Author: Nathaniel Grill

This xv6 file system checker performs the following checks when provided a file
system image:
    -each inode is unallocated or a valid type
    -each address used by a valid inode points to a datablock within the image
    -root directory exists, is properly numbered, and points to itself
    -each directory contains a . and .. entry, with . pointing to itself
    -each address used by a valid inode is marked in-use in the bitmap
    -each block marked in-use in the bitmap is actually in use somewhere
    -each direct address is only used once
    -each indirect address is only used once
    -each inode marked in-use is referred to in at least one directory
    -each inode referred to in a valid directory is actually marked in use
    -reference counts for regular files = number of times file is referred to
    -each directory only appears in one other directory

Additionally, the following extra checks are performed as well:
    -each .. entry in a directory refers to the proper parent node
    -there are no loops in the directory tree

Finally, the file system checker will repair an image that contains lost inodes
  (i.e. an inode is marked in-use but not found in a directory). Each lost inode
  is placed in a lost and found directory (titled lost_found). An image will
  only be repaired if the '-r' flag is specified. Any other flag will cause the
  checker to exit without doing anything.

Implementing the checker generally involved looping through different aspects of
  the file system (inodes, directories, datablocks, etc.), performing multiple
  checks on each part.

Overall, this impementation could be vastly more efficient, particularly with
  checks 6-8, which each loop through all inodes. On the contrary, checks 1-5
  and 9-12 are performed, respectively, on an individual inode before moving on
  to the next one. Efficiency could perhaps also be found by using data
  structures with faster lookups (e.g. hash tables) rather than looping through
  lists.