[PARISC] Handle wrapping in expand_upwards()
authorHelge Deller <deller@gmx.de>
Tue, 19 Dec 2006 18:28:33 +0000 (19:28 +0100)
committerKyle McMartin <kyle@minerva.i.cabal.ca>
Thu, 21 Jun 2007 21:46:20 +0000 (17:46 -0400)
Function expand_upwards() did not guarded against wrapping
around to address 0. This fixes the adjtimex02 testcase from
the Linux Test Project on a 32bit PARISC kernel.

[expand_upwards is only used on parisc and ia64; it looks like it does
 the right thing on both. --kyle]

Signed-off-by: Helge Deller <deller@gmx.de>
Cc: Tony Luck <tony.luck@intel.com>
Signed-off-by: Kyle McMartin <kyle@parisc-linux.org>
mm/mmap.c

index 68b9ad2ef1d6917c28721419627fa9fd4a29e3b3..906ed402f7cabda336a73d6d0d297b862b0d2932 100644 (file)
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -1536,9 +1536,14 @@ int expand_upwards(struct vm_area_struct *vma, unsigned long address)
         * vma->vm_start/vm_end cannot change under us because the caller
         * is required to hold the mmap_sem in read mode.  We need the
         * anon_vma lock to serialize against concurrent expand_stacks.
+        * Also guard against wrapping around to address 0.
         */
-       address += 4 + PAGE_SIZE - 1;
-       address &= PAGE_MASK;
+       if (address < PAGE_ALIGN(address+4))
+               address = PAGE_ALIGN(address+4);
+       else {
+               anon_vma_unlock(vma);
+               return -ENOMEM;
+       }
        error = 0;
 
        /* Somebody else might have raced and expanded it already */