[linux.git] / mm / mprotect.c

/*
 *  mm/mprotect.c
 *
 *  (C) Copyright 1994 Linus Torvalds
 *  (C) Copyright 2002 Christoph Hellwig
 *
 *  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/shm.h>
#include <linux/mman.h>
#include <linux/fs.h>
#include <linux/highmem.h>
#include <linux/security.h>
#include <linux/mempolicy.h>
#include <linux/personality.h>
#include <linux/syscalls.h>
#include <linux/swap.h>
#include <linux/swapops.h>
#include <linux/mmu_notifier.h>
#include <linux/migrate.h>
#include <linux/perf_event.h>
#include <linux/pkeys.h>
#include <linux/ksm.h>
#include <linux/uaccess.h>
#include <asm/pgtable.h>
#include <asm/cacheflush.h>
#include <asm/mmu_context.h>
#include <asm/tlbflush.h>

#include "internal.h"

/*
 * For a prot_numa update we only hold mmap_sem for read so there is a
 * potential race with faulting where a pmd was temporarily none. This
 * function checks for a transhuge pmd under the appropriate lock. It
 * returns a pte if it was successfully locked or NULL if it raced with
 * a transhuge insertion.
 */
static pte_t *lock_pte_protection(struct vm_area_struct *vma, pmd_t *pmd,
			unsigned long addr, int prot_numa, spinlock_t **ptl)
{
	pte_t *pte;
	spinlock_t *pmdl;

	/* !prot_numa is protected by mmap_sem held for write */
	if (!prot_numa)
		return pte_offset_map_lock(vma->vm_mm, pmd, addr, ptl);

	pmdl = pmd_lock(vma->vm_mm, pmd);
	if (unlikely(pmd_trans_huge(*pmd) || pmd_none(*pmd))) {
		spin_unlock(pmdl);
		return NULL;
	}

	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, ptl);
	spin_unlock(pmdl);
	return pte;
}

static unsigned long change_pte_range(struct vm_area_struct *vma, pmd_t *pmd,
		unsigned long addr, unsigned long end, pgprot_t newprot,
		int dirty_accountable, int prot_numa)
{
	struct mm_struct *mm = vma->vm_mm;
	pte_t *pte, oldpte;
	spinlock_t *ptl;
	unsigned long pages = 0;
	int target_node = NUMA_NO_NODE;

	pte = lock_pte_protection(vma, pmd, addr, prot_numa, &ptl);
	if (!pte)
		return 0;

	/* Get target node for single threaded private VMAs */
	if (prot_numa && !(vma->vm_flags & VM_SHARED) &&
	    atomic_read(&vma->vm_mm->mm_users) == 1)
		target_node = numa_node_id();

	arch_enter_lazy_mmu_mode();
	do {
		oldpte = *pte;
		if (pte_present(oldpte)) {
			pte_t ptent;
			bool preserve_write = prot_numa && pte_write(oldpte);

			/*
			 * Avoid trapping faults against the zero or KSM
			 * pages. See similar comment in change_huge_pmd.
			 */
			if (prot_numa) {
				struct page *page;

				page = vm_normal_page(vma, addr, oldpte);
				if (!page || PageKsm(page))
					continue;

				/* Avoid TLB flush if possible */
				if (pte_protnone(oldpte))
					continue;

				/*
				 * Don't mess with PTEs if page is already on the node
				 * a single-threaded process is running on.
				 */
				if (target_node == page_to_nid(page))
					continue;
			}

			ptent = ptep_modify_prot_start(mm, addr, pte);
			ptent = pte_modify(ptent, newprot);
			if (preserve_write)
				ptent = pte_mkwrite(ptent);

			/* Avoid taking write faults for known dirty pages */
			if (dirty_accountable && pte_dirty(ptent) &&
					(pte_soft_dirty(ptent) ||
					 !(vma->vm_flags & VM_SOFTDIRTY))) {
				ptent = pte_mkwrite(ptent);
			}
			ptep_modify_prot_commit(mm, addr, pte, ptent);
			pages++;
		} else if (IS_ENABLED(CONFIG_MIGRATION)) {
			swp_entry_t entry = pte_to_swp_entry(oldpte);

			if (is_write_migration_entry(entry)) {
				pte_t newpte;
				/*
				 * A protection check is difficult so
				 * just be safe and disable write
				 */
				make_migration_entry_read(&entry);
				newpte = swp_entry_to_pte(entry);
				if (pte_swp_soft_dirty(oldpte))
					newpte = pte_swp_mksoft_dirty(newpte);
				set_pte_at(mm, addr, pte, newpte);

				pages++;
			}
		}
	} while (pte++, addr += PAGE_SIZE, addr != end);
	arch_leave_lazy_mmu_mode();
	pte_unmap_unlock(pte - 1, ptl);

	return pages;
}

static inline unsigned long change_pmd_range(struct vm_area_struct *vma,
		pud_t *pud, unsigned long addr, unsigned long end,
		pgprot_t newprot, int dirty_accountable, int prot_numa)
{
	pmd_t *pmd;
	struct mm_struct *mm = vma->vm_mm;
	unsigned long next;
	unsigned long pages = 0;
	unsigned long nr_huge_updates = 0;
	unsigned long mni_start = 0;

	pmd = pmd_offset(pud, addr);
	do {
		unsigned long this_pages;

		next = pmd_addr_end(addr, end);
		if (!pmd_trans_huge(*pmd) && !pmd_devmap(*pmd)
				&& pmd_none_or_clear_bad(pmd))
			continue;

		/* invoke the mmu notifier if the pmd is populated */
		if (!mni_start) {
			mni_start = addr;
			mmu_notifier_invalidate_range_start(mm, mni_start, end);
		}

		if (pmd_trans_huge(*pmd) || pmd_devmap(*pmd)) {
			if (next - addr != HPAGE_PMD_SIZE) {
				__split_huge_pmd(vma, pmd, addr, false, NULL);
				if (pmd_trans_unstable(pmd))
					continue;
			} else {
				int nr_ptes = change_huge_pmd(vma, pmd, addr,
						newprot, prot_numa);

				if (nr_ptes) {
					if (nr_ptes == HPAGE_PMD_NR) {
						pages += HPAGE_PMD_NR;
						nr_huge_updates++;
					}

					/* huge pmd was handled */
					continue;
				}
			}
			/* fall through, the trans huge pmd just split */
		}
		this_pages = change_pte_range(vma, pmd, addr, next, newprot,
				 dirty_accountable, prot_numa);
		pages += this_pages;
	} while (pmd++, addr = next, addr != end);

	if (mni_start)
		mmu_notifier_invalidate_range_end(mm, mni_start, end);

	if (nr_huge_updates)
		count_vm_numa_events(NUMA_HUGE_PTE_UPDATES, nr_huge_updates);
	return pages;
}

static inline unsigned long change_pud_range(struct vm_area_struct *vma,
		pgd_t *pgd, unsigned long addr, unsigned long end,
		pgprot_t newprot, int dirty_accountable, int prot_numa)
{
	pud_t *pud;
	unsigned long next;
	unsigned long pages = 0;

	pud = pud_offset(pgd, addr);
	do {
		next = pud_addr_end(addr, end);
		if (pud_none_or_clear_bad(pud))
			continue;
		pages += change_pmd_range(vma, pud, addr, next, newprot,
				 dirty_accountable, prot_numa);
	} while (pud++, addr = next, addr != end);

	return pages;
}

static unsigned long change_protection_range(struct vm_area_struct *vma,
		unsigned long addr, unsigned long end, pgprot_t newprot,
		int dirty_accountable, int prot_numa)
{
	struct mm_struct *mm = vma->vm_mm;
	pgd_t *pgd;
	unsigned long next;
	unsigned long start = addr;
	unsigned long pages = 0;

	BUG_ON(addr >= end);
	pgd = pgd_offset(mm, addr);
	flush_cache_range(vma, addr, end);
	set_tlb_flush_pending(mm);
	do {
		next = pgd_addr_end(addr, end);
		if (pgd_none_or_clear_bad(pgd))
			continue;
		pages += change_pud_range(vma, pgd, addr, next, newprot,
				 dirty_accountable, prot_numa);
	} while (pgd++, addr = next, addr != end);

	/* Only flush the TLB if we actually modified any entries: */
	if (pages)
		flush_tlb_range(vma, start, end);
	clear_tlb_flush_pending(mm);

	return pages;
}

unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
		       unsigned long end, pgprot_t newprot,
		       int dirty_accountable, int prot_numa)
{
	unsigned long pages;

	if (is_vm_hugetlb_page(vma))
		pages = hugetlb_change_protection(vma, start, end, newprot);
	else
		pages = change_protection_range(vma, start, end, newprot, dirty_accountable, prot_numa);

	return pages;
}

int
mprotect_fixup(struct vm_area_struct *vma, struct vm_area_struct **pprev,
	unsigned long start, unsigned long end, unsigned long newflags)
{
	struct mm_struct *mm = vma->vm_mm;
	unsigned long oldflags = vma->vm_flags;
	long nrpages = (end - start) >> PAGE_SHIFT;
	unsigned long charged = 0;
	pgoff_t pgoff;
	int error;
	int dirty_accountable = 0;

	if (newflags == oldflags) {
		*pprev = vma;
		return 0;
	}

	/*
	 * If we make a private mapping writable we increase our commit;
	 * but (without finer accounting) cannot reduce our commit if we
	 * make it unwritable again. hugetlb mapping were accounted for
	 * even if read-only so there is no need to account for them here
	 */
	if (newflags & VM_WRITE) {
		/* Check space limits when area turns into data. */
		if (!may_expand_vm(mm, newflags, nrpages) &&
				may_expand_vm(mm, oldflags, nrpages))
			return -ENOMEM;
		if (!(oldflags & (VM_ACCOUNT|VM_WRITE|VM_HUGETLB|
						VM_SHARED|VM_NORESERVE))) {
			charged = nrpages;
			if (security_vm_enough_memory_mm(mm, charged))
				return -ENOMEM;
			newflags |= VM_ACCOUNT;
		}
	}

	/*
	 * First try to merge with previous and/or next vma.
	 */
	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	*pprev = vma_merge(mm, *pprev, start, end, newflags,
			   vma->anon_vma, vma->vm_file, pgoff, vma_policy(vma),
			   vma->vm_userfaultfd_ctx);
	if (*pprev) {
		vma = *pprev;
		VM_WARN_ON((vma->vm_flags ^ newflags) & ~VM_SOFTDIRTY);
		goto success;
	}

	*pprev = vma;

	if (start != vma->vm_start) {
		error = split_vma(mm, vma, start, 1);
		if (error)
			goto fail;
	}

	if (end != vma->vm_end) {
		error = split_vma(mm, vma, end, 0);
		if (error)
			goto fail;
	}

success:
	/*
	 * vm_flags and vm_page_prot are protected by the mmap_sem
	 * held in write mode.
	 */
	vma->vm_flags = newflags;
	dirty_accountable = vma_wants_writenotify(vma, vma->vm_page_prot);
	vma_set_page_prot(vma);

	change_protection(vma, start, end, vma->vm_page_prot,
			  dirty_accountable, 0);

	/*
	 * Private VM_LOCKED VMA becoming writable: trigger COW to avoid major
	 * fault on access.
	 */
	if ((oldflags & (VM_WRITE | VM_SHARED | VM_LOCKED)) == VM_LOCKED &&
			(newflags & VM_WRITE)) {
		populate_vma_page_range(vma, start, end, NULL);
	}

	vm_stat_account(mm, oldflags, -nrpages);
	vm_stat_account(mm, newflags, nrpages);
	perf_event_mmap(vma);
	return 0;

fail:
	vm_unacct_memory(charged);
	return error;
}

/*
 * pkey==-1 when doing a legacy mprotect()
 */
static int do_mprotect_pkey(unsigned long start, size_t len,
		unsigned long prot, int pkey)
{
	unsigned long nstart, end, tmp, reqprot;
	struct vm_area_struct *vma, *prev;
	int error = -EINVAL;
	const int grows = prot & (PROT_GROWSDOWN|PROT_GROWSUP);
	const bool rier = (current->personality & READ_IMPLIES_EXEC) &&
				(prot & PROT_READ);

	prot &= ~(PROT_GROWSDOWN|PROT_GROWSUP);
	if (grows == (PROT_GROWSDOWN|PROT_GROWSUP)) /* can't be both */
		return -EINVAL;

	if (start & ~PAGE_MASK)
		return -EINVAL;
	if (!len)
		return 0;
	len = PAGE_ALIGN(len);
	end = start + len;
	if (end <= start)
		return -ENOMEM;
	if (!arch_validate_prot(prot))
		return -EINVAL;

	reqprot = prot;

	if (down_write_killable(&current->mm->mmap_sem))
		return -EINTR;

	/*
	 * If userspace did not allocate the pkey, do not let
	 * them use it here.
	 */
	error = -EINVAL;
	if ((pkey != -1) && !mm_pkey_is_allocated(current->mm, pkey))
		goto out;

	vma = find_vma(current->mm, start);
	error = -ENOMEM;
	if (!vma)
		goto out;
	prev = vma->vm_prev;
	if (unlikely(grows & PROT_GROWSDOWN)) {
		if (vma->vm_start >= end)
			goto out;
		start = vma->vm_start;
		error = -EINVAL;
		if (!(vma->vm_flags & VM_GROWSDOWN))
			goto out;
	} else {
		if (vma->vm_start > start)
			goto out;
		if (unlikely(grows & PROT_GROWSUP)) {
			end = vma->vm_end;
			error = -EINVAL;
			if (!(vma->vm_flags & VM_GROWSUP))
				goto out;
		}
	}
	if (start > vma->vm_start)
		prev = vma;

	for (nstart = start ; ; ) {
		unsigned long mask_off_old_flags;
		unsigned long newflags;
		int new_vma_pkey;

		/* Here we know that vma->vm_start <= nstart < vma->vm_end. */

		/* Does the application expect PROT_READ to imply PROT_EXEC */
		if (rier && (vma->vm_flags & VM_MAYEXEC))
			prot |= PROT_EXEC;

		/*
		 * Each mprotect() call explicitly passes r/w/x permissions.
		 * If a permission is not passed to mprotect(), it must be
		 * cleared from the VMA.
		 */
		mask_off_old_flags = VM_READ | VM_WRITE | VM_EXEC |
					ARCH_VM_PKEY_FLAGS;

		new_vma_pkey = arch_override_mprotect_pkey(vma, prot, pkey);
		newflags = calc_vm_prot_bits(prot, new_vma_pkey);
		newflags |= (vma->vm_flags & ~mask_off_old_flags);

		/* newflags >> 4 shift VM_MAY% in place of VM_% */
		if ((newflags & ~(newflags >> 4)) & (VM_READ | VM_WRITE | VM_EXEC)) {
			error = -EACCES;
			goto out;
		}

		error = security_file_mprotect(vma, reqprot, prot);
		if (error)
			goto out;

		tmp = vma->vm_end;
		if (tmp > end)
			tmp = end;
		error = mprotect_fixup(vma, &prev, nstart, tmp, newflags);
		if (error)
			goto out;
		nstart = tmp;

		if (nstart < prev->vm_end)
			nstart = prev->vm_end;
		if (nstart >= end)
			goto out;

		vma = prev->vm_next;
		if (!vma || vma->vm_start != nstart) {
			error = -ENOMEM;
			goto out;
		}
		prot = reqprot;
	}
out:
	up_write(&current->mm->mmap_sem);
	return error;
}

SYSCALL_DEFINE3(mprotect, unsigned long, start, size_t, len,
		unsigned long, prot)
{
	return do_mprotect_pkey(start, len, prot, -1);
}

#ifdef CONFIG_ARCH_HAS_PKEYS

SYSCALL_DEFINE4(pkey_mprotect, unsigned long, start, size_t, len,
		unsigned long, prot, int, pkey)
{
	return do_mprotect_pkey(start, len, prot, pkey);
}

SYSCALL_DEFINE2(pkey_alloc, unsigned long, flags, unsigned long, init_val)
{
	int pkey;
	int ret;

	/* No flags supported yet. */
	if (flags)
		return -EINVAL;
	/* check for unsupported init values */
	if (init_val & ~PKEY_ACCESS_MASK)
		return -EINVAL;

	down_write(&current->mm->mmap_sem);
	pkey = mm_pkey_alloc(current->mm);

	ret = -ENOSPC;
	if (pkey == -1)
		goto out;

	ret = arch_set_user_pkey_access(current, pkey, init_val);
	if (ret) {
		mm_pkey_free(current->mm, pkey);
		goto out;
	}
	ret = pkey;
out:
	up_write(&current->mm->mmap_sem);
	return ret;
}

SYSCALL_DEFINE1(pkey_free, int, pkey)
{
	int ret;

	down_write(&current->mm->mmap_sem);
	ret = mm_pkey_free(current->mm, pkey);
	up_write(&current->mm->mmap_sem);

	/*
	 * We could provie warnings or errors if any VMA still
	 * has the pkey set here.
	 */
	return ret;
}

#endif /* CONFIG_ARCH_HAS_PKEYS */
Commit	Line	Data
1da177e4 LT	1	/*
	2	* mm/mprotect.c
	3	*
	4	* (C) Copyright 1994 Linus Torvalds
	5	* (C) Copyright 2002 Christoph Hellwig
	6	*
046c6884	7	* Address space accounting code <[email protected]>
1da177e4 LT	8	* (C) Copyright 2002 Red Hat Inc, All Rights Reserved
	9	*/
	10
	11	#include <linux/mm.h>
	12	#include <linux/hugetlb.h>
1da177e4 LT	13	#include <linux/shm.h>
	14	#include <linux/mman.h>
	15	#include <linux/fs.h>
	16	#include <linux/highmem.h>
	17	#include <linux/security.h>
	18	#include <linux/mempolicy.h>
	19	#include <linux/personality.h>
	20	#include <linux/syscalls.h>
0697212a CL	21	#include <linux/swap.h>
0697212a CL	22	#include <linux/swapops.h>
cddb8a5c	23	#include <linux/mmu_notifier.h>
64cdd548	24	#include <linux/migrate.h>
cdd6c482	25	#include <linux/perf_event.h>
e8c24d3a	26	#include <linux/pkeys.h>
64a9a34e	27	#include <linux/ksm.h>
7c0f6ba6	28	#include <linux/uaccess.h>
1da177e4 LT	29	#include <asm/pgtable.h>
1da177e4 LT	30	#include <asm/cacheflush.h>
e8c24d3a	31	#include <asm/mmu_context.h>
1da177e4 LT	32	#include <asm/tlbflush.h>
1da177e4 LT	33
36f88188 KS	34	#include "internal.h"
36f88188 KS	35
1ad9f620 MG	36	/*
	37	* For a prot_numa update we only hold mmap_sem for read so there is a
	38	* potential race with faulting where a pmd was temporarily none. This
	39	* function checks for a transhuge pmd under the appropriate lock. It
	40	* returns a pte if it was successfully locked or NULL if it raced with
	41	* a transhuge insertion.
	42	*/
	43	static pte_t lock_pte_protection(struct vm_area_struct vma, pmd_t *pmd,
	44	unsigned long addr, int prot_numa, spinlock_t **ptl)
	45	{
	46	pte_t *pte;
	47	spinlock_t *pmdl;
	48
	49	/* !prot_numa is protected by mmap_sem held for write */
	50	if (!prot_numa)
	51	return pte_offset_map_lock(vma->vm_mm, pmd, addr, ptl);
	52
	53	pmdl = pmd_lock(vma->vm_mm, pmd);
	54	if (unlikely(pmd_trans_huge(pmd) \|\| pmd_none(pmd))) {
	55	spin_unlock(pmdl);
	56	return NULL;
	57	}
	58
	59	pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, ptl);
	60	spin_unlock(pmdl);
	61	return pte;
	62	}
	63
4b10e7d5	64	static unsigned long change_pte_range(struct vm_area_struct vma, pmd_t pmd,
c1e6098b	65	unsigned long addr, unsigned long end, pgprot_t newprot,
0f19c179	66	int dirty_accountable, int prot_numa)
1da177e4	67	{
4b10e7d5	68	struct mm_struct *mm = vma->vm_mm;
0697212a	69	pte_t *pte, oldpte;
705e87c0	70	spinlock_t *ptl;
7da4d641	71	unsigned long pages = 0;
3e321587	72	int target_node = NUMA_NO_NODE;
1da177e4	73
1ad9f620 MG	74	pte = lock_pte_protection(vma, pmd, addr, prot_numa, &ptl);
	75	if (!pte)
	76	return 0;
	77
3e321587 AK	78	/* Get target node for single threaded private VMAs */
	79	if (prot_numa && !(vma->vm_flags & VM_SHARED) &&
	80	atomic_read(&vma->vm_mm->mm_users) == 1)
	81	target_node = numa_node_id();
	82
6606c3e0	83	arch_enter_lazy_mmu_mode();
1da177e4	84	do {
0697212a CL	85	oldpte = *pte;
0697212a CL	86	if (pte_present(oldpte)) {
1da177e4	87	pte_t ptent;
b191f9b1	88	bool preserve_write = prot_numa && pte_write(oldpte);
1da177e4	89
e944fd67 MG	90	/*
	91	* Avoid trapping faults against the zero or KSM
	92	* pages. See similar comment in change_huge_pmd.
	93	*/
	94	if (prot_numa) {
	95	struct page *page;
	96
	97	page = vm_normal_page(vma, addr, oldpte);
	98	if (!page \|\| PageKsm(page))
	99	continue;
10c1045f MG	100
	101	/* Avoid TLB flush if possible */
	102	if (pte_protnone(oldpte))
	103	continue;
3e321587 AK	104
	105	/*
	106	* Don't mess with PTEs if page is already on the node
	107	* a single-threaded process is running on.
	108	*/
	109	if (target_node == page_to_nid(page))
	110	continue;
e944fd67 MG	111	}
e944fd67 MG	112
8a0516ed MG	113	ptent = ptep_modify_prot_start(mm, addr, pte);
8a0516ed MG	114	ptent = pte_modify(ptent, newprot);
b191f9b1 MG	115	if (preserve_write)
b191f9b1 MG	116	ptent = pte_mkwrite(ptent);
4b10e7d5	117
8a0516ed MG	118	/* Avoid taking write faults for known dirty pages */
	119	if (dirty_accountable && pte_dirty(ptent) &&
	120	(pte_soft_dirty(ptent) \|\|
	121	!(vma->vm_flags & VM_SOFTDIRTY))) {
	122	ptent = pte_mkwrite(ptent);
4b10e7d5	123	}
8a0516ed MG	124	ptep_modify_prot_commit(mm, addr, pte, ptent);
8a0516ed MG	125	pages++;
0661a336	126	} else if (IS_ENABLED(CONFIG_MIGRATION)) {
0697212a CL	127	swp_entry_t entry = pte_to_swp_entry(oldpte);
	128
	129	if (is_write_migration_entry(entry)) {
c3d16e16	130	pte_t newpte;
0697212a CL	131	/*
	132	* A protection check is difficult so
	133	* just be safe and disable write
	134	*/
	135	make_migration_entry_read(&entry);
c3d16e16 CG	136	newpte = swp_entry_to_pte(entry);
	137	if (pte_swp_soft_dirty(oldpte))
	138	newpte = pte_swp_mksoft_dirty(newpte);
	139	set_pte_at(mm, addr, pte, newpte);
e920e14c MG	140
e920e14c MG	141	pages++;
0697212a	142	}
1da177e4 LT	143	}
1da177e4 LT	144	} while (pte++, addr += PAGE_SIZE, addr != end);
6606c3e0	145	arch_leave_lazy_mmu_mode();
705e87c0	146	pte_unmap_unlock(pte - 1, ptl);
7da4d641 PZ	147
7da4d641 PZ	148	return pages;
1da177e4 LT	149	}
1da177e4 LT	150
7d12efae AM	151	static inline unsigned long change_pmd_range(struct vm_area_struct *vma,
	152	pud_t *pud, unsigned long addr, unsigned long end,
	153	pgprot_t newprot, int dirty_accountable, int prot_numa)
1da177e4 LT	154	{
1da177e4 LT	155	pmd_t *pmd;
a5338093	156	struct mm_struct *mm = vma->vm_mm;
1da177e4	157	unsigned long next;
7da4d641	158	unsigned long pages = 0;
72403b4a	159	unsigned long nr_huge_updates = 0;
a5338093	160	unsigned long mni_start = 0;
1da177e4 LT	161
	162	pmd = pmd_offset(pud, addr);
	163	do {
25cbbef1 MG	164	unsigned long this_pages;
25cbbef1 MG	165
1da177e4	166	next = pmd_addr_end(addr, end);
5c7fb56e DW	167	if (!pmd_trans_huge(pmd) && !pmd_devmap(pmd)
5c7fb56e DW	168	&& pmd_none_or_clear_bad(pmd))
88a9ab6e	169	continue;
a5338093 RR	170
	171	/* invoke the mmu notifier if the pmd is populated */
	172	if (!mni_start) {
	173	mni_start = addr;
	174	mmu_notifier_invalidate_range_start(mm, mni_start, end);
	175	}
	176
5c7fb56e	177	if (pmd_trans_huge(pmd) \|\| pmd_devmap(pmd)) {
6b9116a6	178	if (next - addr != HPAGE_PMD_SIZE) {
fd60775a	179	__split_huge_pmd(vma, pmd, addr, false, NULL);
337d9abf	180	if (pmd_trans_unstable(pmd))
6b9116a6 KS	181	continue;
6b9116a6 KS	182	} else {
f123d74a	183	int nr_ptes = change_huge_pmd(vma, pmd, addr,
e944fd67	184	newprot, prot_numa);
f123d74a MG	185
f123d74a MG	186	if (nr_ptes) {
72403b4a MG	187	if (nr_ptes == HPAGE_PMD_NR) {
	188	pages += HPAGE_PMD_NR;
	189	nr_huge_updates++;
	190	}
1ad9f620 MG	191
1ad9f620 MG	192	/* huge pmd was handled */
f123d74a MG	193	continue;
f123d74a MG	194	}
7da4d641	195	}
88a9ab6e	196	/* fall through, the trans huge pmd just split */
cd7548ab	197	}
25cbbef1	198	this_pages = change_pte_range(vma, pmd, addr, next, newprot,
0f19c179	199	dirty_accountable, prot_numa);
25cbbef1	200	pages += this_pages;
1da177e4	201	} while (pmd++, addr = next, addr != end);
7da4d641	202
a5338093 RR	203	if (mni_start)
	204	mmu_notifier_invalidate_range_end(mm, mni_start, end);
	205
72403b4a MG	206	if (nr_huge_updates)
72403b4a MG	207	count_vm_numa_events(NUMA_HUGE_PTE_UPDATES, nr_huge_updates);
7da4d641	208	return pages;
1da177e4 LT	209	}
1da177e4 LT	210
7d12efae AM	211	static inline unsigned long change_pud_range(struct vm_area_struct *vma,
	212	pgd_t *pgd, unsigned long addr, unsigned long end,
	213	pgprot_t newprot, int dirty_accountable, int prot_numa)
1da177e4 LT	214	{
	215	pud_t *pud;
	216	unsigned long next;
7da4d641	217	unsigned long pages = 0;
1da177e4 LT	218
	219	pud = pud_offset(pgd, addr);
	220	do {
	221	next = pud_addr_end(addr, end);
	222	if (pud_none_or_clear_bad(pud))
	223	continue;
7da4d641	224	pages += change_pmd_range(vma, pud, addr, next, newprot,
4b10e7d5	225	dirty_accountable, prot_numa);
1da177e4	226	} while (pud++, addr = next, addr != end);
7da4d641 PZ	227
7da4d641 PZ	228	return pages;
1da177e4 LT	229	}
1da177e4 LT	230
7da4d641	231	static unsigned long change_protection_range(struct vm_area_struct *vma,
c1e6098b	232	unsigned long addr, unsigned long end, pgprot_t newprot,
4b10e7d5	233	int dirty_accountable, int prot_numa)
1da177e4 LT	234	{
	235	struct mm_struct *mm = vma->vm_mm;
	236	pgd_t *pgd;
	237	unsigned long next;
	238	unsigned long start = addr;
7da4d641	239	unsigned long pages = 0;
1da177e4 LT	240
	241	BUG_ON(addr >= end);
	242	pgd = pgd_offset(mm, addr);
	243	flush_cache_range(vma, addr, end);
20841405	244	set_tlb_flush_pending(mm);
1da177e4 LT	245	do {
	246	next = pgd_addr_end(addr, end);
	247	if (pgd_none_or_clear_bad(pgd))
	248	continue;
7da4d641	249	pages += change_pud_range(vma, pgd, addr, next, newprot,
4b10e7d5	250	dirty_accountable, prot_numa);
1da177e4	251	} while (pgd++, addr = next, addr != end);
7da4d641	252
1233d588 IM	253	/* Only flush the TLB if we actually modified any entries: */
	254	if (pages)
	255	flush_tlb_range(vma, start, end);
20841405	256	clear_tlb_flush_pending(mm);
7da4d641 PZ	257
	258	return pages;
	259	}
	260
	261	unsigned long change_protection(struct vm_area_struct *vma, unsigned long start,
	262	unsigned long end, pgprot_t newprot,
4b10e7d5	263	int dirty_accountable, int prot_numa)
7da4d641	264	{
7da4d641 PZ	265	unsigned long pages;
7da4d641 PZ	266
7da4d641 PZ	267	if (is_vm_hugetlb_page(vma))
	268	pages = hugetlb_change_protection(vma, start, end, newprot);
	269	else
4b10e7d5	270	pages = change_protection_range(vma, start, end, newprot, dirty_accountable, prot_numa);
7da4d641 PZ	271
7da4d641 PZ	272	return pages;
1da177e4 LT	273	}
1da177e4 LT	274
b6a2fea3	275	int
1da177e4 LT	276	mprotect_fixup(struct vm_area_struct vma, struct vm_area_struct *pprev,
	277	unsigned long start, unsigned long end, unsigned long newflags)
	278	{
	279	struct mm_struct *mm = vma->vm_mm;
	280	unsigned long oldflags = vma->vm_flags;
	281	long nrpages = (end - start) >> PAGE_SHIFT;
	282	unsigned long charged = 0;
1da177e4 LT	283	pgoff_t pgoff;
1da177e4 LT	284	int error;
c1e6098b	285	int dirty_accountable = 0;
1da177e4 LT	286
	287	if (newflags == oldflags) {
	288	*pprev = vma;
	289	return 0;
	290	}
	291
	292	/*
	293	* If we make a private mapping writable we increase our commit;
	294	* but (without finer accounting) cannot reduce our commit if we
5a6fe125 MG	295	* make it unwritable again. hugetlb mapping were accounted for
5a6fe125 MG	296	* even if read-only so there is no need to account for them here
1da177e4 LT	297	*/
1da177e4 LT	298	if (newflags & VM_WRITE) {
84638335 KK	299	/* Check space limits when area turns into data. */
	300	if (!may_expand_vm(mm, newflags, nrpages) &&
	301	may_expand_vm(mm, oldflags, nrpages))
	302	return -ENOMEM;
5a6fe125	303	if (!(oldflags & (VM_ACCOUNT\|VM_WRITE\|VM_HUGETLB\|
cdfd4325	304	VM_SHARED\|VM_NORESERVE))) {
1da177e4	305	charged = nrpages;
191c5424	306	if (security_vm_enough_memory_mm(mm, charged))
1da177e4 LT	307	return -ENOMEM;
	308	newflags \|= VM_ACCOUNT;
	309	}
	310	}
	311
1da177e4 LT	312	/*
	313	* First try to merge with previous and/or next vma.
	314	*/
	315	pgoff = vma->vm_pgoff + ((start - vma->vm_start) >> PAGE_SHIFT);
	316	pprev = vma_merge(mm, pprev, start, end, newflags,
19a809af AA	317	vma->anon_vma, vma->vm_file, pgoff, vma_policy(vma),
19a809af AA	318	vma->vm_userfaultfd_ctx);
1da177e4 LT	319	if (*pprev) {
1da177e4 LT	320	vma = *pprev;
e86f15ee	321	VM_WARN_ON((vma->vm_flags ^ newflags) & ~VM_SOFTDIRTY);
1da177e4 LT	322	goto success;
	323	}
	324
	325	*pprev = vma;
	326
	327	if (start != vma->vm_start) {
	328	error = split_vma(mm, vma, start, 1);
	329	if (error)
	330	goto fail;
	331	}
	332
	333	if (end != vma->vm_end) {
	334	error = split_vma(mm, vma, end, 0);
	335	if (error)
	336	goto fail;
	337	}
	338
	339	success:
	340	/*
	341	* vm_flags and vm_page_prot are protected by the mmap_sem
	342	* held in write mode.
	343	*/
	344	vma->vm_flags = newflags;
6d2329f8	345	dirty_accountable = vma_wants_writenotify(vma, vma->vm_page_prot);
64e45507	346	vma_set_page_prot(vma);
d08b3851	347
7d12efae AM	348	change_protection(vma, start, end, vma->vm_page_prot,
7d12efae AM	349	dirty_accountable, 0);
7da4d641	350
36f88188 KS	351	/*
	352	* Private VM_LOCKED VMA becoming writable: trigger COW to avoid major
	353	* fault on access.
	354	*/
	355	if ((oldflags & (VM_WRITE \| VM_SHARED \| VM_LOCKED)) == VM_LOCKED &&
	356	(newflags & VM_WRITE)) {
	357	populate_vma_page_range(vma, start, end, NULL);
	358	}
	359
84638335 KK	360	vm_stat_account(mm, oldflags, -nrpages);
84638335 KK	361	vm_stat_account(mm, newflags, nrpages);
63bfd738	362	perf_event_mmap(vma);
1da177e4 LT	363	return 0;
	364
	365	fail:
	366	vm_unacct_memory(charged);
	367	return error;
	368	}
	369
7d06d9c9 DH	370	/*
	371	* pkey==-1 when doing a legacy mprotect()
	372	*/
	373	static int do_mprotect_pkey(unsigned long start, size_t len,
	374	unsigned long prot, int pkey)
1da177e4	375	{
62b5f7d0	376	unsigned long nstart, end, tmp, reqprot;
1da177e4 LT	377	struct vm_area_struct vma, prev;
	378	int error = -EINVAL;
	379	const int grows = prot & (PROT_GROWSDOWN\|PROT_GROWSUP);
f138556d PK	380	const bool rier = (current->personality & READ_IMPLIES_EXEC) &&
	381	(prot & PROT_READ);
	382
1da177e4 LT	383	prot &= ~(PROT_GROWSDOWN\|PROT_GROWSUP);
	384	if (grows == (PROT_GROWSDOWN\|PROT_GROWSUP)) /* can't be both */
	385	return -EINVAL;
	386
	387	if (start & ~PAGE_MASK)
	388	return -EINVAL;
	389	if (!len)
	390	return 0;
	391	len = PAGE_ALIGN(len);
	392	end = start + len;
	393	if (end <= start)
	394	return -ENOMEM;
b845f313	395	if (!arch_validate_prot(prot))
1da177e4 LT	396	return -EINVAL;
	397
	398	reqprot = prot;
1da177e4	399
dc0ef0df MH	400	if (down_write_killable(&current->mm->mmap_sem))
dc0ef0df MH	401	return -EINTR;
1da177e4	402
e8c24d3a DH	403	/*
	404	* If userspace did not allocate the pkey, do not let
	405	* them use it here.
	406	*/
	407	error = -EINVAL;
	408	if ((pkey != -1) && !mm_pkey_is_allocated(current->mm, pkey))
	409	goto out;
	410
097d5910	411	vma = find_vma(current->mm, start);
1da177e4 LT	412	error = -ENOMEM;
	413	if (!vma)
	414	goto out;
097d5910	415	prev = vma->vm_prev;
1da177e4 LT	416	if (unlikely(grows & PROT_GROWSDOWN)) {
	417	if (vma->vm_start >= end)
	418	goto out;
	419	start = vma->vm_start;
	420	error = -EINVAL;
	421	if (!(vma->vm_flags & VM_GROWSDOWN))
	422	goto out;
7d12efae	423	} else {
1da177e4 LT	424	if (vma->vm_start > start)
	425	goto out;
	426	if (unlikely(grows & PROT_GROWSUP)) {
	427	end = vma->vm_end;
	428	error = -EINVAL;
	429	if (!(vma->vm_flags & VM_GROWSUP))
	430	goto out;
	431	}
	432	}
	433	if (start > vma->vm_start)
	434	prev = vma;
	435
	436	for (nstart = start ; ; ) {
a8502b67	437	unsigned long mask_off_old_flags;
1da177e4	438	unsigned long newflags;
7d06d9c9	439	int new_vma_pkey;
1da177e4	440
7d12efae	441	/* Here we know that vma->vm_start <= nstart < vma->vm_end. */
1da177e4	442
f138556d PK	443	/* Does the application expect PROT_READ to imply PROT_EXEC */
	444	if (rier && (vma->vm_flags & VM_MAYEXEC))
	445	prot \|= PROT_EXEC;
	446
a8502b67 DH	447	/*
	448	* Each mprotect() call explicitly passes r/w/x permissions.
	449	* If a permission is not passed to mprotect(), it must be
	450	* cleared from the VMA.
	451	*/
	452	mask_off_old_flags = VM_READ \| VM_WRITE \| VM_EXEC \|
	453	ARCH_VM_PKEY_FLAGS;
	454
7d06d9c9 DH	455	new_vma_pkey = arch_override_mprotect_pkey(vma, prot, pkey);
7d06d9c9 DH	456	newflags = calc_vm_prot_bits(prot, new_vma_pkey);
a8502b67	457	newflags \|= (vma->vm_flags & ~mask_off_old_flags);
1da177e4	458
7e2cff42 PBG	459	/* newflags >> 4 shift VM_MAY% in place of VM_% */
7e2cff42 PBG	460	if ((newflags & ~(newflags >> 4)) & (VM_READ \| VM_WRITE \| VM_EXEC)) {
1da177e4 LT	461	error = -EACCES;
	462	goto out;
	463	}
	464
	465	error = security_file_mprotect(vma, reqprot, prot);
	466	if (error)
	467	goto out;
	468
	469	tmp = vma->vm_end;
	470	if (tmp > end)
	471	tmp = end;
	472	error = mprotect_fixup(vma, &prev, nstart, tmp, newflags);
	473	if (error)
	474	goto out;
	475	nstart = tmp;
	476
	477	if (nstart < prev->vm_end)
	478	nstart = prev->vm_end;
	479	if (nstart >= end)
	480	goto out;
	481
	482	vma = prev->vm_next;
	483	if (!vma \|\| vma->vm_start != nstart) {
	484	error = -ENOMEM;
	485	goto out;
	486	}
f138556d	487	prot = reqprot;
1da177e4 LT	488	}
	489	out:
	490	up_write(&current->mm->mmap_sem);
	491	return error;
	492	}
7d06d9c9 DH	493
	494	SYSCALL_DEFINE3(mprotect, unsigned long, start, size_t, len,
	495	unsigned long, prot)
	496	{
	497	return do_mprotect_pkey(start, len, prot, -1);
	498	}
	499
c7142aea HC	500	#ifdef CONFIG_ARCH_HAS_PKEYS
c7142aea HC	501
7d06d9c9 DH	502	SYSCALL_DEFINE4(pkey_mprotect, unsigned long, start, size_t, len,
	503	unsigned long, prot, int, pkey)
	504	{
	505	return do_mprotect_pkey(start, len, prot, pkey);
	506	}
e8c24d3a DH	507
	508	SYSCALL_DEFINE2(pkey_alloc, unsigned long, flags, unsigned long, init_val)
	509	{
	510	int pkey;
	511	int ret;
	512
	513	/* No flags supported yet. */
	514	if (flags)
	515	return -EINVAL;
	516	/* check for unsupported init values */
	517	if (init_val & ~PKEY_ACCESS_MASK)
	518	return -EINVAL;
	519
	520	down_write(&current->mm->mmap_sem);
	521	pkey = mm_pkey_alloc(current->mm);
	522
	523	ret = -ENOSPC;
	524	if (pkey == -1)
	525	goto out;
	526
	527	ret = arch_set_user_pkey_access(current, pkey, init_val);
	528	if (ret) {
	529	mm_pkey_free(current->mm, pkey);
	530	goto out;
	531	}
	532	ret = pkey;
	533	out:
	534	up_write(&current->mm->mmap_sem);
	535	return ret;
	536	}
	537
	538	SYSCALL_DEFINE1(pkey_free, int, pkey)
	539	{
	540	int ret;
	541
	542	down_write(&current->mm->mmap_sem);
	543	ret = mm_pkey_free(current->mm, pkey);
	544	up_write(&current->mm->mmap_sem);
	545
	546	/*
	547	* We could provie warnings or errors if any VMA still
	548	* has the pkey set here.
	549	*/
	550	return ret;
	551	}
c7142aea HC	552
c7142aea HC	553	#endif /* CONFIG_ARCH_HAS_PKEYS */