RFR [9] 8177536: Avoid Apple Peer-to-Peer interfaces in networking tests

[Chris Hegarty]

> On 12 Apr 2017, at 16:17, Roger Riggs <Roger.Riggs at oracle.com> wrote:
> ..
> B6558853.java: 28:  an extra "*"
> 
> A few tests do nothing unless an interface is found; they could use either output that says what
> interface is tested or a message that no interface was available.
> (In case someone is reviewing the logs and wants to know what was tested).

Thanks. Done.

> On 13 Apr 2017, at 12:15, Michael McMahon <michael.x.mcmahon at oracle.com> wrote:
> 
> Nice tidy up Chris. I guess we can add additional variants later to
> NetworkConfiguration::probe which return interfaces that are point to point or are down etc.

Sure. I expect this infrastructure to evolve over time. I just what
was needed for now.

> Also, trivially, the two tests os system property "os.name" could be made consistent.
> I think it's one that is always guaranteed to be set, or if not, then both accesses could
> return the same "unknown" string.

Done .

I added a non-failing “test" that captures the basic network configuration.
This can be used when analysing any failures in the :jdk_net group, and
can be expanded to capture more exotic configuration over time.

Updated webrev:
  http://cr.openjdk.java.net/~chegar/8177536.01/

-Chris