[CRIU] [PATCH 4/5] inet: remember ipv6 connections' ifindex for restore

[Andrey Wagin]

2015-11-24 18:39 GMT+03:00 Tycho Andersen <tycho.andersen at canonical.com>:
> On Tue, Nov 24, 2015 at 02:08:16PM +0300, Andrew Vagin wrote:
>> On Mon, Nov 23, 2015 at 01:55:27PM -0700, Tycho Andersen wrote:
>> > For some ipv6 scope types (link local, amongo others), we need to tell the
>> > kernel the ifindex of the interface whose address we want to bind to as
>> > well as the address itself.
>> >
>>
>> Can we use SO_BINDTODEVICE to get this information?
>
> Oh, probably. I didn't know about this flag. That would make things
> much simpler. I'll check and see.
>
>> diff --git a/test/zdtm/live/static/bridge.c b/test/zdtm/live/static/bridge.c
>> index 06bff50..5bb9c40 100644
>> --- a/test/zdtm/live/static/bridge.c
>> +++ b/test/zdtm/live/static/bridge.c
>> @@ -77,6 +77,13 @@ int main(int argc, char **argv)
>>                 goto out;
>>         }
>>
>> +       {
>> +               char name[1024];
>> +               socklen_t len = 1024;
>> +               getsockopt(sk, SOL_SOCKET, SO_BINDTODEVICE, &name, &len);
>> +               pr_err("%s\n", name);
>> +       }
>> +
>>         /* Here, we grep for inet because some of the IPV6 DAD stuff can be
>>          * racy, and all we really care about is that the bridge got restored
>>          * with the right MAC, since we know DAD will succeed eventually.
>>
>> [root at fc22-vm criu]# cat test/zdtm/live/static/bridge.out
>> 10:58:52.252:     4: ERR: bridge.c:84: zdtmbr0
>> 10:58:52.261:     4: PASS
>
> Not sure what this means?

I use your test to check that SO_BINDTODEVICE does what we need. Looks
like we get the right answer by using this option.

>
> Tycho
> _______________________________________________
> CRIU mailing list
> CRIU at openvz.org
> https://lists.openvz.org/mailman/listinfo/criu