GSoC Idea: per-process filesystem namespaces for FreeBSD

[Dieter BSD]

Someone typed:
> A kind of related task; FreeBSD could benefit from : Fixing and
> improving unionfs / nullfs. There are some weird issues with the current
> unionfs
...
> try to unmount it in the wrong order or thrash a unionfs with
> lots of writes on top of a tmpfs and things break

Suggestion: fix the unionfs so that it works.

1) Code review.  Read the code looking for problems.

2) Hunt down whatever filesystem tests exist.  Run them.  Do they pass?

3) Are the tests complete?  Assuming that they are not:
   Complete the test suite:
   A) Functional tests (A filesystem should be able to do x, y, and z.
      Test that it can.  Include various stress tests, and things like
      a kernel panic, disconnecting a disk's data cable, various types
      of raid, with/without encryption, various newfs options, labels, ...)
   B) Code tests (Write test cases to exercise every line of code,
      and verify that you get the expected result.)

4) Check for open PRs against the filesystem.  Can your expanded test suite
   duplicate the problem?

5) Fix any problems found.

6) Don't forget the filesystem support tools, like newfs, fsck,
   dump/restore, ...

7) Is the documentation complete, and does it match the code?
   Remember that the test suite needs documentation too.

This should be a good first step towards the Plan 9 functionality.