<net-dev> Java_java_net_Inet6AddressImpl_isReachable0 is returning false for InetAdress 0.0.0.0
Deven You
youdwei at linux.vnet.ibm.com
Tue May 8 20:36:26 PDT 2012
Hi Alan,
Since this patch is for Unix like platforms, I didn't test the test case
on Windows yet.
However when I tested it on Windows I found it will fail on Windows.
I also use ping command on Windows to test both 0.0.0.0 and ::0, they
all fail. So I want to know if we need further investigation to see why
these 2 addresses can not be reachable or we just think it is the proper
behavior because JDK's behavior is consistent with ping on Windows?
Thanks a lot!
On 05/07/2012 03:46 PM, Alan Bateman wrote:
> On 07/05/2012 03:29, Deven You wrote:
>> Hi Alan and Chris,
>>
>> I have updated the webrev[1] according to your suggestions.
>>
>> For the test case, I did two things:
>>
>> 1. add the @run main/othervm -Djava.net.preferIPv4Stack=true PingThis
>>
>> 2. Only when !preferIPv4Stack and hasIPv6(), ::0 will be added into
>> the address list.
>>
>> I think it's enough for the test case, please review it.
>>
>> I just have another concern, how we can configure the test env so
>> this test case can be run under root (may use sudo) privilege?
>>
>> [1] http://cr.openjdk.java.net/~littlee/OJDK-217/webrev.04/
>> <http://cr.openjdk.java.net/%7Elittlee/OJDK-217/webrev.04/>
> I think we are close to the finish line on this one.
>
> One thing I notice in the webrev is that net_util_md.h hasn't been
> updated to defined the prototype for NET_IsZeroAddr. It would be good
> to add that (no need to generate another webrev just for this one).
>
> The test case looks okay to me. Minor comment is that the while loop
> could be replaced with for (String: addr: addrs) { ...}. No need to
> re-generate the webrev if you decide to take up this suggestion.
>
> I don't think we should change the test to attempt to run it via sudo,
> that would just complicate the test as there is no guarantee that the
> user running the tests is in sudoers. Also one wouldn't want the test
> prompting for a password. it would also require a script as it would
> need to be skipped on platforms such as Windows.
>
> So I think I'm happy with the test as is and I assume you will run
> with regular user/root before pushing. I also assume you will run it
> on Windows to make sure that it passes there too.
>
> -Alan.
--
Best Regards,
Deven
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/net-dev/attachments/20120509/11ddd800/attachment.html
More information about the net-dev
mailing list