Doing the VM_MERGEABLE check after the page == kpage check won't provide
any meaningful benefit.  The !vma->anon_vma check of find_mergeable_vma is
the only superfluous bit in using find_mergeable_vma because the !PageAnon
check of try_to_merge_one_page() implicitly checks for that, but it still
looks cleaner to share the same find_mergeable_vma().
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Acked-by: Hugh Dickins <hughd@google.com>
Cc: Petr Holasek <pholasek@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
        if (page == kpage)                      /* ksm page forked */
                return 0;
 
-       if (!(vma->vm_flags & VM_MERGEABLE))
-               goto out;
        if (PageTransCompound(page) && page_trans_compound_anon_split(page))
                goto out;
        BUG_ON(PageTransCompound(page));
        int err = -EFAULT;
 
        down_read(&mm->mmap_sem);
-       if (ksm_test_exit(mm))
-               goto out;
-       vma = find_vma(mm, rmap_item->address);
-       if (!vma || vma->vm_start > rmap_item->address)
+       vma = find_mergeable_vma(mm, rmap_item->address);
+       if (!vma)
                goto out;
 
        err = try_to_merge_one_page(vma, page, kpage);