Revert "Ignore madvise(MADV_WILLNEED) for hugetlbfs-backed regions"
authorLinus Torvalds <torvalds@linux-foundation.org>
Wed, 13 May 2009 15:29:12 +0000 (08:29 -0700)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 13 May 2009 15:29:12 +0000 (08:29 -0700)
This reverts commit a425a638c858fd10370b573bde81df3ba500e271.

Now that the previous commit removed the "readpage" actor for hugetlb
files, read-ahead will no longer mess up the mapping, and there's no
longer any reason to treat hugetlbfs mappings specially.

Tested-and-acked-by: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
mm/madvise.c

index 36d6ea2b63405a8bd682b1e225bb3f748de2eda5..b9ce574827c8a2a48b972f0261decc234edf9e27 100644 (file)
@@ -112,14 +112,6 @@ static long madvise_willneed(struct vm_area_struct * vma,
        if (!file)
                return -EBADF;
 
-       /*
-        * Page cache readahead assumes page cache pages are order-0 which
-        * is not the case for hugetlbfs. Do not give a bad return value
-        * but ignore the advice.
-        */
-       if (vma->vm_flags & VM_HUGETLB)
-               return 0;
-
        if (file->f_mapping->a_ops->get_xip_mem) {
                /* no bad return value, but ignore advice */
                return 0;