mm: larger stack guard gap, between vmas

commit 1be7107fbe18eed3e319a6c3e83c78254b693acb upstream.

Stack guard page is a useful feature to reduce a risk of stack smashing
into a different mapping. We have been using a single page gap which
is sufficient to prevent having stack adjacent to a different mapping.
But this seems to be insufficient in the light of the stack usage in
userspace. E.g. glibc uses as large as 64kB alloca() in many commonly
used functions. Others use constructs liks gid_t buffer[NGROUPS_MAX]
which is 256kB or stack strings with MAX_ARG_STRLEN.

This will become especially dangerous for suid binaries and the default
no limit for the stack size limit because those applications can be
tricked to consume a large portion of the stack and a single glibc call
could jump over the guard page. These attacks are not theoretical,
unfortunatelly.

Make those attacks less probable by increasing the stack guard gap
to 1MB (on systems with 4k pages; but make it depend on the page size
because systems with larger base pages might cap stack allocations in
the PAGE_SIZE units) which should cover larger alloca() and VLA stack
allocations. It is obviously not a full fix because the problem is
somehow inherent, but it should reduce attack space a lot.

One could argue that the gap size should be configurable from userspace,
but that can be done later when somebody finds that the new 1MB is wrong
for some special case applications.  For now, add a kernel command line
option (stack_guard_gap) to specify the stack gap size (in page units).

Implementation wise, first delete all the old code for stack guard page:
because although we could get away with accounting one extra page in a
stack vma, accounting a larger gap can break userspace - case in point,
a program run with "ulimit -S -v 20000" failed when the 1MB gap was
counted for RLIMIT_AS; similar problems could come with RLIMIT_MLOCK
and strict non-overcommit mode.

Instead of keeping gap inside the stack vma, maintain the stack guard
gap as a gap between vmas: using vm_start_gap() in place of vm_start
(or vm_end_gap() in place of vm_end if VM_GROWSUP) in just those few
places which need to respect the gap - mainly arch_get_unmapped_area(),
and and the vma tree's subtree_gap support for that.

Original-patch-by: Oleg Nesterov <oleg@redhat.com>
Original-patch-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Tested-by: Helge Deller <deller@gmx.de> # parisc
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
[wt: backport to 4.11: adjust context]
[wt: backport to 4.9: adjust context ; kernel doc was not in admin-guide]
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

3 files changed, 89 insertions, 103 deletions

diff --git a/mm/gup.c b/mm/gup.c
index ec4f82704b6f..c63a0341ae38 100644
--- a/mm/gup.c
+++ b/mm/gup.c
@@ -370,11 +370,6 @@ static int faultin_page(struct task_struct *tsk, struct vm_area_struct *vma,
 	/* mlock all present pages, but do not fault in new pages */
 	if ((*flags & (FOLL_POPULATE | FOLL_MLOCK)) == FOLL_MLOCK)
 		return -ENOENT;
-	/* For mm_populate(), just skip the stack guard page. */
-	if ((*flags & FOLL_POPULATE) &&
-			(stack_guard_page_start(vma, address) ||
-			 stack_guard_page_end(vma, address + PAGE_SIZE)))
-		return -ENOENT;
 	if (*flags & FOLL_WRITE)
 		fault_flags |= FAULT_FLAG_WRITE;
 	if (*flags & FOLL_REMOTE)
diff --git a/mm/memory.c b/mm/memory.c
index cbb1e5e5f791..e6a5a1f20492 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -2699,40 +2699,6 @@ out_release:
 }
 
 /*
- * This is like a special single-page "expand_{down|up}wards()",
- * except we must first make sure that 'address{-|+}PAGE_SIZE'
- * doesn't hit another vma.
- */
-static inline int check_stack_guard_page(struct vm_area_struct *vma, unsigned long address)
-{
-	address &= PAGE_MASK;
-	if ((vma->vm_flags & VM_GROWSDOWN) && address == vma->vm_start) {
-		struct vm_area_struct *prev = vma->vm_prev;
-
-		/*
-		 * Is there a mapping abutting this one below?
-		 *
-		 * That's only ok if it's the same stack mapping
-		 * that has gotten split..
-		 */
-		if (prev && prev->vm_end == address)
-			return prev->vm_flags & VM_GROWSDOWN ? 0 : -ENOMEM;
-
-		return expand_downwards(vma, address - PAGE_SIZE);
-	}
-	if ((vma->vm_flags & VM_GROWSUP) && address + PAGE_SIZE == vma->vm_end) {
-		struct vm_area_struct *next = vma->vm_next;
-
-		/* As VM_GROWSDOWN but s/below/above/ */
-		if (next && next->vm_start == address + PAGE_SIZE)
-			return next->vm_flags & VM_GROWSUP ? 0 : -ENOMEM;
-
-		return expand_upwards(vma, address + PAGE_SIZE);
-	}
-	return 0;
-}
-
-/*
  * We enter with non-exclusive mmap_sem (to exclude vma changes,
  * but allow concurrent faults), and pte mapped but not yet locked.
  * We return with mmap_sem still held, but pte unmapped and unlocked.
@@ -2748,10 +2714,6 @@ static int do_anonymous_page(struct fault_env *fe)
 	if (vma->vm_flags & VM_SHARED)
 		return VM_FAULT_SIGBUS;
 
-	/* Check if we need to add a guard page to the stack */
-	if (check_stack_guard_page(vma, fe->address) < 0)
-		return VM_FAULT_SIGSEGV;
-
 	/*
 	 * Use pte_alloc() instead of pte_alloc_map().  We can't run
 	 * pte_offset_map() on pmds where a huge pmd might be created
diff --git a/mm/mmap.c b/mm/mmap.c
index 1af87c14183d..26542b346229 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -183,6 +183,7 @@ SYSCALL_DEFINE1(brk, unsigned long, brk)
 	unsigned long retval;
 	unsigned long newbrk, oldbrk;
 	struct mm_struct *mm = current->mm;
+	struct vm_area_struct *next;
 	unsigned long min_brk;
 	bool populate;
 
@@ -228,7 +229,8 @@ SYSCALL_DEFINE1(brk, unsigned long, brk)
 	}
 
 	/* Check against existing mmap mappings. */
-	if (find_vma_intersection(mm, oldbrk, newbrk+PAGE_SIZE))
+	next = find_vma(mm, oldbrk);
+	if (next && newbrk + PAGE_SIZE > vm_start_gap(next))
 		goto out;
 
 	/* Ok, looks good - let it rip. */
@@ -251,10 +253,22 @@ out:
 
 static long vma_compute_subtree_gap(struct vm_area_struct *vma)
 {
-	unsigned long max, subtree_gap;
-	max = vma->vm_start;
-	if (vma->vm_prev)
-		max -= vma->vm_prev->vm_end;
+	unsigned long max, prev_end, subtree_gap;
+
+	/*
+	 * Note: in the rare case of a VM_GROWSDOWN above a VM_GROWSUP, we
+	 * allow two stack_guard_gaps between them here, and when choosing
+	 * an unmapped area; whereas when expanding we only require one.
+	 * That's a little inconsistent, but keeps the code here simpler.
+	 */
+	max = vm_start_gap(vma);
+	if (vma->vm_prev) {
+		prev_end = vm_end_gap(vma->vm_prev);
+		if (max > prev_end)
+			max -= prev_end;
+		else
+			max = 0;
+	}
 	if (vma->vm_rb.rb_left) {
 		subtree_gap = rb_entry(vma->vm_rb.rb_left,
 				struct vm_area_struct, vm_rb)->rb_subtree_gap;
@@ -350,7 +364,7 @@ static void validate_mm(struct mm_struct *mm)
 			anon_vma_unlock_read(anon_vma);
 		}
 
-		highest_address = vma->vm_end;
+		highest_address = vm_end_gap(vma);
 		vma = vma->vm_next;
 		i++;
 	}
@@ -539,7 +553,7 @@ void __vma_link_rb(struct mm_struct *mm, struct vm_area_struct *vma,
 	if (vma->vm_next)
 		vma_gap_update(vma->vm_next);
 	else
-		mm->highest_vm_end = vma->vm_end;
+		mm->highest_vm_end = vm_end_gap(vma);
 
 	/*
 	 * vma->vm_prev wasn't known when we followed the rbtree to find the
@@ -854,7 +868,7 @@ again:
 			vma_gap_update(vma);
 		if (end_changed) {
 			if (!next)
-				mm->highest_vm_end = end;
+				mm->highest_vm_end = vm_end_gap(vma);
 			else if (!adjust_next)
 				vma_gap_update(next);
 		}
@@ -939,7 +953,7 @@ again:
 			 * mm->highest_vm_end doesn't need any update
 			 * in remove_next == 1 case.
 			 */
-			VM_WARN_ON(mm->highest_vm_end != end);
+			VM_WARN_ON(mm->highest_vm_end != vm_end_gap(vma));
 		}
 	}
 	if (insert && file)
@@ -1783,7 +1797,7 @@ unsigned long unmapped_area(struct vm_unmapped_area_info *info)
 
 	while (true) {
 		/* Visit left subtree if it looks promising */
-		gap_end = vma->vm_start;
+		gap_end = vm_start_gap(vma);
 		if (gap_end >= low_limit && vma->vm_rb.rb_left) {
 			struct vm_area_struct *left =
 				rb_entry(vma->vm_rb.rb_left,
@@ -1794,7 +1808,7 @@ unsigned long unmapped_area(struct vm_unmapped_area_info *info)
 			}
 		}
 
-		gap_start = vma->vm_prev ? vma->vm_prev->vm_end : 0;
+		gap_start = vma->vm_prev ? vm_end_gap(vma->vm_prev) : 0;
 check_current:
 		/* Check if current node has a suitable gap */
 		if (gap_start > high_limit)
@@ -1821,8 +1835,8 @@ check_current:
 			vma = rb_entry(rb_parent(prev),
 				       struct vm_area_struct, vm_rb);
 			if (prev == vma->vm_rb.rb_left) {
-				gap_start = vma->vm_prev->vm_end;
-				gap_end = vma->vm_start;
+				gap_start = vm_end_gap(vma->vm_prev);
+				gap_end = vm_start_gap(vma);
 				goto check_current;
 			}
 		}
@@ -1886,7 +1900,7 @@ unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info)
 
 	while (true) {
 		/* Visit right subtree if it looks promising */
-		gap_start = vma->vm_prev ? vma->vm_prev->vm_end : 0;
+		gap_start = vma->vm_prev ? vm_end_gap(vma->vm_prev) : 0;
 		if (gap_start <= high_limit && vma->vm_rb.rb_right) {
 			struct vm_area_struct *right =
 				rb_entry(vma->vm_rb.rb_right,
@@ -1899,7 +1913,7 @@ unsigned long unmapped_area_topdown(struct vm_unmapped_area_info *info)
 
 check_current:
 		/* Check if current node has a suitable gap */
-		gap_end = vma->vm_start;
+		gap_end = vm_start_gap(vma);
 		if (gap_end < low_limit)
 			return -ENOMEM;
 		if (gap_start <= high_limit && gap_end - gap_start >= length)
@@ -1925,7 +1939,7 @@ check_current:
 				       struct vm_area_struct, vm_rb);
 			if (prev == vma->vm_rb.rb_right) {
 				gap_start = vma->vm_prev ?
-					vma->vm_prev->vm_end : 0;
+					vm_end_gap(vma->vm_prev) : 0;
 				goto check_current;
 			}
 		}
@@ -1963,7 +1977,7 @@ arch_get_unmapped_area(struct file *filp, unsigned long addr,
 		unsigned long len, unsigned long pgoff, unsigned long flags)
 {
 	struct mm_struct *mm = current->mm;
-	struct vm_area_struct *vma;
+	struct vm_area_struct *vma, *prev;
 	struct vm_unmapped_area_info info;
 
 	if (len > TASK_SIZE - mmap_min_addr)
@@ -1974,9 +1988,10 @@ arch_get_unmapped_area(struct file *filp, unsigned long addr,
 
 	if (addr) {
 		addr = PAGE_ALIGN(addr);
-		vma = find_vma(mm, addr);
+		vma = find_vma_prev(mm, addr, &prev);
 		if (TASK_SIZE - len >= addr && addr >= mmap_min_addr &&
-		    (!vma || addr + len <= vma->vm_start))
+		    (!vma || addr + len <= vm_start_gap(vma)) &&
+		    (!prev || addr >= vm_end_gap(prev)))
 			return addr;
 	}
 
@@ -1999,7 +2014,7 @@ arch_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
 			  const unsigned long len, const unsigned long pgoff,
 			  const unsigned long flags)
 {
-	struct vm_area_struct *vma;
+	struct vm_area_struct *vma, *prev;
 	struct mm_struct *mm = current->mm;
 	unsigned long addr = addr0;
 	struct vm_unmapped_area_info info;
@@ -2014,9 +2029,10 @@ arch_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
 	/* requesting a specific address */
 	if (addr) {
 		addr = PAGE_ALIGN(addr);
-		vma = find_vma(mm, addr);
+		vma = find_vma_prev(mm, addr, &prev);
 		if (TASK_SIZE - len >= addr && addr >= mmap_min_addr &&
-				(!vma || addr + len <= vma->vm_start))
+				(!vma || addr + len <= vm_start_gap(vma)) &&
+				(!prev || addr >= vm_end_gap(prev)))
 			return addr;
 	}
 
@@ -2151,21 +2167,19 @@ find_vma_prev(struct mm_struct *mm, unsigned long addr,
  * update accounting. This is shared with both the
  * grow-up and grow-down cases.
  */
-static int acct_stack_growth(struct vm_area_struct *vma, unsigned long size, unsigned long grow)
+static int acct_stack_growth(struct vm_area_struct *vma,
+			     unsigned long size, unsigned long grow)
 {
 	struct mm_struct *mm = vma->vm_mm;
 	struct rlimit *rlim = current->signal->rlim;
-	unsigned long new_start, actual_size;
+	unsigned long new_start;
 
 	/* address space limit tests */
 	if (!may_expand_vm(mm, vma->vm_flags, grow))
 		return -ENOMEM;
 
 	/* Stack limit test */
-	actual_size = size;
-	if (size && (vma->vm_flags & (VM_GROWSUP | VM_GROWSDOWN)))
-		actual_size -= PAGE_SIZE;
-	if (actual_size > READ_ONCE(rlim[RLIMIT_STACK].rlim_cur))
+	if (size > READ_ONCE(rlim[RLIMIT_STACK].rlim_cur))
 		return -ENOMEM;
 
 	/* mlock limit tests */
@@ -2203,17 +2217,30 @@ static int acct_stack_growth(struct vm_area_struct *vma, unsigned long size, uns
 int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 {
 	struct mm_struct *mm = vma->vm_mm;
+	struct vm_area_struct *next;
+	unsigned long gap_addr;
 	int error = 0;
 
 	if (!(vma->vm_flags & VM_GROWSUP))
 		return -EFAULT;
 
 	/* Guard against wrapping around to address 0. */
-	if (address < PAGE_ALIGN(address+4))
-		address = PAGE_ALIGN(address+4);
-	else
+	address &= PAGE_MASK;
+	address += PAGE_SIZE;
+	if (!address)
 		return -ENOMEM;
 
+	/* Enforce stack_guard_gap */
+	gap_addr = address + stack_guard_gap;
+	if (gap_addr < address)
+		return -ENOMEM;
+	next = vma->vm_next;
+	if (next && next->vm_start < gap_addr) {
+		if (!(next->vm_flags & VM_GROWSUP))
+			return -ENOMEM;
+		/* Check that both stack segments have the same anon_vma? */
+	}
+
 	/* We must make sure the anon_vma is allocated. */
 	if (unlikely(anon_vma_prepare(vma)))
 		return -ENOMEM;
@@ -2257,7 +2284,7 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
 				if (vma->vm_next)
 					vma_gap_update(vma->vm_next);
 				else
-					mm->highest_vm_end = address;
+					mm->highest_vm_end = vm_end_gap(vma);
 				spin_unlock(&mm->page_table_lock);
 
 				perf_event_mmap(vma);
@@ -2278,6 +2305,8 @@ int expand_downwards(struct vm_area_struct *vma,
 				   unsigned long address)
 {
 	struct mm_struct *mm = vma->vm_mm;
+	struct vm_area_struct *prev;
+	unsigned long gap_addr;
 	int error;
 
 	address &= PAGE_MASK;
@@ -2285,6 +2314,17 @@ int expand_downwards(struct vm_area_struct *vma,
 	if (error)
 		return error;
 
+	/* Enforce stack_guard_gap */
+	gap_addr = address - stack_guard_gap;
+	if (gap_addr > address)
+		return -ENOMEM;
+	prev = vma->vm_prev;
+	if (prev && prev->vm_end > gap_addr) {
+		if (!(prev->vm_flags & VM_GROWSDOWN))
+			return -ENOMEM;
+		/* Check that both stack segments have the same anon_vma? */
+	}
+
 	/* We must make sure the anon_vma is allocated. */
 	if (unlikely(anon_vma_prepare(vma)))
 		return -ENOMEM;
@@ -2339,28 +2379,25 @@ int expand_downwards(struct vm_area_struct *vma,
 	return error;
 }
 
-/*
- * Note how expand_stack() refuses to expand the stack all the way to
- * abut the next virtual mapping, *unless* that mapping itself is also
- * a stack mapping. We want to leave room for a guard page, after all
- * (the guard page itself is not added here, that is done by the
- * actual page faulting logic)
- *
- * This matches the behavior of the guard page logic (see mm/memory.c:
- * check_stack_guard_page()), which only allows the guard page to be
- * removed under these circumstances.
- */
+/* enforced gap between the expanding stack and other mappings. */
+unsigned long stack_guard_gap = 256UL<<PAGE_SHIFT;
+
+static int __init cmdline_parse_stack_guard_gap(char *p)
+{
+	unsigned long val;
+	char *endptr;
+
+	val = simple_strtoul(p, &endptr, 10);
+	if (!*endptr)
+		stack_guard_gap = val << PAGE_SHIFT;
+
+	return 0;
+}
+__setup("stack_guard_gap=", cmdline_parse_stack_guard_gap);
+
 #ifdef CONFIG_STACK_GROWSUP
 int expand_stack(struct vm_area_struct *vma, unsigned long address)
 {
-	struct vm_area_struct *next;
-
-	address &= PAGE_MASK;
-	next = vma->vm_next;
-	if (next && next->vm_start == address + PAGE_SIZE) {
-		if (!(next->vm_flags & VM_GROWSUP))
-			return -ENOMEM;
-	}
 	return expand_upwards(vma, address);
 }
 
@@ -2382,14 +2419,6 @@ find_extend_vma(struct mm_struct *mm, unsigned long addr)
 #else
 int expand_stack(struct vm_area_struct *vma, unsigned long address)
 {
-	struct vm_area_struct *prev;
-
-	address &= PAGE_MASK;
-	prev = vma->vm_prev;
-	if (prev && prev->vm_end == address) {
-		if (!(prev->vm_flags & VM_GROWSDOWN))
-			return -ENOMEM;
-	}
 	return expand_downwards(vma, address);
 }
 
@@ -2487,7 +2516,7 @@ detach_vmas_to_be_unmapped(struct mm_struct *mm, struct vm_area_struct *vma,
 		vma->vm_prev = prev;
 		vma_gap_update(vma);
 	} else
-		mm->highest_vm_end = prev ? prev->vm_end : 0;
+		mm->highest_vm_end = prev ? vm_end_gap(prev) : 0;
 	tail_vma->vm_next = NULL;
 
 	/* Kill the cache */