From: Serge Belyshev <belyshev@xxxxxxxxxxxxxxxxx> Date: Sat, 19 Apr 2008 02:44:48 +0400 > David Miller <davem@xxxxxxxxxxxxx> writes: > > > From: Serge Belyshev <belyshev@xxxxxxxxxxxxxxxxx> > > Date: Sat, 19 Apr 2008 02:24:04 +0400 > > > >> But my socket is unbound and I'm connecting to a loopback address, so > >> I expect the connection to fail if nobody listens on the port I am > >> connecting to. Yet the connections succeedes on the rare occasion when > >> the kernel chooses the source port number that matches the destination. > > > > At the point you do a connect() you are "listening" on that > > port and can be connected to. > > Okay, so to reliably detect the case when nobody else but me is > "listening" on that port I should do bind() (with a different port) > instead of relying on the kernel, right? If you bind to a local port, then connect to the same local port it's still going to work. You're only going to avoid the situation if you bind to a local port and specifically connect to a different port. -- To unsubscribe from this list: send the line "unsubscribe linux-net" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html