[omniORB] suspicios code in tcpTransportImpl.cc, omniORB4

[Harri Pasanen]

Following up on my own post, investigating problems on Windows NT...

Harri Pasanen wrote:

> In tcpTransportImpl.cc, function win32_get_ifinfo,
> 
> line 329:
>     if ( WSAIoctl(sock,SIO_ADDRESS_LIST_QUERY,
>                   NULL,0,
>                   (LPVOID)ifr,(DWORD)len,(LPDWORD)&retlen,
>                   NULL,NULL) == SOCKET_ERROR ) {
> 
>       if ( WSAGetLastError() != WSAEFAULT || lastlen != 0 ) {
>         if ( omniORB::trace(1) ) {
>           omniORB::logger log;
>           log << "Warning: WSAIoctl SIO_ADDRESS_LIST_QUERY failed. Unable
>           to
> obtain the list of all interface addresses.\n";
>           return;
>         }
>       }
> 
> Looks like only if it is doing tracing it will return at this point.
> Perhaps the return statement should be outside of the inner braces?
> 
> I've been getting this:
> "omniORB: Warning: WSAIoctl SIO_ADDRESS_LIST_QUERY failed. Unable to
> obtain the list of all interface addresses." all the time on one system,
> that's how I stumbled on this one.
> 

Hmm... in fact it seems to work only if tracing is enabled.  Otherwise the 
client, just hangs there forever.

So 

  nameclt list -ORBtraceLevel 0

will never return on Windows NT.

  nameclt list -ORBtraceLevel 1

will first print out the warning, then function normally and exit 
gracefully.

But that will not initialise ifaddrs correctly and may cause problems later?

Regards,

Harri