Re: [BEHAVE] (no subject)

[ivan c <[e-mail]>]

On Thu, 27 Jun 2013 16:56:44 +0200, Simon Perreault
<[e-mail]> wrote:
> Le 2013-06-27 16:51, ivan c a écrit :
>> This is taken from "The Open Group Base Specifications" (or informally
>> The
>> Single Unix Specification), by the IEEE.
>> Volume "System Interfaces", function "bind()":
>>
>>
>> "The bind() function shall fail if:
>>
>> [EADDRINUSE]
>>      The specified address is already in use."
> 
> Unless I misunderstand, there is no call to bind() in the sample that 
> you assert is prohibited by POSIX:
> 
> s1 = socket(..., SOCK_DGRAM, ...);
> s2 = socket(..., SOCK_DGRAM, ...);
> connect(s1, ...);
> connect(s2, ...);
> 
> About that code, I said: "The kernel is free to use the same local 
> endpoint for those two sockets as long as the remote endpoint is
> different."
> 
> To which you replied: "Your pseudo-code will never bind the sockets on 
> the same local endpoint. It is forbidden POSIX behavior."
> 
> Simon

Yes, the kernel will actually call bind() in the connect() function from
within the kernel (implicit bind()). strace wouldn't show it as it is a
kernel-mode call.


-- 
_Ivan Chollet_