[Bug 193246] Bug in IPv6 multicast join(), uncovered by Jenkins

[Andrey V. Elsukov]

On 03.09.2014 17:39, Craig Rodrigues wrote:
> On Wed, Sep 3, 2014 at 5:19 AM, Andrey V. Elsukov <ae at freebsd.org> wrote:
>> On 03.09.2014 14:05, bugzilla-noreply at freebsd.org wrote:
>>
>> Hi,
>>
>> you said that this code works in linux. I looked in the linux kernel
>> source, and I think it should return EINVAL too.
>> net/ipv6/mcast.c:ipv6_sock_mc_join:
>>
>>  154         if (!ipv6_addr_is_multicast(addr))
>>  155                 return -EINVAL;
> 
> 
> The code does work in Linux.  However, you need to look at the
> JDK source, not the Linux kernel source.
> 
> In this file:
> http://hg.openjdk.java.net/jdk7/jdk7/jdk/file/9b8c96f96a0f/src/solaris/native/java/net/PlainDatagramSocketImpl.c
> 
> in the mcast_join_leave() function, there are two code paths: (1)
> Linux, (2) Solaris.
> 
> It looks like on Solaris, they support IPv4-mapped multicast addresses for IPV6,
> and things work when they create an IPv6 socket, and then put an
> IPv4-mapped multicast address in it.  For Linux, they have specific
> code paths in that function which seem to force creating an IPv4
> socket.

Yes, I have illumos-gate's code and I can confirm, that IPv4-mapped IPv6
addresses have special handling in multicast code.

-- 
WBR, Andrey V. Elsukov