Date: Thu, 16 Feb 2006 22:44:15 +0100 From: Arne Juul <Arne.Juul@europe.yahoo-inc.com> To: freebsd-java@freebsd.org Subject: SO_REUSEADDR should not also mean SO_REUSEPORT Message-ID: <43F4F22F.1060402@europe.yahoo-inc.com>
next in thread | raw e-mail | index | archive | help
Description by Håvard Pettersen
(Havard.Pettersen@europe.yahoo-inc.com) > Ticket Opened: 2006-02-02
13:37 PST
>
> In the FreeBSD Java port, setting the reuse_address feature on a
> server socket also sets the reuse_port feature, which is not a good
> thing, since it allows multiple server sockets to bind to the same
> local port.
>
> A possible work-around is to also disable reuse_address, but this
> will disallow listening to an unused local port if a connection
> established against that port is still active, which is not good
> either.
I don't know why this code was put into the bsd jdk patches, it just
seems plain wrong to me. It makes java subtly non-portable in a very
strange manner; I suggest this code is removed. Below is the patch
I'm currently using :-)
--- ../../j2se/src/solaris/native/java/net/net_util_md.c.orig Thu Feb
16 20:24:41 2006
+++ ../../j2se/src/solaris/native/java/net/net_util_md.c Thu Feb
16 20:23:24 2006
@@ -1019,17 +1019,20 @@
*bufsize = maxsockbuf;
}
}
}
+#if 0
+this is just plain wrong!
/*
* If SO_REUSEADDR option requested, unconditionally set SO_REUSEPORT.
*/
if (level == SOL_SOCKET && opt == SO_REUSEADDR) {
addopt = SO_REUSEPORT;
setsockopt(fd, level, addopt, arg, len);
}
+#endif
/*
* Don't allow SO_LINGER value to be too big.
* Current max value (240) is empiric value based on tcp_timer.h's
* constant TCP_LINGERTIME, which was doubled.
- Arne H. J.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?43F4F22F.1060402>