RR	27	#include "internal.h"
ba470de4 RR	28
7be7a546	29	static pmd_t get_old_pmd(struct mm_struct mm, unsigned long addr)
1da177e4 LT	30	{
	31	pgd_t *pgd;
	32	pud_t *pud;
	33	pmd_t *pmd;
	34
	35	pgd = pgd_offset(mm, addr);
	36	if (pgd_none_or_clear_bad(pgd))
	37	return NULL;
	38
	39	pud = pud_offset(pgd, addr);
	40	if (pud_none_or_clear_bad(pud))
	41	return NULL;
	42
	43	pmd = pmd_offset(pud, addr);
37a1c49a	44	if (pmd_none(*pmd))
1da177e4 LT	45	return NULL;
1da177e4 LT	46
7be7a546	47	return pmd;
1da177e4 LT	48	}
1da177e4 LT	49
8ac1f832 AA	50	static pmd_t alloc_new_pmd(struct mm_struct mm, struct vm_area_struct *vma,
8ac1f832 AA	51	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
8ac1f832	66	VM_BUG_ON(pmd_trans_huge(*pmd));
c74df32c	67
7be7a546	68	return pmd;
1da177e4 LT	69	}
1da177e4 LT	70
7be7a546 HD	71	static void move_ptes(struct vm_area_struct vma, pmd_t old_pmd,
	72	unsigned long old_addr, unsigned long old_end,
	73	struct vm_area_struct new_vma, pmd_t new_pmd,
	74	unsigned long new_addr)
1da177e4 LT	75	{
	76	struct address_space *mapping = NULL;
	77	struct mm_struct *mm = vma->vm_mm;
7be7a546	78	pte_t old_pte, new_pte, pte;
4c21e2f2	79	spinlock_t old_ptl, new_ptl;
1da177e4 LT	80
	81	if (vma->vm_file) {
	82	/*
	83	* Subtle point from Rajesh Venkatasubramanian: before
25d9e2d1	84	* moving file-based ptes, we must lock truncate_pagecache
25d9e2d1	85	* out, since it might clean the dst vma before the src vma,
1da177e4 LT	86	* and we propagate stale pages into the dst afterward.
	87	*/
	88	mapping = vma->vm_file->f_mapping;
3d48ae45	89	mutex_lock(&mapping->i_mmap_mutex);
1da177e4	90	}
1da177e4	91
4c21e2f2 HD	92	/*
	93	* We don't have to worry about the ordering of src and dst
	94	* pte locks because exclusive mmap_sem prevents deadlock.
	95	*/
c74df32c	96	old_pte = pte_offset_map_lock(mm, old_pmd, old_addr, &old_ptl);
ece0e2b6	97	new_pte = pte_offset_map(new_pmd, new_addr);
4c21e2f2 HD	98	new_ptl = pte_lockptr(mm, new_pmd);
4c21e2f2 HD	99	if (new_ptl != old_ptl)
f20dc5f7	100	spin_lock_nested(new_ptl, SINGLE_DEPTH_NESTING);
6606c3e0	101	arch_enter_lazy_mmu_mode();
7be7a546 HD	102
	103	for (; old_addr < old_end; old_pte++, old_addr += PAGE_SIZE,
	104	new_pte++, new_addr += PAGE_SIZE) {
	105	if (pte_none(*old_pte))
	106	continue;
7b6efc2b	107	pte = ptep_get_and_clear(mm, old_addr, old_pte);
7be7a546 HD	108	pte = move_pte(pte, new_vma->vm_page_prot, old_addr, new_addr);
7be7a546 HD	109	set_pte_at(mm, new_addr, new_pte, pte);
1da177e4	110	}
7be7a546	111
6606c3e0	112	arch_leave_lazy_mmu_mode();
4c21e2f2 HD	113	if (new_ptl != old_ptl)
4c21e2f2 HD	114	spin_unlock(new_ptl);
ece0e2b6	115	pte_unmap(new_pte - 1);
c74df32c	116	pte_unmap_unlock(old_pte - 1, old_ptl);
1da177e4	117	if (mapping)
3d48ae45	118	mutex_unlock(&mapping->i_mmap_mutex);
1da177e4 LT	119	}
1da177e4 LT	120
7be7a546 HD	121	#define LATENCY_LIMIT (64 * PAGE_SIZE)
7be7a546 HD	122
b6a2fea3	123	unsigned long move_page_tables(struct vm_area_struct *vma,
1da177e4 LT	124	unsigned long old_addr, struct vm_area_struct *new_vma,
	125	unsigned long new_addr, unsigned long len)
	126	{
7be7a546 HD	127	unsigned long extent, next, old_end;
7be7a546 HD	128	pmd_t old_pmd, new_pmd;
7b6efc2b	129	bool need_flush = false;
1da177e4	130
7be7a546 HD	131	old_end = old_addr + len;
7be7a546 HD	132	flush_cache_range(vma, old_addr, old_end);
1da177e4	133
7b6efc2b AA	134	mmu_notifier_invalidate_range_start(vma->vm_mm, old_addr, old_end);
7b6efc2b AA	135
7be7a546	136	for (; old_addr < old_end; old_addr += extent, new_addr += extent) {
1da177e4	137	cond_resched();
7be7a546	138	next = (old_addr + PMD_SIZE) & PMD_MASK;
ebed4846	139	/* even if next overflowed, extent below will be ok */
7be7a546	140	extent = next - old_addr;
ebed4846 AA	141	if (extent > old_end - old_addr)
ebed4846 AA	142	extent = old_end - old_addr;
7be7a546 HD	143	old_pmd = get_old_pmd(vma->vm_mm, old_addr);
	144	if (!old_pmd)
	145	continue;
8ac1f832	146	new_pmd = alloc_new_pmd(vma->vm_mm, vma, new_addr);
7be7a546 HD	147	if (!new_pmd)
7be7a546 HD	148	break;
37a1c49a AA	149	if (pmd_trans_huge(*old_pmd)) {
	150	int err = 0;
	151	if (extent == HPAGE_PMD_SIZE)
	152	err = move_huge_pmd(vma, new_vma, old_addr,
	153	new_addr, old_end,
	154	old_pmd, new_pmd);
	155	if (err > 0) {
	156	need_flush = true;
	157	continue;
	158	} else if (!err) {
	159	split_huge_page_pmd(vma->vm_mm, old_pmd);
	160	}
	161	VM_BUG_ON(pmd_trans_huge(*old_pmd));
	162	}
	163	if (pmd_none(*new_pmd) && __pte_alloc(new_vma->vm_mm, new_vma,
	164	new_pmd, new_addr))
	165	break;
7be7a546 HD	166	next = (new_addr + PMD_SIZE) & PMD_MASK;
	167	if (extent > next - new_addr)
	168	extent = next - new_addr;
	169	if (extent > LATENCY_LIMIT)
	170	extent = LATENCY_LIMIT;
	171	move_ptes(vma, old_pmd, old_addr, old_addr + extent,
	172	new_vma, new_pmd, new_addr);
7b6efc2b	173	need_flush = true;
1da177e4	174	}
7b6efc2b AA	175	if (likely(need_flush))
	176	flush_tlb_range(vma, old_end-len, old_addr);
	177
	178	mmu_notifier_invalidate_range_end(vma->vm_mm, old_end-len, old_end);
7be7a546 HD	179
7be7a546 HD	180	return len + old_addr - old_end; /* how much done */
1da177e4 LT	181	}
	182
	183	static unsigned long move_vma(struct vm_area_struct *vma,
	184	unsigned long old_addr, unsigned long old_len,
	185	unsigned long new_len, unsigned long new_addr)
	186	{
	187	struct mm_struct *mm = vma->vm_mm;
	188	struct vm_area_struct *new_vma;
	189	unsigned long vm_flags = vma->vm_flags;
	190	unsigned long new_pgoff;
	191	unsigned long moved_len;
	192	unsigned long excess = 0;
365e9c87	193	unsigned long hiwater_vm;
1da177e4	194	int split = 0;
7103ad32	195	int err;
1da177e4 LT	196
	197	/*
	198	* We'd prefer to avoid failure later on in do_munmap:
	199	* which may split one vma into three before unmapping.
	200	*/
	201	if (mm->map_count >= sysctl_max_map_count - 3)
	202	return -ENOMEM;
	203
1ff82995 HD	204	/*
	205	* Advise KSM to break any KSM pages in the area to be moved:
	206	* it would be confusing if they were to turn up at the new
	207	* location, where they happen to coincide with different KSM
	208	* pages recently unmapped. But leave vma->vm_flags as it was,
	209	* so KSM can come around to merge on vma and new_vma afterwards.
	210	*/
7103ad32 HD	211	err = ksm_madvise(vma, old_addr, old_addr + old_len,
	212	MADV_UNMERGEABLE, &vm_flags);
	213	if (err)
	214	return err;
1ff82995	215
1da177e4 LT	216	new_pgoff = vma->vm_pgoff + ((old_addr - vma->vm_start) >> PAGE_SHIFT);
	217	new_vma = copy_vma(&vma, new_addr, new_len, new_pgoff);
	218	if (!new_vma)
	219	return -ENOMEM;
	220
	221	moved_len = move_page_tables(vma, old_addr, new_vma, new_addr, old_len);
	222	if (moved_len < old_len) {
948f017b AA	223	/*
	224	* Before moving the page tables from the new vma to
	225	* the old vma, we need to be sure the old vma is
	226	* queued after new vma in the same_anon_vma list to
	227	* prevent SMP races with rmap_walk (that could lead
	228	* rmap_walk to miss some page table).
	229	*/
	230	anon_vma_moveto_tail(vma);
	231
1da177e4 LT	232	/*
	233	* On error, move entries back from new area to old,
	234	* which will succeed since page tables still there,
	235	* and then proceed to unmap new area instead of old.
	236	*/
	237	move_page_tables(new_vma, new_addr, vma, old_addr, moved_len);
	238	vma = new_vma;
	239	old_len = new_len;
	240	old_addr = new_addr;
	241	new_addr = -ENOMEM;
	242	}
	243
	244	/* Conceal VM_ACCOUNT so old reservation is not undone */
	245	if (vm_flags & VM_ACCOUNT) {
	246	vma->vm_flags &= ~VM_ACCOUNT;
	247	excess = vma->vm_end - vma->vm_start - old_len;
	248	if (old_addr > vma->vm_start &&
	249	old_addr + old_len < vma->vm_end)
	250	split = 1;
	251	}
	252
71799062	253	/*
365e9c87 HD	254	* If we failed to move page tables we still do total_vm increment
	255	* since do_munmap() will decrement it by old_len == new_len.
	256	*
	257	* Since total_vm is about to be raised artificially high for a
	258	* moment, we need to restore high watermark afterwards: if stats
	259	* are taken meanwhile, total_vm and hiwater_vm appear too high.
	260	* If this were a serious issue, we'd add a flag to do_munmap().
71799062	261	*/
365e9c87	262	hiwater_vm = mm->hiwater_vm;
ab50b8ed	263	vm_stat_account(mm, vma->vm_flags, vma->vm_file, new_len>>PAGE_SHIFT);
71799062	264
1da177e4 LT	265	if (do_munmap(mm, old_addr, old_len) < 0) {
	266	/* OOM: unable to split vma, just get accounts right */
	267	vm_unacct_memory(excess >> PAGE_SHIFT);
	268	excess = 0;
	269	}
365e9c87	270	mm->hiwater_vm = hiwater_vm;
1da177e4 LT	271
	272	/* Restore VM_ACCOUNT if one or two pieces of vma left */
	273	if (excess) {
	274	vma->vm_flags \|= VM_ACCOUNT;
	275	if (split)
	276	vma->vm_next->vm_flags \|= VM_ACCOUNT;
	277	}
	278
1da177e4 LT	279	if (vm_flags & VM_LOCKED) {
	280	mm->locked_vm += new_len >> PAGE_SHIFT;
	281	if (new_len > old_len)
ba470de4 RR	282	mlock_vma_pages_range(new_vma, new_addr + old_len,
ba470de4 RR	283	new_addr + new_len);
1da177e4 LT	284	}
	285
	286	return new_addr;
	287	}
	288
54f5de70 AV	289	static struct vm_area_struct *vma_to_resize(unsigned long addr,
	290	unsigned long old_len, unsigned long new_len, unsigned long *p)
	291	{
	292	struct mm_struct *mm = current->mm;
	293	struct vm_area_struct *vma = find_vma(mm, addr);
	294
	295	if (!vma \|\| vma->vm_start > addr)
	296	goto Efault;
	297
	298	if (is_vm_hugetlb_page(vma))
	299	goto Einval;
	300
	301	/* We can't remap across vm area boundaries */
	302	if (old_len > vma->vm_end - addr)
	303	goto Efault;
	304
982134ba LT	305	/* Need to be careful about a growing mapping */
	306	if (new_len > old_len) {
	307	unsigned long pgoff;
	308
	309	if (vma->vm_flags & (VM_DONTEXPAND \| VM_PFNMAP))
54f5de70	310	goto Efault;
982134ba LT	311	pgoff = (addr - vma->vm_start) >> PAGE_SHIFT;
	312	pgoff += vma->vm_pgoff;
	313	if (pgoff + (new_len >> PAGE_SHIFT) < pgoff)
	314	goto Einval;
54f5de70 AV	315	}
	316
	317	if (vma->vm_flags & VM_LOCKED) {
	318	unsigned long locked, lock_limit;
	319	locked = mm->locked_vm << PAGE_SHIFT;
59e99e5b	320	lock_limit = rlimit(RLIMIT_MEMLOCK);
54f5de70 AV	321	locked += new_len - old_len;
	322	if (locked > lock_limit && !capable(CAP_IPC_LOCK))
	323	goto Eagain;
	324	}
	325
	326	if (!may_expand_vm(mm, (new_len - old_len) >> PAGE_SHIFT))
	327	goto Enomem;
	328
	329	if (vma->vm_flags & VM_ACCOUNT) {
	330	unsigned long charged = (new_len - old_len) >> PAGE_SHIFT;
191c5424	331	if (security_vm_enough_memory_mm(mm, charged))
54f5de70 AV	332	goto Efault;
	333	*p = charged;
	334	}
	335
	336	return vma;
	337
	338	Efault: /* very odd choice for most of the cases, but... */
	339	return ERR_PTR(-EFAULT);
	340	Einval:
	341	return ERR_PTR(-EINVAL);
	342	Enomem:
	343	return ERR_PTR(-ENOMEM);
	344	Eagain:
	345	return ERR_PTR(-EAGAIN);
	346	}
	347
ecc1a899 AV	348	static unsigned long mremap_to(unsigned long addr,
	349	unsigned long old_len, unsigned long new_addr,
	350	unsigned long new_len)
	351	{
	352	struct mm_struct *mm = current->mm;
	353	struct vm_area_struct *vma;
	354	unsigned long ret = -EINVAL;
	355	unsigned long charged = 0;
097eed10	356	unsigned long map_flags;
ecc1a899 AV	357
	358	if (new_addr & ~PAGE_MASK)
	359	goto out;
	360
	361	if (new_len > TASK_SIZE \|\| new_addr > TASK_SIZE - new_len)
	362	goto out;
	363
	364	/* Check if the location we're moving into overlaps the
	365	* old location at all, and fail if it does.
	366	*/
	367	if ((new_addr <= addr) && (new_addr+new_len) > addr)
	368	goto out;
	369
	370	if ((addr <= new_addr) && (addr+old_len) > new_addr)
	371	goto out;
	372
ecc1a899 AV	373	ret = do_munmap(mm, new_addr, new_len);
	374	if (ret)
	375	goto out;
	376
	377	if (old_len >= new_len) {
	378	ret = do_munmap(mm, addr+new_len, old_len - new_len);
	379	if (ret && old_len != new_len)
	380	goto out;
	381	old_len = new_len;
	382	}
	383
	384	vma = vma_to_resize(addr, old_len, new_len, &charged);
	385	if (IS_ERR(vma)) {
	386	ret = PTR_ERR(vma);
	387	goto out;
	388	}
	389
097eed10 AV	390	map_flags = MAP_FIXED;
	391	if (vma->vm_flags & VM_MAYSHARE)
	392	map_flags \|= MAP_SHARED;
9206de95	393
097eed10 AV	394	ret = get_unmapped_area(vma->vm_file, new_addr, new_len, vma->vm_pgoff +
	395	((addr - vma->vm_start) >> PAGE_SHIFT),
	396	map_flags);
ecc1a899	397	if (ret & ~PAGE_MASK)
097eed10 AV	398	goto out1;
	399
	400	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	401	if (!(ret & ~PAGE_MASK))
	402	goto out;
	403	out1:
	404	vm_unacct_memory(charged);
ecc1a899 AV	405
	406	out:
	407	return ret;
	408	}
	409
1a0ef85f AV	410	static int vma_expandable(struct vm_area_struct *vma, unsigned long delta)
1a0ef85f AV	411	{
f106af4e	412	unsigned long end = vma->vm_end + delta;
9206de95	413	if (end < vma->vm_end) /* overflow */
f106af4e	414	return 0;
9206de95	415	if (vma->vm_next && vma->vm_next->vm_start < end) /* intersection */
f106af4e AV	416	return 0;
	417	if (get_unmapped_area(NULL, vma->vm_start, end - vma->vm_start,
	418	0, MAP_FIXED) & ~PAGE_MASK)
1a0ef85f	419	return 0;
1a0ef85f AV	420	return 1;
	421	}
	422
1da177e4 LT	423	/*
	424	* Expand (or shrink) an existing mapping, potentially moving it at the
	425	* same time (controlled by the MREMAP_MAYMOVE flag and available VM space)
	426	*
	427	* MREMAP_FIXED option added 5-Dec-1999 by Benjamin LaHaise
	428	* This option implies MREMAP_MAYMOVE.
	429	*/
63a81db1 AV	430	SYSCALL_DEFINE5(mremap, unsigned long, addr, unsigned long, old_len,
	431	unsigned long, new_len, unsigned long, flags,
	432	unsigned long, new_addr)
1da177e4	433	{
d0de32d9	434	struct mm_struct *mm = current->mm;
1da177e4 LT	435	struct vm_area_struct *vma;
	436	unsigned long ret = -EINVAL;
	437	unsigned long charged = 0;
	438
63a81db1 AV	439	down_write(&current->mm->mmap_sem);
63a81db1 AV	440
1da177e4 LT	441	if (flags & ~(MREMAP_FIXED \| MREMAP_MAYMOVE))
	442	goto out;
	443
	444	if (addr & ~PAGE_MASK)
	445	goto out;
	446
	447	old_len = PAGE_ALIGN(old_len);
	448	new_len = PAGE_ALIGN(new_len);
	449
	450	/*
	451	* We allow a zero old-len as a special case
	452	* for DOS-emu "duplicate shm area" thing. But
	453	* a zero new-len is nonsensical.
	454	*/
	455	if (!new_len)
	456	goto out;
	457
1da177e4	458	if (flags & MREMAP_FIXED) {
ecc1a899 AV	459	if (flags & MREMAP_MAYMOVE)
	460	ret = mremap_to(addr, old_len, new_addr, new_len);
	461	goto out;
1da177e4 LT	462	}
	463
	464	/*
	465	* Always allow a shrinking remap: that just unmaps
	466	* the unnecessary pages..
	467	* do_munmap does all the needed commit accounting
	468	*/
	469	if (old_len >= new_len) {
d0de32d9	470	ret = do_munmap(mm, addr+new_len, old_len - new_len);
1da177e4 LT	471	if (ret && old_len != new_len)
	472	goto out;
	473	ret = addr;
ecc1a899	474	goto out;
1da177e4 LT	475	}
	476
	477	/*
ecc1a899	478	* Ok, we need to grow..
1da177e4	479	*/
54f5de70 AV	480	vma = vma_to_resize(addr, old_len, new_len, &charged);
	481	if (IS_ERR(vma)) {
	482	ret = PTR_ERR(vma);
1da177e4	483	goto out;
119f657c	484	}
1da177e4	485
1da177e4	486	/* old_len exactly to the end of the area..
1da177e4	487	*/
ecc1a899	488	if (old_len == vma->vm_end - addr) {
1da177e4	489	/* can we just expand the current mapping? */
1a0ef85f	490	if (vma_expandable(vma, new_len - old_len)) {
1da177e4 LT	491	int pages = (new_len - old_len) >> PAGE_SHIFT;
1da177e4 LT	492
5beb4930 RR	493	if (vma_adjust(vma, vma->vm_start, addr + new_len,
	494	vma->vm_pgoff, NULL)) {
	495	ret = -ENOMEM;
	496	goto out;
	497	}
1da177e4	498
d0de32d9	499	vm_stat_account(mm, vma->vm_flags, vma->vm_file, pages);
1da177e4	500	if (vma->vm_flags & VM_LOCKED) {
d0de32d9	501	mm->locked_vm += pages;
ba470de4	502	mlock_vma_pages_range(vma, addr + old_len,
1da177e4 LT	503	addr + new_len);
	504	}
	505	ret = addr;
	506	goto out;
	507	}
	508	}
	509
	510	/*
	511	* We weren't able to just expand or shrink the area,
	512	* we need to create a new one and move it..
	513	*/
	514	ret = -ENOMEM;
	515	if (flags & MREMAP_MAYMOVE) {
ecc1a899 AV	516	unsigned long map_flags = 0;
	517	if (vma->vm_flags & VM_MAYSHARE)
	518	map_flags \|= MAP_SHARED;
	519
	520	new_addr = get_unmapped_area(vma->vm_file, 0, new_len,
93587414 AV	521	vma->vm_pgoff +
	522	((addr - vma->vm_start) >> PAGE_SHIFT),
	523	map_flags);
ecc1a899 AV	524	if (new_addr & ~PAGE_MASK) {
	525	ret = new_addr;
	526	goto out;
1da177e4	527	}
ecc1a899	528
1da177e4 LT	529	ret = move_vma(vma, addr, old_len, new_len, new_addr);
	530	}
	531	out:
	532	if (ret & ~PAGE_MASK)
	533	vm_unacct_memory(charged);
1da177e4 LT	534	up_write(&current->mm->mmap_sem);
	535	return ret;
	536	}