From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from eggs.gnu.org ([2001:4830:134:3::10]:44471)
	by lists.gnu.org with esmtp (Exim 4.71)
	(envelope-from <paolo.bonzini@gmail.com>) id 1Zigpz-0006YX-ST
	for qemu-devel@nongnu.org; Sun, 04 Oct 2015 06:48:16 -0400
Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71)
	(envelope-from <paolo.bonzini@gmail.com>) id 1Zigpz-0000Z9-00
	for qemu-devel@nongnu.org; Sun, 04 Oct 2015 06:48:15 -0400
Sender: Paolo Bonzini <paolo.bonzini@gmail.com>
References: <9B0F7F7B0E031F4A99FC8A50E22E3AA360FC1AFD@EU-MBX-01.mgc.mentorg.com>
	<CAFEAcA8CY2T0-gOMPphDYqHGqasH62EFSyxHvbd+Z5oG0gzAFA@mail.gmail.com>
	<9B0F7F7B0E031F4A99FC8A50E22E3AA360FC1B8E@EU-MBX-01.mgc.mentorg.com>
	<CAFEAcA_tYB6j-APrvPyDKAdNTjZay0-b-kC2LTQhRpDJ0pPxag@mail.gmail.com>
From: Paolo Bonzini <pbonzini@redhat.com>
Message-ID: <561103DF.5000704@redhat.com>
Date: Sun, 4 Oct 2015 12:47:59 +0200
MIME-Version: 1.0
In-Reply-To: <CAFEAcA_tYB6j-APrvPyDKAdNTjZay0-b-kC2LTQhRpDJ0pPxag@mail.gmail.com>
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 8bit
Subject: Re: [Qemu-devel] [Qemu-discuss] TCP options ipv4 and ipv6 have no
	effect
List-Id: <qemu-devel.nongnu.org>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
To: Peter Maydell <peter.maydell@linaro.org>, "Sair, Umair" <Umair_Sair@mentor.com>
Cc: QEMU Developers <qemu-devel@nongnu.org>, "qemu-discuss@nongnu.org" <qemu-discuss@nongnu.org>

On 03/10/2015 00:36, Peter Maydell wrote:
> 
> I agree about the (!ipv4 || !ipv6) condition though.
> The three states I listed above ought to correspond
> to "qemu_opt not set", "qemu_opt set to false" and
> "qemu_opt set to true",

The problem is that the underlying QemuOpts-based code treats "qemu_opt not set" and "qemu_opt set to false" the same way:

  ipv4   ipv6
  Y      Y       PF_INET6
  Y      N       PF_INET
  N      Y       PF_INET6
  N      N       PF_UNSPEC

We want:

 ipv4 ipv6
 Y    Y     PF_INET6
 Y    N     PF_INET
 Y    -     PF_INET         (ipv6 = N)
 N    Y     PF_INET6
 N    N     PF_UNSPEC
 N    -     PF_INET6        (ipv6 = Y)
 -    Y     PF_INET6        (ipv4 = N)
 -    N     PF_INET         (ipv4 = Y)
 -    -     PF_UNSPEC

I think this patch gets the desired semantics:

diff --git a/util/qemu-sockets.c b/util/qemu-sockets.c
index 2add83a..fdcf3fa 100644
--- a/util/qemu-sockets.c
+++ b/util/qemu-sockets.c
@@ -586,12 +586,15 @@ fail:
 
 static void inet_addr_to_opts(QemuOpts *opts, const InetSocketAddress *addr)
 {
-    bool ipv4 = addr->ipv4 || !addr->has_ipv4;
-    bool ipv6 = addr->ipv6 || !addr->has_ipv6;
+    bool ipv4 = addr->has_ipv4 && addr->ipv4;
+    bool ipv6 = addr->has_ipv6 && addr->ipv6;
 
-    if (!ipv4 || !ipv6) {
+    if (ipv4 || ipv6) {
         qemu_opt_set_bool(opts, "ipv4", ipv4, &error_abort);
         qemu_opt_set_bool(opts, "ipv6", ipv6, &error_abort);
+    } else if (addr->has_ipv4 || addr->has_ipv6) {
+        qemu_opt_set_bool(opts, "ipv4", !addr->has_ipv4, &error_abort);
+        qemu_opt_set_bool(opts, "ipv6", !addr->has_ipv6, &error_abort);
     }
     if (addr->has_to) {
         qemu_opt_set_number(opts, "to", addr->to, &error_abort);


The first if handles the "default to N" case, the second handles
"default to Y", the (absent) else case handles "default to
PF_UNSPEC".

Paolo