Re: [BEHAVE] (no subject)

[ivan c <[e-mail]>]

On Thu, 27 Jun 2013 14:23:48 +0200, Simon Perreault
<[e-mail]> wrote:
> Le 2013-06-27 13:29, ivan c a écrit :
>>> s1 = socket(..., SOCK_DGRAM, ...);
>>> s2 = socket(..., SOCK_DGRAM, ...);
>>> connect(s1, ...);
>>> connect(s2, ...);
>>>
>>> The kernel is free to use the same local endpoint for those two
sockets
>>> as long as the remote endpoint is different.
>>>
>>> If you really want to force reusing the same local endpoint, you do
the
>>> SO_REUSEADDR+bind() dance before calling connect().
>>
>> Your pseudo-code will never bind the sockets on the same local
endpoint.
>> It is forbidden POSIX behavior.
> 
> I'm ready to accept that, with a proper citation.

Without the option SO_REUSEADDR, this is forbidden POSIX behavior


>> The option SO_REUSEADDR won't help either, this option is used to bind
>> over an *already closed* socket in TIME_WAIT state, that's all.
> 
> I humbly believe that you are mistaken. It works, try it. 

It surely *works* on platforms that support this use of SO_REUSEADDR, we
discussed it numerous times already, including on the phone.
I also made the point in numerous posts already that SO_REUSEADDR is used
to bypass the TIME_WAIT state on closed sockets, and even this is not
specified by POSIX. See
http://www.ietf.org/mail-archive/web/behave/current/msg10876.html
In another post, I explained that if you use SO_REUSEADDR on all your TCP
sockets you expose them to the old duplicate problem, and the worst is that
this crosses applications boundaries. See
http://www.ietf.org/mail-archive/web/behave/current/msg10906.html

However, applications should be free to use it if they wish.
I also made the point numerous times that applications should have as many
ways as possible to cross NATs.




-- 
_Ivan Chollet_