[SR-Users] force_send_socket and IPv6

[Daniel-Constantin Mierla]

Hello,

On 12/15/11 1:18 PM, marius zbihlei wrote:
> On 12/14/2011 07:28 PM, Andreas Granig wrote:
>> Hey Daniel,
>>
>> On 12/14/2011 06:00 PM, Daniel-Constantin Mierla wrote:
>>> is the ipv6 address in between square brackets (can you give exact 
>>> usage
>>> of the foce_send_socket())? It should work as the bridging ipv4-ipv6 is
>>> using the same mechanism -- may be an issue with the exported function
>>> to config, though. Can you try to set $fs as alternative to
>>> force_send_socket()?
>> I tried both ways, like force_send_socket(x::x:5060); and
>> force_send_socket([x::x]:5060); - both resulting in the same error.
>>
>> However, doing $fs="[x::x]:5060"; works perfectly fine, thanks for 
>> the hint!
>>
>> A fix for force_send_socket would maybe still be good to not confuse 
>> people.
>>
>> Thanks,
>> Andreas
>>
>
> Hello,
>
> Actually, looking at the code I've observed some weird behavior :  for 
> example to fix the force_send_socket IPv6 address, a call to str2ip6() 
> (from resolve.h) is done (via resolvehost()). The function tries to 
> manually parse the IP address, so it makes me wonder why not use a 
> library function like inet_pton(AF_INET6... to get the correct structure.
>
> A possible bug is that with the current code, IPv4 mapped to IPv6 
> (x:x:x:x:x:x:d.d.d.d) will not be parsed correctly.
>
> Why don't we use inet_pton(AF_INET6 ... . Any reasons?

have you checked if inet_pton() doesn't break the internal ip_addr 
structure and the use of it? If all is ok, I guess inet_pton() is 
portable enough through most unix/linux distro's to make use of it.

Cheers,
Daniel

-- 
Daniel-Constantin Mierla -- http://www.asipto.com
http://linkedin.com/in/miconda -- http://twitter.com/miconda