mm/mmap.c: do not blow on PROT_NONE MAP_FIXED holes in the stack
authorMichal Hocko <mhocko@suse.com>
Mon, 10 Jul 2017 22:49:51 +0000 (15:49 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 10 Jul 2017 23:32:33 +0000 (16:32 -0700)
Commit 1be7107fbe18 ("mm: larger stack guard gap, between vmas") has
introduced a regression in some rust and Java environments which are
trying to implement their own stack guard page.  They are punching a new
MAP_FIXED mapping inside the existing stack Vma.

This will confuse expand_{downwards,upwards} into thinking that the
stack expansion would in fact get us too close to an existing non-stack
vma which is a correct behavior wrt safety.  It is a real regression on
the other hand.

Let's work around the problem by considering PROT_NONE mapping as a part
of the stack.  This is a gros hack but overflowing to such a mapping
would trap anyway an we only can hope that usespace knows what it is
doing and handle it propely.

Fixes: 1be7107fbe18 ("mm: larger stack guard gap, between vmas")
Link: http://lkml.kernel.org/r/20170705182849.GA18027@dhcp22.suse.cz
Signed-off-by: Michal Hocko <mhocko@suse.com>
Debugged-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Ben Hutchings <ben@decadent.org.uk>
Cc: Willy Tarreau <w@1wt.eu>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Hugh Dickins <hughd@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/mmap.c

index 7f8cfe9..d190287 100644 (file)
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -2244,7 +2244,8 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
                gap_addr = TASK_SIZE;
 
        next = vma->vm_next;
-       if (next && next->vm_start < gap_addr) {
+       if (next && next->vm_start < gap_addr &&
+                       (next->vm_flags & (VM_WRITE|VM_READ|VM_EXEC))) {
                if (!(next->vm_flags & VM_GROWSUP))
                        return -ENOMEM;
                /* Check that both stack segments have the same anon_vma? */
@@ -2328,7 +2329,8 @@ int expand_downwards(struct vm_area_struct *vma,
        if (gap_addr > address)
                return -ENOMEM;
        prev = vma->vm_prev;
-       if (prev && prev->vm_end > gap_addr) {
+       if (prev && prev->vm_end > gap_addr &&
+                       (prev->vm_flags & (VM_WRITE|VM_READ|VM_EXEC))) {
                if (!(prev->vm_flags & VM_GROWSDOWN))
                        return -ENOMEM;
                /* Check that both stack segments have the same anon_vma? */