moving to the right side
- ext3_ext_find_goal() takes into account offset of a block to be allocated
  (thanks to Andreas for smart idea)

  cache-aliasing problems
- ext3_forget() should be used in conjuction with ext3_free_blocks()

NOTE: 2.4.24 + extents + batching block allocation passes data-consistency
      tests now (at least I haven't seen problems for many hours of testing)

  possible deadlock
- locking in extent's ioctl added
- ext3_ext_walk_space() drops bh's references in path before finding
  next extent

NOTE: dbench and syntetic tests work OK in vmware for many hours

  - fixed goal selection
  - more bits of batching allocation

- bug in extents fixed: it could cause tree corruption
- minor bug in extents fixed: during tree growing it could possible
  to get suboptimal size for index block

- support for extents in EA
- access to extents stored in EA via ioctl