RFR: 8080819: Inet4AddressImpl regression caused by JDK-7180557

[Rob McKenna]

Thanks Michael,

A test would be pretty tricky given the configuration specific nature of 
the problem. (Mac, VPN etc.)

	-Rob

On 03/06/15 16:22, Michael McMahon wrote:
> Seems reasonable Rob. I guess a test isn't feasible?
>
> Michael.
>
> On 03/06/15 14:25, Rob McKenna wrote:
>> Hi folks,
>>
>> 7180557 used getifaddrs as a way of determining the local hosts ip
>> address on Mac OSX in order to fix a problem with OSX's naming system.
>> When fixing this we decided to place that call before the call to
>> getaddrinfo thus taking the naming system out of the equation.
>>
>> Unfortunately when running over VPN we return the wrong ifaddr causing
>> a failure.
>>
>> The solution is to place the getifaddrs call below getaddrinfo and to
>> only call it if there is an error with getaddrinfo as there was in the
>> original (7180557) case.
>>
>> http://cr.openjdk.java.net/~robm/8080819/webrev.01/
>>
>>     -Rob
>