mm: fix ENOSPC returned by handle_mm_fault()
authorHugh Dickins <hughd@google.com>
Mon, 6 Jun 2011 05:03:13 +0000 (22:03 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Mon, 6 Jun 2011 09:00:27 +0000 (18:00 +0900)
Al Viro observes that in the hugetlb case, handle_mm_fault() may return
a value of the kind ENOSPC when its caller is expecting a value of the
kind VM_FAULT_SIGBUS: fix alloc_huge_page()'s failure returns.

Signed-off-by: Hugh Dickins <hughd@google.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/hugetlb.c

index f33bb319b73f093ce38e828a265b24d9c18e0525..6402458fee38ca5ba2109882e92ebbf63e97c787 100644 (file)
@@ -1033,10 +1033,10 @@ static struct page *alloc_huge_page(struct vm_area_struct *vma,
         */
        chg = vma_needs_reservation(h, vma, addr);
        if (chg < 0)
-               return ERR_PTR(chg);
+               return ERR_PTR(-VM_FAULT_OOM);
        if (chg)
                if (hugetlb_get_quota(inode->i_mapping, chg))
-                       return ERR_PTR(-ENOSPC);
+                       return ERR_PTR(-VM_FAULT_SIGBUS);
 
        spin_lock(&hugetlb_lock);
        page = dequeue_huge_page_vma(h, vma, addr, avoid_reserve);