[openjdk-x] Bind to a multicast address fails

[Chris Hegarty]

Vyom,

On 12/09/18 10:09, vyom tewari wrote:
> Hi Alan,Chris,
> 
> I imported the Andre's patch and our all existing internal tests are clean.

Ok, great. This is as expected.

Next step would be to create a few tests to verify the
change. Maybe look as some existing tests that use join,
and refactor them to use bind instead. One starting point
may be the java/nio/channels/DatagramChannel
/MulticastSendReceiveTests.java. I think that if this test
was copied, that it should be relatively straight forward
to update it to test this change. Similar could be done on
the java.net side too.

-Chris.



> Thanks,
> 
> Vyom
> 
> 
> On Monday 10 September 2018 10:58 PM, Chris Hegarty wrote:
>>
>> On 10/09/18 15:35, Chris Hegarty wrote:
>>> ...
>>>
>>> Maybe the native implementation of these methods can be
>>> updated to accept a scope, rather than poking around in
>>> the InetSocketAddress from JNI code.
>>
>> Rolling back on this comment. I overlooked the fact that
>> the additional checks are multicast-specific,
>> IN6_IS_ADDR_MC_NODELOCAL and IN6_IS_ADDR_MC_LINKLOCAL.
>> Given this, I am less concerned about the proposed patch.
>>
>> -Chris.
>