do_mremap() untangling, part 3
authorAl Viro <viro@zeniv.linux.org.uk>
Tue, 24 Nov 2009 12:43:18 +0000 (07:43 -0500)
committerAl Viro <viro@zeniv.linux.org.uk>
Fri, 11 Dec 2009 11:30:22 +0000 (06:30 -0500)
Take the check for being able to expand vma in place into a separate
helper.

Acked-by: Russell King <rmk+kernel@arm.linux.org.uk>
Acked-by: Hugh Dickins <hugh.dickins@tiscali.co.uk>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
mm/mremap.c

index 5f346178f16f5492034f59ad6c40e5584d40109c..90e422c9f410462652ca06d56b19f404752dd515 100644 (file)
@@ -366,6 +366,17 @@ out:
        return ret;
 }
 
+static int vma_expandable(struct vm_area_struct *vma, unsigned long delta)
+{
+       unsigned long max_addr = TASK_SIZE;
+       if (vma->vm_next)
+               max_addr = vma->vm_next->vm_start;
+       if (max_addr - vma->vm_end < delta)
+               return 0;
+       /* we need to do arch-specific checks here */
+       return 1;
+}
+
 /*
  * Expand (or shrink) an existing mapping, potentially moving it at the
  * same time (controlled by the MREMAP_MAYMOVE flag and available VM space)
@@ -430,11 +441,8 @@ unsigned long do_mremap(unsigned long addr,
        /* old_len exactly to the end of the area..
         */
        if (old_len == vma->vm_end - addr) {
-               unsigned long max_addr = TASK_SIZE;
-               if (vma->vm_next)
-                       max_addr = vma->vm_next->vm_start;
                /* can we just expand the current mapping? */
-               if (max_addr - addr >= new_len) {
+               if (vma_expandable(vma, new_len - old_len)) {
                        int pages = (new_len - old_len) >> PAGE_SHIFT;
 
                        vma_adjust(vma, vma->vm_start,