> From memory, this was one of the killers for SOCKS5 UDP under Linux - the
> socket would die with what they considered an invalid error code and
> they would ignore it, close the socket and drop all UDP packets coming back
> to that socket. And since it's not a documented error code, the SOCKS5
> people just said "it's a Linux bug" and ignored the bug report.
It's documented, rfc1122.
The broken BSD behaviour is available using setsockopt(..):
#ifdef SO_BSDCOMPAT
{
int one = 1;
setsockopt(sock_fd,SOL_SOCKET,SO_BSDCOMPAT,&one,sizeof(one));
}
#endif
> It should either be documented or not be there.
Most of the networking man pages are hopelessly out of date. I
started updating them a year or so back but got sidetracked learning
groff.
> Either libc5 and glibc documentation is broken or the kernel is
> returning an invalid error code - what does POSIX have to say?
No idea.. what POSIX says, anyone? Alan?
-cw
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/