src/patches/suse-2.6.27.31/patches.fixes/ocfs2-return-0-in-page_mkwrite-to-let-VFS-retry.patch

   1 From: Tao Ma <tao.ma@oracle.com>
   2 Date: Mon, 6 Oct 2008 16:59:55 +0800
   3 Subject: ocfs2: return 0 in page_mkwrite to let VFS retry.
   4
   5 In ocfs2_page_mkwrite, we return -EINVAL when we found the page mapping
   6 isn't updated, and it will cause the user space program get SIGBUS and
   7 exit. The reason is that during race writeable mmap, we will do
   8 unmap_mapping_range in ocfs2_data_downconvert_worker. The good thing is
   9 that if we reuturn 0 in page_mkwrite, VFS will retry fault and then
  10 call page_mkwrite again, so it is safe to return 0 here.
  11
  12 Signed-off-by: Tao Ma <tao.ma@oracle.com>
  13 Signed-off-by: Mark Fasheh <mfasheh@suse.com>
  14 ---
  15  fs/ocfs2/mmap.c |    6 +++++-
  16  1 files changed, 5 insertions(+), 1 deletions(-)
  17
  18 Index: linux-2.6.27-ocfs2/fs/ocfs2/mmap.c
  19 ===================================================================
  20 --- linux-2.6.27-ocfs2.orig/fs/ocfs2/mmap.c
  21 +++ linux-2.6.27-ocfs2/fs/ocfs2/mmap.c
  22 @@ -113,7 +113,11 @@ static int __ocfs2_page_mkwrite(struct i
  23          * ocfs2_write_begin_nolock().
  24          */
  25         if (!PageUptodate(page) || page->mapping != inode->i_mapping) {
  26 -               ret = -EINVAL;
  27 +               /*
  28 +                * the page has been umapped in ocfs2_data_downconvert_worker.
  29 +                * So return 0 here and let VFS retry.
  30 +                */
  31 +               ret = 0;
  32                 goto out;
  33         }
  34