rpms/libreswan
7
0
Fork 0
Code Issues Pull Requests Releases Activity
7f8fe26d5b
libreswan/libreswan-3.19-uniqueid_replace.patch
Paul Wouters ee061f381a * Fri Feb 03 2017 Paul Wouters <pwouters@redhat.com> - 3.19-2 
- Resolves: rhbz#1392191 libreswan: crash when OSX client connects
- Improved uniqueid and session replacing support
- Test Buffer warning fix on size_t
- Re-introduce --configdir for backwards compatibility
2017-02-03 17:51:18 -05:00

196 lines
7.3 KiB
Diff
Raw Blame History

diff -Naur libreswan-3.19-orig/programs/configs/d.ipsec.conf/uniqueids.xml libreswan-3.19/programs/configs/d.ipsec.conf/uniqueids.xml
--- libreswan-3.19-orig/programs/configs/d.ipsec.conf/uniqueids.xml 2017-01-15 14:34:34.000000000 -0500
+++ libreswan-3.19/programs/configs/d.ipsec.conf/uniqueids.xml 2017-02-03 17:41:48.703575771 -0500
@@ -1,17 +1,20 @@
<varlistentry>
<term><emphasis remap='B'>uniqueids</emphasis></term>
<listitem>
-<para>whether a particular participant ID should be kept unique,
-with any new (automatically keyed)
-connection using an ID from a different IP address
-deemed to replace all old ones using that ID.
-Acceptable values are <emphasis remap='B'>yes</emphasis>
-(the default)
-and
-<emphasis remap='B'>no</emphasis>.
-Participant IDs normally <emphasis remap='I'>are</emphasis> unique,
-so a new (automatically-keyed) connection using the same ID is
-almost invariably intended to replace an old one.</para>
-
+<para> Whether IDs should be considered identifying remote parties
+uniquely. Acceptable values are <emphasis remap='B'>yes</emphasis> (the
+default) and <emphasis remap='B'>no</emphasis>. Participant IDs normally
+are unique, so a new connection instance using the same remote ID is
+almost invariably intended to replace an old existing connection.
+</para>
+<para> When the connection is defined to be a server (using xauthserver=)
+and the connection policy is authby=secret, this option is ignored (as
+of 3.20) and old connections will never be replaced. This situation is
+commonly known as clients using a "Group ID".
+</para>
+<para> This option may disappear in the near future. People using identical
+X.509 certificates on multiple devices are urged to upgrade to use seperate
+certificates per client and device.
+</para>
</listitem>
</varlistentry>
diff -Naur libreswan-3.19-orig/programs/pluto/ikev1_aggr.c libreswan-3.19/programs/pluto/ikev1_aggr.c
--- libreswan-3.19-orig/programs/pluto/ikev1_aggr.c 2017-01-15 14:34:34.000000000 -0500
+++ libreswan-3.19/programs/pluto/ikev1_aggr.c 2017-02-03 17:41:42.320610486 -0500
@@ -906,6 +906,9 @@
*/
set_ph1_iv_from_new(st);
DBG(DBG_CONTROL, DBG_log("phase 1 complete"));
+
+ ISAKMP_SA_established(st->st_connection, st->st_serialno);
+
#ifdef USE_LINUX_AUDIT
linux_audit_conn(st, LAK_PARENT_START);
#endif
@@ -1012,6 +1015,9 @@
*/
set_ph1_iv_from_new(st);
DBG(DBG_CONTROL, DBG_log("phase 1 complete"));
+
+ ISAKMP_SA_established(st->st_connection, st->st_serialno);
+
#ifdef USE_LINUX_AUDIT
linux_audit_conn(st, LAK_PARENT_START);
#endif
diff -Naur libreswan-3.19-orig/programs/pluto/ikev2_child.c libreswan-3.19/programs/pluto/ikev2_child.c
--- libreswan-3.19-orig/programs/pluto/ikev2_child.c 2017-01-15 14:34:34.000000000 -0500
+++ libreswan-3.19/programs/pluto/ikev2_child.c 2017-02-03 17:41:37.984634068 -0500
@@ -1271,6 +1271,9 @@
ikev2_derive_child_keys(cst, role);
+ /* Check to see if we need to release an old instance */
+ ISAKMP_SA_established(pst->st_connection, pst->st_serialno);
+
/* install inbound and outbound SPI info */
if (!install_ipsec_sa(cst, TRUE))
return STF_FATAL;
diff -Naur libreswan-3.19-orig/programs/pluto/initiate.c libreswan-3.19/programs/pluto/initiate.c
--- libreswan-3.19-orig/programs/pluto/initiate.c 2017-01-15 14:34:34.000000000 -0500
+++ libreswan-3.19/programs/pluto/initiate.c 2017-02-03 17:41:48.704575766 -0500
@@ -929,73 +929,58 @@
bool uniqueIDs = FALSE; /* --uniqueids? */
/*
- * Called by main_inI3_outR3_tail() and ikev2_child_sa_respond() which is called for
- * initiator and responder alike! So this function should not be in initiate.c.
- * It is also not called in IKEv1 Aggressive Mode!
+ * Called by IKEv1 and IKEv2 when the IKE SA is established.
+ * It checks if the freshly established connection needs is
+ * replacing an established version of itself.
+ *
+ * The use of uniqueIDs is mostly historic and might be removed
+ * in a future version. It is ignored for PSK based connections,
+ * which only act based on being a "server using PSK".
*/
void ISAKMP_SA_established(struct connection *c, so_serial_t serial)
{
c->newest_isakmp_sa = serial;
- if (uniqueIDs && !c->spd.this.xauth_server &&
- (c->policy & POLICY_AUTH_NULL) == LEMPTY) {
- /*
- * for all connections: if the same Phase 1 IDs are used
- * for different IP addresses, unorient that connection.
- * We also check ports, since different Phase 1 ID's can
- * exist for the same IP when NAT is involved.
- */
- struct connection *d;
-
- for (d = connections; d != NULL; ) {
- /* might move underneath us */
- struct connection *next = d->ac_next;
-
- /*
- * We try to find duplicate instances of same
- * connection to clean up old ones when uniqueids=yes
- *
- * We are testing for all of:
- * 1: an appropriate kind to consider
- * 2: same ids, left and right
- * 3: same address family
- * 4: same connection name
- * 5: but different IP address or port
- * 6: differing dnsnames (sort of)
- *
- * DHR (2014-10-29):
- * Is the sense of the last clause inverted?
- * The logic kind of suggests that in fact the
- * same dnsnames should be the same, not different.
- *
- * Let's make 6 clearer:
- * if BOTH have dnsnames, they must be unequal.
- *
- * I suspect that it should be:
- * if BOTH have dnsnames, they must be equal.
- *
- * In other words the streq result should be negated.
- */
- if ((d->kind == CK_PERMANENT ||
- d->kind == CK_INSTANCE ||
- d->kind == CK_GOING_AWAY) &&
- (c->name == d->name) &&
- same_id(&c->spd.this.id, &d->spd.this.id) &&
- same_id(&c->spd.that.id, &d->spd.that.id) &&
- addrtypeof(&c->spd.that.host_addr) ==
- addrtypeof(&d->spd.that.host_addr) &&
- (!sameaddr(&c->spd.that.host_addr,
- &d->spd.that.host_addr) ||
- c->spd.that.host_port !=
- d->spd.that.host_port) &&
- !(c->dnshostname != NULL &&
- d->dnshostname != NULL &&
- streq(c->dnshostname,
- d->dnshostname))) {
- release_connection(d, FALSE);
- }
- d = next;
+ /* NULL authentication can never replaced - it is all anonnymous */
+ if (LIN(POLICY_AUTH_NULL, c->policy) ||
+ ( c->spd.that.authby == AUTH_NULL || c->spd.this.authby == AUTH_NULL)) {
+
+ DBG(DBG_CONTROL, DBG_log("NULL Authentication - all clients appear identical"));
+ return;
+ }
+
+ /*
+ * If we are a server and use PSK, all clients use the same group ID
+ * Note that "xauth_server" also refers to IKEv2 CP
+ */
+ if (c->spd.this.xauth_server && LIN(POLICY_PSK, c->policy)) {
+ DBG(DBG_CONTROL, DBG_log("We are a server using PSK and clients are using a group ID"));
+ return;
+ }
+
+ if (!uniqueIDs) {
+ DBG(DBG_CONTROL, DBG_log("uniqueIDs disabled, not contemplating releasing older self"));
+ return;
+ }
+
+ /*
+ * for all existing connections: if the same Phase 1 IDs are used,
+ * unorient that (old) connection - This is a replacement.
+ */
+ struct connection *d;
+
+ for (d = connections; d != NULL; ) {
+ /* might move underneath us */
+ struct connection *next = d->ac_next;
+
+ if (c != d && c->kind == d->kind && streq(c->name, d->name) &&
+ (same_id(&c->spd.this.id, &d->spd.this.id) &&
+ same_id(&c->spd.that.id, &d->spd.that.id)))
+ {
+ DBG(DBG_CONTROL, DBG_log("Unorienting old connection with same IDs"));
+ release_connection(d, FALSE);
}
+ d = next;
}
}
Reference in New Issue View Git Blame Copy Permalink