From: "Daniel P. Berrange" <berrange@redhat.com>
To: qemu-devel@nongnu.org
Cc: Peter Maydell <peter.maydell@linaro.org>,
"Daniel P. Berrange" <berrange@redhat.com>
Subject: [Qemu-devel] [PULL v1 3/5] sockets: ensure we don't accept IPv4 clients when IPv4 is disabled
Date: Tue, 11 Jul 2017 13:44:09 +0100 [thread overview]
Message-ID: <20170711124411.10499-4-berrange@redhat.com> (raw)
In-Reply-To: <20170711124411.10499-1-berrange@redhat.com>
Currently if you disable listening on IPv4 addresses, via the
CLI flag ipv4=off, we still mistakenly accept IPv4 clients via
the IPv6 listener socket due to IPV6_V6ONLY flag being unset.
We must ensure IPV6_V6ONLY is always set if ipv4=off
This fixes the following scenarios
-incoming tcp::9000,ipv6=on
-incoming tcp:[::]:9000,ipv6=on
-chardev socket,id=cdev0,host=,port=9000,server,nowait,ipv4=off
-chardev socket,id=cdev0,host=,port=9000,server,nowait,ipv6=on
-chardev socket,id=cdev0,host=::,port=9000,server,nowait,ipv4=off
-chardev socket,id=cdev0,host=::,port=9000,server,nowait,ipv6=on
which all mistakenly accepted IPv4 clients
Acked-by: Gerd Hoffmann <kraxel@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
---
util/qemu-sockets.c | 40 +++++++++++++++++++++++++++++++---------
1 file changed, 31 insertions(+), 9 deletions(-)
diff --git a/util/qemu-sockets.c b/util/qemu-sockets.c
index f4ddcaf..37d386f 100644
--- a/util/qemu-sockets.c
+++ b/util/qemu-sockets.c
@@ -104,17 +104,16 @@ NetworkAddressFamily inet_netfamily(int family)
* f t PF_INET6
* t - PF_INET
* t f PF_INET
- * t t PF_INET6
+ * t t PF_INET6/PF_UNSPEC
*
* NB, this matrix is only about getting the necessary results
* from getaddrinfo(). Some of the cases require further work
* after reading results from getaddrinfo in order to fully
- * apply the logic the end user wants. eg with the last case
- * ipv4=t + ipv6=t + PF_INET6, getaddrinfo alone can only
- * guarantee the ipv6=t part of the request - we need more
- * checks to provide ipv4=t part of the guarantee. This is
- * outside scope of this method and not currently handled by
- * callers at all.
+ * apply the logic the end user wants.
+ *
+ * In the first and last cases, we must set IPV6_V6ONLY=0
+ * when binding, to allow a single listener to potentially
+ * accept both IPv4+6 addresses.
*/
int inet_ai_family_from_address(InetSocketAddress *addr,
Error **errp)
@@ -124,6 +123,23 @@ int inet_ai_family_from_address(InetSocketAddress *addr,
error_setg(errp, "Cannot disable IPv4 and IPv6 at same time");
return PF_UNSPEC;
}
+ if ((addr->has_ipv6 && addr->ipv6) && (addr->has_ipv4 && addr->ipv4)) {
+ /*
+ * Some backends can only do a single listener. In that case
+ * we want empty hostname to resolve to "::" and then use the
+ * flag IPV6_V6ONLY==0 to get both protocols on 1 socket. This
+ * doesn't work for addresses other than "", so they're just
+ * inevitably broken until multiple listeners can be used,
+ * and thus we honour getaddrinfo automatic protocol detection
+ * Once all backends do multi-listener, remove the PF_INET6
+ * branch entirely.
+ */
+ if (!addr->host || g_str_equal(addr->host, "")) {
+ return PF_INET6;
+ } else {
+ return PF_UNSPEC;
+ }
+ }
if ((addr->has_ipv6 && addr->ipv6) || (addr->has_ipv4 && !addr->ipv4)) {
return PF_INET6;
}
@@ -213,8 +229,14 @@ static int inet_listen_saddr(InetSocketAddress *saddr,
port_max = saddr->has_to ? saddr->to + port_offset : port_min;
for (p = port_min; p <= port_max; p++) {
#ifdef IPV6_V6ONLY
- /* listen on both ipv4 and ipv6 */
- int v6only = 0;
+ /*
+ * Deals with first & last cases in matrix in comment
+ * for inet_ai_family_from_address().
+ */
+ int v6only =
+ ((!saddr->has_ipv4 && !saddr->has_ipv6) ||
+ (saddr->has_ipv4 && saddr->ipv4 &&
+ saddr->has_ipv6 && saddr->ipv6)) ? 0 : 1;
#endif
inet_setport(e, p);
#ifdef IPV6_V6ONLY
--
2.9.4
next prev parent reply other threads:[~2017-07-11 12:44 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-07-11 12:44 [Qemu-devel] [PULL v1 0/5] Merge sockets 2017/07/11 Daniel P. Berrange
2017-07-11 12:44 ` [Qemu-devel] [PULL v1 1/5] sockets: ensure we can bind to both ipv4 & ipv6 separately Daniel P. Berrange
2017-07-11 12:44 ` [Qemu-devel] [PULL v1 2/5] sockets: don't block IPv4 clients when listening on "::" Daniel P. Berrange
2017-07-11 12:44 ` Daniel P. Berrange [this message]
2017-07-11 12:44 ` [Qemu-devel] [PULL v1 4/5] io: preserve ipv4/ipv6 flags when resolving InetSocketAddress Daniel P. Berrange
2017-07-11 12:44 ` [Qemu-devel] [PULL v1 5/5] tests: add functional test validating ipv4/ipv6 address flag handling Daniel P. Berrange
2017-07-11 14:21 ` [Qemu-devel] [PULL v1 0/5] Merge sockets 2017/07/11 no-reply
2017-07-12 9:13 ` Daniel P. Berrange
2017-07-12 11:30 ` Daniel P. Berrange
-- strict thread matches above, loose matches on Subject: below --
2017-06-07 17:54 [Qemu-devel] [PULL v1 0/5] Merge sockets 2017/06/07 Daniel P. Berrange
2017-06-07 17:54 ` [Qemu-devel] [PULL v1 3/5] sockets: ensure we don't accept IPv4 clients when IPv4 is disabled Daniel P. Berrange
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170711124411.10499-4-berrange@redhat.com \
--to=berrange@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=qemu-devel@nongnu.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).