xfs: xfs_mru_cache_insert() should use GFP_NOFS

xfs_mru_cache_insert() can be called from within transaction context
during block allocation like so:

write(2)
  ....
    xfs_get_blocks
      xfs_iomap_write_direct
        start transaction
        xfs_bmapi_write
          xfs_bmapi_allocate
            xfs_bmap_btalloc
              xfs_bmap_btalloc_filestreams
                xfs_filestream_new_ag
                  xfs_filestream_pick_ag
                    xfs_mru_cache_insert
                      radix_tree_preload(GFP_KERNEL)

In this case, GFP_KERNEL is incorrect and can potentially lead to
deadlocks in memory reclaim. It should use GFP_NOFS allocations to
avoid lock recursion problems.

[dchinner: rewrote commit message]

Signed-off-by: Byoungyoung Lee <blee@gatech.edu>
Signed-off-by: Sanidhya Kashyap <sanidhya.gatech@gmail.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Signed-off-by: Dave Chinner <david@fromorbit.com>
This commit is contained in:
Byoungyoung Lee 2015-03-25 14:57:53 +11:00 committed by Dave Chinner
parent 65dd297ac2
commit 20dafeefac

View file

@ -437,7 +437,7 @@ xfs_mru_cache_insert(
if (!mru || !mru->lists)
return -EINVAL;
if (radix_tree_preload(GFP_KERNEL))
if (radix_tree_preload(GFP_NOFS))
return -ENOMEM;
INIT_LIST_HEAD(&elem->list_node);