Attention is currently required from: plaisthos.
Hello plaisthos,
I'd like you to reexamine a change. Please visit
http://gerrit.openvpn.net/c/openvpn/+/1768?usp=email
to look at the new patch set (#21).
Change subject: oob: advertise a connect_lifetime in the probe reply
......................................................................
oob: advertise a connect_lifetime in the probe reply
A server answering an out-of-band SERVER_PROBE now also advertises a
connect_lifetime in the PROBE_REPLY: how long, in seconds, a probing
client may use the reply as the server's HARD_RESET when it starts a
handshake.
The value is inferred, not configurable: the reply is only usable as
that reset while its stateless SYN-cookie is valid, i.e. the guaranteed
cookie window of ~handshake_window (2 quantised buckets; see
check_session_hmac_and_pkt_id), so the server advertises exactly that. A
probe that arrived as P_CONTROL_OOB_WKC_V1 also gets
OOB_PROBE_REPLY_FLAG_RESEND_WKC, telling the client to resend the WKc
when completing the handshake, since the server keeps no state.
Only the wire advertisement is added here -- the server puts both values
into the probe_reply it builds for an accepted probe. The client side
that acts on them follows.
Change-Id: Ib2b6c2246f9d9c0a505292ee8d879f714901ffae
Signed-off-by: Lev Stipakov <[email protected]>
---
M doc/man-sections/server-options.rst
M src/openvpn/mudp.c
M src/openvpn/oob.h
3 files changed, 27 insertions(+), 0 deletions(-)
git pull ssh://gerrit.openvpn.net:29418/openvpn refs/changes/68/1768/21
diff --git a/doc/man-sections/server-options.rst
b/doc/man-sections/server-options.rst
index e77e85c..9192fcd 100644
--- a/doc/man-sections/server-options.rst
+++ b/doc/man-sections/server-options.rst
@@ -700,6 +700,12 @@
a small budget, a twentieth of ``--connect-freq-initial``, so a client
with a wrong clock can still probe while a replayed probe gets little.
+ For a client that supports it, the reply also serves as this server's
+ reset packet, letting the client start its handshake from it and save a
+ round trip. The reply carries how long it may be used for that, derived
+ from ``--hand-window``: lowering ``--hand-window`` shortens that window
+ as well.
+
--stale-routes-check args
Remove routes which haven't had activity for ``n`` seconds (i.e. the ageing
time). This check is run every ``t`` seconds (i.e. check interval).
diff --git a/src/openvpn/mudp.c b/src/openvpn/mudp.c
index a460634..078d769 100644
--- a/src/openvpn/mudp.c
+++ b/src/openvpn/mudp.c
@@ -266,12 +266,27 @@
return false;
}
+ /* A tls-crypt-v2 client must resend the WKc if it later uses this
reply
+ * to start a handshake, since we keep no state. */
+ uint32_t reply_flags =
+ (verdict == VERDICT_VALID_OOB_WKC_V1) ?
OOB_PROBE_REPLY_FLAG_RESEND_WKC : 0;
+
+ /* The client's third packet validates only while its SYN-cookie does,
so
+ * the advertised connect_lifetime is inferred (not configurable): the
+ * guaranteed cookie window of ~handshake_window (2 quantised buckets;
see
+ * check_session_hmac_and_pkt_id). Advertising more would make the
client
+ * trust an already-expired cookie. (RFC: connect_lifetime is how long
the
+ * server considers the reply valid.) */
+ int connect_lifetime = min_int(2 * ((handwindow + 1) / 2), 0xffff);
+
/* the echo of the peer's session id, plus what we advertise */
struct oob_probe_reply reply = {
.peer_session_id = state->peer_session_id,
.priority = (uint16_t)m->top.options.server_probe_reply_priority,
.weight = (uint16_t)m->top.options.server_probe_reply_weight,
.max_latency_diff =
(uint16_t)m->top.options.server_probe_reply_max_latency_diff,
+ .connect_lifetime = (uint16_t)connect_lifetime,
+ .flags = reply_flags,
};
/* Our session id is a stateless SYN cookie (the same HMAC the
three-way
diff --git a/src/openvpn/oob.h b/src/openvpn/oob.h
index d41953b..09780c8 100644
--- a/src/openvpn/oob.h
+++ b/src/openvpn/oob.h
@@ -159,6 +159,12 @@
*/
bool oob_timestamp_in_window(uint64_t probe_ts, uint64_t now, uint64_t
window_secs);
+/* probe_reply flags (the reply TLV's 32-bit flags field) */
+/* bit 0: the client must resend the wrapped client key (via P_CONTROL_WKC_V1)
+ * when it completes the handshake started from this reply. Set
+ * by a tls-crypt-v2 server, which is stateless and discarded the WKc. */
+#define OOB_PROBE_REPLY_FLAG_RESEND_WKC 0x1
+
enum oob_probe_verdict
{
OOB_PROBE_INVALID, /**< no valid probe_parameter: drop */
--
To view, visit http://gerrit.openvpn.net/c/openvpn/+/1768?usp=email
To unsubscribe, or for help writing mail filters, visit
http://gerrit.openvpn.net/settings?usp=email
Gerrit-MessageType: newpatchset
Gerrit-Project: openvpn
Gerrit-Branch: master
Gerrit-Change-Id: Ib2b6c2246f9d9c0a505292ee8d879f714901ffae
Gerrit-Change-Number: 1768
Gerrit-PatchSet: 21
Gerrit-Owner: stipa <[email protected]>
Gerrit-Reviewer: plaisthos <[email protected]>
Gerrit-CC: openvpn-devel <[email protected]>
Gerrit-Attention: plaisthos <[email protected]>
_______________________________________________
Openvpn-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/openvpn-devel
